AGARD-LS..

I SO (Revised)

....en
Q

a::

.
AGARD LECTURE SERIES No.1 SO

Design Methods in Solid Rocket

Motors
Revised Version 1988

DISTRIBUTION AND AVAILABILITY

ON BACK COVER

AGARD-LS-150
(Revised Version 1988)

NORTH ATLANTIC TREATY ORGANIZATION

ADVISORY GROUP FOR AEROSPACE RESEARCH AND DEVELOPMENT
(ORGANISATION DU TRAITE DE L'ATLANTIQUE NORD)

AGARD Lecture Series No.150

REVISED VERSION 1988
DESIGN METHOD IN SOLID ROCKET MOTORS

The material in this revised publication was assembled to support a Lecture Series under the sponsorship
of the Propulsion and Energetics Panel and the Consultant and Exchange Programm.e of AGARD
presented on 18-19 April 1988 in London, United Kingdom, 21-22 April 1988 in
Saint-Aubin de Medoc, France, 25-26 April 1988 in Neubiberg, Germany and on 28-29 April 1988 in
Rome, Italy.

THE MISSION OF AGARD

According to its Charter, the mission of AGARD is to bring together the leading personalities of the NATO nations in
the fields of science and technology relating to aerospace for the following purposes:
- Recommending effective ways for the member nations to use their research and development capabilities for the
common benefit of the NATO community;
- Providing scientific and technical advice and assistance to the Military Committee in the field of aerospace research
and development (with particular regard to its military application);
- Continuously stimulating advances in the aerospace sciences relevant to strengthening the common defence posture;
- Improving the co-operation among member nations in aerospace research and development;
- Exchange of scientific and teGhnical information;
- Providing assistance to member nations for the purpose of increasing their scientific and technical potential;
- Rendering scientific and technical assistance, as requested, to other NATO bodies and to member nations in
connection with research and development problems in the aerospace field.
The highest authority within AGARD is the National Delegates Board consisting of officially appointed senior
representatives from each member nation. The mission of AGARD is carried out through the Panels which are composed of
experts appointed by the National Delegates, the Consultant and Exchange Programme and the Aerospace Applications
Studies Programme. The results of AGARD work are reported to the member nations and the NATO Authorities through
the AGARD series of publications of which this is one.
Participation in AGARD activities is by invitation only and is normally limited to citizens of the NATO nations.

The content of this publication has been reproduce j

directly from material supplied by AGARD or the authors.

Published April 1988

Copyright AGARD 1988
All Rights Reserved
ISBN 92-835-0454-2

Printed by Specialised Printing Sen1ices Limited

40 Chigwell Lane, Loughton, Essex IG10 3TZ
ii

PREFACE
This Lecture Series will try to summarize the current state-of-the-art in designing solid rocket motors and their
components. The aim is to collect the experience of several countries in using new technologies and new methods which have
been developed over the past ten years.
Specific sessions will deal with propellant grain, cases, nozzle, internal thermal insulations; the question of the general
optimization of a solid rocket motor will be emphasized.
This Lecture Series, sponsored by the Propulsion and Energetics Panel of AGARD, has been implemented by the
Consultant and Exchange Programme.

***
Ce cycle de conferences presente une synthese de l'etat de I'art dans Ie domaine de la conception des moteurs-fusees a
propergol solide et de leurs elements constitutifs. Les conferences presentees rassemblent les connaissances et Ie savoir-faire
de plusleurs pays dans la mise en oeuvre des nouvelles technologies et de nouvelles methodes qui ont ete developpees au cours
de la derniere decennie.
Les exposes specifiques portent sur Ie bloc de poudre, son enveloppe, les tuyeres et la protection thermique interne, en
mettant I'accent sur l'optimisation globale des moteurs fusees a propergol solide.
Ce cycle de conferences est presente dans Ie cadre du Programme des Consultants et des Echanges, sous I'egide du Panel
de Propulsion et d'Energetique de I'AGARD.

iii

LIST OF AUTHORS/SPEAKERS
Lecture Series Director:

Mr l'Ing. eng Chef D.Reydellet

Service Technique des Engins Balistiques
Direction des Engins
26, Boulevard Victor
75996 Paris Armees
France

AUTHORS/SPEAKERS
MrH.Badham
Royal Ordnance Summerfield
Kidderminster Worcs DY11 7RZ
United Kingdom

DrB.Lucas
BP No.2 Le Bouchet
91710 Vert Le Petit
France

Mr lP.Denost
Aerospatiale/Aquitaine
Boite Postale 11 Issac
33165 Saint-Medard en Jalles Cedex
France

MrS.Scippa
SNIAIBPD
Corso Garibaldi 22
00034 Colleferro (Roma)
Italy

Dr PREvans
Hercules Corp.
Aerospace Products Division
Alleghanys Ballistics Laboratory
P.O. Box 210, Rocket Center,
WV26726
USA

Mr G.P.Thorp
Royal Ordnance Summerfield
Kidderminster, Worcs DY11 7RZ
United Kingdom
Mr D.Thrasher
Air Force Rocket Propulsion Laboratory
Edwards Air Force Base
California 93523
USA

Mr lH.Hildreth
Air Force Rocket Propulsion Laboratory
Edwards Air Force Base
California 93523
USA

!vIr A.Tmchot
Agence Spatiale Europeenne
Division STS-AR
8-10, rue Mario Nikis
75738 Paris Cedex 15
France

Mr A.Lampert
Bayern Chemie GmbH
Postfach 220
80120ttobrunn
Germany

DrB.Zelier
SNPE/CRB
BP No.2 Le Bouchet
91710 Vert Ie Petit
France

MrLeMerer
SEP/DPPC
Les Cinq Chemins ~ Le Haillan
33165 Saint-Medard-en-Jalles
France

iv

CONTENTS

Page
PREFACE

iii

LIST OF AUTHORS/SPEAKERS

iv
Reference

INTRODUCTION
by D.Reydellet

ADVANCES IN SOLID ROCKET NOZZLE DESIGN AND ANALYSIS TECHNOLOGY IN THE

UNITED STATES SINCE 1970
by J.H.Hildreth

DESIGN AND ANALYSIS OF SOLID ROCKET MOTOR NOZZLE

by A.Truchot

COMPOSITE MOTOR CASE DESIGN

by P.R.Evans

4A

OVERALL DESIGN OF SOLID ROCKET MOTORS

by A.Lampert

4B"

DESIGN OF FILAMENTWOUND ROCKET CASES

by J.P.Denost

CONSIDERATIONS FOR DESIGNERS OF CASES FOR SMALL SOLID PROPELLANT

ROCKET MOTORS
by H.Badham and G.P.Thorp

PROPELLANT GRAIN DESIGN

by S.Scippa

SOLID PROPELLANT GRAIN DESIGN

by B.Zeller

STATE OF THE ART OF SOLID PROPELLANT ROCKET MOTOR GRAIN DESIGN IN THE
UNITED STATES
by D.I.Thrasher

DESIGN AND ANALYSIS OF SOLID ROCKET MOTOR INTERNAL INSULATION

by A.Truchot

10

OVERALL OPTIMIZATION OF SOLID ROCKET MOTOR

by A.Truchot

11

ROUND TABLE DISCUSSIONS

RTD

BIBLIOGRAPHY

"Not available at time of printing.

1-1

Introduction to the lecture series 150

by D. REYDELLET
"Ingenieur en Chef de l'Armement" (Colonel, Armament Corps)
Direction des Engins (Technical Directorate for Missiles)
26, boulevard Victor
75996 PARIS-ARMEES
FRANCE

I.

FOREWORD
When looking back to the evolution of the solid rocket motor (SRM) propulsion for the past twenty
years, the following fact is obvious : this evolution has been very similar to the one of the others
techniques,despite of its specific features.
Eventually, the same revolutions has appeared in the materials field (tendancy to the general use
of composite materials), and in the modelization methods (general use of finite elements codes and,
more recently, design methods with computer aids). The only area that is still relatively traditionnal,
is the formulation of propellants, though the trend to specific performances (increasing of specific
impulse or reducing of signature) has driven the technicians to formulate compositions that are closer
to the explosives than to the conventional propellants.
On the other hand, it is obvious that the new materials for SRM propulsion are very specific. In
addition their behaviour is often very difficult to modelize from a thermomechanical point of view.
As a result, it has been necessary to adapt the general methods for the characterization of materials,
the finite elements modelization, the validation of these modelizations, the qualification tests, and,
generally speaking, all the techniques that are used in a development program. It is obviously necessary
to make a periodical synthesis of these questions. In addition, there are very few synthetic publications
in this field.,
So we have to acknowledge AGARD once more for the opportunity we have to achieve this goal.

2.

GOAL OF THE SESSION

The purpose is to collect the most important points concerning the design and computation methods used
for SRMS. Of course, it is not possible in such a short time to deal completely with all the problems.
However, I hope that the fundamental ideas and the most critical or difficult problems will be emphasized.
I invite you to make a synthesis on this point at the end of the session.
Beside the question of the general optimization of SRM (that is a specific subject), every presentation
will try to summarize all the problems that are to be solved by an office in charge of designing a new
SRM or a component of SRM : choice of technologies, modelization, material characterization, choice
of margin of safety, identification of critical points, computation validation, qualification tests.
For the design of every main component, you will mainly attend two or three presentations by speakers
from different countries.
This is not a mistake, but a choice : it will allow you to look at the same question from several points
of view, since the different lecturers have lived various experiences on a same subject.
Once more time, in invite you to try to make a synthesis to morrow during the round table.
On the other hand, you will not find in these presentations a detailed paper dealing with the prediction
by theoritical methods of the specific impulse of the propellants used for SRMS. Once more, it is not an
error, but a choice: this question has already been dealt with by the WORKING GROUP nr. 17 of AGARD PEP
(AGARD Report nO 230). For any question on this specific problem, please consult this document: you
will find an evaluation of the different methods for predicting specific impulse of propellants with high
Aluminium loads, from the most sophisticated analytical methods to the most simple semi-empirical methods;
the question of the measurement accuracy of this parameter (especially under altitude simulation) is
approached, and, at the end, the group recommends the use of a semi-empirical method as a common reference
between the AGARD countries.
A last, beyond the presentations, it is obvious that this session has a secondary goal : that is to put
together different specialists in SRM propulsion. Persons in the audience of this session must be also,
if possible, active players. I do hope, from my point of view, that profitable conversations will be
initiated, that you will not hesitate to ask questions in case you want to make clear by the speaker a
difficult point, or if you want to share your experience on a special point.

3.

ORGANIZATION OF THE SESSION

During this session, eight speakers will present the result of their experience in the field of their
speciality. Theyare all professionalpeople, whose skill is well known in the world of SRM and solid
propellants.
In their country, they have all high responsability in important civil or military programs. Despite that,
they have devoted a lot of time to the preparation of their papers. I want now to address my deepest
thanks to them. I think their best reward will probably be the interest in their work you are going to show
through the questions that you will ask them, every time you will have the opportunity to do so, that is,

1-2
I hope, as often as possible. I will ask the speakers to be very watchful of the timing during the presentations, in order to save a sufficient time for questions at the end of each presentation. In addition, the
session will end tomorrow with a round table, with all speakers; during this round table, you will be able
to ask all the questions and make clear the points that you think the most important.
In don't want now to detail all the papers that are going to be presented to you. I only want to set a
list of the subjects and try to point out some important features.
3.1. Design and Computation of Nozzles
This topic will be presented by two speakers: Mr. J. HILDERTH (USA) and Mr. TRUCHOT (FRANCE).
Obviously this component is very specific in comparison with other nozzles used in other propulsion
techniques, specifically because of the very high exhaust gas temperature, their chemical and physical
characteristics (especially with two phases).
At first, one can say that the problem is to choose the materials, along with their place in the nozzle
(in a subsonic, sonic or supersonic flow). Then, the purpose is to solve a problem of design from a
thermo-mechanical point of view (computation of the stresses as a result of the thermal expension of
the parts under the huge temperature differences everywhere). In addition, one must predict the erosion
rate.
This technique has followed the general evolution during this last fiften years : at first, only high
temperature alloys (such as tungsten) or bulk graphite were used. Currently, the carbon-carbon composite
are more and more used. On the other hand, the computation methods have experienced important changes :
the general use of finite elements code allows now to perform the computation of mechanical margins of
safety for every part of the nozzle.
With the recent progresses concerning the materials that are called "thermostructural", an important
simplification in nozzle design can be hoped, since cold structural parts could be removed. However, a
lot of technical and economical problems are still to be solved : quality insurance, non destructive
tests, complete characterization of materials, verification of margin of safety
3.2. Design and Computation of Cases

This topic will be presented by three speakers: Dr. P. EVANS (USA), Mr. DENOST (FRANCE) et Mr BADHAM
(or Mr. THORP) (U.K.). The two first lecturers will present you the technologies and the design of
composite cases for large SRM. In this field, progresses have been spectacular during these last twenty
years : the first filament wound cases were using glass fiber (whose performances were increasing
continuously), then an aramide fiber (trade mark: KEVLAR), and at last carbon (or graphite) fiber.
At the same time, the development of computation methods was considerable; many difficulties were
encountered for modelizing filament wound cases by finite elements codes because of the very specifical
arrangements (successive monodirectionnal layers) of this composite material. The question was especially
critical at the bonding between the vessel and the rest of the missile (design of skirts), and at the
aft and forward polar fittings.
Nowadays ,the state of the art allows to perform precise mechanical computations; however, the higher
performances are, the more severe "small mistakes in designing details" are, in the same way, small
defects can become very critical. As a result, many progresses are still to be made for quality insurance
(in a general meaning, that is the problem of discovering and assessing critical defects),
The third speaker will deal with technologies and design methods used for cases of small SRM for tactical
missiles. The concerns for this applications are very different, since other critical loads than the
internal pressure (for instance, the problems of rigidity for very slender SRM) can be encountered.
In addition, economical constraints can be preponderant. As a result, the materials that are choosen are
often metallic alloys, although some composite reinforcement can be used in some cases.
3.3.

Design and Computation of Propellant Grains

Three speakers will deal with this point: Dr. SCIPPA (ITALY), Dr. ZELLER (FRANCE), Mr. THRASHER (USA).
The most important point for that subject is obviously the continuous evolutions of propellants :
double-base, conventional composites, double-base composites Every kind of propellant has a very
particular ballistic and mechanical behaviour. A very general evolution is the use of cast bonded
grains, that is the technique of casting the propellant directly in the case (which is already
equipped with its internal thermal protections). The question of the bonding between propellant and
thermal insulation is very critical. On the other hand, the techniques used for obtaining various
shapes, by machining or integral molding have provided the designer with many degrees of freedom.
Besides the conventionnal star cylindrical and axisymetric grains, one can now use three dimensional
shapes (for exemple : finocyl grains).
These solutions lead to optimal pressure history and high volumetrie loading. This last point become
possible from a mechanical point of view with the enhancement of the mechanical properties of the
propellant. Modern propellant compositions are optimized for sophisticated requirements : high specific
impulse and density, reduced signature (actually this requirement covers many different concerns).
The methods used for mechanical computations of propellant grains were at first very simple. Nowadays,
the most sophisticated finite elements codes are used (in certain case with three-dimensionnal
capabilities). Dynamic loads can be modelized by taking in account the most various rheological
behaviours (typically: linear or non-linearviscoelasticity). These methods allow the identification
of the critical mechanical points,at low temperature and during firing as well. By using failure
criteria, the margin of safety can be assessed. Since the critical points in the grain it self are
rather well predicted, one must point out that the prediction of margin of safety at the bonding
between propellant and thermal protection is still difficult. Some progresses are still to be made
in that field,
A last, I want to remind you the questions of investigating the influence of aging, the quality
insurance and the verification of margin of safety.

1-3
3.4.

Design and Computation of Internal Thermal Insulations

This question will be analysed by only one paper presented by Mr. TRUCHOT (FRANCE). This is not a
negligible problem, in reference to the relative magnitude of the thermal insulation's mass in the total
inert mass of a modern SRM. Internal thermal insulations use elastomeric materials, with various loads
(powders or fibers), that are designed to enhance the material's properties. The most important requirements are : erosion performances, low diffusivity, and (if possible) low density. In this fmlds, progresses have been,spectacular these last years.
The design of internal thermal insulation is very difficult, especially concerning the prediction of
the erosion rate
the role of the experimental knowledge is still very important.

3.5.

General optimization of SRMS

Mr. TRUCHOT will present also a paper dealing with general optimization methods for SRMS, that have
been developped extensively these last years. As a matter of fact, the emergence of computer aids for
design has produced some applications for SRMS. These methods allow to generate SRM predesigns
automatically; as a result, one can make better choices between different technologies and main
geometrical parameters. These codes have some computation subroutines for components; these subroutines
are more and more sophisticated; the designers are now able to provide a lot of predesign studies very
quickly.

4.

CONCLUSION
At the end of these two days, we will try to draw some conclusions together. I am sure that you will
be convinced that solid rocket motor propulsion shows an extraordinary maturity.
I will not try to remind you the classical comparisons between various propulsion techniques. However,
1 would like to point out a few points about the future of solid propellants.
I'm strongly convinced that this kind of propulsion has still an important role to play in the future.
It will be, of course, still present in its usual uses, such as military applications, for tactical
or strategical missiles as well. The continuous progresses in performances, along with the very simple
operational use, make the adequacy of SRM propulsion to that kind of application almost perfect.
In addition, it is obvious that solid propellants have not yet developed all their qualities for the
space applications, and especially for large boosters; by the way, it seems to me that spectaculary
progresses are to be forecasted in the field or cost reduction. However, a recent and tragic accident
make someone to doubt about the intrinsic reliability of SRMS. Some people may have the temptation
to conclude that they are incompatible with manned space flights. This conclusion is obviously
excessive, but this dramatic exemple must invite us to be careful.
It is very dangerous -to believe that SRMS are so simple that nothing severe can happen,
especially after a long success ful series. In any case, the burning gases of solid propellants are
always extremely dangerous; the huge mass of propellant grain is alwaysa source of hazard, because
of their very high specific energy. As a result, the safety studies (design, justifications, insurance
and maintenance of safety) must always be looked at with the greatest care, under the direct authority
of the program manager.
At last, I think the most important thing is to be aware of this fact (that could seem completely
obvious) : a SRM can work only one time and, consequently, it is impossible to verify the operation
of every specimen by an acceptance test.
The whole quality insurance rests on the margins of safety, but also on the process controls and on
the performances of non destructive inspection tests.
If all the people involved in a program keep these simple ideas in their mind, I'm convinced that it
is possible to reach a considerable level of reliability and safety; this level will lead to a very low
failure probability.
As a result of all these elements, it is for me obvious that SRMS will be present for many applications,
for after the year 2000.

1-4

Introduction a la lecture serie 150

par D. REYDELLET
Ingenieur en Chef de l'Armement
Direction des Engins
26, boulevard Victor
75996 PARIS-ARMEES
FRANCE

1. - GENERALITES
Si l'on observe l'evolution de la propulsion a propergol solide dans les vingt dernieres annees, on
est frappe par Ie fait que, malgre la specificite de ce domaine, l'evolution a ete assez conforme a
celIe des autres techniques. En effet, les memes revolutions sont apparues dans Ie domaine des materiaux (tendance a l'utilisation generalisee des materiaux composites), et dans les methodes de dimensionnement (utilisation a grande echelle des codes aux elements finis, et, plus recemment des outils
de conception assistes par ordinateurs). Le seul secteur relativement reste traditionnel est celui de
la formulation des propergols, encore que, pour satisfaire a certaines performances (impulsion specifique elevee ou recherche de discretion) les techniciens soient de plus en plus tentes de faire appel
a des compositions ayant plus de parente avec les explosifs qu'avec les propergols traditionnels.
II est vrai par ailleurs que les nouveaux materiaux de la propulsion a poudre sont des materiaux tres
specifiques. De plus, leur comportement est souvent difficile a modeliser au plan thermomecanique. lIs
sont souvent utilises dans des gammes de sol licitation tres inusitees. Crest pourquoi, les methodes
generales de caracterisation des materiaux, de modelisation par elements finis, de validation des
modeles de calcu1, d'essai de qualification en un mot toutes les techniques utilisees lors de la mise
au point des materiels ont dfi etre adaptees. La necessite de faire Ie point de maniere periodique sur
ces prob1emes est absolument evidente. On doit ajouter qu'il existe tres peu d'ouvrages de synthese
dans ce domaine.
C'est pourquoi, il convient de remercier une fois de plus l'AGARD pour l'occasion qui nous est donnee
de realiser cet objectif.
2. - OBJECTIF DE LA SESSION
II s'agit de rassembler les elements les plus importants relatifs aux regles de conception et de dimensionnement des propulseurs a poudre. Bien entendu, i1 n'est pas possible dans un temps aussi limite d'a1ler completement jusqu'au fond des choses. Cependant, j'espere que seront degagees au moins
1es idees fondamenta1es et soulignes 1es points les plus sensibles ou difficiles. Je vous engage a
bien vouloir en faire 1e bi1an en fin de session.
Si on excepte la question de l'optimisation generale des propulseurs, qui constitue un sujet a part,
chaque expose tentera de vous resumer de maniere succincte l'ensemble des problemes qui doivent etre
resolus par un bureau d'etude 10rs de 1a conception d'un nouveau propulseur ou de sous-ensemble de
propulseur : choix des diffetentes technologies possibles, mode1isation, caracterisation des materiaux,
choix des coefficients de securite, inventaire des points critiques, verification des ca1cu1s, essai
de qualification, etc
Pour Ie dimensionnement de chaque sous-ensemble majeur, vous assisterez genera1ement a deux ou trois
presentations d'orateurs de pays differents. Ce fait ne constitue pas une erreur de programmation,
mais un choix de1ibere pour vous permettre d'aborder Ie meme prob1eme de maniere differente, dans la
mesure ou 1es differents intervenants auront une experience differente sur un meme sujet. Je vous
invite, une fois de plus, a tenter d'en faire la synthese 10rs de table ronde de demain.
Par ailleurs, vous ne trouverez pas dans ces presentations de papiers detailles portant sur la determination par des methodes theoriques de l'impulsion specifique des propergols utilises dans 1es
moteurs. Ce fait ne constitue pas un oubli, mais un choix delibere : Ie sujet a deja ete traite par Ie
groupe de travail nO 17 de l'AGARD (AGARD report nO 230). Pour toute question portant sur ce prob1eme
particulier, vous voudrez bien vous reporter a ce document ou sont eva1uees 1es differentes methodes
de prevision de l'impulsion specifique des propergo1s fortement a1uminisesdepuis 1es methodes analytiques les plus sophistiquees jusqu'aux methodes semi-empiriques 1es plus simples; 1e prob1eme de
la precision de mesure de ce parametre (notamment en altitude simu1ee) est aborde, et fina1ement, 1e
groupe preconise 1 'utilisation d'une methode semi-empirique, en tant que reference parmi 1es pays de
1 'AGARD.
Enfin, au-dela de 1a matiere-meme des conferences, il va sans dire que cette manifestation a aussi
pour but de mettre en contact divers specialistes de la propulsion a propergol solide. Les auditeurs
de cette session doivent egalement, si possible, en etre les acteurs actifs. Je souhaite personnel1ement que des echanges fructueux soient inities, et que vous n'hesitiez pas a prendre la parole soit
pour faire ec1aircir par l'orateur quelque point de1icat, soit pour faire part de votre experience
propre sur un point particu1ier.

1-5
3. -

DEROULEMENT DE LA SESSION
Au cours de cette session, huit conferenciers vont vous presenter Ie fruit de leur reflexion dans Ie
domaine de leur specialite. Ce sont tous de grand professionnels, dont la competence est reconnue
dans Ie monde de la propulsion a poudre et des propergols.
lIs ont tous dans leur pays des responsabilites elevees dans de grands programmes civils ou militaires. Malgre cela, ils ont bien voulu consacrer beaucoup de temps a la preparation de leurs exposes.
Qu'ils en soient des maintenant collectivement et chaleureusement remercies. Leur meilleure recompense sera sans doute les marques d'interet que vous voudrez bien manifester pour leur travail, grace
aux questions que vous ne manquerez pas de leur poser, chaque' fois que vous en aurez l'occasion,
c'est-a-dire, j'espere, Ie plus souvent possible. A cet effet, je demanderai aux conferenciers de
bien vouloir etre vigilents sur Ie respect de l'horaire lors des presentations, afin de menager un
temps suffisant a la fin de chaque seance pour les questions. Par ailleurs, la session sera cloturee
demain par une table ronde reunissant tous les conferenciers, au cours de laquelle vous pourrez
poser les questions et voir traiter les sujets qui vous tiennent a coeur.
II n'entre pas dans mon propos de vous detailler la teneur des papiers qui vous seront exposes. Je
me bornerai done a lister les sujets abordes, en tentant d'en souligner les caracteristiques les plus
importantes a mon sens.

3.1.

conception et dimensionnement des tuyeres

Ce sujet vous sera expose par deux orateurs : Mr J. HILDERTH (USA) et Mr TRUCHOT (FRANCE). II va sans
dire que la conception de ce sous-ensemble presente une originalite considerable par rapport aux
tuyeres utilisees dans les autres techniques de propulsion, par la temperature considerable des gaz,
leur nature chimique, leur caractere biphasique.
On peut dire que, dans un premier temps, il s'agit d'un probleme de choix de materiaux, diversifie
suivant leur position dans la veine (zone sub, trans ou supersonique). Dans un deuxieme temps, il
s'agit de resoudre un probleme d'architecture au plan thermomecanique (calculs de contrainte) liees
aux dilatation produites par les enormes gradients de temperature dans toutes les pieces). II faut
egalement evaluer les performances en erosion.
Cette technique a subi l'evolution generale au cours des quinze dernieres annees : au depart, seuls
les metaux refractaires (tungstene) ou Ie graphite polycristallin etaient utilises au col, Ie reste
de la veine ayant recours au carbone ou a la silice phenolique. Actuellement, les materiaux carbonecarbone se generalisent. Par ailleurs, les methodes de calcul ont evolue de maniere considerable : la
generalisation des calculs aux elements finis permet desormais d'effectuer des calculs de marge de
securite mecanique pourchacune des pieces des tuyeres.
Les recents progres de materiaux dits "thermostructuraux".permettent d'esperer une simplification
considerable dans les architectures, par elimination des pieces structurales froides.
Cependant, de nombreux problemes techniques et economiques restent a resoudre : assurance de la qualite, controle non destructif, caracterisation complete des materiaux, verification des marges de
securite etc

3.2.

Conception et dimensionnement des structures

Ce sujet vous sera expose par trois orateurs : Mr M. EVANS (USA), Mr DENOST (FRANCE) et Mr BADHAM
ou Mr THORP (G.B.).
Les deux premiers vous presenteront les technologies et la conception des structures composites pour
propulseurs de grande dimension. Dans ce comaine, les progres ont ete spectaculaires durant les vingt
dernieres annees : les premiers enroules filamentaires utilisaient de la fibre de verre, dont les
performances n'ont fait qu'augmenter, puis une fibre polyamide (nom commercial : KEVLAR), at enfin
des fibres de carbone. Parallelement, les methodes de calcul ont connu un developpement considerable;
de nombreuses difficultes ont ete rencontrees dans la modelisation des structures roving par element
finis, compte tenu de la nature specifique, en couches successives, de ce composite. Les problemes
sont particulierement delicats aux liaisons entre Ie bidon et la virole du missile (conception des
jupes) et aux embases avant et arriere.
L'etat de l'artpermet a l'heure actuelle d'effectuer des dimensionnements precis; cependant, plus les
performances sont elevees, plus les"petites erreurs locales de dimensionnement" sont graves. De meme,
les defauts locaux peuvent devenir extremement critiques. C'est pourquoi, de nombreux progres sont
encore a attendre dans l'assurance de la qualite au sens large, c'est-a-dire dans la determination et
la detection des defauts critiques.
Le troisieme orateur vous parlera des technologies et des conceptions de structure de propulseur de
petite dimension adaptees aux missiles tactiques. Les preoccupations sont nettement differentes, puisque d'autres facteurs dimensionnants en dehors de la pression interne peuvent intervenir (par exemple
la rigidite pour les propulseurs tres elances). Par ailleurs, les considerations d'ordre economique
peuvent etre preponderantes. Dans ces conditions, les materiaux utilises peuvent etre plutot
metalliques, bien que pouvant avoir recours a des renforcements composites.

3.3.

Conception et dimensionnement des chargements propulsifs

Trois orateurs vous developperont ce point : Dr SCIPPA (ITALY) Dr ZELLER (FRANCE) et Mr D. THRASHER
(USA)
Ce sujet est bien entendu domine par les evolutions des formulations des propergols : double-base,
composites, double-base composite. Chaque famille a un comportement balistique et mecanique tres
particulier. Une evolution tout a fait generale est egalement la generalisation des blocs moulescolles, dont Ie propergol est directement coule a l'interieur de la structure equipee de ses protections thermiques. Le probleme du collage protection-thermique propergol est des plus crucial. Par
ailleurs, les techniques d'obtention de formes variees, soit par usinage, soit par moulage integral
ont donne au concepteur de nombreux degres de liberte. A cote des blocs a forme classique cylindrique etoiles, et des profils de revolution, on peut desormais utiliser des formes tridimensionnelles

1-6

(par exemple : bloc finocyl) conduisant a des evolutions balistiques optimales associees a des
coefficients de remplissage eleves, rendus egalement possibles au plan mecanique par l'amelioration
des proprietes mecaniques du propergol.
Les formulations actuelles ont ete adaptees a des exigences fonctionnelles de plus en plus importantes : forte impulsion specifique, forte densite, eventuellement discretion (en distinguant d'ailleurs
de nombreuses sous-categories dans cette specification.
Les methodes de calcul mecanique des blocs initialement tres rudimentaires, font actuellement appel
aux codes
aux elements finis les plus elabores (dans certains cas tri-dimensionnels), pouvant
travailler en dynamique en prenant en compte des rheologies de propergol les plus variees (typiquement : viscoelasticite plus ou moins lineaire).
Ces methodes permettent d'acceder aux points les plus contraints mecaniquement soit a basse temperature, soit en tir et, parl~ntermediaire d'un critere de rupture, evaluer les marges de dimensionnement. Si les points critiques du cote du propergol sont assez bien evalues de maniere classique,
on peut cependant souligner que Ie dimensionnement des marges de securite aux collage sont connUes
avec beaucoup moins de precision, et la verification de ces marges encore plus problematiques. II
y a sans doute encore des progres a faire dans ce domaine.
Enfin, soulignons les problemes souleves par l'evaluation des caracteristiques en vieillissement,
l'assurance de la qualite et la verification des marges de securite.
3.4.

Conception et dimensionnement des protections thermiques internes

Ce probleme fera l'objet d'un seul papier, presente par M. TRUCHOT (FRANCE).
Cette question est loin d'etre negligeable, si l'on se refere a l'importance actuelle du poste
"protection thermique" dans Ie devis-masse d'un propulseur moderne. Les protections thermiques
internes des propulseurs utilisant des elastomeres, comportent des charges soit pulverulentes, soit
fibreuses, destinees a ameliorer leurs proprietes. Les proprietes requises sont : la resistance a
l'ablation, Ie caractere isolant (faible diffusivite) et si possible, la faible densite. La aussi,
les progres ont ete spectaculaires ces dernieres annees.
Le dimensionnement des protections thermiques est tres delicat, notamment les previsions de comportement en erosion et la part experimentale est encore tres importante.

3.5.

Optimisation generale des propulseurs a poudre

Mr. TRUCHOT vous presentera egalement une communication sur les methodes d'optimisation generales des
propulseurs, qui ont fait l'objet de developpement importants ces dernieres annees. En effet, l'avenement des methodes de conception assistee par ordinateur a egalement eu des applications pour les
propulseurs a poudre. Ces methodes permettent de generer de maniere automatique des avant-projets de
propulseur, et donc de faire des choix ec1aires en ce qui concerne les differentes technologies et
les parametres geometriques fondamentaux. Ces programmes comportent des modules de dimensionnement
de sous-ensembles de plus en plus sophistiques, et permettent donc d'augmenter considerablement
l'aptitude des bureaux d'etudes a fournir rapidement des avant-projets et de multiples variantes.

4. -

CONCLUSION
A l'issue de ces deux jours, nous essayerons de tirer quelques conclusions. Je suis certain que vous
serez tous convaincus du degre de maturite extraordinaire atteint a ce jour par la propulsion a poudre.
Je ne tenterai pas les comparaisons classiques que l'on fait habituellement entre les divers modes de
propulsion. Cependant, je voudrais souligner quelques points ayant trait a l'avenir du propergol
solide.
Je suis intimement convaincu que ce mode de propulsion est encore promis a un grand avenir.
Elle sera bien entendu toujours presente dans ses domaines priviligies, a savoir les utilisations
militaires, que ce soit pour les missiles tactiques ou strategiques. Les progres dans les performances, associes aux caracteristiques de simplicite operationnelle, en font une technique parfaitement
adaptee a ce type de besoin.
En outre, il est certain que les propergols solides n'ont pas dit leur dernier mot dans Ie domaine
du lancement spatial, et principalement pour les etages d'appoint des gros lanceurs; en effet, il me
semble que des progres spectaculaires sont a attendre dans Ie domaine de la reduction des couts.
Cependant, une actualite recente et dramatique a peut-etre pu amener a douter de la fiabilite intrinseque des propulseurs a poudre. Certains sont memes tentes de conclure qu'ils sont incompatibles avec
les vols spatiaux habites. Cette conclusion est bien evidemment excessive, mais ce malheureux exemple
est la pour nous rappeler a la prudence. II ne faut jamais considerer que les propulseurs a propergol
solide sont des objets tellement simples que rien de vraiment dramatique ne peut arriver, surtout apres
une longue serie de succes. Quoi qu'il en soit, les gaz de combustion des propergols sont toujours
extremement dangereux; les masses considerables des chargements de propergols sont un facteur de
risques tres importants compte tenu de leur e.nergie specifique tres levee. C' est pOllrquoi, les etudes
liees a la securite (construction, justification, assurance et maintien de la securite) doivent tOlljours atre menees avec la plus grande rigueur, sous l'autorite directe du responsable du programme.
Enfin, il me semble que Ie plus important est de bien avoir conscience du fait suivant totalement
evident "a priori" : un propulsellr ne peut servir qu'une seule fois, et, en consequence, il est
impossible de verifier son bon fonctionnement en essai de recette. Toute l'assurance de la qualite reside
dans les marges de dimensionnement, mais aussi dans la maitrise des pro cedes de fabrication et les
performances des moyens de contrale non destructifs.
Si ces idees simples sont presentes a l'esprit de toutes les personnes impliquees dans le programme,
il n'y a aucune raison pour qu'un niveau de fiabilite et de securite considerable ne puisse atre
atteint, et ceci quelle que soit la severite de l'objectif en terme de probabilite.
Compte tenu de ces elements, il me semble evident que l'on trouvera des propulseurs a propergol
solide dans de multiples applications bien apres l'an 2000.

2-1

Advances in Solid Rocket Nozzle Design and Analysis Technology in the United States Since 1970
Joseph H. Hildreth
Structural Analyst
United States Air Force Rocket Propulsion Laboratory
Edwards Air Force Base, California, USA
ABSTRACT
Since 1970 design and analysis technologies for sol id rocket nozzles in the United States have
experienced significant changes.
These changes have been dictated primarily by increasingly severe
operating environments. To a smaller but still significant degree, these changes have also been caused
by outside influences.
The environmental hazards of using materials incorporating asbestos is one
example of these outside influences.
This paper examines the evolution of the nozzle design and
analysis in the US over the past decade. The discussion begins with a brief summary of general ized
design requirements and follows with a description of the design evolution from tungsten and graphite
based designs to the current carbon-carbon designs.
The main causes for the incorporation of new
technologies in the designs are identified.
A modestly technical description of the analysis processes that are employed in the design cycle is
given! The portions of the analysis activities where increased technology has been incorporated are
identified as well as the current trends for improved capabilities. The verification of the analysis
procedures is a vital aspect of improving the analysis capabil ities.
The paper concludes with a
description of some of the activities which have occurred to demonstrate the val idity of the new
analysis technology.
These activities include code versus code comparisons as well as code versus
laboratory measurement efforts.
The code versus measurement efforts include investigations into
threaded interfaces between the throa t and ex it cone parts and the behavi or 0 f carbon-carbon i nvo 1ute
exit cones subjected to mu1tiaxial loads.
I NTRODUCTI ON
The technologies used in the design and development of solid rocket nozzles have been substantially
improved over the last decade. The areas of improvement include materials, analysis techniques, testing
techniques for measurement of material properties, and supporting computer technology. The advances in
these technologies have made it possible to develop nozzles capable of surviving the more severe
operati ng envi ronments generated by current high performance motors.
Converse1y, these improvements
also permit us to uti1 ize the more estab1 ished approaches more efficiently and with more confidence.
The evolving technologies are based primarily on the advances in fiber reinforced carbon composite
materials, generally referred to as carbon-carbon materials, but these technologies have also had a
significant impact on our capacity to make use of the various phenolic nozzle materials as well.
The advances in technology are driven by the ever increasing demands of those who set the mission
requirements for a particular system. The demands on sol id rocket nozzle designs fall under two major
headings; Payload and Range (or Delta V in the case of space motors). Air launch designs are also
driven by a requirement for low observability, for example, minimum radar cross section and infrared
signature. Another global requirement for the design of any nozzle is the need for high re1 iabil ity.
From these requirements, a number of additional design requirements begin to appear depending upon how
the vehicle is sized and its specific mission, for example, motor operation time, overall vehicle length
constraints, directional control of the vehicle (spin stabilized or thrust vector control), and overall
system cost. General i zed desi gn requi rements for booster, space, and air 1aunch systems are summari zed
in the following table:

TABLE 1
GENERALIZED NOZZLE DESIGN REQUIREMENTS
Booster
---

Air Launch

Burn Time (seconds)

120

120

<

20

Erosion (mils/sec)

15-20

15

<

15

10

100+

10-20

Sometimes

Sometimes

Sometimes

Expansion Ratio
TVC

Vehicle Length (relative importance)

Modera te

Nozzle Weight (relative importance)

Moderate

Cost (relative importance)

Moderate

Production Time
Reliability

Moderate
Moderate-High

Moderate
Moderate-Low

Low

High

Low

Low

Moderate

High

High

High

2-2

Although they are not mission related requirements, environmental considerations have played a -role in
nozzle design. Materials containing asbestos, such as adhesives and phenolic insulators, are being
eliminated due to health hazards to the manufacturers. Similarly, other materials have changed somewhat
because of environmental restrictions on waste products produced during their manufacture.
The items listed in Table 1 have a significant impact upon the technologies to be considered for a
particular design and thus upon the design techniques that must be used in the design process. For
booster and space motor applications, the requirements on burn time, erosion, and, to some extent, TVC
have greatly impacted the forward region of the nozzle. This is particularly true for the throat and the
approach region to the throat.
The nozzle weight and expansion ratio requirements have produced
substantial changes in the exit cone. These requirements have resulted in the need to use very thin
wa11 ed constructi ons that do not requ i re a surrounding structure to support the mechani ca1 loads. In
the air launch arena, the geometry of the nozzle has not been changed as drastically as have space
nozzle designs, but the materials selection has been significantly affected. The design considerations
of burn time, pulsed and nonpulsed operation, and production time and cost result in constant trade-offs
between materials. For example, low cost and rapidly producible materials generally do not perform very
well in terms of erosion performance. The increased thermal loads occurring in pulsed operation have
dictated the use of carbon-carbon materials in the throat.
DESIGN HISTORY
Over the past decade, the space nozzles have evolved from all graphite and phenol ic struc'tures to
the use of tungsten or molybdenum throats surrounded by graphite and phenol ic to throats of pyrolytic
graphite washers surrounded by graphite and phenolics, finally, to a design consisting of a single piece
of carbon-carbon for the throat/entrance, and either an all phenol ic or a carbon-carbon exit cone.
Booster designs, on the other hand, have changed very little. The design evolution for air launch
nozzles has been very similar to that of the space nozzles. The air launched nozzles have progressed
from tungsten throats to pyrolytic graphite washers and, most recently, to 3 directional and 4
directional carbon-carbon inserts for the throat. Materials surrounding the throat insert consist of
tape wrapped ablatives and monolithic graphites. Exit cones for air launched nozzles are typically tape
wrapped phenolics.
The design trends for nozzles are illustrated in Figures 1 through 6. Figure 1 is a conceptual
drawing of an early 70's booster nozzle. The significant feature of this design i,s that all of the
flame side parts are tape wrapped phenolics. Carbon and graphite cloth phenolic were the most common of
these.
The exit cone required a metal shell for structural support because the charring phenol ic
materials had insufficient strength to withstand the applied loads. Consequently, additional insulative
material had to be added to keep the metal shell from melting.
In the mid 70's, space nozzles
incorporated a pyrolytic graphite washer pack in the throat and a low density carbon phenolic exit cone
liner to reduce weight. This design represented a significant increase in the severity of the operating
environment over a phenolic based design. The use of the pyrolytic graphite material in the throat
permitted operating in a more severe environment while maintaining a reasonable throat erosion rate. On
the negative side, these throat packs experienced the formation of rather large steps and gaps along the
nozzle surface.
These irregularities were caused by enhanced erosion at the interfaces between
individual washers and differential erosion rates between the washer pack and the surrounding materials.
The design shown in Figure 2 approached a new generation in nozzle design beyond the technology of
the space nozzle designs in the mid 70's. This design retains the pyrolytic graphite washer pack in the
throat to minimize the throat erosion, but a single forward and aft piece of carbon-carbon material was
used for throat retention. Additionally, the exit cone was made of a carbon-carbon involute material
instead of an ablative.
This meant that the exit cone no longer required a structural shell. A
significant weight savings was real ized because the metal shell was el iminated and the exit cone 1iner
was much thinner. Some back surface insulation was still required to protect equipment surrounding the
nozzle, but this was satisfied with the use of a light weight graphite felt. The nozzle shown in Figure
3 represents another major step forward.
This nozzle uses a one piece 3 directional carbon-carbon
throat and entrance piece, referred to as the ITE, to minimize the number of components in the design.
A similar evolutionary process for air launch nozzles is shown in Figures 4 through 6. The nozzle
in Figure 4 represents a typical air launch configuration from the mid 1970's. The inlet and exit
materials are carbon phenolics. The throat region is assembled from a tungsten throat piece surrounded
by monolithic graphite. The entire structure is surrounded by a metal shell which supplies all of the
structural support for the nozzle.
Figure 5 shows a design from the late 70's.
This design
incorporates a pyrolytic graphite throat pack surrounded by phenolics and bulk graphite. The pyrolytic
graphite washers reduced the erosion in the throat of the air launched design, but incurred some of the
same penalty as the space nozzles due to the formation of relatively large steps and gaps. Figure 6 is a
more modern design incorporating a carbon-carbon throat piece which provides better erosion performance
and results in fewer parts. The thermal loads on the throat during pul se motor operation require the
use of a carbon-carbon material for survival.
STRUCTURAL ANALYSIS TECHNOLOGY
The evolution in the nozzle designs has required a similar change in the design and analysis
As the requirements have become more severe, the analytical methods have become more
technologies.
complex, and the area of analytical emphasis has changed. The all-ablative 'designs, such as in Figure
I, are critical in terms of the amount of erosion and the char depth. Char refers to the phase change
in the material when the temperature has reached the point where the phenolic resin pyrolizes. A great
deal of effort was expended to understand the materials' thermochemistry so that the amount of erosion
and the char depth could be determined prior to testing. The results of these efforts were subsequently

2-3

incorporated into the analysis techniques.

With the appearance of 'pyrolytic graphite washers, a new
emphasis came into being. Pyrolytic graphite has a greatly increased resistance to erosion compared to
the ablative materials. This produced a substantial differential in the eroded surface across material
boundaries and produced a highly irregular surface comprised of steps and gaps between components. As a
result, a significant effort was undertaken to determine the effects of these steps and gaps on the
efficiency of the nozzle and structural integrity.
Pyrolytic graphite is also highly anisotropic in
strength and modulus. This severely challenged the structural analysis techniques.
The search for a solution to the performance losses caused by the steps and gaps led to the use of
more erosion-resistant materials to surround the washer throat packs.
Figure 3 shows a design using
both a 3 directional carbon-carbon material and a 2 directional carbon-carbon involute material to
control the amount of erosion. At this same time, the carbon-carbon involute exit cone made its debut.
These materials produced a new emphasis on the analytical technology. The 3 directional carbon-carbon
materials experienced most of their difficulties during processing, so many of the efforts"'with 3
directional materials were focused on processing issues, but not necessarily from an analytical view
point.
Most of these efforts were approached empirically.
The 2 directional involute materials
represented an analytical challenge which had not been addressed previously. This was due to the unique
torsional displacement response to even axisymmetric loads on the exit cone such as internal pressure.
The standard axisymmetric formulations of commonly used computer programs could not account for this and
were in substantial error predicting stresses and strains. After the space nozzle designs progressed to
a configuration such as in Figure 3, the concern over erosion effectively disappeared.
The 3
directional and 2 directional carbon-carbon erosion rates, at the expansion ratio where the materials
interface, are similar enough that significant steps do not form. Also, since the entrance and throat
are now a single piece of material, no steps form here either. This greatly reduced the analytical
difficulties relating to erosion prediction. However, the design and analysis tasks were compl icated
due to the presence of a threaded interface between the ITE and exit cone.
I n the next several
paragraphs, the capabilities being used in the various computer tools will be described.
The structural assessment of a new nozzle design must consider a number of issues which will
determine the approach that has to be used. In addition to the environmental loads, temperature and
pressure, these issues include interfaces between materials, the specific materials to be used, and
operational considerations such as the requirement for thrust vector control (TVe). Nearly all of the
nozzle designs produced over the past several years have been designed assuming the problem to be
strictly axisymmetric. These methods have been shown to be incorrect for involute exit cones and are
obviously incorrect for loads and boundary conditions that are not axisymmetric, as in the case of a
vectored nozzl e. The need for hi gher performance requ i res that the des'i gns be more e ffeci ent with the
use of materials which, in turn, requires that the design and analysis methods be much more accurate.
Three dimensional analysis approaches are being used more often in order to achieve the accuracy. The
most obvious example of this is the analysis approach to the involute exit cone that has evolved over
the last five or six years. The involute exit cone will twist under an internal pressure load causing
the problem to be 3 dimensional. Until fairly recently it was bel ieved that an axisymmetric analysis
was sufficient. It has now been demonstrated that, not only are the results of an axisymmetric analysis
incorrect in magnitude, they are frequently incorrect in type, predicting compression instead of tension
or vice versa. Figure 7 is a plot of the axial stress distribution through the thickness of the part
from a representative structural analysis of an involute exit cone. This analysis was conducted using
an axisymmetric approach and a 3 dimensional approach. It is clear from this figure that the use of an
improper model can generate very wrong answers.
This realization led to a series of programs that
resulted in a 3 dimensional finite element model which properly accounts for the material orientation
angles and the torsional behavior of an involute part.
The analysis of a nozzle in a vectored condition is also a 3 dimensional problem which is normally
treated using an axisymmetric model. Typically, this problem is addressed by performing two or three
different axisymmetric analyses.
One model is used to investigate the unvectored or normal nozzle
position, a second model to incorporate the loads seen by one "side" of the vectored nozzle in the plane
of vectoring, and a third model to investigate the other "side" of the vectored nozzle. These models of
the vectored nozzle are axisymmetric approximations of real 3 dimensional problem, and the intent is to
bracket the actual conditions. If the analyses can show that the nozzle will survive at these 1imits,
then the nozzle is considered structurally sound. As will be discussed later, determining the gas flow
conditions in the nozzle is an important part of the analysis procedures. At this time, a 3 dimensional
flow ana lys i s methodology does not ex i st that can model the flow condit ions in appropri ate deta i 1 and
still fit on existing computer facilities.
Therefore, this part of the analysis also involves a
substantial amount of approximation with one dimensional and axisymmetric tools.
Another issue that has recei ved a sign ifi cant amount of attention is the interfaces between
dissimilar materials, particularly the threaded interfaces.
The design depicted in Figure 3 is
relatively simple in terms of the configuration of the flame surface components. This simpl icity is
ach i eved by us i ng a threaded a ttachment between the ITE and ex it cone.
The integrity of the threaded
interface in the pictured design is critical, particularly so since it is a vectorable design and the
vectoring loads must be transferred through the threaded joint. Only recently have there been serious
efforts to correctly model the joint. Most often the thread area is modeled as a continuous material
with estimated properties to account for all of the phenomena that occur. Thread shear tests would then
be run at room temperature to verify the structural strength of the interface. Figure 8 shows a finite
element model that uses this approach to model ing the thread response. Figure 9 is an enlargement of
the thread area in Figure 8 with the "thread" elements shaded.
An accurate approach to model the
behavior of the threaded joint requires a nonlinear solution that can account for the opening and
closing of gaps between the dissimilar materials as well as the frictional loads that are developed, see
Figure 10.
Additionally, since the joint is frequently filled with adhesive, the effects of the
adhesive as it degrades with temperature must also be included.
Due to the complexity of the model
required to capture these highly nonlinear responses, this type of analysis is generally performed using
only a portion of the overall nozzle structure, while the analysis of the total nozzle assembly

2-4

incorporates a crude approximation on the thread area.

Excluding the metal components of the nozzle. the materials used in current nozzle designs are
orthotropic in behavior. This means that there are three mutually perpendicular material directions.
Depending upon the specific material. the material properties can be significantly different in all
three directions. In addition, the effective properties in a single direction are different depending
upon. whether the 1oadi ng is tens il e or compress i ve (see Fi gure 11) and upon the tempera tUI'e of the
material at a particular location in the nozzle. Generally speaking. all of the materials exhibit some
degree of nonl inearity in their stress-strain responses. The typical structural analysis code stores
the material property information in a series of tables based on the temperature at which the properties
were measured.
Figure 12 illustrates these data in terms of the stress-strain curves.
The code
accounts for the temperature dependence by 1i nea rly interpol ati ng between the values stored in the
tables. the circled temperatures in Figure 12. Nearly all of the codes model the difference in the
properties due to the type of loading with a bilinear curve fit to the data. Figure 11. Some codes will
also approximate the nonlinearity of the tension only and compression only portions of the'curve with an
additional bilinear fit. Figure 13. More recently. with the trend toward using 3 dimensional analyses.
the nonlinear response is modelled as an actual curve. There are two approaches to modeling the actual
curve. The first is to input the stress-strain coordinates explicitly using a sufficient number of
coordinates to describe the actual curve. This is the simplest approach to implement and is being used.
However. thi s approach requ i res a si gnifi cant amount of data to properl y model the response since the
program will linearly interpolate between the input values. A scattered set of data will cause large
flat spots along the curve and produce a bad model. The second approach is to fit some form of
nonlinear curve through the measured data and then use this equation for the required information. This
approach is much more difficult to implement and the results are dependent upon the quality of the curve
fit to the actual response. The Jones-Nelson-Morgan model (Ref. 1) is an example of the latter method.
This model fits a curve of the form:
c
Property = A(l - BU )

through the measured data. Figure 14. In this equation A. B. and C are curve fit coefficients and U is
the strain energy of the material. "Property" is the value of the material property for a given value of
U. This represents a state variable approach since the variable U is the strain energy of the material
(energy stored in the body due to deformation). This method requires transitioning to a different type
of curve, usually 1inear, at some point because the form of the curve fit does not prevent the model
from going negative instead of reaching a 1imiting value.
If the fit is to one of the material's
moduli. then the implied stress-strain curve will appear as shown in Figure 15. In this case. the model
transitions into a straight line with a positive slope.
The goal of the analyses is to determine if the design will survive before the hardware is actually
built.
To assess survivability. the analysis results are judged against a predetermined failure
criteria. For the metal parts this does not present a problem. a Von Mises or Maximum Stress or Maximum
Strain criteria seems to work quite well. particularly if the metal part is considered to fail when it
reaches its yield point.
However, a failure criterion for carbon-carbon materials has yet to be
validated. The common practice is to use a maximum stress or strain based criterion or to use some form
of a polynomial curve fit through measured strength data. One approach to a polynomial curve fit is the
Tsai-Wu formulation or some variation of it (Ref. 2). This is a tensor equation involving second order
terms of the form:
Fio i + FijOiO j

=1

i.j

= 1.2 .. 6

Figure 16 represents a hypothetical Tsai-Wu curve in the sigma 1. sigma 2 plane.

Determining the
coefficients. F. in this equation is difficult if not impossible in general. therefore. the form is
typically limited to two directions (i.j = 1.2). The coefficients can then be related to the various
materi a1 strengths in the 1.2 plane. Recentl y. a sign ifi cant effort has been expended to develop a
criterion that accounts for the interaction between simultaneous loads in several directions on the
ultimate load carrying capabil ity for 2D involute exit cones. This effort involved a series of flat
panel. cylinder. and conical specimens loaded uniaxially and multiaxially to develop the data to define
the failure surface. Some test results from this effort will be discussed later.
THERMAL ANALYSIS TECHNOLOGY
The other half of the design/analysis effort. is the prediction of the thermal behavior of the
nozzle. The objectives of the thermal analyses are to determine the in-depth temperature throughout the
nozzle. to determine the material recession. and to define the surface pressures. all as functions of
the burn time of the motor.
This information is used separately by the designer to assess the
insulative qualities of the nozzle for the protection of the low temperature parts and to confirm that
there is sufficient material remaining at the end of motor operation to maintain nozzle integrity. In
addition. this information serves as input data to the structural analyses to predict the
thermo/mechanical response of the design and evaluate the survivability of the nozzle under the imposed
thermal and mechanical loads. The thermal analysis techniques have undergone an evolution similar to
the structural analysis methods.
The more current approaches are gravitating toward 3 dimensional
formulations to provide the sophistication necessary to address all of the appropriate analysis issues.
The thermal analysis methodology consists of a number of steps. each requiring its own approach and
consequently its own computer code. The procedures perform a series of chemical and energy balances
beginning with the free stream characteristics in the motor chamber. progressing into supersonic flow.
then determining the momentum and energy transfer through the boundary layer and finally considering the

2-5

conduction of the thermal energy through the various materials of the nozzle. The various methodologies
can consider conduction, convection, radiation, thermally degrading materials, mass removal, two-phase
gas kinetics, and particle impact.
The analyses are generally performed as a series of sequential steps.
Each step develops
information that becomes part of the input to the next step. The process begins with the free stream
calculations.
These calculations use the propellant formulation, heats of formation, and the motor
chamber conditions and produce the velocity, pressure, and temperature profiles at the outer edge of the
boundary layer down the length of the nozzle. The first analytical approaches to this problem used a 10
isentropic expansion model. This approach is applicable only in the supersonic region of the nozzle and
ignores any two dimensional effects in the flow field. The most recent development activities invol ve
this part of the thermal analysis cycle. The use of aluminized propellants result in the presence of
aluminum droplets in the exhaust gas, commonly referred to as two-phase flow. The early isentropic flow
analyses ignored the presence of these droplets. Current methods have included the effects of two-phase
flow into their methodologies and have extended the analysis model to include the subsonic region
upstream of the throat as well as the effects of submergence into the motor chamber.
The next step in the thermal analysis procedures considers the effects of the boundary layer.
Using the velocities, pressures, and temperatures previously calculated, this step determines the heat
and mass transfer coeffecients at specific points along the nozzle surface. The early methodologies for
these calculations assumed that the boundary layer formed on a smooth. wall. It has since been shown
that the surface roughness of the nozzle wall has a significant effect in the form of an augmentation of
the heat transfer to the wall. In the newer approaches, methods have been implemented to account for
the surface roughness, transpiration, and acceleration as well as the boundary layer properties. At
this point the chemical kinetics on the nozzle surface must be determined.
The computerized methods
that accompl ish this are based on a predetermined number of chemical reactions such as the reaction
between carbon and water at high temperature to yield carbon monoxide and hydrogen. The accuracy of the
results of the kinetic calculations are thus dependent upon the selection of reactions built into the
computer codes. The sel ected reactions have had to change over time because of the use of different
lOa teri a1s for nozzl e components and the increase in surface temperatures (greater than 5500 0 R) due to
the use of hotter burning propellants. If the nozzle design incorporates ablative materials, either on
the surface or as backup insulators, the release of the pyrolysis gases and the subsequent change in the
material's composition and thermal properties must also be taken into account.
Once all of the free stream, boundary 1ayer, and surface cal cu1 ati ons are made, the pred i ct i on of
the in-depth temperatures can be made. All of the previous calculations develop information that serve
as input data to the in-depth temperature predictions and establish the boundary conditions for the
problem. Early on, all of the thermal analysis methods were one dimensional. This required the analyst
to perform the calculations at a number of different axial stations and manually smooth the data between
the ana 1ys i s poi nts. More recently, some of these procedures have been expanded to a two dimens i ona 1
formulation. Now, in the case of the in-depth temperature calculations, the procedures are gravitating
toward 3 dimensional techniques.
In addition, the in-depth temperature calculations, which were
. traditionally based on finite difference solution methods, are being reformulated using finite element
methods. Using the finite element formulations makes the transfer of data between the thermal analysis
and the structural anal ys i s much eas i er and more accurate.
The fi nite difference approaches used
geometry models that were optimized for that type of solution and did not have as much general ity in
terms of the app1 ied boundary conditions. The finite element formulations for the thermal analyses
permit the use of a geometry model that is exactly the same as the model used in the structural analyses
without significantly impacting the accuracy of the thermal calculations.
While it is not necessarily obvious, another rapidly developing technology that has had a major
effect on all of the nozzle analysis capabilities is the improvement in computational capacity that has
happened over the last decade. Indeed, it has been mainly due to the development of the 32 bit super
mini computers that has allowed us to incorporate the increased complexity into the analysis approaches.
A1so, due to the development of these computers, the industry has been able to develop the computer
aided design and model ing (CAD/CAM) software that permits the human mind to deal effectively with the
magnitude of the input and output data for the analyses. While the computer technology has made huge
advances that allow more accurate and complex analyses to be attempted, further advances are needed to
permit solving the current problems. For example, significant increases in machine capacity and speed
are required to model the 3 dimensional flow problem and to sufficiently model the thread region of the
nozzle as an integral part of the entire nozzle assembly.
VERIFICATION
Development of more capable analytical techniques is important.
However, these new or improved
techniques must be verified before much faith can be placed in the results they produce. The most
positive way of verifying the analytical methods is to compare the predictions directly to measured
data. This can only be accomplished for cases where accurate measurements can be made. For instance,
one cannot expect to verify high temperature strain calculations because it is nearly impossible to
measure the data. Another way that some limited verification for an analysis technique can be obtained
is to compare its predictions against predictions made using other analysis tools. While this does not
provide assurances that the predicted numbers are real istic, it does determine how well predictions
based on one computer code compare with the same predictions made with another code. The assumption
that is made in using this approach is that the second code has already achieved some credibil ity for
correctly performing the analyses.
Both of these verification techniques have been used to gain
confidence with a specific analysis tool.
The first example of comparing one analysis against another is shown in Figure 17.
a hypothetical ITE insert subjected to a highly specialized set of boundary conditions.

The problem is
The purpose of

2-6

the exercise was to assess the sensitivity of the mathematical models to numerical approximations, and
not to analyze an ITE subjected to real istic firing conditions.
The boundary conditions for this
exercise permitted formulating an exact elasticity solution to the problem, against which typical finite
element solutions could be compared. Likewise, the material properties, shown in Table 2, were tailored
so the exact solution was available.
Figures 18 and 19 show examples of the results obtained through
this exercise.
Figure 18 is a plot of the circumferential stress through roughly the midplane of the
model.
The various data sets, A,B ,C ,etc., represent results submitted by several different
organizations. In this figure, all analyses appear to be quite good. Figure 19 is a plot of the axial
stress for this same problem, and, as can be readily seen, the various solutions do not compare nearly
as well.
Evaluation of the various answers showed that the main reason for the descrepancies in the
results was the care given to the modeling of the traction boundary conditions. The traction conditions
that were suppl ied with the problem description were not smooth.
If the analyst was not careful, the
variation in the tractions, coupled with the numerical approximations inherent in the finite element
models, could produce serious problems with the solution. This exercise was a clear reminder that too
much confidence is often placed in the computer to give the correct answer without applying the
engineering jUdgement to check the results.
TABLE 2
ITE ROUND ROBIN MATERIAL PROPERTIES
Engineering Moduli
Ez = 105,OOO,OOO./r psi
Ea = 70,OOO,OOO./r psi

Grz = 2,100,OOO./r psi

\lz a = \l zr = 'Jar = .05

Er = 35,OOO,OOO./r psi
Where 'Jij = -ej/ei under uniaxial stress 0i
and

= radial

coordinate

Free Thermal Strains

ez

.00001146666

ea
er

.000007
.000015

*
*

r
r

Figure 20 shows the geometry for another code versus code modeling exercise that occurred recently.
In this case, the problem was developed around an involute exit cone and was intended to assess the
capabil ity of the various finite element and other approaches for model ing the involute response.
Fi gures 21 and 22 show comparisons of the resul ts for three di fferent model ing approaches. The dashed
1ine in Figure 21 and the sol id 1ine in Figure 22 represent the results from a variational approach
incorporating a finite difference equation solver.
The open symbols in both figures are the results
from a fully three dimensional finite element approach, and the remaining data came from a 20
axisymmetric approach that had been modified to account for the circumferential degree of freedom. The
results from all three types of analyses agree very well. Since the 3 dimensional finite element model
used in these analyses had previously achieved some credibility based on comparison with measured data,
the other two models have now achieved some level of credibility as well.
Earl ier, it was mentioned that the analysis methods had to be improved to properly predict the
behavior of an involute exit cone.
Part of that series of efforts addressed nonl inear response and
multiaxial load behavior.
Figures 23 and 24 show two cylinder specimens that were tested to failure
under combi ned compress ion and internal pressure loads.
The measured hoop stress-strain behavi or for
these two specimens is shown in Figure 25.
These two tests were part of a test matrix to verify an
improved analysis model.
The predicted response using the new model is shown by the open circles
connected by the solid line in the figure. As can be seen, the model's predicted behavior agrees quite
well with the measured data.
It is also apparent from this figure that predictions using a 1inear
analysis methodology would be in substantial error. The argument could be raised that this is data on
an involute cyl inder, and that conclusions drawn from this data might not apply to actual exit cones.
Figure 26 shows the test configuration for a subscale involute cone.
This test setup was used to
measure the room temperature response of an actual exit cone geometry under combined loads. Figure 27
shows a cone specimen that was tested to failure under combined internal pressure and axial compression.
Figures 2B and 29 are plots of the measured test data for the hoop and axial directions and the pretest
predicted behavior using the new model. Here, again, the correlation is quite good. It would seem safe
to conclude that the new model represents a significant improvement over the previous methods. However,
this level of verification is still not definitive.
The model's ability to predict high temperature
response remains unproven, as well as the existence of any scaling effects for full size parts.
Another important finding came from the nonlinear model development and test effort. The material
properties used in these analyses were measured using specimens cut from flat panels made from the same
raw materials and coprocessed with the cylinders and cones. The performance of the model to accurately
predict the response of the cylinders and cones has provided some hard evidence that properties measured
from flat panel specimens are, in fact, representative of the actual response of the exit cone. This is
very important because property measurements using specimens excised from actual exit cones and tag end
rings are questionable, due to the curvature of the specimens.
It is also extremely difficult to
measure the cross ply response using excised specimens due to the small radial dimension of the part.
Carefully
made
flat
panels
can
be
constructed
to
provide
specimens
of

2-7

sufficient thickness to determine cross ply properties, and do not have any curvature to corrupt the
data.
The threaded joint between the ITE and exit cone is a very important interface and the ability to
ana1yze it is 1ess than sat is factory.
EffOi'ts have been undertaken to understand and pro perl y model
this interface. The capability to predict the response of threaded metal parts was developed a number
of years ago to address drilling technology for the oil industry. These same methods, when applied to
sol id rocket nozzles, do not adequately solve the problem. Research and development efforts have been
sponsored to address the threaded interface for nozz1 es.
Fi gure 30 shows two photoe1 astic specimens
using two of the most commonly used thread forms for nozzles.
The objective of these tests was to
assess the performance of existing analysis tools for solving the nozzle thread interface problem.
Figures 31 and 32 show the results of these photoe1astic tests compared against the analytical
predictions for both thread types.
The predictions were made using three different finite e1 ement
codes. These predictions are labeled as Code I, 2, and 3 in the figures. All three of the predictions
were made using an axisymmetric model. Code 1 represented the industry standard for nozzle analysis at
the time.
Using this approach, the threads were either ignored or, as was done for these analyses,
pseudo-thread properties were fabricated to estimate the response for the thread elements.
Code 2 is a descendant of Code 1 incorporating a specialized interface element. This element is a
4 node quadrilateral element of zero width that carries only 'compressive and shear loads along its
1ength. Code 3 is a general purpose three dimens i ona1 code with an axi symmetri c interface capabil ity.
The interface "element"
consi sts of two nodes, each associ ated with one side of the interface and an
ang1 e that represents the slope of the interface at the' 1oca t i on of the nodes.
Thi s element ca rri es
only compressive and shear loads. From Figures 31 and 32, the predictions agree reasonably well with
the photoe1astic measurements.
One would have hoped for better agreement because this configuration
represents the "i deal" case; an i sotropi c, well characteri zed materi a1 at constant temperature.
A
This test used real nozzle materials and
second type of thread shear test is shown in Figure 33.
attempted to measure data at each thread root on the 3 directional part.
Figure 34 is a plot of the
room temperature strain measurements against the pretest predictions.
The strain measurements were
made with small gauges located at the root of each thread as indicated by the filled dots marked Al thru
AS. The predictions were made using the same codes used for the photoe1astic case. The scatter in the
measured data might be due in part to experimental error, but variations in the thread clearances most
probably represent the largest effect.
The analyses assumed that the threads were perfect thus
producing a smooth variation in load.
This illuminates one of the biggest challenges to accurately
analyzing the behavior of the threads, how to incorporate the as-built geometry.
In a follow-on program an effort was made to determine the important mechanisms of thread behavior
in carbon-carbon parts. Knowing these mechanisms, it should be possible to develop appropriate computer
models that properly predict the response of the threaded interface. Part of the testing in this effort
used spec imens tha t represented segments of the cy1 i ndri ca1 interface regi on.
These spec imens cou1 d
either be curved sections, excised from the cylindrical interface, or specially machined rectangular
pi eces removed from the cy1 i ndri ca1 carbon-carbon parts, see Fi gure 35.
Ei ther specimen represented
approximately 15 degrees of the 360 degree interface. A special test fixture had to be built to test
these specimens (Ref. 3).
Figure 36 is a schematic of the test fixture which installs on a standard
load frame.
Using this specimen design allowed gathering the maximum amount of data from a 1imited
amount of material and permitted making photographic records of the exposed face of the joint while the
test was in progress. Figure 37 shows a set of these photographs for a sawtooth thread form with a 30
degree thread surface angle. Figure 38 shows the load/deflection curves for this same thread form. In
this specific case the specimen included an adhesive bond that had been precharred. In Figure 39, the
load/deflection curves from Figure 38 are simpl ified and represented by the dashed 1ines. The other
lines in the figure represent various degrees of modification to the finite element model in an attempt
to match the measured behavior. The modifications to the model were cumulative in nature and ended with
the following assumptions:
1.
2.
3.
4.

Shear yield stress set to 500 psi,

As built thread engagement (use correct number of active threads),
ITE's shear and Young's moduli set to 60 ksi,
Tension side of threads assumed to be unbonded.

The order of these assumptions matches the designations Mod 1 through Mod 4 on Figure 39.
The last
modification caused the computer results to lie within the bounds of the initial slopes of the measured
responses.
It should be noted here that this model ing exercise was not intended to match the complete
Further
1oad/def1 ecti on curve, but instead to uncover the major i nf1 uences on the thread behavi or.
analysis cases were run to capture the joint behavior after the threads became totally unbonded and
could move relative to each other.
To accomplish this, an average load level was chosen for total
unbondi ng based on the measured data, the open square in Fi gure 40.
Then different coefficients of
friction were tried for the interface element input. The extension lines from the open square in Figure
40 are for friction coefficients of 0.2 and 0.5. Estimates of effective coefficients of friction were
made from the measured axial and lateral loads.
The calculated effective coefficients of friction
ranged from 0.18 to 0.79 for the 30 degree sawtooth threads with a charred bond.
The resul ts of the segmented thread specimen tests which are most meaningful to model ing issues
indicate that:
a) Bonding improves the load carrying capacity of the joint even after charring.
For the ACME
threads the increase was about a factor of 2. This means that an effective model of the joint cannot
ignore the adhesive bond.
b) The effective coefficient of friction at the interface in the presence of charred C34 cement is
greater than 0.2.

2-8

SUMMARY
In summary, the design approaches for sol id rocket nozzles and the analytical methods used to
develop these des i gns have changed substanti ally duri ng the past decade. Although the nozzl e desi gns
have tended to become somewhat more simpl Hied, the demands that the design requirements place on the
new nozzle designs force the use of greater analytical complexity. The nozzle designs have progressed
toward minimizing the number of parts and reducing the surface erosion as much as possible. The most
significant changes have been the incorporation of 3 directional carbon-carbon throats for space and air
launch nozzle types, and the use of the 20 carbon-carbon involute exit cone in space nozzle designs.
The continuing trend for the design of nozzles is to incorporate higher performance materials by part
replacement without substantially altering the configuration. The most significant improvement in the
analytical techniques is the capacity to effectively model the behavior of the involute exit cone. The
current trend in nozzle analysis is to make wider use of nonlinear and 3 dimensional analytical
techniques.
The work which most needs to be done to improve the nozzle design capabilities is' in the area of
material technology.
The science and understanding of the manufacturing processes of the nozzle
materials is one of the most important areas to be researched. We must understand the influences of
changes in the raw materials and manufacturing processes on the ultimate performance of the nozzle part.
In terms of analytical capabilities, two problems come to the forefront as the greatest challenges. One
is the proper modeling of the threaded interface between the throat and exit cone. This joint is a key
feature of current nozzles and must be properly designed. The second is the capabil ity to analyze
ablative materials. These materials decompose with- time in the nozzle environment evolving gas and
drastically changing their character in terms of thermal and mechanical response.
All of these
phenomena must be considered to accurately predict nozzle performance.
REFERENCES
1)
Jones, Robert M., Mechanics of Composite
Compression, AFOSR TR-78-1400, July 1978.
2)

Materials

With

Different

Moduli

in

Tension

and

Jones, Robert M., Mechanics of Composite Materials, McGraw-Hill, 1975, pp 76-83.

3)
Jortner, J. and Pfeifer, W.H., "Experimental Study of Strength, Deformation, and Failure Modes of
Threaded Joints Between 20 and 3D Carbon-Carbons", ASME Booklet AD-II, 1986.

All Fl..... P.,t_ ar. Carhft Cloth Phenolic

Fig. 1.

C-C -

Modern phenolic based nozzle.

CARBONCARBON COMPOSITES

GP - GRAPHITE PHENOLIC
SP - SILICA PHENOLIC

Fig. 2.

Space nozzle - late 1970's.

<D

SUlc. Cloth Phenolic

SWeone Rubber

CarbonClothPbtnoJlc

Pyrolytlc Graphite Wuhera

Graphltt Cloth Ph.noUc

<V

Steel

@
@)

Fig. 3.

Fig. 4.

Space nozzle - early 1980's.

CDSto.'

Graphlt.

o Cllrb.. Ptl.....

SlUe. Phtnollc

@Graphlte

@BClndlln.

Carb.. Phenolic

@
@

Carha Ph,notlc

@8ondLln.

CarbonPbtnolic

@8ondUn.

@ Ste.J Jal.t
@ Gllrhe Phentllc

Silica PheoUe
@Ti g.t..

Gr.,hlh

Fig. 5.

Air launch nozzle - mid 1970'.

(i)

Silica Phtlltlle

@GI... Pht"tUc

Bond Lin.

Air launch nozzle - late 1970's.

!
~

I.
2.

J.
4.
5.

6.
7.

1.i)ZZL atTAHUNC IlINC

tlOZZLE5HElL
[:tncal/ELINER
ITt
THllOA:r Hl5ULATOR
Ai'1'1I0ACIlIHSULAtOR
/iOLDEDIN5UU,10R

AL.UIIlj!jUIol
ITEEL

CAUOH/fIlENOllt, TAl'EIIR.\I'HU
J-Otl\ CARIION~CAR801" Ill-Del( "'LAVE
CARIION/rtlEtlOLlC,TArt>lRAPPED
CAJUIOtl/I'IlEIIOL!C, TAPrUl<ArfEU

-20(10'

/iOLDEDSILtCAPIiENOLlC

0.00

0.20

0.40

0.60

o.eo

I !

1.00

1.20

CBNE a.D. 16 FLAl1E SURFRCE

Fig. 6.

Air launch nozzle - early 1980's.

Fig. 7.

Comparison of results between involute and

axisymmetric models.

Fig. 8.

Typical finite element model to analyze

nozzle assembly.
N

'"

t;J
.....

a- - 81UNEAR APPROXIMAnON
- - - Ac:nJAL IEHAVIOR

~~>'II

IlThnadll Elementl an Shaded

~T"ANSlTlON

REGION

I
Fig. 9.

Enlargement of pseudo thread-elements in

nozzle assembly model.

Fig. 10.

Deformed geometry for detailed thread

analysis model.

Fig. 11.

Bilinear material model versus actual

stress-strain behavior.

Circled Tallparatur.. Indicate

Int.rpolat:a40ata

"

Umolrcld Ta.p.ratune Indicate

W.aaurad Data

30

25

en
en
w

ACTUAL STRESS-STRAIN CURVE

MECHANICAL
PROPERTT

IX
l-

en

~MECHANlCA1

0.7

Umax

U*

ii

PROPERTY A(J~BuCl

STRAIN

Fig. 12.

Typical stress-strain versus temperature

data.

Fig. 13.

Bilinear fit to tension only stress-strain

curve.

Fig. 14.

Jones-Nelson curve fit to typical mechanical

property.

Ul WARP STRESS

u:l FILL STRESS

POINT

~_ <....---'"f'
RANSITION

acr'

1)2

*'--*

LeXTRAPOlATED
STRESS - STRAIN CURVE

_--'/"?

er.

0"1_0

__

INPUT DATA POINTS

(J:j'

IMPLIED STRESS - STRAIN CURVE

,,"

*"

*'*

Fig. 15. Implied stress-strain curve

from Jones-Nelson property model.
Fig. 16.

x -

Unll.llial Tut.

*-

81alll_1 T"tII

Hypothetical Teal-Wu curve fit to warp-fill

stress plane.

Fig. 17.

ITE code comparison problem geometry.

-I

r-

.A
oB

.c

00

-E
.F
.G

yH

I.

:l!

ff.~
$

b~

-I

;;i

I;'

-2

.0
..

-3~
.,;>.y
-I

7.2

8,0

8.8

9.6

/
a

'b

10.4

. ..
..
.

IJ

"

.....

: .

00

-12

8.0

BB

96

104

IJ r,in

Fig. 20.
Fig. 18.

ITE code comparison circumferential

stress results.

Fig. 19.

ITE code comparison axial stress

results.

5~/6J
.1 I

ALL DIMENSIONS IN eM.

'G
YH

-4

"II r,in.

9""

-E
F

22.72

~ 9"77

m~~

"

.. "'. ,0 ,

14.89

.. .

r
7"03_1'-l
'-\\'2

7.68

Exit cone code comparison problem

geometry.
N

.....
.....
I

 :".'J

~
.....

V.riatl...l Af,nach

"
"
20

l:!

.;

0 0

// \\
,:

3D Finite EI._nt Approacla

- - V"latl.... "Hruca
o

[] 3D fiMt. EJ t .......aacb

YD4I'I"Axla:t. . .trlcFlnlt.

Mo4lfl.4 A.la:t_tric FWt,

E'e_.nt

,/.~--\

El....t Alp""c'
fiOU

"', \\,

tv

""t_c.

"
R.

c"

-10

~!<O

~ou

Fig. 21.

Exit cone code comparison R-Z shear

stress results.

Fig. 22.

Exit cone code comparison

mid-section stresses.

Fig. 23.

Involute cylinder (#585) failed under

biaxial loads (axial compression and
int~rnal pressure).

INVOLUTE CYLINDER
COMBINED LOAD
AXIAL COMPRESSIONI
INTERNAL PRESSURE

HOOP STRAIN

Fig. 24.

Involute cylinder (#587) failed under

biaxial loads (axial compression and
internal pressure).

Fig. 25.

Measured versus predicted response for

cylinders under biaxial loads.

Fig. 26.

Involute core specimen for biaxial

load test.

Involute Mod.,

TEST CONE 103

COMBINED LOAD
AXIAL COMPRESSIONI
INTERI';Al PRESSIJRE

HOOP STRAIN

Fig. 27.

Failed cone after biaxial test.

Fig. 28.

Measured versus predicted hoop

response for cone under biaxial loads

TEST CONE 103

COMBINED LOAD
AXIAL CQM?RESSION,
INTERNAL PRESSURE

AXIAL STRAIN

Fig. 29.

c,",,,,,,,

~"'''''''''''~

w.l-Qi)-"'~~.:lI-~ ~~'h ;~V-<>-"

""u:-'
J;,
':~~::..'c..t1, \
"--fa~

Measured versus predicted axial

response for cone under biaxial loads.

,~.

--;::-;-

' .....

"1---1-

oC.d.3

o Co

401.1

Fig. 30.

Photoelastic specimens for two thread

forms.

Fig. 31.

+f
.-8"0.

locuto~

~ ..

"'-0

-,~:
1 ~:~

o Cod. 2
(:, Photo.lutlc

~.t~j

o Cod.:3

o
6.'1

,,,,fie.

Measured versus predicted results for

ACME thread form.

.h'

Cod. I
I

'II

5.0

(.0

I
J.'1

A_i.l Lo,.t1on -

Fig. 32.

!
2.0

- ............

tn~hu

Measured versus predicted results for

lamp bulb thread form.
'?'

"'"

t;>
>-'

-I>-

LOAD

"0

,.,,,r,IO'"
~.
... "".."..
LOAD
,":",,'
LOAD~
"

~ n}---~

\!

.,

DC." 2.

"k

V \,rl

o Cod. I. F=73.5

F=I'6'~
~~

Geode 3. F =l 022

.Coll_ 3. F:; 340& Ii

$~:3~UO'ltt

eM..."rt4
F= 6600 Ibt

Fig. 33.

All carbon-carbon thread test

specimen.

(81 RECTANGUlAASEGMENT

IAI CYllHDRICAlSEGMEt1T

fL/:--=--tI--+i
_1~~,
,;nlnYn1nin
.l
.l
.l. .1 l
.

."" ,

Fig. 34.

Fig. 35.

Segment specimens for testing threaded

joint configurations.

Measured versus predicted response in

thr'ead roots.

~ INSH:ON
CROSS dEAD

<.>-_-,--_-+-_,.-~~

(i';ifi;~

\,hh:!

\~'i'i~;

J,~".l

'l:I"J;"'i~

Il~:,i,.d

I~W:f

J~(j

'In:!

AXIAL
LOAD
(LBl
1500

>, 1..
i';.';

',iL'

"j ,

l>JOO

t
',l

LATERI\L
LOAD CELL
(OPTIONAl)

sao

I,

LOWER PLATEN

I'IlI II SI

Fig. 36.

Test fixture for thread segment

specimens.

BEFORE
nONIl IlIU ftK
377, ULI.

Fig. 37.

ftFTFR
IImlll IlIUftK
23% ULT.

99'IIlLlIMftlE

Photographs of deformations for 30

degree V thread test.

rnST-tiLTII'lftlE

.07

.06

,as

.04

.03

.02

.01

CROSS HEAD TRAVEL. Ill.

B8r, ULI.

Fig. 38.

Load-displacement results for 30

degree V thread tests.

2-15

'GOO

...... ';:?'J
"
......
, .. ..,' ..
............: ........
' "
~ ..'

.0

"JJ

"

....

.~

J/'
0.0

~
, ... , ....

...., ..

........

.--0 .... ,...._

~Pftjc ... I . . I ... I....

1-

--6"*... ' . I.... I .... ~=U

.........

~Pn6:'"'

I .... p..=U

I
10.0

20.0

30.0

40.0

50.0

60.0

OISPLACEMENT CMILSI

Fig. 40.

Measured versus calculated response

for model with friction.

"II" TIlCJIO

ClIMf(

n:n

~'r

TI_'__1IlI/'

Ul!Dfl4lDQ,.IIWooIllQIIIII'l

IPtCIJIU

CI Q"UIIIIll1..I,.KU,

.A ctMI.t

DI\..",....
"'I
1.. "l1li11.1 'OllI l*IPOT

l .,111(' _(I. cu_ll

CIIII,.,............. nn ... :llIa
+ . . "*1. a-,. nn WlClAO!

....

-.

1I..u .....1

3000

25DD

2000

noD

1000

:--."'."'.,:--.=-='.,

,.1:: - - ".."'
' : - - ".."'
' : --,-."'- -."'."'

aUr'LACf"fIfT (niLS)

Fig. 39. Measured versus calculated response

for various model assumptions.

3-1

DESIGN AND ANALYSIS OF SOLID ROCKET MOTOR NOZZLE

by

Mr ATruchot
Agence Spatiale Europeenne
Division STS-AR
8-10, rue Mario Nikis
75738 Paris Cedex 15
France
ABSTRACT

The objective of this paper is to present the different methods used to design and analyse solid
rocket motor nozzles. In a preliminary phase, the typical requirements and parameters required to design a
nozzle are discussed. Afterwards the general process of the design is detail led : nozzle configuration,
aerodynamic contour design, selection and sizing of liners and insulators, etc... In a second step, the
different systems used to control the motor thrust or to improve the performance, such as extendible exit
cone, are presented. To conclude the presentation, the different steps of the nozzle analysis are
presented.
For each step (aerodynamic, thermal and structural) the different computer programs used are
presented (principle, possibilities, limitations).
1. - INTRODUCTION

The main function of a solid rocket motor nozzle is to

coming from the chamber, thus creating the motor thrust.

channel and control expansion of hot gases

The scale and complexity of a nozzle are dependant on the application of the motor (ballistic,
space
or tactical) and the level of performance required. Thus for tactical applications, with low burning time
and reduced scale, the nozzle is often a simple device made up of metal or classical reinforced plastic.
However for ballistic or space applications, the scale and complexity are of more importance due to the
greater dimensions (greater than one meter), the more important burning time (1 to 2 minutes) and the
higher level of performance required.
This paper deals mainly with large solid rocket motor nozzles
missile applications will be presented.

f'fm

Pressure

however some examples

of tactical

(/J col

throat diameter

~s

exit cone expansion ratio

-f'fm

exit cone half angle

(/J ouv.

aft opening diameter of the case

x/l

nozzle submergence

f(t}

(/Jcol

(/Jouv.

Fig. 1 -

NOZZLE PARAMETERS

In order to design a nozzle, a certain number of specific parameters are required such as throat
diameter, exit cone expansion ratio and half angle, aft opening diameter of the case and nozzle submergence. A great number of these parameters are a result of the overall- optimization of the motor based upon the
general requirements of the propulsion system. Some of these parameters are left to the discretion of the
designer or imposed by the storage or operating environment. These different requirements and parameters
are summarized below.
Operating conditions (pressure, burn time)
The mechanical design is usually performed with
thermal design with the maximum burn time.

the

maximum

expected

operating

pressure

and

the

3-2
Throat area - Throat area variation
The throat area variation can be compensated by the grain design. The throat.erosion is a function of
several parameters such as chamber pressure, burn time, nozzle configuration and grain design.
On this
point several iterations between grain design and nozzle design may be required.
Propellant type :
The characteristics of the propellant such as combustion temperature, erosion and oxidation properties
are of great importance to select nozzle materials (liners and insulators).
Envelope limit :
The envelope limit is usually a vehicle designer requirement.
restricted by guidance or navigation equipment.

In some systems,

the envelope

can

be

Exit cone expansion ratio :

The optimum exit cone expansion ratio is dependant on the external pressure. The optimum efficiency of
the nozzle is obtained with a static pressure at the exit plane equal to the external pressure.
Thus for
first stage, expansion ratios from 7 to 15 are usual. Expansion ratios from 15 to 80 may be used for upper
stage or high altitude vehicles.
Nozzle submergence :
For a same exit cone expansion ratio, an increase of the submergence allows a decrease of the external part of the nozzle, thus leading to an improvement of performance for limited length motors. The nozzle submergence is usually a result of the overall optimization of the motor.
Thrust vector control
The thrust vector angle is a vehicle designer requirement. The most usual device currently used for
ballistic or space applications is the flexible bearing allowing thrust vector angles from 3 to 15 degrees.
Case interface :
The motor inert weigth is minimized when the aft opening diameter of the case is minimum.
minimum diameter may be required in order to cast the propellant into the motor case.

Sometimes a

Performance - Cost - Reliability - Development time :

Nozzle technologies, such as materials or thrust vector control systems are functions of the performance,
cost,
reliability and development time objectives. These considerations can restrict the USe of
certain efficient technologies, because they are not sufficiently demonstrated.
Environment - Storage - Life time :
These different considerations must be taken into account for the materials selection
elastomers are incompatible with a low storage temperature.

thus,

certain

3-3
3. - NOZZLE DESIGN
3.1. Configuration
The nozzle can be external to the motor or submerged into the combustion chamber

EXTERNAL NOZZLE

SUBMERGED NOZZLE

Fig. 2 -

NOZZLE CONFIGURATIONS

These two configurations lead to different internal flow fields. The distance to the grain of an
external nozzle is increased, thus giving a lower erosion of the throat. However the aft dome internal
insulation erosion is greater due to the importance of the gas velocity near the wall.
An external configuration is usual with fixed nozzle and non limited length motors, thus giving a
more simple design at a lower cost.
The submerged configuration allows an increase of the performance for limited length motors.
However this configuration leads to a more complex design and a higher cost. A submerged configuration is
usual with movable nozzles; thus, in the case of a flexible bearing, the articulation is located in a low
mach number area (see figure 3).

Propellant

Flexible bearing

Fig. 3 -

MOVABLE NOZZLE

3-4
An other possible configuration is the four nozzle design. This configuration has been used on
the first generations of solid rocket motors to provide thrust vector control.
With the development of
omniaxial thrust vector control systems,
the four nozzle configuration has considerably declined in
usefullness. Single nozzle systems provide a higher efficiency at a lower cost.

Fig. 4 -

FOUR NOZZLE CONFIGURATION

An usual configuration for tactical applications is the blast tube nozzle. This design is used to
accommodate guidance or navigation equipment.

Fig. 5 -

NOZZLE WITH BLAST TUBE

3-5
3.2. Aerodynaaic contour design

The general configuration of the nozzle aerodynamic contour is presented in figure 6.

entrance area ratio

throat diameter
upstream radius
downstream radius
initial angle
exit angle
mean half angle
exit cone expansion ratio
deflection angle

ke
(j)col

Rl
R2
Oi
Os

Om
kS
Oi- Os

Fig. 6 -

AERODYNAMIC CONTOUR

In the subsonic area, contraction ratios from 2.5 to 3.5 and nOSe tip lengths greater than the
throat radius are usual. These rules are based on both aerodynamic and thermal considerations (see
reference C 1 J). In the exit area a contoured shape is desirable in order to limit divergence losses. However in the case of aluminized propellants the deflection angle is limited by particules impacting on the
tip of the exit cone. Typical values of deflection angles are in the range of 9 to 14 degrees.
For the same deflection angle and same length of the exit cone, an increase of the half mean
angle leads to an increase of both the expansion ratio and divergence losses. The optimum v~lue of the half
mean angle is a compromise between these two aspects.
3.3. Theraal liner and insulator materials

After designing the aerodynamic contour, the next step of the design consists in selecting and
s1z1ng thermal liners and insulators. The thermal liner is exposed directly to the hot gases and forms the
aerodynamic contour of the nozzle. The insulator serves to protect structural parts of the nozzle. Sometimes, a same material can achieve different functions at the same time; it is the case of phenolic materials, which can serVe as both liners and insulators.

Fig. 7 -

liner

insulator

nozzle structure

chamber structure

NOZZLE MATERIALS

In a first step, the nozzle thermal design is performed taking into account experience obtained
in firing tests, certain empirical correlations and one dimensionnal simplified thermal analyses.
This
preliminary design is afterwards refined during the thermal analysis of the nozzle (see 6.2. Thermal
analysis).

3-6

3.3.1. Thermal liner materials

a) Refractory metals :
These metals have been used on the first generations of solid rocket motor nozzles.
Tungsten with its very high melting point (34l0C) has been used as throat material. The main drawback of
these metals is the density (19.3 in the case of tungsten). For this reason, these metals are rarely used
at the present time.
b) Graphites
Two classes of graphite have been used: the polycristalline graphite and the pyrolitic graphite. These materials are able to withstand very high temperatures (greater than 2600C) but present a low thermal shock resistance and an unpredictible thermomechanical behavior. For these reasons the
use-of graphite has considerably declined.
An example

of a nozzle

using

graphite

washers in

the throat

area is

presented in

figure 8.

Fig. 8 -

NOZZLE WITH PYROLYTIC GRAPHITE WASHERS IN THE THROAT

c) Ablative materials :
An ablative material is made up of a refractory reinforcement and a polymer resin.
Different sorts of reinforcement are possible such as carbon, graphite, silica, glass or asbestos. phenolic
matrices are extensively used for this application because they give off high levels of char during the
pyrolisis which occures during the motor combustion. The pyrolisis starts approximately at 300C and
increases the thermal protection effect; indeed most of the chemical reactions involved are endothermic
thus absorbing energy. The pyrolisis is accompanied with an important o~tgassing, which must be taken into
account for the design.
Carbon, graphite or silica phenolics are the most commonly used ablative
Typical thermal and mechanical properties of these materials are summarized in table 1.

materials.

CARBON-PHENOLIC

GRAPHITE-PHENOLIC

SILICA-PHENOLIC

1.40 - 1.50

1.35 - 1.54

1. 70 - 1.80

Tensile strength (MPa)

110 - 140

40 - 150

80 - 190

Ablation temperature (C)

2500

2500

1700

Density

Heat diffusivity (m 2 Is)

0.5 - 0.8 10 -6
TABLE 1

1 - 1.5 10 -6

0.2 - 0.3 10 -6

3-7
These materials present low density, good insulating properties and high ablation temperatures.
The use of a carbon or graphite reinforcement gives a higher ablation temperature than that of silica.
silica phenolics can be used in areas less exposed to thermal loads such as the downstrearm part of the
exit cone in order to decrease the cost; indeed the price per kilogram of silica fibers is approximately
one third that of carbon fibers.
Carbon or graphite phenolics can be used in the thoat area. This technology is used at the present time for the nozzle of the Space Shuttle Solid Rocket ~ooster; at the development time, a carboncarbon throat would have been too expensive because of the large size of the nozzle. However this technology leads to the following drawbacks, a higher ablation rate and a more complex design. Moreover, the
throat must be segmented in several parts in order to achieve the proper thermomechanical behavior of the
nozzle.
Ablative parts can be molded.
This process is well suited for small parts produced in large
quantities. For larger sizes and reduced quantities, tape wrap or lay-up (see figures 9 and 10) are less
expensive.
For parts such as exit cone,
tape wrape is usually the most economical solution. The
reinforcement can be parallel to the center line or inclined at a certain angle.

MOLDED PART

ROSETTE LAY-UP
Fig. 9 -

MOLDED OR LAID-UP PARTS

ex

PARALLEL TO AXIS

BlAIS TAPE
Fig. 10 -

TAPE WRAP

3-8
d) Carbon-carbon
Carbon-carbon materials present many advantages compared to other liner materials : low
density, good thermal stock behavior, excellent erosion resistance and more repr.oductible thermomechanical
behavior.
Most modern nozzles use one-piece carbon-carbon integral throat and entrance.
These
monobloc parts are heavely loaded by the thermal gradients during the motor combustion. This type of onepiece design presented in figure 11, has enabled great simplications and an improvement of reliability by
eliminating throat failure.

CARBON-CARBON ITE

Fig. 11 -

CARBON-CARBON INTEGRAL THROAT AND ENTRANCE

The fabrication of a carbon-carbon part involves two phases

Preform construction :
The preform construction can be performed by different processes such as carbon clothes lay-up, multidirectionnal weaving and pultruded carbon rods arrangement
Densification process :
The densification consists to fill up the preform porosity by a carbon matrix. There
are basically two methods to deposit the matrix into the substate. In the liquid impregnation method, a
resin or a pitch is injected into the preform and charred in order to create the carbon matrix. In the chemical vapor deposition process a deposit of pyrocarbon is obtained by cracking a gazeous hydrocarbure.
The different densification cycles are repeated until the required density of the
part is obtained. Lastly a final graphitization of the matrix is performed at temperatures greater than
2000'C. Typical densities in the throat area may vary from 1.9 to 2.0 ; however, for weight saving, lower
densities are desirable in less severe erosion environments such as exit cones (1.5 to 1.6 glee).
3.3.2. Insulators
Nozzle insulators can be divided into the three following classes
a) Elastomers :
These materials are mainly used for internal insulation. They can be used also for nozzles in subsonic areas exposed to very low mach numbers.
b) Phenolic insulators :
Phenolic
insulators have already
Ablative materials). These materials present excellent
tion effect is increased by the ablation process, which
mical reactions involved are endothermic thus absorbing

been presented as liner materials (see 3.3.1. c)

thermal insulation properties. The thermal protecoccures during the firing; indeed most of the cheenergy.

silica phenolic diffusivity is two to three times less than the diffusivity of
phenolic, but its use is limited in temperature because of the lower ablation temperature (1700'C).

carbon

3-9
Phenolic insulators present some limitations. The degradation of the matrix, which occures during firing leads to poor mechanical properties at elevated temperatures. Outgassing of pyrolisis
gasses may induce stresses in other parts of the nozzle; moreover the modelisation of phenolic insulators
for thermomechanical analysis is difficult due to the chemical change in the material.
c) Thermostable insulators :
An insulator is said to be thermostable when its reinforcement and matrix are both refractory. Different combinaisons of matrix and reinforcement are possible using carbon or ceramics. These
new materials currently in development are very promising (see 3.5. Multifunction materials).
3.4. Structural materials
The structural materials form the mechanical component of the nozzle. The most
are metals (aluminium, steel, titanium) and reinforced plastics (carbon or glass epoxy).

usual

materials

3.5. Multifunction materials

Some composite materials have the capability to acnieve different functions at the same time. It
is the case of carbon-carbon for exit cones and thermostable insulators. They can serve at the same time as
liner, insulator and structural materials.
3.5.1. Carbon-carbon exit cone
A freestanding carbon-carbon exit cone presents the following advantages : a lower inert
mass thus increasing performance and a reduction of the number of parts thus decreasing nozzle complexity.
The nozzle of the Apogee Boost Motor MAGE II is presented in figure 12. This nozzle, which
has been developed in 1980 by SEP, uses a carbon-carbon exit cone. 15 firing tests have already been successfully performed.

Carbon-carbon throat

Fig. 12 -

MAGE 2 NOZZLE

The main drawback of carbon-carbon exit cones is currently the cost; indeed for the different operational applications, the preform construction is hand-performed by an involute process (STAR
30, IUS, MAGE II).
However the development of new methods enabling the construction of large multidirectionnal reinforced preforms will increase the applications of carbon-carbon exit cones by decreasing the cost.
Thus for freestanding exit cones, SEP has developed a new carbon-carbon material called NOVOLTEX. This new
material is characterized by a fully automated preform construction and a very fine spacing texture.
3.5.2. Thermostable insulators
A thermostable insulator is made up of a multidirectionnal carbon or ceramic reinforcement
and a carbon or ceramic matrix. The final mechanical and thermal properties are a main function of the combinaison of reinforcement and matrix. These new materials currently in development are presented in reference [6]. Some typical thermal and structural properties for these materials are summarized in table 2.

3-10

Reinforcement ~
~

Density

Ultimate tensile
strength

2.3

100

Matrix

Carbon-alumina

------------------- ------------------Alumina-carbon
1.7
55
------------------------------------~----------------------Alumina
+

carbon

1.55

Heat diffusivity
10 -6 m 2 /s

Temperature of
use

1,8

1800C

0.8

~----------------1500C

~----------------- ~-----------------

55

> 2000C

0.7

carbon

---------------------------------------------------------------------------------------------------Carbon-carbon
1 .60
70
3.0
> 2000 C
------------------------------------------------------------------------------------------------Carbon-zirconia
2.3
100
1.1
2000C
TABLE 2
----

Compared

to

phenolic

insulators,

thermostable

insulators

present

the

following

advantages :
no outgassing
excellent mechanical properties at elevated temperature
more predictible thermomechanical behavior
design simplications
decrease of the nozzle assembly time.
A nozzle using thermostable insulators is presented in figure 13. Design
of this nozzle, fired in 1982, are presented with more details in reference [5 J.

Fig. 13 -

CSD - SEP MOTOR NOZZLE

and development

3-11

Vectoring loads can be transmitted by thermostable insulators. An example is presented in

figure 14. The vectoring loads are applied directly to the exit cone and transmitted to the flexible bearing by a monobloc part, which achieves at the same time a mechanical function (transmission of vectoring
loads) and a thermal function to protect the flexible bearing,

Non degrable insulator

Flexible bearing

Carbon-carbon

Fig. 14

NON DEGRADABLE INSULATORS APPLICATION

4. - THRUST VECTOR CONTROL

In many solid rocket motors thrust vector control is required. The nozzle itself can provide Tve; it
is the case of movable nozzles hinged by a flexible bearing, a ball and socket, or a hydraulic bearing
joint. The nozzle can also be combined to an attached device acting on the nozzle supersonic stream : fluid
injection, jet vanes, etc

4.1. Movable nozzles

a) Four nozzle configuration
The four nozzle configuration has been used on the first generation of
The thrust is controlled by a selective orientation of each canted nozzle.

Ball-bearing
O-Ring
Fig. 15 -

ROTATING CANTED NOZZLE - PRINCIPLE

solid

rocket

motors.

3-12
The use of this very complex technology has considerably decreased over the past years. Single
nozzle systems provide a higher efficiency at a lower cost.

A flexible bearing is made up of alternate spherical segments of elastomeric and rigid shims
(see figure 16).

Elastomer

Shims

Fig. 16 -

FLEXIBLE BEARING - PRINCIPLE

The flexible bearing enables control of both pitch and yaw attitudes of the missile. If a roll
movement control is necessary, it is controlled by a other system.

Flexible bearing

Fig. 17 -

NOZZLE HINGED BY A FLEXIBLE BEARING

3-13
Two designs are possible depending on the position of the pivot
tions are called upstream and downstream pivot point configurations.

Upstream pivot point

Downstream pivot point

Fig. 18 -

point. These two configura-

FLEXIBLE BEARING CONFIGURATIONS

The flexible bearing is the most widely used device in modern nozzles for ballistic or space
applications. The use of a downstream pivot point allows a better protection of the joint which is located
in a low mach number area.
c) Hydraulic bearing
The principle of a hydraulic bearing in presented in figure 19. The joint consists of a fluid
contained in a rubber membrane, thus reacting to the nozzle blow-off load. This technology is used at the
present time for the motors of the orbital transfer vehicle IUS (see figure 19).

--~----

-------4 - -------Fig. 19 -

HYDRAULIC BEARING - PRINCIPLE

3-14
A low torque value is associated to the hydraulic bearing
in the IUS, this characteristic
allows the transmission of vectoring loads directly by the carbon-carbon exit cone and the phenolic insulators.

d) Ball and socket :

The ball and socket joint consists of two spherical sliding surfaces, thus g1v1ng an omnidirectionnal orientation to the nozzle. Low friction coefficient materials are desirable in order to m1n1m1ze
the torque value. If the joint is made of teflon, it is called "cold ball and socket", and must be protected by thermal insulators, because the use of teflon is limited by temperature. Carbon-carbon can also be
used ; the joint is called "hot ball and socket" and can withs tand higher temperatures in the range of 1000
to 2000C (see figure 20).

Ball and socket

Fig. 20 -

BALL AND SOCKET - PRINCIPLE

e) Supersonic splitline nozzles

In most movable nozzle designs, the splitline between the fixed and movable portion of the
nozzle is located in the subsonic area of the nozzle. This splitline can also be located in the supersonic
area. An example is presented in figure 21.

Structure

Carbon-carbon
Fig. 21 -

Supersonic splitline nozzle

SUPERSONIC SPLITLINE NOZZLE

This nozzle is made up of a carbon-carbon throat and exit cone hinged by a hot ball and
socket.
This nozzle has been developed by SEP and fired for a cooperation program with the compagny ARC.
The test motor was representative of a tactical missile motor with a low burning time and a non aluminized
propellant.
The firing test was successful and demonstrated an 18 thrust vector angle. The supersonic
splitline nozzle presents two advantages :
a high vector angle capability: the exit cone steering angle
to 1.7 created by the supersonic deflection effect,

is amplified by a factor ranging from 1.2

an important design simplication by reducing the number of parts.

However the concept presents limitations with aluminized propellants because of the erosion by
particules impacting on the exit cone.

3-15
4.2. Attached syste.s acting on the nozzle flow
Two categories of attached systems acting on the nozzle flow have been used to
solid rocket motors :

provide

TVC

for

a) fluid injection
b) solid devices introduced into the nozzle supersonic stream.
4.2.1. Fluid injection
result
flow.

of

both

When a jet of fluid is injected into a supersonic stream, a side force is created as a
momentum of the jet and the dynamic and chemical interactions of the jet with the primary

Separation shock

Injection

Fig. 22 -

Injectant-gas
Mixing and reaction

FLUID INJECTION - PRINCIPLE

The main advantage of fluid injection is the design simplification of the nozzle compared
to that of movable nozzles. The main drawback is the increase of inert mass due to the transported fluid.
For this reason the use of this technology has considerably declined over the past years. However the current development of hot gas valves using modern ceramic-ceramic materials enables the controlled injection
of hot gases coming directly from the combustion chamber. This gives new interest to this technology.
4.2.2. Solid device introduced into the nozzle supersonic stream
Several systems have been used mainly for tactical applications ; two examples are presented in figures 23 and 24. The system presented in figure 24 consists of 4 jet vanes, which are able to control pitch, yaw and roll attitudes of the missile. This category of systems present some limitations by the
thermal and mechanical loads induced on the device by the nozzle flow.

Fig. 23 -

MECHANICAL DEVICE INTRODUCED fNTO THE NOZZLE FLOW

3-16

4 jet vanes

Fig. 24 -

JET VANES

5. - EXTENDIBLE EXIT CONE

The principle of an extendible exit cone is to increase the exit cone expansion ratio in a limited
external volume of the nozzle. This concept is well suited for limited length motors, such as submarine
based missile motors and allows on increase of the performance by 6 to 8 percent.

After motor ignition

Before motor ignition

Fig. 25 -

EXTENDIBLE EXIT CONE - PRINCIPLE

The experience of SEP in the field of extendible exit cone deployed after motor ignition is summarized in reference [8]. This demonstration program has enabled investigation of four concepts incorporating
carbon-carbon cones and petals (see figure 26).

\.---~~.
~/

FOLDED PETALS

~.

NESTED CONES

L..~i~--------:;7
"rY

HINGED PETALS AND CONE

~.

Jv

~.

Fig. 26 -

HINGED PETALS

lc-'EC CONCEPTS

3-17

6. - NOZZLE ANALYSIS
The nozzle analysis is usually conducted in 3 steps.
Aerodynamic analysis :
The goal of this analysis is to predict the performance and calculate the loads induced by the aerodynamic flow.
Thermal analysis :
This step consists to calculate the in-depth temperature distribution in the nozzle in order to verify
the structural integrity of the mechanical parts.
Thermomechanical analysis
This analysis is the last step of the nozzle analysis process. This step uses the results
during the two previous steps : the aerodynamic pressure and in depth temperature distribution.

obtained

The results obtained at each step and the recommandations of analysis people are taken into account by
the nozzle designer. Several iterations are often required between the design and the analysis in order to
obtain the optimum configuration of the nozzle.

6.1. Aerodynamic analysis

In a preliminary phase, the thermodynamic properties of exhaust products are evaluated by using a
slightly modified version of the Nasa Lewis thermochemical analysis. This analysis is based on a freeenergy calculation with the following assumptions :
one dimensional flow
non viscous and adiabatic flow
chemical equilibrium
velocity and thermal equilibrium of condensed and gas phases.
The inputs required are the following
elemental composition of the propellant
heat of formation of the propellant at the operating grain temperature
motor operating pressure (average)
expansion ratio of the exit cone.
This analysis leads to a maximum theoretical specific impulse (Isv th). However the effective
expansion achieved by the nozzle is different from this ideal case. The different performance losses are
the following :
Flow divergence
This effect depends on the exit cone geometry.
angle and the deflection angle.

The main influence parameters are the half mean

Two phase flow

In the case of aluminized propellant, the two phase flow effect may represent a loss ranging from
3 to 5 %.

Viscosity effect :
The friction on the boundary layer may induce a loss ranging from 1 to 1.5 %.
Heat loss
This loss is induced by the heat transfer at the nozzle wall.
Chemical non equilibrium :
This loss is due to the finite rate of chemical reactions, that do not allow the total energy of
these reactions to be released during the expansion process through the nozzle.
Nozzle erosion
This loss is mainly
diameter during the firing.

reflected by the expansion ratio decrease due to the' change of the throat

3-18
The method used at SEP for performance prediction is described with more details in reference
cumulating these losses in order to obtain the
effective nozzle efficiency. The results obtained by this method are in good accordance with experimental
results. Some typical values of specific impulse losses are given in table 3.
~-3]. It consists of evaluating separatly each effect, and

Isv th losses

Typical value %

Flow divergence

2 - 3 %

Viscosity loss

1 - 1.5 %

Heat transfer

0.5 %

----------------------------------------------------------------Two phase flow

3 - 4 %

-------------------------------------~---------------------------Chemical non equilibrium

0.5 %
----------------------------------------------------------------Throat erosion
0.5 %
TOTAL

9 - 10 %

Aa shown in table 3, the flow divergence and two phase flow effects are preponderant and represent more than 60 % of the total loss. They are currently evaluated in two steps. Initially a bidimensionnal monophasic program enables the calculation of flow divergence loss.
Afterwards,
a bidimensionnal
diphasic program gives by a comparison with the results obtained previously the two phase flow effects.
These programs consider only the trans - and supersonic stream of the nozzle and use stationnary methods
for instance the supersonic stream is analyzed by a method of characteristics.
A new computer program presented in reference L:9] is currently under development. This program
will be able to analyse directly a two dimensional two phase flow. This program is based upon a non steady
state approach solving the Euler equations with a Mac Cormack's explicite finite difference scheme. This
new program will be able to take into account the subsonic area of the nozzle and to improve the performance prediction. An example of results of a flow field analysis performed with this new program is presented
in figure 27. In the first case, the flow is supposed to be monophasic; as shown the results are very
different from those obtained by the second calculation assuming a two phase flow.

Flow
One phase

Flow
Two phase
Fig. 27 -

NOZZLE MACH NUMBER CONTOUR

3-19
6.2. Thermal analysis
The initial sizing of liner and insulator thicknesses is usually performed with material erosion
data and one dimensional thermal calculation. This design is afterwards analysed with more refined techniques using two dimensional computer programs taking into account the ablation.
6.2.1. Thermal analysis characteristics
The thermal analysis presents the following characteristics
the anisotropic and temperature dependant materials properties
the very high convective and radiative heat flux (> 10 MW/m 2 )
the analysis must predict and take into account the ablation by hot gases.
6.2.2. Boundary conditions
Convection : the convective heat flux is usually preponderant in the total heat exchange.
Radiation: the radiative heat flux is important during the first seconds of the
tion, for which the difference of temperature between the wall and the hot gases is maximum.

combus-

Mass transfer
the heat flux resulting of this effect may represent 10 % of the total
heat flux. The chemical reactions taken into account are the following
C + H20

->
<-

CO + H 2

C + CO 2

-->
<-

2CO

2C + H2

->
<-

C2H2

The boundary conditions are determined in the three following steps :

(1) Calculation of transport properties (specific heat, viscosity,
and mass diffusion coefficients) for use in boundary layer thermodynamic analysis.

thermal conductivity

(2) Calculation of chemical surface recession rate considering the chemical reaction
ween the material and the combustion products.

bet-

6.2.3. Conduction calculation

The transient heat conduction problem is solved by a two dimensional finite difference
scheme taking into account the surface recession and the thermochemical boundary conditions at the wall.
Figure 28 shows the typical analytical results for a nozzle at a given combustion time.
Temperatures

Ablation

Fig. 28 -

IN-DEPTH TEMPERATURE DISTRIBUTION EVALUATION

423

473

573

673

773

873

1073

1273

1573

10

2073

11

2573

12

3073

3-20
6.2.4. Developments underway
The following developments are underway
improvement of boundary layer analysis
in depth temperature calculation with a finite element scheme in order to perform the thermal and thermomechanical analysis in one step. This development will enable transient thermomechanical analysis taking
into account "the history of the load".

The structural analysis is performed at given combustion times.

the following characteristics :

~he

structural analy.i. presents

very severe thermal gradients

anisotropic and temperature dependant material. properties
non linearity of stress-strain curves
geometry non linearities (gaps, etc ).
6.3.1. Characteristics of materials
Nozzle materials are usually considered orthotropic. The structural behavior evaluation
requires for each direction, the stress-strain curves, the Poisson's effect, the shear moduli and expansion
ratios.
These values must be measured at different temperatures. The characterization must be performed
with samples fabricated with industrial conditions representative of the nozzle part process. An example of
stress-strain curves for a carbon-carbon is presented in figure 29.

3
Stress (MPa)

2
TRACTION
2600C
Strain (%)

COMPRESSION
2600 0 C

Fig. 29 -

EXAMPLE OF STRESS-STRAIN CURVES FOR A CARBON-CARBON

6.3.2. Calculation of stress and strain

The analysis is performed with a finite element computer program (code MARC see reference
The loads applied to the nozzle can be aerodynamic pressure, thermal gradients or external loads
such as actuation loads, vibration.

1:5 J).

Two to four combustion times are selected, for which the parameters are the most significative (thermal gradient, chamber pressure, temperature), The modeling must take into account the surface
recession due to ablation. The stress and strain calculation is performed with an incremental iterative
technique in order to consider the different non linearities (non linearity of stress-strain curves,
gaps
in the nozzle as.embly). An example of stress distribution in a carbon-carbon exit cone subjected to actuation loads is presented in figure 30.

3-21
Fig. 30 -

STRESS DISTRIBUTION IN A CARBON-CARBON EXIT CONE

14.R<J

Ill. 78

5.57
1:1.36

-4.85
-113.136

-15.27

-20.48
-25.69
-29.613

The nozzle structure factor of safety is determined with maximum stress or maximum strain
criteria.
These criteria consist of comparing for each direction the calculated stress or strain with the
maximum value obtained on samples.

6.3.3. Further development

The use of the same finite element modeling for thermal and structural analysis will enable the achievement of transient thermomechanical analyses with a better adaptation of the modeling to the
thermal gradients. An example of automatic remodeling between different combustion times is presented in
figure 31. These developments enable the taking into account of inelastic behavior of the materials, the
history of load" and the effects of dammage.

3-22

INITIAL MODELING

TIME

3s

TIME

lOs

THROAT ABLATION

Fig, 31 -

0,8 mm

THERMOMECHANICAL ANALYSIS WITH AUTOMATIC REMODELING

7. - CONCLUSION
A lot of improvements have been performed in the field of nozzle materials over the past ten years.
Most modern nozzles use one piece carbon-carbon integral throat and entrance. The current development of
new methods allowing the construction of large muldirectional reinforced preforms, will increase the applications of carbon-carbon exit cones. Ceramic-ceramic insulators are emerging. They will greatly simplify
the design of nozzles, thus leading to an improvement of both performance and reliability.
In parallele, the analysis methods have been enhanced. Thus the different computer programs, which have
been developed have .111owed great t!TIprovements of performance predLction and thermostructuraL behavior evaluation.

4A-l

COMPOSITE MOTOR CASE DESIGN

by
Dr. P. R. Evans
Department Manager, Structures & Analysis
Hercules Incorporated (Aerospace Products Group)
Allegany Ballistics Laboratory
P.O. Box 210
Rocket Center, WV 26726
USA

SUMMARY
Information is presented on the design of composi te cases for rocket motors.
Typical design requirements are discussed, including pressure, axial and bending loads
and thermal loading.
Material technology discussed includes filament winding resins
and glass, Kevlar and graphite fibers.
Design features discussed include the pressure
vessel body, skirts, polar adapters and other attachments.
Typical qualification
testing and inspection techniques are presented.

OBJECTIVE
The objective of this lecture is to provide an overview of the design and verification of filament wound motor cases for use in solid rocket motors.
BACKGROUND
The motor case for a solid propulsion rocket motor serves:
(a) to protect and
"store" the propellant grain until the motor is used, (b) as a combustion chamber for
high pressure, high temperature burning of the grain during motor operation, (c) to
mechanically/structurally interface with other motor components like the nozzle,
igniter, internal insulation, handling/carrying brackets, etc., and (d) usually as a
primary airframe during missile flight.
Motor cases may be very large like the Space
Shuttle Solid Rocket motor or quite small man-operated tactical units.
The design
requirements dictated by the missile I s size and mission profile encompass wide ranges
of mechanical and thermal loadings which can occur during the manufacture, stot"age,
operational deployment, and flight history of a given missile.
Since the motor case is an inert or non-energy-contributing missile component, the
design objective is to make the case as lightweight as possible, within the bounds of
technology and cost.
This will result in a higher motor mass fraction (propellant
weight/total motor weight) and high motor and missile "performance" (range, delta
velocity, etc.).
Case designs utilizing structural composites, - laminates of high
strength fibrous materials like fiberglass, Kevlar and graphite (carbon) with matrix
materials like epoxies, typically result in significantly lower weights than metallic
designs.
The nature and behavior of composite materials dictate distinctly di fferent
design and quality verification methodologies than would be required for metal case
designs.
The "typical" solid rocket motor case is basically a double-domed right circular
cylinder with openings In both domes and cylindrical extensions called "skirts" (Figure
1). The aft opening interfaces with the nozzle.
The forward opening accommodates the
igniter and safe/arm.
The forward opening may be absent altogether if an aft igniter
is utilized.
The "skirts" are integral extensions of the case body which serve as
structurally-capable attachment interfaces for other missile stages or hardware
packages.
The motor case interior is covered with an insulator which provides thermal
protection and flow erosion resistance in those regions where the propellant burns to
the wall before the entire grain is consumed.
A case bond system chemically attaches
the grain to the case wall interior and insulator.
Additionally, numerous conduits,
ducts, and attach/carry features may be required on the case.
A particular design
challenge present on most, large intercontinental-capable motors is thrust terminaton
ports--specific regions of the case pressure vessel which must be penetrated or
destructed in a controlled manner on command, to depressurize and extinguish the
propellant grain.
The solid rocket motor case is essentially a minimum-weight pt"essure vessel, one
whose design is complicated by the presence of significant levels of thrust and bending
as well as locally concentrated loads ft"om the skit"ts and any attachments. Solid rocket
motors typically opera'te from 500 to 3500 psi.
Appreciable thermal loads are also
present, since the propellant grain combusts at 5-6000F and aeroheating can result in
exterior surface temperatures in excess of l200F.
CASE DESIGN APPROACH
Mechanical and thermal lOads and geometry constraints are usually supplied to the
designer.
Factors of safety or load factot"s typically vary somewhat from customer-tocustomer (Navy, Army, AF, NASA, etc.) and fot" a specific missile.
Intet"nal pt"essure is
often the dominant load.
A typical factot" of safety requires the minimum vessel but"st

4A-2

stl;'ength to be 1.25 times greater than maximum expected operating pressure, MEOP, ("hot
firing", maximum pressure-increasing tolerances, manufacturing variability).
For some
air-launched applications or man-portable motors, this factor may be increased to 1,.4
or 1.5 for the flight period while the motor is within a specified distance of the
launch point.
other mechanical load factors like bending and thrust may have values
higher or lower than that selected for pressure, depending upon the precision withwhich their limit magnitudes can be predicted and their criticality.
Thermal loads
require particular attention since the mechanical properties of composite degrade
appreciable at temperatures well below those affecting metals.
Such loads may be
accommodated by imposing maximum (or minimum) temperature limits on the case structural
region and requiring insulation thicknesses adequate to insure that the temperature
limits are never exceeded.
Even so, the mechanical strengths and stiffnesses of the
stt"uctural materials must be known across the allowable temperature range for all of
the other envi ronmental bounds imposed and for the design Ii fe of the motor.
Since
composi te materials are sensi ti ve to temperature, moisture and other "environmental"
exposures and since composites "age"--particularly the resin matrix materials--the
subject of material properties characterization is pivotal to effecient and effective
composite case design.
Consider the "typical" motor case (Figure 2).
Since pressure loads usually
dominate the design we will, for preliminary design sizing, treat the case as a "thin
shell, a membrane"1 i.e., one in which the stresses do not vary through the thicKness.
At any point on the vessel, the in-plane membrane loads are a combination of:
(a) those
due to internal pressure (determinable if the local radii of curvatures of the membr.ane
are known), and (b) those due to thrust and bending of the shell body (determinable if
the shell thickness is known).
For example, in the simplified case shown in Figure 2,
in the cylindrical section the hoop and axial membrane loads are given by Eq.'s (1) and
(2) :

PR, lb/in.

(1 )

Hoop

Line Load

NH (lb/in.)

NHP

(2)

Axial Line Load

NA (lb/in. )

NAP + NAM - NAT'

M
PR
2 + 1T R2

where:

motor design pressure

motor case radius, in.

T
lb/in.
21TR'

(FS) (MEOP), psi

case body bending moment

motor design thrust (FS)(Limit Thrust), Ib

(FS) (Limit Bending), in.-lb

If the motor case were of metal, the wall thickness could be quickly determined by Eq.
(3) :

(3 )

t
N

max.
crallow.
.

m1n.

max.
crallow.

case wall design thickness

N or N
H
A
allowable metal design stress

since metals are homogeneous and isotropic1 i.e.,

same at all points and uniform in all directions.

their properties are essentially the

Composites are highly inhomogeneous and anisotropic. Their properties vary considerably with location and direction, variations which may often be used to the designer's
advantage.
COMPOSITE MATERIALS AND CONSTRUCTION
Composites are "laminates" of unidit"ectional "lamina" (Figure 3). The lamina, the
basic building block, is a ply of continuous fibers lying side-by-side, parallel, and
essentially one fiber thick. The fiber is bound together by a resin matrix. The resin
matrix performs the critical functions of holding the fibers in place and transferring
stresses from fiber-to-fiber through shear. The structural behavior of this "composite"
of fiber and resin is dominated by the fiber.
Pt"operties are highly directional, as compared with metals (Tables I and II).
Unlike metals, composites are non-yielding and quite unforgiving of stress risers or
local overloads.
The composite lamina is very strong and stiff in the direction of the fiber but is
comparatively quite weak in other directions where only the resin matrix is available
to resist loads.
Accordingly, the designer must be sure to place adequate amounts of
fiber in all load directions but has the distinct advantage of having to place at a
given poi~only the amounts and in the directions required by the load state at that
point. That is, the composite structure may be optimally designed.

4A-3

A "laminate" consists of stacking plies of lamina to achieve a desired strength or

sti.ffness.
The resulting directional strength and stiffnEj!ss is very dependent on the
stacku~ (Figure 4).
Tne composite motor case designer/analyst requires an extensive material property
data base, one which recognizes that the strengths and stiffnesses of a given composite
can vary substantially with temperature, moisture content, and time.
The theoretical strength of a given fiber--as measured by a strand or "tow" test-is seldom realized in a pressure vessel (primarily due to less than perfect fiber
collimation).
The "delivered strength" often varies significantly with changes in
winding resin.
Since the "deliverable fiber strength" must be known before even a
preliminary design can be developed, various laboratory specimens have been utilized to
provide design data. Resin coated strands, unidirectional "dogbones" cut from layed-up
plates, filament wound rings and actual pressure vessels have been employed. Only the
latter have provided reasonably satisfactory results and even then significant size
scale effects occur and verification typically requires the bursting of full scale
cases.
For detailed analysis purposes,
the constitute stiffnesses Ell' E22'
G12,
and 1112 are requirep to be known for each lamina material at each temperature which
must be evaluated. Corresponding strengths are required to allow calculation of margins
of safety. These properties may be generated by means of fairly large numbers of small
scale laboratory specimens like" dogbones and plates. Since most such specimens are not
really "filament windable" , press cured plates are often used, resulting in some
parameters like compressive and shear stresses and strengths of the specimen differing
somewhat from those in the wound vessel.
The complexity of determining allowable strengths, in tension, compression and
shear, and encompassing all environmental exposure conditions, results in major emphasis
being placed on:
(a) in-process control during case manufacture to assure unit-to--unit
variability is minimized, and (b) proof and ultimate structural testing of full scale
cases to verify design margin.
FILAMENT WOUND MOTOR CASES
The basic approach used for fabricating composite motor cases is "filament winding" (Figure 5), placing alternating layers of "helix" or "polar" wraps with "90" or
"hoop" wraps onto a mandrel shaped to form the desired case interior profile. The helix
or polar wraps carryall of the axial membrane load and, to an extent depending on the
"winding" or "wrap" angle, CI, a portion of the hoop membrane load. That portion of the
hoop load beyond the hoop-capacity of the helix wraps is then carried by the hoop wraps.
The helix layers encompass the entire vessel surface--cylinder and ends--like a cocoon,
while the hoops cover only the cylindrical section. Once all helices and hoops are in
place, the matrix resin is cured, either thermally of by microwave, and the mandrel
removed, leaving a rigid lightweight vessel.
The fiber is placed or "wound" as a continuous "band" consisting of several "tows"
or "rovings" of fiber placed side-by-side--<me tow coming from each spool of fiber as
packaged by the fiber manufacturer.
The bandwidth and "band density" (thickness) is a
design variable which may be selected to result in thick or thin layers and wide or
narrow bands.
The fiber may be "wet wound": Le., passed as a dry fiber through a
resin bath while the winding operation is occurring.
An alternate approach is to use
"prepregged" tow--fiber which has already been saturated (impregnated) with matrix
material (resin) and the resin partially cured (B-staged) to a handleable condition but
being under cured to the extent that it will reflow during elevated temperaeure cure.
This latter characteristic allows for uniform fiber tensioning (collimation) due to
mandrel expansion before resin set.
Domes and the Helix Angle - The partially or totally closed ends of filament wound
motor cases are called "domes" and are very efficient structural closures. Proper dome
design is a major design feature of filament wound cases and is the determining factor
in defining what the helix winding angle must be for the cylindrical section. The polar
opening diameter in relation to the cylinder diameter sets the angle which the helix
band must have as it crosses the "dome tangent" the junction of cylinder and dome
(Figure 6).
The larger the polar opening and the wider the bandwidth in relation to
the cylinder diameter, the higher the helix angle at the tangent must be. Large polar
openings (greater than 60-70% of the vessel diameter) are severely stressed and structurally untenable.
Crossing the tangent at any other angle than the theoretically correct one defined
by Eq. (4):
(4)
CI

= arc

sin

will result in the fiber band wanting to slip sideways since the theoretically correct
angle is also the "geodesic" or friction-free path angle. Further, a unique dome shape
(Figure 7) may be defined which is "geodesic" along its entire path and also t"esults in
an "isotensoid" condition--constant tension stress in all fibers at all points when the
vessel is pt"essure loaded. This unique dome shape and specific winding angle cannot be

4A-4

substantially deviated from if a high strength structurally well-behaved composite dome

is to result.
The isotensoid condition drives most or all of the membrane load into
the strong and stiff fiber and away from the much weaker resin.
The properly designed
qome utilizes the variations in local hoop and meridional radii of curvature to offset
fiber' thickness build up and angle change (low to high--90 at the pole) to maintain
constant fiber stress.
This is attained by increasing the meridional radius of curvature to offset the propensity for decreased stress with decreasing diameter and
increasing fiber thickness in moving from the tangent to the pole.
Since the typical case has a different size polar opening at the two ends and a
constant body diameter, a helix angle perfect for one dome cannot be perfect for the
other unless it changes along the body--a condition which can be utilized to only a
qui te limited extent unless wasteful bui lding of helix thickness at one end can be
tolerated.
In practice, within limits, the helix angle crossing the tangent can be one
which is "deviated"; Le., different from the theoretical, by a few degrees. The result
will be a variation in stress in each individual filament as it traces its path over the
dome, the variation in fiber or filament stress being accommodated by shear stresses in
the resin.
Therefore, both domes are typically "deviated" by small amounts.
If the
cylinder is long rela~ive to the diameter, a gradual change in helix angle from one end
to the other may also be utilized.
To the designer, this means that he is not free to
choose one polar opening size independent of the other.
The detailed design and analysis of practical domes--particularly as complicated
by the presence of skirts causing local large stiffness gradients and bending stresses-constitute much of the design challenge for composite cases.
Skirts - Skirts, cylindrical extensions of the case body, typically must carry
apprecra:bI"e mechanical loads.
They are basically rigid tUbes or rings which must be
attached to the relatively flexible case body "membrane" without creating unacceptable
stress risers due to locally increased stiffness and in-plane bending (Figure 8).
Their behavior, even with the aid of sophisticated finite element design/analysis codes,
remains quite difficult to predict.
Empiricism in design and full scale testing are
still much used for this feature.
DESIGNING THE CASE WALL
When the nominal cylindrical winding angle is established, the thicknesses of hoop
and helix composi te necessary to carry the body membrane loads may be determined by
"netting design"; i.e., by assuming that all of the membrane load is
carried by the
fibers alone (E22, G12' ~12 = 0). Any required reinforcements are similarly preliminarily designed for regions where concentrated loads or local discontinuities occur
(skirts, attach lugs, T.T. ports, pole piece regions, etc.).
Once the preliminary
design is established, iterative design/analysis utilizing powerful finite element
computational codes and extensive materials models and data bases are used to refine the
design.
The analysis codes, materials models and data bases are typically proprietary
to the designing company.
For a vessel composed of helix layers at a winding angle of a and hoop layers at
90, the required thicknesses of hoop fiber t '90 and helix fiber t 'a of strength
0'90 and o'a may be predicted by Eq. 's (5) and (6):
(5 )

t' a

0' a

(6 )

t'90
where:

t'

cos a

total equivalent helix fiber thickness, in.

t'90

total equivalent hoop fiber thickness, in.

0'

helix fiber tensile strength, psi

0'90

hoop fiber tensile strength, psi

meridional (axial) line load, lb/in.
hoop line load, Ib/in.

When the total required fiber thicknesses are ascertained, the designer determines
the number of hoop and helix composite layers by selecting layer thicknesses. Note that
the values t' are for "fiber only" and do not include resin.
Total composite thickness
will depend on the final in-situ resin content which is dependent on processing variables like winding tension, diameter, and winding angle.
A typical value is 40% resin
by volume.
A DESIGN EXAMPLE
Determine the required hoop and helix thicknesses for a 12-in. diameter cylindrical
composite vessel, subjected to pressure only, to achieve a burst pressure of 4000 psi.

4A-5

Assume the helix winding angle is 25 and a 60% fiber volume (Vf) is achieved.
The
fiber is graphite with a hoop fiber strength of 600 ksi and a helix fiber strength of
500 ksi.
(8),

For pressure only,

(9), and (10):

the axial and hoop membrane loads are as shown by Eq.' s

PR

(7 )

axial membrane load

2
(8)

hoop membrane load

PR

(9)

t'

(7),

PR
el

a
'el

cos

2a' el cos

el

el

2
(4000 lb/in. ) (6 in.)

(10)

0.029 in.

PR - PR (tan

2
el)

2'

a'90

PR (1-0.5 tan
a'
90

2
el)

2
2
(4000 lb/in. )(6 in.)(1-0.5 tan 25)
2
(600,000 lb/in. )

0.036 in.
For a 60% fiber volume, the corresponding composite hoop and helix thicknesses are
shown by Eq. 's (11) and (12):
(11)

tel

0.029/0.6

0.049

(12)

t90

0.036/0.6

0.060
0.109 in.

The density of the composite, Pc' is given by Eq. (13):

0.056 lb/in.

and the weight per inch of cylinder is given by Eq. (14):

(14)

~Dtp

(~)(12)(0.109)(0.056)

= 0.23

lb/in.

A comparable high strength steel case of, say, D6AC steel at an ultimate strength
of 225 ksi would require a wall thickness as given by Eq. (15):
(15)

PR/a

= 4000

x 6/225,000

= 0.107

in.

which would result in a weight per inch of cylinder as given by Eq. (16):

(16)

(~)(12)(0.283)(0.107)

= 1.14

lb/in.

or five times as much.

COMPOSITE MATERIALS TECHNOLOGY
A wide range of fiber and matrix materials are available.
Fibers include glass,
Kevlar and graphite/carbon (Table III). An epoxy resin is typically used as the matrix
material but emerging requirements for structural performance at higher temperatures is
creating interest in more exotic resins like the polyimides for some uses.
Fiberglass was used for most earlier applications (Polaris, c3 Poseidon). It is
relatively high density but inexpensive.
Kevlar, much lower in, density and higher in
specific strength, is presently in wide use (Trident C4, Pershing II, Peacekeeper). The
graphite/carbon fibers have improved dramatically in recent years and, while slightly
higher in density than Kevlar, their substantially higher delivered fiber strength in
pressure vessels and consequently higher speci fic strength makes them the frequent
materials of choice today.
Conventional bisphenol-A based epoxies continue to dominate matrix te,chnology.
They are inexpensive and easily processed either as a prepreg or a wet wind.
These
resins do soften at temperatures beyond 300-350F. Vessel pressure carrying capability
is reasonably well retained even somewhat beyond these temperatures. However, composite
compression and shear strengths essentially disappear at or somewhat beyond the "Heat
Distortion Temperature", the temperature at 'tlhich a cured resin experiences a sharp

4A-6

drop-off of mechanical propertie~.

External and internal insulators are
maintain case wall temperatures low enough to assure structural performance.

used

to

Polymeric materials exist which possess superior. high temperature capabilities

(Table IV).
These materials are substantially more difficult to process, typically
requiring the use of an autoclave, and attempts to use them for wound pressure vessels
have resulted i t'l reduced burst strength wi th a given fiber as compared wi th that
attained with a conventional epoxy, so they cannot be considered state-of-the-art today.
QUALITY CONTROL AND VERIFICATION
The hydroproof test is widely utilized as the final determinant of the acceptability of a given motor case.
Of course, during manufacture, material and component
quality and consistency of manufacturing processes ~re rigidly controlled.
Nondestructive testing techniques like X-ray, and to a lesser extent ultrasonics and
acoustic emission, are used to establish unit-to-unit "fingerprints" but the ultimate
determinant of a "good" case is the passing of hydroproof.
Strain gages are widely utilized at key stress points for routine proof testing.
The strains are compared with the body of data from prior units for any significant
deviation.
Hydroproof testing can and often does include the application of limit levels of
other load conditions like thrust and bending.
Once a case design is established by
initial structural tests, typically several more cases are destructively ultimate
tested to establish variability and design margins.
THE CHALLENGE, THE PAY-OFF AND THE FUTURE
The rocket motor designer is always being challenged to minimize the weight of the
inert components.
The case designer is confronted with a wide array of approaches and
choices.
Metals are rugged and durable, have long case histories and extensive data
bases, are relatively simple to design with, and possess the ability to accommodate
local overloads and stress risers through yielding--but metal cases are heavy.
Structural composites provide much lighter cases, but the designer is obligated to understand
and design to accommodate the non-homogenity, anisotropy and non-yielding stress-strain
behavior of the materials.
They are also more temperature and environmental sensitive
than metals. The pay-off is large if the designer understands the material behavior and
is willing to master the design skills.
composite cases have been state-of-the-art in the U.S. for strategic solid rockets
for 30 years--due to their superior performance. They are now beginning to be used for
tactical rocket cas~s because material and fabrication costs are becoming competitive,
the material and design advances allow accommodation of the substantially more severe
use environment, and the performance requirements of emerging systems require the
improved performance.

081GN

4A-7
TABLE I
METAL MATERIAL PROPERTIES
Constituitive Properties
Materi'al

Specific
Strength,
o/p

Tensile
Strength,
0 (ksi)

Density,p
(lb/in.3)

E (msi)

G (msi)

AISI 4130 Steel

29.0

U.O

0.32

0.283

180

6
0.64 x 10-

D6AC Steel

29.0

11.0

0.32

0.283

220

0.78 x 10

7075-T6 Aluminum

10.3

3.9

0.33

0.101

80

0.79 x 10

64 Titanium

16.0

6.2

0.31

0.160

160

\.I

-6
-6

1.00 x 10- 6

TABLE II
FIBER, MATRIX ANDCOMPOSITE MATERIAL PROPERTIES
Constituitive

Strength
0

22
msi

\.1

10.5

10.5

4.2

0.25

280.0

12.6

12.6

5.2

0.22

450.0

Kev1ar-49

18.0

0.6

0.4

0.30

370.0

Graphite, IM

42.0

2.2

3.5

0.20

560.0

0.4

0.4

0.2

0.38

E-G1ass

6.7

1.8

0.8

~ S2 Glass
H

8.8

1.9

0.9

Material

r:l

msi

E-Glass
~
S2
Glass
r<l

Ii<

pro~erties

G
12
msi

12

T
ksi

U/ p
T

22

T
ksi

p
1b/in. 3

12
ksi

10 6 in.

--

0.090

3.1

--

0.092

4.9

---

--

0.052

7.1

--

0.065

8.6

12.0

12.0

7.5

0.043

0.3

0.25

150.0

6.0

6.0

0.070

2.1

0.30

250.0

9.9

8.8

0.072

3.5

---

~ Epoxy

~ Kev1ar-49
H Graphite, IM

11.8

0.8

0.3

0.34

190.0

2.9

5.4

25.2

1.3

0.6

0.30

340.0

9.6

10.2

0.050

3.8

0.56

6.1

TABLE III
REPRESENTATIVE FIBER AND COMPOSITE PROPERTIES
Tow Properties
Material

E'

olp
6
10 in.

$/lb

Pressure Vessel
Hoop
Eaxial
Fiber
Strength PR/t
msi

Unidirectional
Laminate
0

11 c
ksi

12
ksi

E-Glass

300

10.5

0.090

0.75

3.33

280

105

3.2

85.0

12.0

S2 Glass

550

12.6

0.092

5.00

5.98

450

170

3.8

100.0

Kevlar-49

545

18.0

0.052

18.00

10.48

370

140

4.1

Graphite, AS

580

34.0

0.065

21.00

8.92

480

180

8.0

40.0
200.0

13.5
6.0
17.5

Graphite, IM

700

42.0

0.065

31.00

10.77

560

210

9.3

200.0

17.5

TABLE IV
MATRIX RESINS

Material

Designation

Heat
Deflec.
Temp.
OF

Density
3
Ib/in.

-65F

77F

250F

450F

Composite
Compressive Strength, ksi
600F

Uti l i za t iO:1

Bispheno1-A
Epoxy

Hercules
HBRF-55A

260

0.044

200

175

125

--

--

Trident

Epoxy
Nova1ac

Hercules
HARF-7A

400

0.041

155

130

105

--

--

Development
11otors

Po1yimide

PMR-15

680

0.047

75

62

59

50

40

Experimental

4A-8

AFT SKIRT

FIGURE 1. The Typical Rocket Motor Case Is A Double-Domed Cylindrical Pressure Vessel

t t t t

METAL AFT POLE

METAL FWD POLE

IZ

w
:E

:E

PRESSURE

METAL SKIRT
RINGS
AXIAL LOAD - - . .

FIGURE 2.

The Motor Case May Be Treated As A Membrane For Preliminary Design

4A-9

FIGURE 3.

2L

Laminate Composite Construction

Laminate Properties
1

Direction

IIII
~

Ut!

Fiber
Orientation

Ell

E 22

G12

0'11

msi

msi

msi

ksi

25.2

1.3

0.6

340.0

90

1.3

25.2

0.6

9.6

+ 45

2.2

2.2

6.5

170.0

13.3

13.3

0.6

170.0

0, 90

FIGURE 4.

Laminate Properties Depend On Stack"'Up

4A-1O

TOW (ROVING) SPOOL

BAND
HELIX

THE CASE WALL LAMINATE BUILD-UP

FIGURE 6.

Composite Cases Are Wound Of Alternating Layers Of Hoop & Helix Material

1/2 B.W.

POLAR
RADIUS

Rp

FIGURE 6.

The Relative Diameters Of Polar Opening And Case Body Determine

The Helix Winding Angle

4A-ll

400 DOME

FIGURE 7.

The Isotensoid Dome Shape Depends On The Body Helix Angle

AXIAL FIBER CONTINUiTY

AXIAL LOAD
(THRUST & BENDING)

'lIII!I

~...

AXIAL LOAD
(PRESSURE)

.--------+---t-----"\~

SKIRT

CASE BODY (HOOP & HELIX)

AXIAL LOAD (PRESSURE)

FIGURE 8.

Skirts Are Attached By Bond Joints And Fiber Overwraps

5-1

DESIGN OF FILAMENT-WOUND ROCKET CASES

by
J.P. DENOST
Head of Programs for Filament-Wound Structures
AEROSPATIALE
St-Hedard-en-Jalles
33165
FRAl'<CE.

ABSTRACT
Our Company began developing methods for the design and sizing of filament-wound rocket cases in the
early 1970's. Such methods have since evolved to the extent that rapid design and precise verification
means are now available.
The purpose of this document is to outline these two approaches, to criticize them and to evidence
their complementarity. Detailed presentations will be given to enable the reader to grasp the principal rules, and to gain such information as will enable him to develop them in specific fields.
This predominantly theoretical part will be completed by considerations relating to materials, their
implementation and performances, and by a description of the major principles of fabrication. of this type
of structure.
NOTATIONS
A

area of a straight cross-section of the filament.

-+
b

-+

-+

-+

-+

...

tAn.

bs

tAns'

curvature of (f) in M ; c;'

Cl, C2

main centers of curvature of (S) in M.

O.

D
d

symbol of derivation.

thickness of the vessel in M.

-+

action of the mandrel on the filament, by filament unit length.

terrestrial constant of gravitation.

orthogonal projection of M on the revolution axis of (S).

k
L

axial distance between the two polar openings of the vessel.

running points of (f) and of (S).

mass of the vessel.

-+

unit vector of the main positive semi-norm of (f) in M.

number of filaments cut per unit length of a parallel.

-+

ns

unit vector of the internal semi-norm at (S) in M.

value in M of the internal membrane stress in the direction of

parallel
for tensile stress : N;' O.

...

t~

, by unit length of the

...

value in M of the internal membrane stress in the direction of te , by unit length of the
meridian
for tensile stress: Ne;' O.
-+ ......

point of origin of the datum mark (x,y,z)

and the cylindrical part.

value of internal pressure ; p;'

vessel performance factor ,. q

situated in the plane of junction between the head

O.

~
mg

principal radii of curvature of (S) in M

...

ns

(i

1, 2).

curvilinear abscissa of M on cr) oriented.

mean surface of the vessel ; revolution surface where the concavity of the meridian is turned
towards the axis of revolution.

5-2
+

unit vector of the positive semi-tangent of (r) in M

value of tension in the filament; T;.

+
t

+
z

O.

unit vector of the positive semi-tangent of the meridian in M

t~

O.

;.

t~

;. 0
+

unit vector of the positive semi-tangent of the parallel in M oriented by z

p '.

J-

internal volume of the vessel.

x,y,z

coordinates of M in relation to the general datum mark.

p 0

+ + +

direct orthonormal base of the general datum mark ; z : carried by the axis of revolution of

(x,y,z)

(S)

measurement in radians of the angle of the pair of vectors

O<;a:<;

(t~

+
t)

if

(t~

At). n s ;. 0

1t

curve represented by the filament.

measurement in radians of the angle of the pair of vectors (x,HM)

e' being expressed by the relation (36).

+ +

>.

F
F

Il

radii of the polar openings of the two heads.

cJ f

stress in the filament; cr f

=--!.A

titer by volume of fiber in the

compos~te.

measurement in radians of the angle of the pair of vectors (n s

'!!

<;

+ +

if (n s A z)te ;. 0

measurement in radians of the angle of the pair of vectors (z

0<;~<;1t.

0<;

'dz ,

bs '

P Fl ' P F2

ns '

HM.

M
Mo

t.

1t

+
n)

+
+ +
i f (n s A n)t;'

Particular notations :

( )
o
L

symbol of derivation in relation to the variable z.

index affecting all parameters evaluated for z

= O.

INTRODUCTION

Filament-winding technology has been in use now for approximately 25 years in the fabrication of the
rocket cases of the various French programs.
Such programs have been civilian (DIAMANT) and military (MSBS, SSBS).
This technology has essentially made it possible to replace metal by composites, thus reducing the
inert weight to the benefit of the range of the carrier. Furthermore, filament winding is an automated fabrication method providing simultaneously rapidity, economy and quality.
Resin impregnated fibers have evolved over the years. The initial structures were made with glass
fiber. Subsequently, Kevlar appeared, and more recently carbon. Each development has served to further
reinforce the advantage of composites over metal .
While the production facilities installed evolved rapidly to meet the requirements mainly of military
program, design facilities have been perfected more progressively.
In the early 1960's, the definition of a structure was obtained by predominantly empirical methods.
Experimentation made it possible to obtain a product meeting requirements. Some years were necessary for
designers to exert their skills in this field, starting by the establishment of relatively simple methods
in the latter part of the 1960's.
The development of data-processing in France, and later the appearance of calculation programs gradually lea to the sophisticated working methods of the mid-1970's. Finally, today, stress is laid on computerized design (CAD) and more recently on the connection possible between CAD and CAM. Nevertheless, so-called
"light-weight" methods, geared to providing rapid answers to questions which do not immediately require a
high degree of precision and above all to enabling operators not to lose the understanding of the physical
phenomena, are still being developed and used.

5-3
Concomitantly, with calculations requiring an ever-increasing number of characteristics, characterization programs were established by designers for filament-wound composites, giving rise to major research
efforts providing a constant level of activity for Materials and Trials Laboratories.
The main purpose of this document is to present the methods of design and justification of filamentwound rocket cases. A critical analysis of the so-called "light-weight" methods and of computer facilities
will serve to demonstrate how our Company uses them complementarily.
This will take up the major part, but short chapters clearly had to be included to present on the one
hand how problems are posed, generally; and on the other hand to present methods of fabrication and experimental certification.
.
Finally, reverting to calculations, a few basic mathematical notions must be developed, in as condensed
a form as possible, to enable the reader to understand, without wearying him, the aim being to provide the
few basic notions necessary to enable anyone who so wishes to develop models geared to his specific applications.
2. STATEMENT OF THE PROBLEM
2.1 DESCRIPTION OF THE CASE

An illustration of the main sub-assemblies of the case is necessary for a better understanding of what
is to follow. This is given in Figure 1 below.

Front skirt

'Cylinder

Front head

Front polar~_..,'T7" J
mounting

Aft skirt

Aft head

Aft polar
#'#:+--mounting
Rubber connections.

Figure 1
This illustration shows details of :
- the two. skirts, which are two short cylinders, connecting the shell of the case with the rest of the carrier,
- the cylinder and the two heads, both filament-wound, the principles of design and of justification of
which will be presented below,
- the two metal polar mountings, which reinforce the openings and provide connection with other components,
such ~s the igniter plug forward and the nozzle aft,
- rubber connections between the vessel and the skirts on the one hand and between the vessel and the polar
mountings on the other hand.
The vessel (the cylindrical part and the heads) are thus filament-wound. In this phase of fabrication,
two types of winding are employed :
- firstly, so-called "satellite" winding, to make up the two heads and the cylinder, by a succession of
loops tangential to the two openings,
- secondly, so-called "circumferential" winding, on the cylindrical part only in such a way that the filaments are laid practically perpendicular to the generatrices.
2.2 MAIN SPECIFICATIONS

The technical specifications of such a product are given in a detailed and voluminous document from
which we have drawn the main elements.
2.2.1 RESISTANCE TO MECHANICAL STRESS

Mechanical stresses come from the internal pressure exerted by combustion of the powder upon operation
of the stage involved, and from the tensile and flexural stresses born mainly by the skirts during the carrier's lifetime.
Moreover, thermal stresses have to be taken into account during the propulsion phases.
Certification regulations associated with these stresses are such that the structure, as new, must have
a safety coefficient of : K;. 1.4.
Furthermore, the various sub-assemblies must remain in the elastic field for loads equal to 1.15 times
the nominal. loads. This latter specification is verified experimentally for each specimen prior to delivery
to the end-customer.

j-4

2.2.2 AGING
Each specimen manufactured must be capable of fulfilling all its functions with a probability over a
minimum value throughout its operational lifetime.
2.2.3 GEOMETRY
Geometric specifications are laid down in order to :
meet the maximum dimensions set botrr for when at rest and in operation,
- ensure interface conditions with the rest of the carrier.
2.2.4 FILLING WITH POUDER
Structural integrity must be maintained throughout the filling phase. Briefly, this involves both temperature and internal pressure stresses over a fixed period of time.
2.3 RESISTANCE TO INTERNAL OPERATING PRESSURE
In this document, we shall deal with the design and justification of such a structure when submitted to
what is in all cases the greatest stress : that of internal pressure due to the combustion of the powder
during the operation of the stage involved.
Moreover, whilst numerous components are subjected to stress from this internal pressure, the most critical sub-assembly for optimisation is the cylindrical part and the two heads, which is why particular attention will be given in the following paragraphs to detailing their design and justification.
Problems relating to the skirts will not be dealt with, given their low weight and given the conventional notions familiar to any designer. On the other hand, the design of the aft polar mounting will be briefly presented as its mass occasionally makes optimisation necessary.
2.4 STATEMENT OF THE PROBLEM SPECIFIC TO FILAMENT WINDING
The filament-wound cylindrical part is designed very simply, by determining the respective thicknesses
of the two windings involved required to ensure the proper resistance to internal pressure. The calculation
method will be presented in a specific paragraph.
The problem of the design of the two heads is quite different. The shape of each head, and the trajectory of the filaments on the head have to be determined simultaneously to provide for their optimized
fabrication.
The figure below illustrates the notations given at the beginning of this document, and how the problem
is posed.

,-Cylindrica I part

Meridian

, - - - - - - Filament (r)
Mean head
, - - - - surface area (5)

y
Figure 2
The shape of the vessel is defined by the function p (z). The lay-up of the filament during winding,
i.e. the line (I'), is defined by the function a (z).
The problem thus includes two unknown functions. Two conditions involving these two functions will therefore be necessary. The following chapters will reveal how this has led to the development of different
design methods.
3. DESIGN
A separation will be made in design methods between
- the cylindrical filament-wound part,
- the filament-wound part on the heads.
We shall take here the term "net method" to cover the theoretical notions to be developed. With this
method, we shall consider that ~t any given point the filament-wound structure has stiffness and resistance
only in the direction of the filaments, to the exclusion of any other direction.

5-5
3.1 ANALYSIS OF THE CYLINDRICAL PART

Given the notations, we can write

~

(1)

=.N<j>s + N<j>c ,

(2)

Ne=~s+~c

With

Furthermore, the basic hypothesis of the net method leads to

(3)

N<j>c = 0 .
The resultant of the stress

c is easily expressed by
(4)

a fc

Ne c = X e c

For satellite winding, the figure shown below must be taken into account

tga

_.

Figure 3
In view of the notatlons, the following can be expressed as shown in the figure
N tg a

(J

fs cos a

N tg a

(J

fs

(5)

Nes

sin a

(6)

!1oreover, it is possible to express the thickness of composite of the satellite winding by

es

=N X

(7)

cos a

Relations (5), (6) and (7) lead to

2

s = X es

(J

fs cos

(8)

Nes = X e s

(J

fs sin 2 a

(9)

Relations (1) and (2), by using the expressions of

~s' ~s'

PR
-2- =

X es

(J.

fs cos

N<j>c, Nec , lead to

= X e s a fs sin2 a + X e c

PR

po),

(R

(J

fc .'

These two relations make it possible to obtain the expression of stresses in the filaments of the two
windings in the cylindrical part :
PR
2 X e s cos 2 a

(J

fs

(J

~ (1 fc =
X ec

(10)

tg 2 a
2

(11)

The latter relations are widely used in designing the cylindrical part of the rocket tanks. They are
simple and enable a rapid definition 6f the necessary thickness of composite material to obtain a level of
stress imposed in the fibers.

5-6

a - If circumferential winding is eliminated, it can easily be deduced that the winding angle is to be
such that :
tg 2 a = 2
It will be demonstrated later that the value of the winding angle a in the cylindrical part imposes
the dimensions of the openings in the heads. To maintain geometric freedom on this point, it is therefore
always necessary to perform a circumferential winding.
b - If the filaments belonging to both types of winding are subjected on the same level of stress, from
(10) and (11) it can easily be deduced that :

~
e

3 cos 2 a - I .

3.2 ANALYSIS OF THE HEADS

3.2.1 OUTLINE OF PRINCIPLES
The problem of defining thickness in the cylindrical part has been studied in the previous paragraph.
The shape of the heads and the lay-up line of the filaments must now be defined. Paragraph 2.4 evidenced
that two conditions are necessary to address the problem.
Three approaches will be detailed. Their presentation will follow in the same way as their chronological use; this will make it possible to highlight the limits of each approach, the problems concretely encountered and what justified the move to the next method.

3.2.2 "ISOTENSOID" DESIGN

This is based on the following two conditions
- the filaments alone counterbalance the stresses from internal pressure (net method),
- stress in the f.ilaments is constant along their entire length.

3.2.2.1 EXPRESSION OF THE FIRST CONDITION

The heads being only in satellite winding, relations (8) and (9) lead to

Ne _ tg2
N-

ex

(12)

Moreover, it is supposed that the tank is subjected only to extension stresses. The expressions of
Ne and N<jl are then as follows :

PR2

--2-

(13)

(14)

Furthermore, it is easy to demonstrate that

(15)

R2 = P

1/

1 + P ,2

(16)

Thus, the final expression of the first condition (net method) is

R2
Le. ,
2
--

2 +

R1

pp
1 + P ,2

(17)

3.2.2.2 EXPRESSION OF THE SECOND CONDITION

This second condition requires that the stress in the filament be constant. The expression of this
stress is deduced from the relation (8).
cr f s = _ _--=N\>c.!.--;;-_

X e s cos 2 a

(18)

To transform this expression into a suitable form, the local thickness of the tank wall must first be
determined.
Relation (7) makes it possible to write:
=

N _ _S_ _ , (at any given point),

X cos a
N

X cos a

,(at the junction between one of the heads and the cylinder).
0

5-7

Moreover, account is taken of the fact all the filaments which cut the parallel to the junction between
head and cylinder also cut the parallel through the point considered. This is translated by :

P N= 2

')l;

P o No

')l;

This provides the relation which expresses the local thickness

p 0 cos 0: 0
es = e
so
p cos 0:

(19)

Whence, with the help of (13), (18) and (20)

v'

p P 2
1 + P ,2
f s = ---"--"-'-------'--'------"------

(J

2 X e so p

cos

0:

cos

(20)

0:

Requiring that the stress in the filament is constant therefore consists of writing the second condition in the following manner :
1 + P .2

p2

= K

cos

(21)

0:

3.2.2.3 DETERMINATION OF THE MERIDIAN AND OF THE WINDING LAW

The meridian of the surface head (5) and the winding law are defined by the functions p (z) and 0: (z)
respectively.
The two functions are solutions of the differential system consisting of equations (17) and (21) recalled below:
2 + _--LP-"P_"_ = tg 2 0: ,
1 + P ,2

v'

1 + P .2
cos 0:

K.

These two equations make it possible first of all to demonstrate a simple relation. The derivation with
respect to z of equation (21) leads, after simplification, to :

2 p'

Taking into account the expression of

p' sin

+ P ,2

pp"

_-L~_ _

1
2 p ,

P , -----"-P,,,-P_-

+ P ,2

p' (tg 2

+ P cos

0:

or finally

0:

0:

0:

o .

deduced from (17), this becomes

2)

0:

p tg

0: '

p sin

0:

P tg

0:

0:

'

i.e.,

0,

(22)

cte.

This relation between the winding angle and the local radius is very simple and, in this particular cadesign), makes it possible to obtain the winding law when the head is geometrically defined.
As will be demonstrated later, relation (22) is a characteristic property of the geodesic lines belonging to a revolving surface.
Thus, it has been shown that designing a tank head such that the filaments alone counterbalance the
stresses due to internal pressure by being subjected to a constant stress means that these filaments will
be laid following the geodesic lines of the mean surface of the tank.
As for the determination of the geometric shape of the mean line of the heads of a tank, the following
mathematical developments make it possible to specify the function linking the coordinates p and z.
After elimination of the variable 0:, the relations (17) and (22) lead to :
se

(isotenso~d

pp"
+ P ,2

(23)

Another form can be deduced

The integrated form of this latter relation is written as follows :

p 6 (1

P ,2) = K2 P 2

(C = instegration constant).

The constant C can be evaluated by writing the previous relation for z

(p

0'

From
law.

~elation

O.

= 0).

A few, simple mathematical manipulations lead to the final result

cos 0: 0 t 3 dt
z(u)=-po

JV

(24)

V(1

(24) the shape of each of the heada can be found. Relation (22) determines the winding

5-8
3.2.2.4 PRACTICAL APPLICATION
A tank has two heads, and therefore two planes of junction between the cylinder and the head.
The geodesic line on a cylinder is a helix. The winding angle a , which is constant, is the same at
these cwo junctions.
a 01

a 02

At the openings of the two heads, a = 90.

Relation (22) immediately leads to :
P Fl

=P

F2

=P

0 s in a 0

The first finding is that the isotenso~d design only enables the fabrication of tanks with identical
polar openings.
Moreover, relation (17) shows that the meridians of the heads present an inflexion point for a , such
that:

Practically speaking, the following rules are applied :

a .; 54.7 Relation (24) is used to define the shape of the heads.
2. 54.7.; a .; 90 The head is spherical, its radius being equal to the value taken by R2 at the point
of inflexion. The winding law is governed by relation (22). The structure includes a metal reinforcement
called polar mounting (Fig. 1). Consequently, the first condition (equation 12) is no longer verified,
and more complex calculation methods are necessary to study the behavior of this area. The point of
inflexion is chosen to situate the beginning of the metal reinforcement.
In conclusion to the development of this first approach, it should be underscored that only tanks with
equal openings can be made. In practical terms, for rocket cases, overall optimisation systematically leads
to a small opening at the front to reduce the weight of the metal reinforcement with a large opening aft
due to the dimensions of the nozzel.
Consequently, this approach cannot be systematically applied for such cases. On the other hand, it has
the advantage of enabling simple and effective products as the stability of the filament is perfect during
winding and the stress within the filament remains constant. We therefore produce small vesers for characterisation based on these principles.
1. a 0 .;

3.2.3 "BALANCED PLANAR DESIGN"

3.2.3.1 STATEMENT OF THE PROBLEM
It has been demonstrated above that isotenso~d design cannot be used for the production of filamentwound structures meeting the specific requirements of rockets.
Moreover, the production facilities have w~nding machines specifically geared to laying filaments according to a trajectory within a plane. Design departments have therefore developed design methods based on
the f0110wing two essential prerequisites :
1. The filaments alone counterbalance the stresses due to internal pressure.
2. The trajectory of the filaments is contained within a plane.
The figure below describes in a simplified form the configuration of the filament and of the mandrel.
Winding plane
plot (8)

PF1

'Y

Figure 4
It should be noted that as the shapes of the heads and therefore their depths are only determined once
the problem has been solved, the length of the cylindrical part is only known at the end of the
calculation. On the other hand, the angle defining the "winding" plane is determined by :
tg y =

P Fl

+ P F2
L

(p Fl, P F2' and L are fixed parameters for the designer).

The first condition that the parameters P (z) and a (z) must meet is always expressed by relation (17).
The second condition is obtained by saylng that the trajectory of the filament is planar. This is covered
in the following paragraph.

5-9
3.2.3.2 EXPRESSION OF THE SECOND CONDITION
The figure below summarizes the parameters and their geoluetric representations involved in the expression of the condition sought.

Cylinder

Mean surface area

of a head

Trajectory of a filament

Running meridian of the head

Figure 5
Lengthy calculations, that are not detailed here, provide the expression of the winding angle a by
[pp' - (z tg y

P e) tg Y ]2

+ (tg 2

Y - P ,2) [p 2 - (z tg y

P e)2

(25)

+ P ,2) [p 2 - (z tg y + p e)2

(1

This equation, together with equation (17) forms the differential system involving p (z) and a (z).
The elimination of a leads to the second order differential equation of the function p :
[pp

(z tg y + P e) tg y

(2 - tg 2 Y + 3 P ,2 + pp") [p 2 - (z tg y + p e)2]

J2

(26)

3.2.3.3 NUMERIC RESOLUTION

The Runge Kutta method is used to integrate numerically this type of equation.
To use this method, equation (26) of the second order is replaced by the following differential system
of the first order :
p'

u' = _1_
P

The two functions, p and u

[up - (z tg y
p 2 -

= u

P e) tg y

(z tg y

(27)
]2

_ 3 u2 _ 2

p e)2

therefore verify two equations of the type

\ p'

u'

1 (u, P
2 (u, P

z)
z)

+ tg2

(28 )

510
Under such conditions, if the functions P and u have values P i and ui for Z = zi' the values P i + 1
and ui + 1 for Z = zi + 6 Z are determined by performing the series of the following calculation (ref. 2).
fl (ui

P i ' zi)

f2 (ui

P i ' zi)

- P 'i 1

fI (ui +_1_U'i0 6

, P i + _l_p 'iO 6
2

zi +~)
2

U'i l

f2 (ui + _1_ u'iO 6

2

, P i +_1_P'i0 6
2

zi +~)
2

- P'i 2

fl (ui + _1_ u'il 6

2

P i + _l_p 'i 1 6
2

zi +~)
2

U'i 2

f2 (ui + _l_U'i l 6

P i +_l_p 'il 6
2

zi +~)
2

- P 'i 3

fl (ui + U'i 26

P i+P'i 26

zi +6 z)

U'i 3

f2 (ui + U'i 26

P i +P'i 26

zi +6 z)

Z (p
- P HI = P i + -66 -

ui+l

= ui

'i

+ 2 P 'i 1 + 2p'i 2 +P'i 3 )

,
(29)

+ ~ (u'iO + 2 U'i l + 2 U'i 2 + U'i 3 )

6

Knowing that for Z = 0, P = P

and u = P '0 = the calculation can be started and subsequently makes
it possible to determine a serie of discrete values of the functions P and u, and of the function a uslng
equation (25).
The design presented being based in particular on equation (17), the meridian of a head will present in
the same way as for the isotensoid design a point of inflexion as soon as the winding angle reaches the value of 54.7. The design rules given in the isotensoid formulation will again be applied for the shape of
the meridian. On the other hand, winding will be planar beyond the point of inflexion, namely in the area
of the polar me~al reinforcement.
In practical terms, the shape of the,meridian and the winding law will therefore be determined, by the
previously described numeric integration, from the planes of the junctions between the heads and the cylinder, to the openings (a = 90).
D

Figure 6
Nevertheless, it is to be noted that equation (28) involves the parameter P e which positions the winding plane (figure 6). This parameter must be determined for each head so as to obtain the opening
required. This aspect of the problem is evidenced in the figure above.
As the parameters P Fl' P F2 , and L are known, the points A and B and the angle yare fixed.
By imposing a value for the parameter Pe the point D can be situated for a head, representing the beginning of the meridian for the head in question (p
is known).
The meridian can be constructed by numeric integration to the point F, where the winding angle assumes
a value of 90. This point F is only known at the end of the integration of the system (27), (28). The problem will only be solved when point F coincides with point B.
This shows that for each head the value Pe must be found, leading to a final point 'coinciding with A or
B respectively. This will be done iteratively and automatically once the method is programmed.

3.2.3.4 CONCLUSIONS ON THE "BALA.'1CED PLANAR" DESIGN

This method has the advantage, on the one hand, of leading to a design such that the composite material
is only subjected to those stresses that are counterbalanced by the tension of the filaments, and, on the
other hand, of presenting a Winding method particularly well adapted to the power train of simple and fastoperating winding machines.

5-11
However, experience has shown that, during or after the winding phase, the filaments may slip, in certain cases. This leads to a position of equilibrium of the filament that is different from the position
scheduled in the design of the product.
Any excessive occurrence of this phenomenon will bring about a drop in the fabrication quality and will
adversely affect the performance of the structure.
Certain concrete examples have meant that this slippage phenomenon has been taken into account as from
the initial design phase, leading to a third design method for vessels with unequal polar openings (ref.3),
as will be explained in the following paragraphs.
3.2.4 BALANCED WINDING AND CONTROLLED STABILITY DESIGN
As indicated in the title, this design takes into account, on the one hand, the balancing of the forces
due to internal pressure, and, on the other hand, the stability of the filament during or after the winding
phase.
The following paragraph will evidence the parameters characterizing this stability.
3.2.4.1 ANALYSIS OF FILAMENT STABILITY DURING WINDING
During the winding phase, the filament is laid on the mandrel with a tension of T. Figure 7 illustrates
the parameters involved in what follows.
-7

-T

_____ (r)

Figure 7

..,.
The mandrel exerts a force F
the following form :

on the filament, per unit length of the filament, which can be written in
~

F = A t

+ >I.

ns

4-

u bs

By placing an infinitely small element of the filament in'equilibrium, it can easily be demonstrated
that the values A ,>I., u have the following expressions
A =

dT
- <i."S

(30)

- Tc cos '

(31)

- Tc sin '

(32)

..,.
Under the effect of the tension Tj the filament tends to slip in the direction of the vector b s .
Furthermore, this tension "flattens" the filament against the mandrel.
The slippage tendency is thus characterized by :
(33)

This relation shows that the angle' is the value characterizing the stability of winding. Consequently,
in the design presented here, this angle will be taken into account to express the second condition necessary to determine p (z) and a (z).
3.2.4.2 EXPRESSION OF tg'
The method of obtaining the expression of this value is rather long. Only the main intermediary results
will be presented in turn.
4-

-+

l-

4-

The components of the vectors n s and t in the datum mark (x, y, z) have the following expressions
cos e
D

..,.
ns

sin
D

-p-'
D

(34 )

5-12

p' cos a cos

- sin a

sin

e ,

+ sin a cos

e ,

.,.

p' cos a sin

(35)

cos a
D

Expressions (35) are obtained taking into account the following relation

de
dZ

D tg a

(36 )

.,.
Subsequently, a vector parallel to n can be obtained by calculating the components of the vector

dt

cos a cos
D

a' cos a sin

cos a sin
D

p' a' sin a cos

D
sin 2 a cos
cos a

sin2 a sin
cos a

e -

p, p
D3

a' sin a
D

p' sin a sin

p,2 P
D3

D
p

p' a' sin a sin

D

dZ
a' cos a cos

-pD-

p' sin a cos

p
p ,2 p ;,

cos a sin

D3

cos

cos a

(37)

e ,

cos a

The expression of tg' is deduced from relations (34) and (37)

tg' =

(a' p cosa +p' sina) (1 +p.2)

(38)

pp" cos 2 a - (1 + P ,2) sin 2 a

Note:
A characteristic property of the geodesic curves of a revolving surface is that the normal vector at
the surface and the main normal vector at the curve meet at all points.
They thus verify that' = O.
Relation (38) therefore indicates that : a' p cos a + p' sin a = ().
Namely : p sin a = e te
The assertion given in paragraph 3.2.2.3 1s thus established, which demonstrates clearly that designing
a so-called "isotenso'id" vessel is equivalent to designing the same vessel laying up. the filaments along
the geodesic lines.
3 2.4.3 DETERMINATION OF THE GEOMETRY AND WINDING LAW
With the first condition unchanged, the problem is solved by integrating the following differential
system :
2

pp
I + P ,2

tg 2 a

(a' p cos a +p' sina) (1 +p,2)

pp" cos 2 a - (1 + P .2) sin2 a

(39)
k

Moreover, in what follows, designs will be sought such that the slippage tendency of the filaments remains constant along their entire length.
As before, the following first order system will be resolved by the Runge Kutta method :
Kutta :
p' = u
(tg 2 a - 2)

u'

(40)

a'

(2 k cos 2 a + u sin a )

p cos a
The following system is resolved using equations of the same type as relations (29) extended to three
functions f1 (u, p ,a ), f2 (u, P ,a ) , f3 (u, P ,a ).
The first equation of the system (39) shows that this design again leads to the existence of a point of
inflexion when the winding angle reaches the value of 54.7 0 ; Beyond this point, the head has a spherical
shape, with the radius being equal to the value of the second radius of curvature at that point, and the
lay-up line of the filament is geodesic, thus ensuring the continuity of the winding angle.
Finally, the angle e necessary to ensure the adjustment of the winding machine is obtained numerically
using relation (36).

5-13
3.2.4.4 DESIGN OF THE CYLINDRICAL PART
The technique of winding with a constant slippage tendency can be used for the cylindrical part. In
this case, the relation (38) is written as follows
a;

P cos a;

sin 2 a;
Thus, at two points MI and M2 separated

b~

a length 1, the winding angles a; I and a; 2 verify

In the same way, the variation of the angle

2 -

l
I = - k \ Log
(

[~
+
1
P
sin a;
0

kl

1
sin a; 1

sin a; 2

(41 )

is obtained by using relations (36) and (41) :

+ ' / (~+
1
)2 _ I
V P 0 sin a; 1

- Log [

sin a; I

+ tg

I
]
a; 1

(42)

3.2.4.5 APPLICATION OF WINDING WITH A CONSTANT SLIPPAGE TENDENCY

When the problem of designing a new vessel is posed, the parameters P F1 , P F2 and P 0 are imposed.
Moreover, an idea of the criticality of the problems to be solved has to be rapidly obtained. In certain
cases, the technological risks have to be evaluated and the designer must be capable of taking decisions as
to the nature of the composite material adapted to the problem posed. This has technical consequences and
naturally a direct effect on the price of the product sold. This obviously involves an initial evaluation
of the problem not requiring a very great accuracy.
That explains why the method presented above, that is used in our industrial fabrication, has been processed in such a way as to enable the very rapid evaluation of a few essential parameters, as illustrated
in the following diagram that will be discussed below :

ao (degrees)
60
0,70

I - - - - - - - - + - - - - - - - - + - = - , - - - - - - - = ;.....'""""=-----=~...._o<'! 0.65
0,60
0.55

1-----=:=---1""=---==--....-;n.",-----=:=--=--!::;;;--=---:::;;;;>o"'...._o<'! 0.50
0,45
0,40
0.35

1---=--===--t-="'-=----::::;-''''I'i7<=---,='''.....,==--~-==-__:::
......--=-=_'"''''' 0.30
0.25
0.20
0.15
.......c.:"=_==--::;j 0.10

-0,2

-0,1

0.1

~
Po

0.2

tglji

Figure 8
This diagram relates the parameters :
- a; 0 (winding angle at the junctions of the heads and the cylinder),
- P F/P 0 (what is called the opening ratio),
- tg 1Jf (slippage factor).
The tracing of this diagram is obtained by the systematic use of the program numerically resolving the
system (40). Thus, the lines traced in Figure 8 have been drawn point by point.
The following example shows how this diagram can be used :
P Fl/P

0.50

(aft head)

P F2/p

0.25

(front head)

If a helicoidal trajectory is used for laying up the filaments on the cylinder, the winding angles at
the two head/cylinder junctions are equal. In addition, in order to obtain the most stable Winding possible, it will be accepted that the procedure must be such that the slippage factors are equal for both
heads. Under such conditions, the points ml and m2 therefore represent the two heads in the diagram, from
which the follOWing can be obtained :
a;

II 23 0

I tg1Jf 1=

0.13 i.e.

'It

5-14
Another possibility lies in choosing a winding law for the cylindrical part which presents a slippage
tendency. In this case, the winding angles at the two ends of the cylinder may be different.
The points m'l and m'2 always obtained by equalling the slippage tendency at the two heads, will then
present a new solution to the problem.
In the present case, the slippage factor is reduced
tg' =0.10

Le.

' =5.7

Winding stability at the h~ads has thus been improved. By reading off the winding angle values at the
head/cylinder junction from the diagram, the slippage factor can easily be calculated using relation (41).
3.2.4.6 DESIGN OF STRESS IN THE FILAMENTS
Relation (20), which gives the general expression of the local stress in the filaments, can be written

p 2

rJ

VI

with

d rJ fs

dz
d rJ f s

P ,2

The expression of

f s = H -"---c-o-s-o:-"---

2 X e so P

cos

0:

is written below
= H[

dz

pp'

cos

0:

sin 0: 0:'
cos 2 0:

Relation (17) makes it possible to eliminate p". After simplification, this becomes
d rJ fs

dz

= - 2 P

H tg

0:

D . tg '

(43)

By using the principle of equalizing the slippage tendencies on both heads, from Figure 8 it appears
that:
- tg, is negative for the head with the larger opening (aft head),
- tg' is positive for the head with the smaller opening (front head).
Given this, tt results from relation (43) that the stress in the filaments increases from the
head/cylinder junction to the point of inflexion for the aft head, and decreases for the front head.
Therefore, without taking into account the phenomena of flexion that the theories developed hitherto
cannot in fact take into account, it would appear that the aft head should always be thicker than the other
head in order to allow for the maximum stress reached at the point of inflexion.
3.3 CONCLUSION ON DESIGN METHODS
These design methods presented above are today operational and are used almost daily. They are usable
by employing simple equipment and software. The time required to obtained results is short (in the order of
a day). The volume of inputs and outputs is low.
The results are complete, given that it is possible to obtain mainly:
- the overall geometry,
- the weights and volume of the filament-wound shell,
- the winding parameters for the winding machines,
- parameters making it possible to have an idea of the difficulties of fabrication.
Nevertheless, the hypotheses of these methods are such that verification of the extent to which the
structure meets the specification must be undertaken, in particular to ensure that the local over-stresses
are acceptable.
Furthermore,' the metal reinforcements must be designed by' calculations that the methods described cannot handle.
The following paragraph will describe the methods used to provide final justification of the structure
before any irreversible fabrication phases are undertaken.
4. JUSTIFICATION
4.1 PRINCIPLES OF

T~

METHODS OF JUSTIFICATION

The working method consists of using a calculation method based on the technique of finite elements
(S.A.M.C.E.F. developed by the LTAS of the University of Liege).
As the purpose is not to describe the method of finite elements, the major phenomena taken into account
simultaneously to ensure as realistic a study as possible of the behavior of such structures will be listed, namely:
.
- the geometric non-linear behavior, related to the fact that the internal pressure, at equilibrium, is no
longer exerted on the initial geometry but on the distorted geometry, is necessary to obtain displacements and a stress condition corresponding to reality,
- local anisotropy, which varies according to the winding angle and therefore the point taken of the structure, is an element that must be taken into account when creating the data card-indexs representing the
terms of the Hooke matrix along the entire structure,
- elasto-plastic behavior laws are necessary to relate realistically the tensors of stress and distortions
of the polar mountings,
- the existence of gap between polar mountings, blanking systems and their connections cannot be overlooked, and has led to the development of extensions to the SM1CEF software program.

5-15
4.2 PRESENTATION OF THE MAIN RESULTS
The results are printed on'voluminous listings, from which a few representative plates have been taken.
1 - Stresses in the filaments :
These streses are presented in plates of the type illustrated in the figure below:
EVOLUTION OF STRESSES ACCORDING TO FILAMENT, A P = 15,0 MPa.
Stress - MPa
-4000.0

-,

J5OO.0

#+

F"" Iih +

3000.0

'.."

+Int, sat.

~Ext.

-+

I 'h

25000

..t

.
t

'00;)0

1500.0

.. -

1000.0

+-- -_.- ~- -

-~-

sat.

o Int. eire,

e-.

500,0

o
0.5

10.

15

30

40.

45

55

65.

ro

75.

~.

~.

Figure 9
This figure indicates the evolution of the stresses from the middle of the cylinder to the opening of
'one of the two heads. Such stresses affect the internal and external layers in such a way that the local
phenomena are evidenced. A more or less constant evolution can be noted, except :
- at the head/cylinder junction,
- in an area close to the opening.
Only the finite elements method can provide the maximum values reached and justify the reinforcement
that had been defined on the basis of past experience.
2 - Stresses in the polar mountings
Several representations have been used over the past years to illustrate a two-dimensional stress
conditon as from numeric indexs providing on stress tensor per element of the modelization.
Currently, the designer is provided with coloured figures, defining zone by zone the levels reached.
This type of figure is illustrated below :

17'"
lUll

19.51

j:
Y

~'~~d'l"I'''>I''''~U ~~~.~~~~~_~~:~_::'~~_
Figure 10
This figure shows a radial cross-section of a front polar mounting (small opening).
In this example, it can be seen that the area most subject to stress is in the ring marked by the arrow.
Multiple results are obtained and subjected to precise analysis, leading either to modifications. or to
the decision to give the go-ahead for fabrication. Only the principal critical parameters for the behavior
of the structure are mentioned, namely :
- shear between winding layers (possibility recently developed),
- shear of the rubber laminates connecting the winding and the polar mountings,
- general deflection of the structure to evaluate its size.

4.3 CRITICAL ANALYSIS OF DESIGN AND JUSTIFICATION METHODS

It is a rule in our company to consider that the existence of these two methods is necessary. The first
has the qualities of being rapid, effective and able to allow users to retain an intuitive knowledge in regard to the influence of the main parameters on the results obtained. The second, longer and more expensive
to use, is alone suitable for making it possible to draw the conclusion that the structure can be
fabricated.
This is why it is necessary to maintain a balance between the use of these two methods, which we consider as complementary.

5-16

5. CHARACTERIZATION OF MATERIALS
This is a subject which deserves a document to itself and a special presentation. We shall confine ourselves here to presenting some of the more important principles, with a view to highlighting the importance of this characterization in the field of industrial production.
Characterization can be split into two main categories :
work in connection with our knowledge of material properties, with a view to enabling the Designers to
justify the product to be fabricated with access to information I,hich is as complete as possible concerning the characteristics of the materials making up the winding and the composite.
The Calculation Department, for example, needs the laws of behavior linking the tensors of stress and
distorsion. Special tes ts are requi red to de termine the terms of the Hooke I s matrix if linear behavior is
considered representative. This particular case, the simplest that can be chosen, nonetheless requires the
determination of nine different coefficients.

Direction perpendicular to the surface (8)

3
2

Direction of the filaments

Figure 11
The relationships are written as follows

22
(J
33
["
(J 11
12
(J 13
(J 23
(J

Hn

Sym

H12

Hn

H13
H23
H33

0
0
0
C12

0
0
0
0
C13

o0 .

C~3

[j

"
22
33
12
13
23

(44)

Other methods in the 1970s based on the properties of the filaments and the resin made it possible to
determine the elements of the relationship (44). Characterization of the constituents was then necessary.
Specific laws were sometimes necessary when the composite had non-linear relationships ( (J , E ) (e.g.
degradation of the composite by microfracturing of the resin when placed under internal pressure).
Moreover, in addition to the composite, the presence of rubber made it necessary to carry out experiments enabling the stress to be linked to the deformation. Curves of the following type were used :

(J

Figure 12
- The second category involves the characterization required to ensure that the product fabricated complies
with the acceptance specifications for the materials.
There are any types of characterization, but we shall only deal here with those relating to winding.
The fibers are bought from manufacturers who deliver the material in such quantities as belong to a
single fabrication batch. Before their use for the fabrication of a filament wound structure, samples of a
similar design to the rockets are produced on a reduced scale (diameter 304 mm). The samples are subjected
to rupture through internal pressure. The values obtained make it possible to decide whether or not the mat erial can be used.
Less costly tests are carried out on unidirectional tows subjected to tensile strain. The former, more
expensive, type of test, however, is more representative of the biaxial strains (meridian and parallel)
existing in a real structure.
Characteristic material performance values will be given in paragraph 8.
6. FABRICATION
ded
a b c -

This paragraph will be devoted to the fabrication methods for this type of structure. It will be diviinto three parts:
description of the filament-winding machines,
description of the main stages during fabrication,
particular remarks on the method of obtaining the filament wound composite.

5-17
6.1 DESCRIPTION OF THE FILAMENT WINDING MACHINES

The filament winding machines we use are of two types, for which we shall use the following terms
- vertical satellite machine,
- horizontal machine.
6.1. 1 VERTICAL SATELLITE MACHINE

The operating principle of such a machine is represented in the figure below

o--t---- Mandrel

Exit point
of filaments

o y
Figure 13

The mandrel is fixed to a supporting structure with a vertical axis. An arm, turning at a speed Q,bears
the filament routing and includes, at its end, a device whose purpose is to provide an outlet for the filaments prior to being laid on the mandrel.
If the mandrel is fixed, after several rotations of the arm the filaments will be laid by overlaying
on a planar trajectory. For this reason, a slight rotation of the mandrel is necessary in order to cause a
shift in the filament braid after one full rotation of the arm. This is how the mandrel is covered. When
these shifts total 360, one filament wound layer is completed.
The number of layers laid is defined by the thickness of composite required in order to comply with the
resistance specifications (paragraph 3).
The configuration of some structures is such that there is no guarantee of the stability of planar
winding. In order to solve this problem, the technique known as "modified planar filament winding" is used.
This type of winding is achieved by turning the mandrel at a rotation speed w . The filament laying trajectory is no longer planar. When the arm has completed a turn, the braid is no longer laid next to the previous braid. The ratio between the speeds wand Q is chosen in such a manner that there will be
juxtaposition. This gives rise to star pattern winding.
This type of machine does not make it possible to rigorously respect the a (z) laws compatible with the
form of mandrel defined by p (z). Experience shows, however, that it is possible to approach the theoretic
laws.
Moreover, these machines are simple, rapid and reliable. They also exist, and are there to be used.
This is why this type of machine is still greatly used. We shall see below that they do nonetheless have
certain limitations in regard to the type of implementation of the composite at the time of winding.

6.1.2 HORIZONTAL FILAMENT WINDING MACHINE

The operating principle of these machines is shown in the figure below

Mandrel

Winding arm

Support

Figure 14

5-18
In the simplest case, the movements enabling a braid of filaments to be laid are
- rotation of the mandrel,
- displacement of the mandrel holder parallel to the z direction,
- displacement of the winding arm in the y direction.
So, the point M which here symbo~zes the winding head with its filament braid outlet, can be displaced
in any manner on a plane parallel to z.
Improvements can be made in one particular field to enhance the quality of the filament winding.
The braid laid is, in practice, a group of tows themselves made up of several thousand filaments (the
tow is the condition of the material as delivered by the manufacturer). Thus, the braid coming out of the
winding head (M) has a certain width, generally around 20 mm. In order to reduce the length of this phase
in production, we envisage increasing that width to around 50 mm. It is inevitable that the braid is not
planar between point M and the point of tangency with the mandrel. This phenomenon, if over pronounced, may
give rise to poor winding quality. In order to remedy this, a winding head with several degrees of freedom
of rotation about point M is currently under development. These degrees of freedom will be programmed, as
will the main movements described above.
This type of machine makes it possible to comply fully with the theoretical (a (z)) winding laws defined
in paragraph 3. Also, unlike the vertical machines, they make it possible to wind structures whatever the
implementation of the composice (paragraph 6.3).
6.2 MAIN STAGES IN FABRICATION
The purpose of the fabrication of rocket cases is to provide the customer with a case ready to be loaded with powder and fitted with the nozzle prior to becoming a rocket stage ready for assembly with the
rest of the vector.
Production is therefore broken down into the following stages :
- fabrication of the primary parts (frames and polar mountings),
- fabrication of the internal thermal protections fitted to the polar mr'lntings by rubber lay-up in special
molds. Vulcanization is performed by placing in a pressure oven (~utoclave),
- fabrication of the mandrel by introducing sand and binder into molds fitted with protected thermal
protections. The sand is rammed, then hardened by oven drying,
- assembly of the various parts thus produced to provide the mandrel ready for winding!
- securing the mandrel to the filament winding machine, and winding,
- polymerization of the composite in an oven (the skirts may be fabricated during the filament winding phase or after the first polymerization. In the latter case, a second polymerization cycle is required),
- extraction of the sand by adding warm water which disintegrates and washes out the sand,
- machining of the skirts, securint the skirt frames, and finishing operations,
- geometric examination and examination of critical areas such as bonds in particular. An important phase
during inspection is pressure testing the case at an internal pressure equal to 1.15 times the maximum operating pressure.
Finally, the case is delivered.
6.3 IMPLEMENTATION OF THE COMPOSITE
Two methods are used. They involve the method of impregnating the filaments with resin.
The first method (the W method) consists in impregnating the tows during the winding phase immediately
before the filaments are laid on the mandrel. We shall call this the WET technique.
The second (the P method) consists in impregnating the tows prior to the winding phase (sometimes several months earlier), subjecting the impregnated filaments to a thermal field to induce initial polymerization, then storing this product on reels at a temperature of -18C. We shall call this Prepreg winding.
Both methods, of course, have their advantages and their drawbacks.
a - W method
This procedure is less costly than the P method. The stability of the filament during the winding phase, however, is sometimes poor.
b - P method
This procedure provides enhanced stability during the winding phase. It involves, however, increased
cost in fabricating the case because of the extra time required to make Prepreg.
Another source of overcoat linked to quality improvement is the inspection of the product prior to use.
Finally, it should be noted (paragraph 3.2.4) that, in certain configurations, Prepreg is required in
order to be able to wind the case.
7. EXPERIMENTAL QUALIFICATION
In addition to the experimental work whose purpose is to characterize the materials used in order to
fabricate the case, a new case is subjected to destructive testing in order to determine its ultimate
resistance.
The skirts are subjected to predetermined mechanical stresses providing the best possible representation of the severest stress conditions.
The rest of the case is subjected to destructive testing through the application of internal pressure,
generally by introducing water into the wound capacity.
In addition, a mechanical effort applied to the rear link represents the thrust relief. The front skirt
is therefore subjected to compressive stress.
This type of test should demonstrate that the failure pressure PF is such that :

(PMO

Maximum operating pressure)

5-19
The case is placed vertically in a pit. A great number of measurements are made in order to
- make it possible to explain how the failure Was initiated,
enable the calculation team to compare theoretical forecasts with reality,
provide the person in charge of powder loading with the deformation limits of the filament-wound case.
These measurements are carried out using displacement sensors and deformation gages.
The following areas in particular are measured :
the middle of the cylindrical part in order to determine the strain on the circumferential winding,
- the skirt/cylinder bonds, which are areas of bending,
- the areas of composite near the polar mountings (areas of bending),
the metal rear polar mounting which is always optimized as it is heavy due to the wide opening.
These tests are always long because of the dense instrumentation (250 measurement channels in major
tests) and the use of extensive equipment :
recording equipment which requires each measurement channel to be connected to a data acquisition system,
- heavy mechanical equipment in order to comply with safety requirements.
Lastly, systematic video and fast camera (~ 3000 frames per sec.) recordings are made, covering the whole case insofar as possible, in order to be able to review the occurrence of the failure after the test.
The test pit is approximately 8 meters deep and approximately 7 meters in diameter.
The figure below gives more precise information on the details dealt with briefly in this paragraph.

Figure 15

8. RESULTS
In this paragraph, we shall present a number of results which can be used to evaluate
- the performance of filament winding in comparison with certain metals,
- developments in the performance of filament wound products,
- the specific performance related to rocket cases.

8.1 COMPOSITE/METAL COMPARISON

We shall exclude here the performances indicated by suppliers for the purpose of this type of
comparison. Performance obviously depends on the intrinsic qualities of the materials, but it also depends
just as much on the object (geometry, stress) and the quality of the implementation which is determined by
the development of good resins, in association with the proper thermal cycles, the winding quality, etc.
Therefore, the filament wound cOt~osite assessed here has been assessed by using failure stresses obtained on known samples, discussed in paragraph 5.
In addition, the stresses presented are obtained by taking account, not of the cross-sectio~al area of
dry fibers, but of the total cross-3ection (dry fiber + impregnating resin).
Finally, the parameter adopted is a /d (d = density of composite).

:5-20
Results as of today are
MATERIALS

(J

/d

Glass composite

'"

Kevlar composite

'" 1400

Carbon composite (Hercules IM6)

900

'" 1600 to 1800

Steel

'"

200

Light alloy

"

170

Titanium

'"

220

(J

= expressed

in MPa.

The advantages of composites are very considerable, as can be seen from comparison between the values
shown in the above table.
It can also be seen that the fibers successively used - glass, Kevlar and carbon - have produced marked
performance enhancement. The notion of the scale effect should be underlined here.
Experience has shown that it is not possible to judge performance solely on the basis of results obtained with small scale samples (diameter'" 300'mm).
It has been noted with glass and kevlar, that there is a significant decrease in performance as the dimensions of the case increase. This experimental fact, unexplained by our design department, is shown in
the diagram below :

R 1MPa )
3000 r - - - , - - - . - - - , - - - - ,

G.

G.

G.

(-11 %1

K (-23%)
2000 I-----I---+----\-----j

G: Glass

---,----1----

K : Kevlar

1000 1------.jf----I---+---1

'":-:--:-1

0'------'----'--

500

1000

1500

2000 ililmml

Figure 16
The following percentages are thus observed respectively
- 23 % decrease ( = 300 mm to 0 = 2000 mm) for kevlar,
- 11 % decrease ( = 300 mm to = 2000 mm) for glass.
This phenomenon has made it much more difficult to obtain a definition complying with the specifications when these materials were used.
Moreover, the performances observed during preliminary small scale tests being unobtainable on the real
object, the inert mass of the case was increased.
As far as is known today, this phenomenon does not appear to exist in regard to carbon fibers, thus
providing yet another advantage for this material.
8.2 PERFORMANCE FACTOR IN ROCKET CASES
The following magnitude is frequently used amongst industrialists fabricating such cases. It is defined
as follows :
q

=1mg

p
V
m
g

failure pressure.
internal volume.
mass of filament wound capacity.
terrestrial acceleration.

5-21
This factor has the dimension of a length. For countries using the metric system, it is generally expressed in kilometers.
This factor is, indeed, fairly representative of the criteria that fabricators wish to respect.
q increases as :
- p increases (no comment),
- V increases, corresponding to the desire to be able to carry the greatest possible mass of powder,
- m decreases, corresponding to the smallest possible inert mass.
An increase in q, therefore, represents the desire that a given stage should provide the best possible
contribution to the range of the vector.
It should, however, be noted that comparisons between cases should be made taking due account of the
fact that the overall length is imposed by the case fabricator.
For the same quality of fabrication, a quick analysis will show that, all other things being equal, an
increase in length gives rise to an increase in the performance factor.
This being so, we nonetheless present below, for information, the best performances achieved by our
company in this field.
Diameter (mm)

1500

2000

1150

Materials

Glass

Kevlar

Carbon

20

31

q (km)

41

As the range of lengths is fairly close, it is clear to see the development resulting from the evolution of the fibers on the market.
8.3 INFORMATION ABOUT FUTURE DEVELOPMENTS
In regard to carbon fibers, two major producers are offering fibers with ever higher performance
the American HERCULES and the Japanese TORAYCA.
The HERCULES Corporation, after its AS4 and AS6 fibers, is now marketing the IM6 fiber, with a
forecast :
JR

= 5000

MPa (tow test)

Another fiber, IM7, exists but is not widely marketed, with a forecast
JR

5500 MPa (tow test)

Other fibers, currently at the laboratory stage, promise even higher levels.
The TORAYCA Corporation, after the T300 fiber (class AS4), is now marketing its T800 and T800R fibers
which are respectively comparable to the IM6 and IM7 fibers.
More recently, this manufacturer has announced its TlOOO fiber, with:
JR

= 7000

MPa (tow test)

It may therefore be seen that there is rapid development in the materials made available to fabricators
of filament wound structures. This fact makes it all the more difficult to take certain decisions in that,
when a program is launched, it is always necessary to have available the elements making it possible to use
the best fibers, and these must be constantly evaluated and qualified for industrial application.
9. CONCLUSION
Experience in recent years in regard to both design and justification, has been detailed in this document (paragraphs 4 and 5). It is important to note that these methods have been developed in order to
meet extremely precise requirements in regard to rocket cases, outlined in paragraph 2.
We wish to stress, for this is the result of constant effort on the part of our design team, the fact
that we have always attempted to preserve rapidity in the field of pure design and-accuracy in
justification.
It should also be pointed out that there has been a constant effort to produce designs enabling our
production team to fabricate quality structures at the least possible cost. Remarks relating to the choice
between the WET and Prepreg methods, or to filament winding stability are intended to highlight such
considerations.
While the results presented demonstrate the obvious advantages of the filament wound composite over metallic materials, the latter also reveal the advances made by fiber producers. Industrialists today must
constantly take account of the rapid develoments in carbon performances.
Computer aided design, finally, though not dealt with in this document as it is a field in which the
equipment is gradually being implemented, is an extremely important subject for our Plant. It enables staff
to be freed from the repetitive tasks and devote themselves more fully to structure optimization. It should
be pointed out that equipment providing automatic transfer of elements from definition to production is
gradually being set up. In particular, the winding parameters are stored on a magnetic medium immediately
at the design stage, and this can be used directly to control the winding machines. This results in a reduction in labor, in cost and in the risk of error.
Currently, we consider that progress must be made in the field of automation during the design and justification stage, then during the fabrication stage.
In conclusion, we hope we have made it clear that our constant concern has been, in the design field,
to find solutions taking account not only of product performance, but also quality and efficiency during
the fabrication stage.

5-22
References

[1] TIMOSHENKO AND WOINOWSKI-KRIEGER - Theory of plates and shells.

(Second Edition) Mc Graw Hill Company - New York - 1959 - Page 435.
[2] HILDEBRAND - Introduction to numerical Analysis
Mc Graw Hill Company - New York - 1956 - Pages 236 - 237.
[3] J.P. DENOST - New Design concepts for filament wound pressure vessel with unequal openings.
Societe Nationale Industrielle Aerospatiale - 1982 - AIAA (Cleveland).

5-23
CONCEPTION DES STRUCTURES DE PROPULSEUR llOBINEES
par
J.P. DENOST
Chef de Programmes Structures Bobines
AEROSPATIALE
St-Mdard-en-Jalles
33165
FRANCE.

RESUME
Les mthodes de conception et de dimensionnement des structuresde propulseur bobines ont commenc a
atre dveloppes dans notre Socit au d~but des annes 70. Depuis, ces mthodes ont volu et ont ainsi
perrois d'obtenir des moyens de conception rapide et de vrification fine.
Le prsent document aura pour objet de delimiter ces deux approches, de les critiquer et de montrer
leur complmentarit. Elles feront l'objet de prsentations dtailles afin de permettre au lecteur d'en
comprendre les grandes r~gles et de tirer les enseignements qui lui permettront de les dvelopper dans des
domaines particuliers.
Cette partie a dominante theorique sera compltee par des considrations relatives aux materiaux, leurs
mises en oeuvre, leurs performances et une description des grands principes de fabrication de ce type de
structure.
NOTATIONS
A

aire d'une section droite du fil.

tAn.

bs

courbure de (f) en M ; c)

O.

CIt C2

centres de courbure priucipaux de (S) en M.

(1

symbole de drivation.

+ p'2)!

paisseur du rservoir en M.

+
F

action du mandrin sur Ie fil, par unite de longueur de celui-ci.

constante de gravitation terrestre.

projection orthogonale de M sur l'axe de rvolution de (8).

tg'l'

distance axiale entre les deux ouvertures polaires du reservoir.

point courant de (f) et de (8).

masse du reservoir.

+
n

vecteur unitaire de la demi-normale principale positive de (f) en M.

nombre de fils coupes par un plan

Us

paral1~le.

vecteur unitaire de la demi-normale intrieure

(8) en M.
+

N<ll

valeur en M de I' efEortinterne de membrane dans la direction de t<ll ' par unite de longueur du
pour la traction : N<ll) O.

parall~le

Ne

valeur en M de l'effort interne de membrane dans la direction de te , par unite de longueur de

la mridienne ; pour la traction : Ne) O.

origine du rep~re (x,y,z) ; situe dans Ie plan de jonction entre .1e fond et La partie
cyl1ndrique.

valeur de la pression interne ; p)

facteur de performance du reservoir ; q

+ + +

O.
=

~~

rayons de courbure principaux de (8) en M ; Ri

= MCi

1, 2).

abscisse curviligne de M sur (f) orientee.

surface moyenne du reservoir ; surface de revolution dont la concavite de la meridienne est

tournee vers l'axe de revolution.

5-24
~

vecteur unitaire de 'la demi-tangente positive de (f) en H

valeur de la tension dans le fil ; T)

z)

O.

O.
~

vecteur unitaire de la demi-tangente positive de la meridienne en M ; tiP

tiP

;>

te

vecteur unitaire de la demi-tangente positive du parallele en

p'
=

oriente par z

-E.p 0

volume interieur du reservoir.

x,Y,Z
~

~l

coordonnees de M par rapport au repere general.

~

(x,y,z)

base orthonormee directe du repere general; z : porte par l'axe de revolution de (S).

a:

mesure en radians de l'angle du couple de vecteurs (tiP' t) ; si (tiP A t).n s ) 0 : 0 <

courbe representee par le fil.

mesure en radians de l'angle du couple de vecteurs (x,HM)

exprime par la relation (36).

l.l

-)0

j:

-)0

.~

o:/~

'dz e ' etant

t.
~

ns'
~

HM.

P Fl ' P F2

rayons des ouvertures polaires des deux fonds.

a f

-+

bs '

contrainte dans le fil ; a f =

2A

iP

mesure en radians de l'angle du couple de vecteurs (z

titre vo1umique de fibre dans 1e composite.

'f

mesure en radians de l'angle du couple de vecteurs (n s

..

si (n s A z)te ) 0

n)

..

...

si (n s A n)t)

o<

'JI

Notations particu1ieres

( ) ,

symbo1e de derivation par rapport

indice affectant toutes grandeurs evaluees pour z

la variable z.

= O.

1. INTRODUCTION

La techno1ogie du bobinage est uti1isee depuis environ 2S ans pour realiser les structures de propulseurs des differents programmes fran~ais.
Ces programmes ont ete civi1s (DIAMANT) et militaires (MSBS, SSBS).
Cette techno1ogie a essentie11ement permis de remplacer 1e metal par 1e composite et ainsi de diminuer
1a masse inerte au profit de 1a portee du vecteur. De plus, 1e bobinage est une methode de realisation automatique qui assure simu1tanement rapidite, economie et qua1ite.
Les fibres impregnees de resine ont evolue au cours du temps. Les premieres structures ont ete realisees avec du fil de verre. Par la suite, sont apparus 1e Kevlar et plus recemment Ie Carbone. Chacune de
ces evolutions a renforce l'avantage du composite sur 1e metal.
Si 1es moyens de production mis en place ont rapidement evo1ue afin de permettre 1e derou1ement des
programmes militaires principa1ement, nos moyens de conception ont ete perfectionnes de maniere plus progressive.
Au debut des annees 60, 1a definition etait obtenue par des methodes ou l'empirisme etait en position
dominante. La voie experimenta1e permettait d'obtenir un produit repondant aux besoins ; quelques annees se
revelerent necessaires pour permettre aux competences des Bureaux d'Etudes de s'exercer en ce domaine, en
commen~ant d'etablir des methodes re1ativement simples vers 1a fin des annees 60.
Le developpement de l'Informatique en France, et plus tard l'apparition de programmes de ca1cu1s ont
conduit progressivement aux methodes de travail sophistiquees du milieu des annees 70. Enfin, aujourd'hui
l'accent est,porte sur l'automatisation de 1a conception et de la definition (CAO) et recemment sur 1a
liaison qui peut etre faite entre 1a definition et la realisation assistee par ordinateur (FAO). Toutefois.
les methodes dites "legeres", adaptees pour repondre rapidement a des questions ne necessit~nt pas immediatement llne grande precision et surtout permettant au potentiel humain de ne pas perdre la comprehension des
phenomenes physiques, continuent d'etre developpees et utillsees.

1t

5-25
Parall~lement, les moyens de calculs ncessitant des caracteristiques de plus en plus nombreuses, le
concepteur a etabli des programmes de caracterisation des composites bobines. Ces programmes ont conduit
des travaux importants assurant une activit soutenue pour les Laboratoires de Matriaux et d'Essais.

L'objet principal de ce document sera de presenter les mthodes de conception et de justification des
structures de propulseurs bobinees. L'analyse critique des methodes dites "lg~res" et des moyens informatiques sera faite et montrera comment notre Societe les utilise de mani~re complmentaire.
La partie la plus developpee sera celle-ci, mais il nous est apparu invitable de devoir consacrer des
chapitres courts pour prsenter, d'une part, comment en general les probl~mes se posent, d'autre part, les
methodes de fabrication et de qualification exprimentale.
Enfin, revenant aux moyens de calculs, il sera ncessaire de developper quelques notions mathematiques
qui seront condenses autant que possible afin de permettre au lecteur de comprendre, en esprant ne pas le
lasser, l'objectif etant d'ouvrir a qui le souhaite quelques notions suffisantes lui permettant de dvelopper les mod~les adapts a ses applications particuli~res.

2. POSITION DU PROBLEME
2.1 DESCRIPTION DE LA STRUCTURE
Dans le but de permettre une meilleure comprehension de ce qui suivra, une prsentation des principaux
sous-ensembles de 1a structure est necessaire. La figure (1) represente une telle structure.

Jupette avant

Cylindre

Ju pette ardare

Fond arriare

Fond avant

Embase polaire

,~#'I-- arriere

Embase polaire

avant--~VA

Liaisons caoutchoucs

Figure 1
Cette figure permet de dtail1er :
- les deux jupettes qui sont deux cy1indres courts, assurent la liaison de l'enve10ppe de propulseur avec
le reste du vecteur,
- le cylindre et 1es deux fonds qui sont realises par bobinage et dont nous prsenterons plus loin les
principes de conception et de justification,
- les deux embases polaires mtalliques qui renforcent les ouvertures et assurent la liaison avec d'autres
pi~ces (bouchon allumeur a l'avant, tuy~re a l'arri~re),
- des liaisons en caoutchouc entre rservoir et jupettes d'une part, et entre rservoir et embases polaires
d'autre part.
Le reservoir (partie cylindrique et fonds) est donc realise par bobinage ; au cours de cette phase de
fabrication on effectue deux types d'enroulements :
le premier par bobinage dit "satellite" qui constitue 1es deux fonds et le cylindre, par une succession
de boucles tangentes aux deux ouvertures,
- 1e deuxi~me par bobinage dit "circonferentiel" qui n'est depos que sur la partie cylindrique de telle
fa~on que les fils soient quasiment perpendicu1aires aux generatrices.

2.2 PRINCIPALES SPECIFICATIONS

Les specifications techniques d'un tel produit font l'objet d'un document detaille et volumineux, nous
retiendrons ici les principales.

2.2.1 TENUE AUX EFFORTS MECANIQUES

Les sollicitations mecaniques sont la pression interne exercee par la combustion de la poudre lors du
fonctionnement de l'etage consider, et des efforts de traction et flexion supportes principalement par les
jupettes lors de la-vie du vecteur.
De plus, des sollicitations thermiques doivent etre prises en compte lors des phases propulsees.
Les regles de qualification associees aces sollicitations sont telles que la structure a l'etat neuf
doit presenter un coefficient de securite : K > 1,4.
Par ailleurs, les differents sous-ensembles doivent rester dans 1e domaine elastique pour des charges
egales a 1,15 fois les charges nominales. Cette derniere specification est verifiee experimentalement pour
chaque specimen avant livraison au client.

5-26
2.2.2 VIEILLISSEMENT

Chaque sp~cimen fabriqu~ doit ~tre capable d'assurer ses fonctions avec une
une valeur minimale lors de toute sa duree de vie op~rationnelle.

probabilit~ sup~rieure

2.2.3 GEOMETRIE

Des sp~cifications g~om~triques sont impos~es afin de :

- respecter des encombrements maximaux fix~s A l'~tat de repos et lors du fonctionnement,
assurer des conditions d'interface avec Ie reste du vecteur.
2.2.4 CHARGEMENT EN POUDRE
L'integrit~ structurale doit ~tre assuree pendant la phase de chargement. En r~sum~, ceci se traduit
par des sollicitations de temp~rature et de pression interne pendant une dur~e fix~e.

2.3 TENUE A LA PRESSION INTERNE DE FONCTIONNEMENT

Dans ce document nous traiteron~ la conception et la justification d'une telle structure lors de la
so11icitation qui, dans tous les cas, est la plus sev~re : la pression interne due A la combustion de la
poudre pendant Ie fonctionnement de l'etage consid~r~.
De plus, si de nombreux ~l~ments sont contraints par l'effet de la pression interne, Ie sous-ensemble
Ie plus critique dans Ie travail d'optimisation est constitue de la partie cylindrique et des deux fonds.
C'est pourquoi, dans les paragraphes suivants, la conception et la justification de cette partie sera plus
particuli~rement d~taill~e.

Les probl~mes li~s aux jupettes ne seront pas traites du fait de leur masse faible et des notions classiques connues de tout Bureau d'Etude, que nous serions amenes A d~velopper. Par contre, une courte pr~sen
tation particuli~re sera consacr~e au dessin de l'embase polaire arri~re, celle-ci ayant parfois une masse
telle que son optimisation est rendue n~cessaire.
2.4 POSITION DU PROBLEME PARTICULIER AU BOBINAGE

La partie cylindrique bobinee s'etudie de mani~re tr~s simple et il suffit de determiner les ~paisseurs
respectives des deux enroulements la constituant, n~cessaires pour assurer la tenue A la pression interne.
La m~thode de calcul sera pr~sent~e dans un paragraphe sp~cifique.
Les deux fonds posent Ie probl~me de conception de mani~re differente. En effet, il faut d~terminer simultan~ment la forme de chaque fond et la trajectoire des fils sur celle-ci de telle fa90n que ces fonds
soient r~alisables et optimis~s.
La figure ci-dessous illustre les notations pr~sent~es en debut de document et permet de pr~ciser comment Ie probleme se pose.
, . - - - - - - Meridien

Fil (r)

Partie cylindrique ~

Surface moyenne
,....---- du fond (S)

Figure 2
La forme du r~servoir est d~finie par la fonction p (z). La maniere de d~poser Ie fil au cours du bobinage, c'est-A-dire la ligne (f) est definie par la fonction a (z).
Le probleme ainsi pos~ comporte donc deux foncttons inconnues. De ce fait, deux conditions faisant
intervenir ces deux fonctions seront necessaires. Dans les chapitres suivants il apparaltra que ceci a
conduit A developper differentes methodes de conception.
3. CONCEPTION

Dans les methodes de conception no us separerons

- la partie bobinee cylindrique,
- la partie boblnee sur les fonds.

5-27
Nous retiendrons Ie terme "methode du filet" pour designer globalement les notions theoriques qui seront developpees ici. Dans Ie cadre de cette methode nous considererons que, en tout point, la structure
bobinee ne presentera raideur et resistance que dans la direction des fils a l'exclusion de toute autre
direction.

3.1 ANALYSE DE LA PARTIE CYLINDRIQUE

Compte tenu des notations nous pouvons ecrire
(1)

(2)

Avec

De plus, l'hypothese de base de la methode du filet entralne

(3)

La resultante de contrainte Nee s'exprime aisement par

(4)
En ce qui concerne l'enroulement satellite il faut considerer la figure 'presentee ci-dessous

tga - '

Figure 3
Compte tenu des notations il vient comme Ie montre cette figure :
N

tg ex

N tg ex

S rJ fs cos ex

(5 )

S rJfs

(6)

sin ex

De plus, il est possible d'exprimer l'epaisseur de composite de l'enroulement satellite par

es

=N x

(7)

cos ex

Les relations (5), (6) et (7) entralnent :

2

ex

(8)

X e s rJ fs sin 2 ex

(9)

X e s rJ fs cos

Les relations (1) et (2) conduisent en utilisant les expressions de

PR

-2- =

X e s rJ fs cos L ex

(R

N~s'

Nes'

N~c'

Nee a

po) ,

X e s rJ fs sin 2 ex + X e c rJ fc

PR

Ces deux relations permettent d'obtenir les expressions des contraintes dans les fils des deux enroulements en partie cylindrique :
rJ fs
rJ fc

PR

(10)

2 X e s cos 2 ex
PR
=--X e
c

(1 -

tg 2 ex
2

(11)

5-28
Ces derni~res relations sont tr~s utilis~es dans l'~tude de la partie cylindrique des rservoirs de
propulseur. Elles sont simples et permettent de dfinir rapidement les paisseurs de composite necessaires
pour obtenir un niveau de contrainte impos dans les fibres.
Remarques :
a - Si Ie bobinage
tel que

circonf~rentiel

est supprim, on dduit

tg 2 ex = 2

ais~ment

soi t ex = 54,7

que l'angle de bobinage doit etre

II sera expos ultrieurement que la valeur de l'angle de bobinage ex en partie cylindrique impose
les dimensions des ouvertures des fonds. De ce fait, pour prserver la libert~ g~omtrique sur ce point, il
est toujours n~cessaire de r~aliser un bobinage circonfrentiel.
b - Si les fils appartenant aux deux types d'enroulement sont soumis au meme niveau de contrainte on
dduit aisment A l'aide de (10) et (11)

..2=
e

3 cos 2 ex - 1

3.2 ANALYSE DES FONDS

3.2.1 EXPOSE DES PRINCIPES

Le probleme de d~finition des ~paisseurs en partie cylindrique a ~te ~tudi~ dans Ie paragraphe
prcedent. II reste A definir la forme des fonds et la ligne de dep3t des fils. II a ~t mis en evidence
dans Ie paragraphe 2.4 que deux conditions sont necessaires pour traiter Ie probleme ainsi pose.
Trois voies seront detaill~es. Leur pr~sentation se succ~dera de la meme maniere que leur utilisation
chronologique s'est faite. Ceci permettra de mettre en evidence les limites de chacune de ces voies, les
probl~mes rencontrs concr~tement et ce qui a justifie Ie passage A la m~thode suivante.
3.2.2 CONCEPTION "ISOTENSOIDE"

Cette conception est fonde sur les deux conditions suivantes :

- les fils seuls equilibrent les efforts dus A la pression interne (Methode du filet),
- la contrainte dans les fils est constante sur toute leur longueur.
3.2.2.1 EXPRESSION DE LA PREMIERE CONDITION

Les fonds

n'~tant

constitues que du bobinage satellite, les relations (8) et (9) conduisent A

Ne =
N$

tg2 ex

(12)

De plus, i1 est suppos~ que Ie reservoir n'est soumis qu'A des sollicitations d'extension
expressions de ~~ et N~ sont 1es suivantes (reference 1)

Par ailleurs, i1 est ais de

PR2

alors les

N~

--2-

(13)

Ne

R2
PR2
)
- 2 - - (2 - Rl

(14)

d~montrer

:
(15)

(16)

Ainsi, l'expression finale de 1a

premi~re

2 +

condition
R2
-Rl

pp
+ P ,2

(m~thode

tg

ex

du filet) est

soit,

tg 2 a

(17)

3.2.2.2 EXPRESSION DE LA DEUXIEME CONDITION

Cette deuxi~me condition impose que la contrainte dans Ie fil soit constante
contrainte se d~duit de la relation (8)~
(] fs = _ _~N~r.....",.-_
X e s cos 2 ex
Pour transformer cette expression Sous une forme convenable il faut
aeur locale de la paroi du r~servoir.

l'expression de cetce
(18)

pr~a1ablement d~terminer l'~pais

5-29
La relation (7) permet d'ecrire
= N _ _8, - _ , (en un point quelconque),

X cos a:

N _ _.::8_ _ , (a la jonction entre I' un des fonds et Ie cylindre).

X cos a: 0
De plus, on tient compte du fait que tous les fils qui coupent Ie parall~le
coupent aussi Ie parall~le passant par Ie point considere ; ceci se traduit par

la jonction fond-cylindre

On obtient ainsi la relation qui exprime l'epaisseur locale

p 0 cos a: 0
p

D'ou,

(19)

cos a:

l'aide de (13), (18) et (20)

(J

fs

'h+

..I:-"'-_
_
pp2 _...J,..;;...._"'p '2
2 X e so p 0 cos a: 0 cos a:

__

Imposer que la contrainte dans Ie fil est constante consiste donc

suivante :

(20)

ecrire la deuxi~me condition de la

mani~re

cos a:

(21)

3.2.2.3 DETERMINATION DE LA MERID1ENNE ET DE LA L01 DE BOB1NAGE

La meridienne du fond de surface (8) et la 10i de bobinage sont respectivement definies par les foncHons p (z) et a: (z).
Ces deux fonctions sont solutions du syst~me differentiel constitue des equations (17) et (21) rappelees iei :
pp
1 + P ,2

2 +

p2

VI

tg 2 a:

+ p.2

cos a:

Ces deux equations permettent tout d'abord de demontrer une relation simple.
La derivation par rapport a z de l'equation (21) conduit apr~s simplification a
2 p' + p' ---<;;,Pj;,P_- + p tg a
1 + P ,2

Tenant compte de l' expression de _~Pt:.p_ _

+ P ,2

a ' == 0 ;

deduit de (17), il vient

2 p' + p' (tg 2 a: - 2) + p tg a: a: '

p' sin a: +p cos a: a:'

soit,

0,
(22)

ou finalement : p sin a:

Cette relation entre l'angle de bobinage et Ie rayon local est tr~s simple et permet donc d'obtenir
dans ce cas particulier (conception isotenso~de) 1a 10i de bobinage lorsque Ie fond est defini geometriquement.
La relation (22), ceci sera demontre ulterieurement, est une propriete caracteristique des 1ignes geodesiques appartenant a une surface de revolution.
Ainsi, i1 a ete montre que Ie fait de concevoir un fond de reservoir tel que 1es fils seu1s equilibrent
1es efforts dus a 1a pression interne en etant sollicites par une contrainte constante entraine que ces
fils soient deposes suivant des geodesiques de la surface moyenne du reservoir.
Concernant 1a determination de la forme geometrique de 1a ligne moyenne des fonds d'un reservoir, les
deve10ppements mathematiques qui suivent permettent de preciser la fonction qui re1ie les coordonnees p et z.
Les relations (17) et (21) apr~s elimination de la variable a: conduisent a :
3 +

pp

(23)

1 + P ,2

Une autre forme peut se deduire :

La forme integree de cette

derni~re

relation s'ecrit comme suit

p 6 (1 + P ,2) = K2 P 2 + C (C

constante d'integration).

5-30
La

constante C peut etre evaluee en ecrivant la relation precedente pour z

(p

0'

O.

= 0).

Quelques manipulations mathematiques simples conduisent au resultat final

I' v
cos 0: 0 t 3 dt
z (u) = - p 0

} 1

'-In - t 2)

[cos 2

0:

(24)

t 2 (1 + t 2 ) - sin 2

0:

oj

La relation (24) permet done de connaltre la forme de chacun des fonds. La relation (22) fixe la loi de
bobinage.
3.2.2.4 MISE EN APPLICATION PRATIQUE
Un reservoir comporte deux fonds, done deux plans de jonction cylindre/fond.
La geodesique sur un cylindre est une helice, done l'angle de bobinage 0: constant, est le meme aces
deux jonctions.
02

0:

P Fl = P F2 = P

sin

0:

01

0:

Aux ouvertures des deux fonds 0: = 90.

La relation (22) entralne immediatement
0: 0

La premiere constatation est que la conception isotenso~de ne permet de realiser que des reservoirs
ayant des ouvertures polaires identiques.
De plus, la relation (17) montre que les meridiennes des fonds presentent un point d'inflexion pour
tel que :
tg 2 0: = 2

0:

En pratique, les regles suivantes sont appliquees

1. 0: 0 ~ 0: ~ 54,7 La relation (24) est utilisee pour definir la forme des fonds.
2. 54,7 ~ a < 90 . La forme du fond est spherique, son rayon est egal a la valeur prise par K2 au point
d'inflexion. La loi de bobinage est regie par 1a relation (22). La structure comporte un renfort metal1ique appe1e embase polaire (figure 1). En consequence la premiere condition (equation 12) n'est plus
verifiee et des methodes de calculs plus compliquees sont necessaires pour etudier le comportement de
cette zone. Le point d'inflexion est choisi pour situer le debut du renforcement metallique.
En conclusion du developpement de cette premiere voie il convient de souligner que l'on ne peut realiser que des reservoirs presentant des ouvertures egales. En pratique, pour des structures de propulseur,
l'optimisation globa1e conduit systematiquement a une petite ouverture a l'avant afin de diminuer 1a masse
du renfort metallique et a une grande ouverture a l'arriere du fait des dimensions de la tuyere.
En consequence, cette voie ne peut etre appliquee rigoureusement pour de telles structures. Par contre,
elle presente l'avantage de permettre des realisations simples et efficaces car la stabilite du fil est
parfaite 10rs du bobinage et la contrainte dans 1e fil est constante. De ce fait, nous rea1isons des petits
reservoirs de caracterisation fondes sur ces principes.
3.2.3 OlNCEPTION "PLANAIRE EQUILIBREE"
3.2.3.1 POSITION DU PROBLEME
11 a ete demontre que la conception isotenso~de ne permettait pas de realiser des structures bobinees
repondant aux necessites specifiques des propulseurs.
De plus, les unites de production disposent de machines a bobiner particulierement adaptees pour realiser des depots de fil suivant une trajectoire contenue dans un plan. De ce fait, les bureaux d'etud~s ont
developpe des methodes de conception fondees sur les deux conditions necessaires suivantes :
1. Les fils seu1s equilibrent 1es efforts dus a 1a pression interne,
2. La trajectoire des fils est contenue dans un plan.
La figure ci-dessous decrit de maniere simplifiee la configuration du fil et du mandrin.
Trace du plan
de bobinage (8)

Figure 4

5-31
II convient de noter que les formes des fonds et done leurs profondeurs n'~tant d~termin~es que lorsque
Ie probl~me est resolu, la longueur de la partie cylindrique n'est connue qu'en fin de calcul. Par contre,
l'angle dHinissant Ie plan dit "de bobinage" est d~termin par
tg y ~

P Fl + P F2
L

(p Fl' P F2, L sont des param~tres fixes au concepteur).

La premi~re condition que doivent verifier les grandeurs P (z) et a (z) est toujours exprimee par la
relation (17). La deuxi~me condition est obtenue en exprimant que la trajectoire du fil est planaire. Ceci
fait l'objet du paragraphe suivant.
3.2.3.2 EXPRESSION DE LA DEUXIEME CONDITION
La figure ci-dessous resume les grandeurs et leurs
l'expression de la condition recherchee.

repr~sentations

geomtriques qui interviendront dans

Cylindre

Surface moyenne
d'un fond

Trajectoire d'un fil

Plan de bobinage (B)

Meridien courant du fond

Figure 5
Des calculs longs, non detailles ici, permettent d'obtenir l'expression de l'angle de bobinage a par
tg2 a = [pp' - (z tg y + P e) tg Y ]2 + (tg 2 Y - p,2) [p 2 - (z tg y + p e)2 ]
(25)

(1

P ,2) [p 2 - (z tg y

P e) 2 ]

Cette equation constitue avec l'equation (17) Ie syst~me differentiel faisant intervenir p (z) et a (z).
de a conduit a l'equation differentielle du second ordre de la fonction p

L'~limination

[pp' - (z tg y + P e) tg y ]2 = (2 - tg 2 Y + 3 P ,2 + pp") [p 2 - (z tg y + p e)2]

(26)

3.2.3.3 RESOLUTION NUMERIQUE

La methode de Runge Kutta est employee pour integrer numeriquement ce type d'equation.
Afin d'utiliser cette methode, l'equation (26) du second ordre est remplacee par Ie syst~me differentiel du premier ordre suivant :
p' = u ,

u'

(27)

j [up - (z tg y + P e) tg Y ]2
p i p 2 - (z tg y + p e)2

= _1_

_ 3 u2 _ 2 + tg2 Y

Les deux fonctions p et u verifient done deux equations du type

f1 (u, P

f2 (u, P

z)
z)

I
)

(28)

532
Dans ces conditions, si 1es fonctions P et u ont des va1eurs P i et ui pour Z = zi' 1es va1eurs P i + 1
et lii + 1 pour Z = zi + 6 z sont d~termin~es en r~a1isant 1a suite des ca1cu1s suivants (reference 2).

P i

- p.' i 1

f2 (ui +_l_ U'iO 6 Z ,P i +_l_ P 'i O 6 Z , zi +~)

2

- P 'i 2
U'i 2

fl (ui + _1_ U'i 1 6 Z , P i +_._1_ P 'i 16 Z , zi +~)

2
2
2
f2 (ui

+ _1_
2

u'i1 t. Z , P i

Z , zi +~)
2

+.2.- P 'i 16
2

- P 'i 3

fl (ui + U'i 2 6 Z

2
P i +P'i t.z

zi +t. z)

U'i 3

f2 (ui + U'i 2 6 Z

2
P i +P'i t.Z

zi +6 z)

- P 1+1 = Pi +

......:..
6

+......:..
6

(P'iO + 2 P 'i 1

+ 2 p' i 2

+ P 'i 3)

(29)

(U'i O + 2 U'i 1 + 2 U'i 2 + U'i 3 )

Sachant que pour Z = 0, P = poet u = P '0 =

1e ca1cu1 peut etre commence et permet ensuite de determiner ainsi une suite de va1eurs discretes des fonctions P et u, et de 1a fonction a a l'aide de l'equation
(25).
La conception present~e ~tant fondee en particu1ier sur l'equation (17) 1a meridienne d'un fond pr~sen
tera de meme que pour 1a conception isotensolde un point d'inf1exion des que l'ang1e de bobinage atteindra
1a valeur de 54;7. Les regles de conception enoncees lors de 1a formulation isotensolde seront encore app1iquees en ce qui concerne 1a forme de la meridienne ; le bobinage sera par contre p1anaire au-dela du
point d'inflexion, c'est-a-dire dans 1a zone du renfort polaire meta11ique.
En pratique, 1a forme de 1a meridienne et la 10i de bobinage sont done determinees. par 1'integration
numerique decrite precedemment, depuis 1es plans de jonctions fonds/cylindre jusqu'aux ouvertures (a = 90).

Figure 6
Toutefois, il faut noter que l'equation (28) fait intervenir la grandeur P e qui positionne 1e plan de
bobinage (figure 6) ; cette grandeur devra etre determinee pour chaque fond de fa90n a obtenir l'ouverture
souhaitee. Get aspect du prob1eme est mis en evidence sur 1a figure ci-dessus.
Les grandeurs P Fl' P F2 ' L etant connues, 1es points A et B et l'angle y sont fixes.
S'imposer une valeur pour 1a grandeur Pe permet de placer, pour un fond, 1e point D qui represente le
debut de 1a meridienne pour le fond considere (p
est connu).
L'integration numerique permet de construire 1a meridienne jusqu'au point F ou l'ang1e de bobinage
prend 1a valeur de 90 ; ce point F n'est connu qu'en fin d'integration du systeme (27), (28). Le prob1eme
ne sera resolu que lorsque 1e point F colncidera avec 1e point B.
11 apparalt done qu'i1 faudra proceder. pour chaque fond, a une recherche de 1a valeur Pe qui conduise
a un point final colncidant respectivement avec A ou B. Gette recherche se fait de maniere iterative et, de
plus, automatisee lorsque 1a methode est programmee.

3.2.3.4 CONCLUSIONS SUR LA CONCEPTION "PLANAIRE EQUILIBREE"

Gette methode presente l'avantage, d'une part, d'aboutir a une conception tel1e que 1e composite ne
subit que des efforts qui sont equi1ibr@s par 1a tension des fils, d'autre part de presenter une methode de
bobinage particu1ierement bien adaptee a 1a cinematique de machines a bobiner simples et rapides dans leur
f onctionnement.

5-33
Toutefois, l'exp~rience a montr que, pendant ou apr~s la phase de bobinage, les fils peuvent, dans
certains cas, glisser. Ceci se traduit par une position d'~quilibre du fil diffrente de celIe qui est prvue dans Ie dossier de d~finition du produit.
Ce ph~nom~ne, lorsqu'il est trop accentue, entralne unediminution de la qualit de fabrication et une
alteration des performances de la structure.
Certains cas concrets ont conduit ~ prendre en compte ce phenom~ne de glissement d~s la phase de
conception; ceci a abouti ~ l'laboration d'une troisi~me methode de conception de reservoirs ~ ouvertures
polaires ingales (reference 3) ; celle-ci sera expose dans les paragraphes suivants.
3.2.4 CONCEPTION A BOBINAGE EQUILIBRE ET STABILITE CONTROLEE
Le titre de ce paragraphe indique que cette conception prend en compte, d'une part l'equilibrage des
forces dues ~ la pression interne, d'autre part, la stabilit~ du fil pendant ou apr~s la phase de bobinage.
La mise en evidence des parametres caracterisant cette stabilite fera l'objet du paragraphe suivant.
3.2.4.1 ANALYSE DE LA STABIL1TE DU F1L PENDANT LE BOB1NAGE
Pendant la phase de bobinage, Ie fil est dpose sur Ie mandrin avec une tension T
te les grandeurs qui interviennent dans ce qui suit.

la figure 7 presen-

Figure 7
+

Le mandrin exerce sur Ie fil, par unite de longueur de celui-ci, une force F qui peut s'ecrire sous la
forme :
+
F = A t

A,

+
1.1. n s

u bs

II se demontre aisement, en mettant en equilibre un element infiniment petit de fil, que les grandeurs
ant les expressions suivantes :

1.1., u

,(30)

- Tc cos '
u = -

(31)

Tc sin '

(32)
+

Sous l'effet de la tension T, Ie fil a tendance ~ glisser dans la direction du vecteur bs ' par ailleurs, cette meme tension "plaque" Ie fil sur Ie mandrin
De ce fait, la tendance au glissement est caracterisee par la grandeur :
k =

I-}- I = [ tg ' [.

(33)

Cette relation montre q~e l'angle ' est la grandeur qui caracterise la stabilite du bobinage ; en
consequence, dans la conception presentee ici, cet angle sera pris en compte pour exprimer la deuxi~me
condition necessaire pour determiner p (z) et a (z).
3.2.4.2 EXPRESSION DE tg '
L'expression de cette grandeur s'obtient de
intermediaires seront successivement prsentes.
~

Les composantes des vecteurs n s et t dans Ie

cos e

mani~re

rep~re

assez longue. Seuls, les principaux resultats

+

(x, y, z) ant les expressions suivantes

sin
D

(34)

5-34

-+

\
\

p' cos a cos

D

- sin a sin

e,

p' cos a sin

D

,+ sin a cos

(35)

cos a
D

Les expressions (35) ant ete obtenues en tenant compte de 1a relation

de

D tg a
p

dZ
11 est possible ensuite d'obtenir un vecteur

cos a cos
D

a' cos a sin

P cos a sin

dt

a' cos a cos

a' sin a
D
L'expression de tg

p' a' sin a cos

D

sin2 a cos
cos a

sin 2 a sin
cos a
p' p"
D3

-+

n en ca1cu1ant 1es composantes du vecteur

p' sin a sin

P

D
.--p

p' a' sin a sin

D

dz-

para11~le ~

(36)

p,2 p"

cos a

e,

cos

D3

sin a cos
+ p'

-p-

p,2 p

cos a sin

D3

(37)

e,

cos a

se deduit des relations (34) et (37) :

tg

(a' p cos a + p' sin a) (1 + p,2)

pp" cos 2 a - (1 + P ,2) sin 2 a

(38)

Remarque :
Une propriete caracteristique des courbes geodesiques d'une surface de revolution est que Ie vecteur
normal ~ 1a surface et le vecteur normal principal ~ la courbe sont confondus en tout point.
Elles verifient donc ~ = O.
La relation (38) indique donc
a' p cos a + p' sin a = 0
Soit : p sin a = cte
Ainsi est etab1ie l'affirmation enoncee dans 1e paragraphe 3.2.2.3. Ceci montre donc bien que le fait
de concevoir un reservoir dit "isotenso'ide" et concevoir ce mame reservoir en d1\posant les fils suivant des
lignes g1\od1\siques est equivalent.
3.2.4.3 DETERMINATION DE LA GEOMETRIE ET DE LA LOI DE BOBINAGE
La premiere condition etant inchangee, Ie probleme est resolu en integrant 1e systeme differentiel suivant :

2 + _...tP:..tP:--_
+ p ,2
(a' p cos a

+ p' sin 0:) (1 + p' 2)

pp" cos 2 a - (1

+ P ,2) sin 2 a

(39)
k

De plus, dans ce qui suit nous rechercherons des conceptions te11es que 1a tendance des fils ~ glisser
est constante sur toute leur longueur.
De mame que pr1\cedemment, nous resolvons le systeme du premier ordre suivant par la methode de Runge
Kutta :
'
p' = u
u'

1 + u2

(tg 2 a - 2)

(40)

a'

1
p cos

(2 k cos 2 a + u sin a )

Le systeme suivant est reso1u en utilisant des equations du marne type que les relations (29) etendues ~
trois fonctions f1 (u, p , a ), f2 (u, P , a ) , f3 (u, P , a ).
La premiere equation du systeme (39) montre que cette conception entralne de nouveau l'existence d'un
point d'inflexion lorsque l'angle de bobinage atteint la valeur de 54,7. Au-del~ de ce point 1e fond a une
forme spherique, le rayon etant egal ~ 1a valeur du deuxieme rayon de courbure en ce point et la 1igne de
depot du fil est geodesique assurant la continuite de l'angle de bobinage.
Enfin, l'angle e n1\cessaire pour assurer le reglage de 1a machine a bobiner s'obtient num1\riquement a
1'aide de 1a relation (36).

5-35
3.2.4.4 ETUDE DE LA PARTIE CYLINDRIQUE
La technique de bobinage avec une tendance au glissement constante peut
cylindrique. Dans ce cas la relation (38) s'ecrit :
et'

p cos et
sin 2 et

~tre utilis~e

pour la partie

Ainsi, en deux points M1 et M2 distants d'une longueur 1, 1es angles de bobinage et 1 et et 2

De

2 -

m~me

1 =

la variation de l'angle

+{

Log [

~l

v~rifient

kl

1
sin et 1

(41)
0

est obtenue en utilisant les relations (36) et (41) :

1
P
sin
J(~+

+ sin let 1 +

et 1

)2_ 1

1 + tg

- Log [ sin \

~ 1]

(42)

3.2.4.5 APPLICATION DU BODINAGE A TENDANCE AU GLISSEMENT CONSTANT

Lorsque le probl~me de concevoir'un nouveau reservoir se pose, les grandeurs p F1 ' P F2 et p 0 sont
imposees. De plus, il est necessaire de pouvoir se faire une opinion rapide sur la criticite des probl~mes
A resoudre. En effet, il faut pouvoir evaluer dans certains cas les risques technologiques et aussi ~tre
capable de prendre des decisions sur la nature du composite adapte au probl~me pose. Ceci a des consequences techniques et evidemment une incidence sur le prix du produit vendu. Evidemment, ceci necessite une
premi~re evaluation du probl~me n'exigeant pas une precision tr~s fine.
C'est pourquoi la methode presentee dans ces paragraphes et qui est employee dans nos fabrications industrielles, a ete traitee de telle mani~re que quelques param~tres essentiels peuvent ~tre evalues tr~s
rapidement.
A cet effet, nous presentons ci-dessous un diagramme commente plus bas:

o (degnis)
60
0,70

1--------l--------l-=::-----::::::l:-~==---=-""'1
0,65

0,60
0,55

0,45 1
r---=--r=-=~~0,50
0,40

1_____

0,35

1--===--"""':::-+-::::;;o-=='----::::;;o"""'r;:;=:'--:::::oo-=-::::;;0010.....~::::;;oo-=-::::;;oo""'1 0,30
0,25

.-'-::=:::_0<:::..-,,1

-0,2

-0,1

0,1

0,20
0,15
0,10

0,2

tg >It

Figure 8
Ce diagramme relie les grandeurs :
0 (angle de bobinage aux jonctions fonds/cylindre),
- p F/p 0 (ce que l'on appelle rapport d'ouverture),
- tg ~ (facteur de glissement).
Le trace de ce diagramme a et obtenu par l'utilisation systematique du programme resolvant numeriquement le syst~me (40). Les lignes apparaissant sur la figure 8 ont pu ainsi atre tracees point par point.
L'exemple suivant permet de constater comment ce diagramme peut atre utilise.
Soit A realiser un reservoir tel que
- et

p flip

0,50

(fond

p F2/p

0,25

(fond avant)

arri~re)

Si une trajectoire helico~dale est utilisee pour deposer les fils sur Ie cylindre, les angles de bobinage aux deux jonctions fond/cylindre sont egaux. De plus, il sera admis que pour obtenir le bobinage le
plus stable possible, il faudra proceder de telle sorte que les facteurs de glissement soient egaux sur les
deux fonds. Dans ces conditions, les points m1 et m2 representent donc les deux fonds sur le diagramme,
celui-ci permettant alors d'obtenir :
et

I tg

/I 23 0

'I'

I=

0,13 soit

'I'

7,4 0

5-36
Une autre possibilit@ consiste ~ choisir une loi de bobinage en partie cylindrique, qui pr@sente une
tendance au glissement. Dans ce cas, les angles de bobinage aux deux extremites du cylindre peuvent etre
diffl!!rents.
Alors, les points m'l et m'Z obtenus toujours en egalant la tendance au glissement sur les deux fonds,
reprl!!sentent une nouvelle solution du probl~me.
Dans Ie cas present, le facteur de glissement est diminue i
tg'l!~O,lO

soit

'l!~S,7

La stabilite du bobinage sur les fonds a donc @tl!! amelioree. En lisant des valeurs des angles de bobinage aux jonctions fond 9 /cylindre sur Ie diagramme il est aise de calculer Ie facteur de glissement sur Ie
cylindre ~ l'aide de la relation (41).
3.Z.4.6 ETUDE DE LA CONTRAINTE DANS LES FILS
La relation (ZO) qui donne l'expression g@nerale de la contrainte locale dans les fils peut s'ecrire
pZ

Vt+P'Z

f s = H -.<:----::C:':O:::S:-:Cl-"'--

(J

avec
H = -,,----P----Z X eso P 0 cos Cl 0

L'expression de

(J

fs

est ecrite ci-dessous

dz
d

(J

Z pp' D

fs

dz

+ pZ p' p"

cos Cl

La relation (17) permet d'eliminer p".

d

(J

Apr~s

fs

dz

= -

cos Cl.D

sin Cl 0:'
cos Z Cl

simplification il vient
ZP

H tg Cl D tg 'I'

(43)

En utilisant Ie principe d'egaliser les tendances au glissement sur les deux fonds, il apparalt sur Ie
diagramme de la figure 8 que
- tg'l' est negatif pour Ie fond ayant la grande ouverture (fond arriere),
- tg'l! est positif pour Ie fond ayant la petite ouverture (fond avant).
Compte tenu de cette remarque, il decoule donc de la relation (43) que la contrainte dans les fils augmente depuis la jonction fond/cylindre jusqu'au point d'inflexion pour Ie fond arriere et diminue pour Ie
fond avant.
De ce fait, sans prendre en compte les phenomenes de flexion que les theories developees jusqu'~ present ne peuvent prendre en compte, il apparalt que Ie fond arriere devra toujours posseder une epaisseur
superieure a celIe de l'autre fond pour tenir compte de la contrainte maximale atteinte au point
d'inflexion.
3.3 IDNCLUSION SUR LES METHODES DE CONCEPTION
Ces methodes de conception presentees sont aujourd'hui en fonctionnement et font l'objet d'une utilisation quasi quotidienne. Elles sont utilisables en mettant en oeuvre des materiels et des logiciels simples.
Les delais pour obtenir les resultats sont courts (de l'ordre de la journee). Les entrees et les sorties
sont d'un faible volume.
Les resultats sont complets puisque l'on peut obtenir principalement :
- la geometrie d'ensemble,
- les masses et volume de l'enveloppe bobinee,
- les parametres de bobinage des machines ~ bobiner,
- des parametres permettant d'avoir un avis sur les difficultes de realisation.
Toutefois, les hypotheses de ces methodes sont telles qu'une verification de la tenue de la structure
aux specifications doit etre faite pour s'assurer, en particulier, que les surcontraintes locales sont
admissibles.
De plus, les renforts metalliques doivent aussi faire l'objet de calculs que ne peuvent faire les methodes decrites.
Le paragraphe suivant decrira les methodes utilisees pour justifier de maniere definitive la structure
avant que des phases de realisation irreversibles ne soient commencees.
4. JUSTIFICATION
4.1 PRINCIPES DES METHODES DE JUSTIFICATION
La methode de travail consiste ~ utiliser un moyen de calcul fonde sur la technique des elements finis
(S.A.M.C.E.F. mis au point par Ie LTAS de l'Universite de Liege).
L'objet n'etant pas de decrire la methode des elements finis nous enoncerons les grands phenomenes pris
simultanement en compte pour assurer une etude du comportement de telles structures de la maniere la plus
realiste possible :
- Ie comportement non lineaire geometrique lie au fait que la pression interne, a l'equilibre, ne s'exerce
pas sur la geometrie initiale mais sur la geometrie deformee est necessaire pour obtenir des deplacements
et un tat de contrainte correspondant ~ la realite,

5-37
- l'anisotropie locale et variable suivant l'angle de bobinage, done Ie point considere de la structure,
est un fait dont il faut tenir compte en creant des fichiers de donnees representant les termes de la matrice de Hooke tout le long de la structure,
- des lois de comportement lasto-plastique sont necessaires pour relier de mani~re realiste les tenseurs
de contraintes et de deformations des embases polaires,
l'existence des jeux entre embases polaires, systemes d'obturation et leurs liaisons ne peut etre negligee et a conduit ~ developper des extensions du logiciel SAMCEF.
4.2 PRESENTATION DES PRINCIPAUX RESULTATS
Les resultats font l'objet de listings volumineux, nous extrairons de ceux-ci quelques planches
representatives.
1 - Contraintes dans les fils
Ces contraintes sont presentees sur des planches de type de la figure ci-dessous
EVOLUTION DES CONTRAINTES SUIVANT LE FIL, A P = 15,0 MPa.

Contraint_ - MPa

...

I""'" ~

1./

+ Sat. into

l-./

Sat. ext.
oCirc. into

.....
'000'

~l

10,

15.

N.

~,

35.

~~

~.

g,

ro

85

Figure 9
Cette figure indique l'evolution des contraintes depuis le milieu du cylindre jusqu'~ l'ouverture d'un
des deux fonds. Ces contraintes concernent les couches interne et externe de telle maniere que les phenomenes locaux soient mis en evidence. On cons tate principalement une evolution sensiblement constante excepte :
- ~ la jonction fond/cylindre,
- dans une zone proche de l'ouverture.
Seule la methode aux elements finis peut fournir les valeurs maximales atteintes et justifier le renforcement qui avait ete defini en se fondant sur les experiences passees.
2 - Contraintes dans les embases polaires
Plusieurs representations ont ete utili sees dans les annees pas sees pour illustrer un etat de contrainte bidimensionnel ~ partir de fichiers numeriques fournissant un tenseur de contrainte par element de la
modelisation.
Actuellement il est retenu de fournir au concepteur des figures en couleur definissant par zone les niveaux atteints. Le type de figure est represente ci-dessous

,
I

".52

14.81

18.$1

'l ;"
--.

C,.,., ...... 'nLe.

"'lU1_hl"~.S

d.. von 111"",

Figure 10
lei est dessinee une coupe radiale d'une embase polaire avant (petite ouverture).
On cons tate sur cet exemple que la zone la plus sollicitee est situee dans l'arrondi indique par la
fleche.
De multiples resultats sont obtenus et font l'objet d'une analyse precise conduisant soit ~ des modifications, soit ~ une decision de donner accord pour effectuer la realisation. Nous ne citerons que les principaux parametres critiques pour la tenue de la structure :
- cisaillement entre couches du bobinage (possibilite recemment mise au point),
- cisaillement des lames de caoutchouc assurant la liaison bobinage/embases polaires,
- deformee generale de la structure pour juger de son encombrement.

5-38

4.3 ANALYSE CRITIQUE DES METHODES DE CONCEPTION ET DE JUSTIFICATION

11 est de regle dans notre Socit de considrer que l'existence de ces deux mthodes est ncessaire.
La premiere possede les qualits de rapidit, d'efficacit et la capacit de permettre a leurs utilisateurs
de garder une connaissance intuitive en ce qui concerne l'influence des principaux parametres sur les rsultats obtenus. La deuxieme, plus longue d'emploi, plus chere est seule adapte pour permettre de tirer la
conclusion de pouvoir fabriquer la structure.
C'est pourquoi il est ncessaire de prserver un quilibre entre l'utilisation de ces deux mthodes que
nous considrons comme complmentaires.

5. CARACTERISATION DES MATERIAUX

Ce sujet mriterait a lui seul un document et une prsentation particuliere. Nous limiterons ici l'expos a quelques grands principes ayant pour objet de mettre en vidence l'importance de cette caractrisation dans le domaine de production industrielle.
La caractrisation peut etre dcompose en deux grandes catgories :
- travaux lis a la connaissance des proprits dans le but de permettre au Bureau d'Etude de raliser la
justification du produit a fabriquer en ayant les renseignements les plus complets possibles concernant
les caractristiques des matriaux constituant le bobinage et Ie composite.
Par exemple, le Bureau de Calcul a besoin de lois de comportement liant les tenseurs de contraintes et
dformations. Des .essais particuliers sont ncessaires pour dterminer les termes de la matrice de Hooke
si un comportement linaire est considr comme reprsentatif. Ce cas particulier, Ie plus simple qui puisse etre choisi, ncessite toutefois la dtermination de neuf coefficients.

Direction perpendiculaire ilia surface (5)

Direction des fils

Figure 11
Les relations s'crivent :

1""1

cJ 22
cJ 33
cJ 12
cJuJ
L cJ 23

Hll

Hl2
H22

0
0
0

HU
HZ3
H33

Gl2
Sym

0
0
0
0

GU

I
['
"
1
1]
E
E

22
33

Y 12
Y 13
Y 23

(44)

D'autres mthodes, dans les annes 70, partant des proprits des fils et de la rsine permettaient de
dterminer les lments de la relation (44). Alors la caractrisation des constituants tait necessaire.
Des lois specifiques ont t parfois necessaires lorsque Ie composite avait des relations ( cr , E ) non
linaires (degradation du composite, par microfracturation de la resine pendant la mise en pression
interne).
Par ailleurs, en plus du composite, la prsence de caoutchouc a ncessite des exprimentations permettant de relier la contrainte a la dformation. Des courbes du type suivant ont t utilises :

Figure 12
- La deuxieme catgorie concerne la caractrisation ncessaire pour assurer la conformit du produit ralis aux spcifications de rception des matriau::.
Les caracterisations sont multiples, nous ne dtaillerons ici que ce qui est attach au bobinage.
La fibre est achete a des fabricants qui livrent la matiere en quantites telles qu'elles appartiennent
a un lot de leur fabrication. Avant utilisation pour fabrication d'une structure bobine, des prouvettes
de conception semblable aux propulseurs sont ralises a chelle rduite (diametre 304 mm). Les prouvettes
sont soumises a rupture sous pression interne. Les valeurs obtenues permettent de decider si Ie matriau
peut ou non etre utilise.

5-39
Des essais moins onereux sont faits sur des meches unidirectionnelles soumises a des efforts de
traction. Toutefois, Ie premier type d'essai, plus cher, est plus representatif des sollicitations biaxiales (meridien, parallele) qui existent sur une structure reelle.
Des valeurs caracteristiques des performances des materiaux seront indiquees dans Ie paragraphe 8.

6. FABRICATION
Ce paragraphe sera consacre aux modes de fabrication de ce type de structure. Trois parties seront detaillees :
a - description des machines a bobiner,
b - description des principales phases de. fabrication,
c - commentaires particuliers sur la maniere d'obtenir Ie composite bobine.

6.1 DESCRIPTION DES MACHINES A BOBINER

Les machines a bobiner que nous utilisons sont de deux types, nous retiendrons les denominations suivantes :
- machine satellite verticale,
- machine horizon tale.

6.1.1 MACHINE SATELLITE VERTICALE

Le principe de fonctionnement d'une telle machine est represente par la figure ci-dessous

Mandrin

Point de sortie
des fils -+-----~

Figure 13
Le mandrin est fixe sur un support, 1'axe etant vertical. Un bras tournant a une vitesse Q supporte le
cheminement des fils et comporte a son extremite, un dispositif destine a assurer la sortie de ceux-ci
avant la depose sur Ie mandrin.
Si Ie mandrin est immobile, les fils apres plusieurs rotations du bras, seront deposes par superposition suivant une trajectoire p1anaire. C'est pourquoi une petite rotation du mandrin est necessaire afin de
provoquer un decalage de l'ensemble des fils (nappe) a l'issue d'un tour de rotation du bras. C'est ainsi
que l'on assure Ie recouvrement du mandrin. Lorsque l'ensemble de ces decalages represente 360, une couche du bobinage est constituee.
Le nombre de couches deposees est defini par l'epaisseur de composite necessaire pour respecter les
specifications de tenue (paragraphe 3).
La configuration de certaines structures est telle que la stabilite du bobinage planaire n'est pas
assuree. Pour resoudre ce probleme, la technique du "bobinage planaire modifie" est utilisee. Ce type de
bobinage est realise en faisant tourner le mandrin a une vitesse de rotation w La trajectoire de depot drr
fil n'est plus planaire. Lorsque le bras a fait un tour, la nappe ne se depose plus a cote de la precedente
depose. Le rapport des vitesses w et Q est choisi de telle sorte que la juxtaposition puisse se produire.
On obtient ainsi des bobinages etoiles.
Ce type de machine ne permet pas de respecter en toute rigueur les lois a (z) compatibles avec la forme
du mandrin definie par p (z). Toutefois, l'experience montre qu'il est possible d'approcher les lois
theoriques.
Par ailleurs, ces machines sont simples, rapides et fiables. De plus, elles existent et il convient de
les utiliser. C'est pourquoi ce type de machine est encore tres utilise. Nous verrons ulterieurement
qu'elles presentent toutefois des limitations en ce qui concerne Ie type de mise en oeuvre du composite au
moment du bobinage.

5-40
6.1. 2 MACHINE A BOBINER HORIZONTALE

Le principe de fonctionnement de telles machines est

repr~sente

par la figure ci-dessous

Mandrin

Bras de bobinage

Support

Figure 14
Dans Ie cas Ie plus simple, les mouvements permettant de realiser Ie depot de la nappe de fils sont
- une rotation du mandrin,
- un d~p1acement paral1~le a 1a direction z du support,
- un d~p1acement du bras de bobinage dans 1a direction y.
Donc, Ie point M symbolisant ici 1a t~te de bobinage d'ou sort 1a nappe de fils peut se dep1acer de ma-

ni~re quelconque dans un plan paral1~le a ~.

Des perfectionnements peuvent ~tre apportes dans un domaine particulier pour ame1iorer 1a qualite du
bobinage.
La nappe d~posee est, en pratique, un ensemble de meches qui elles-memes sont un ensemble de p1usieurs
milliers de filaments (la m~che est l'etat du mat~riau livre par 1e fabricant). De ce fait, 1a nappe qui
sort de 1a t~te de bobinage (M) a une 1argeur qui, en general, est environ 20 mm. Pour diminuer 1a duree de
cette phase de realisation i1 est envisage d'augmenter cette 1argeur jusqu'a environ 50 mm. 11 est inevitable que 1a nappe n'est pas naturellement p1anaire entre le point M et 1e point de tangence avec 1e mandrin.
Ce phenomene, s ' i1 est trop accentue, peut provoquer une mauvaise qua1it~ de bobinage. Afin d'y remedier,
i1 est d~ve10ppe actue11ement une t~te de bobinage possedant des degres de 1iberte de rotation autour du
point M. Ces degres de liberte seront programmes de m~meque 1es mouvements principaux decrits au d~but.
Ce type de machine permet de respecter parfaitement 1es lois de bobinage theoriques (a (z definies au
paragraphe 3. De plus, e11es permettent, contrairement aux machines vertica1es, de bobiner 1es structures
que11e que soit 1a mise en oeuvre du composite (paragraphe 6.3).

6.2 PRINCIPALES PHASES DE FABRICATION

La realisation des structures de propu1seur a pour objet de livrer au client une structure pr~te a ~tre
chargee de poudre et equip~e de 1a tuy~re avant de pouvoir devenir un etage propu1sif pr~t a ~tre assemble
avec 1e reste du vecteur.
De ce fait, 1a realisation se decompose comme suit :
- realisation des pi~ces primaires (cadres et embases po1aires),
realisation des protections thermiques internes assemblees aux embases polaires par drapage de caoutchouc
dans des moules specifiques. La vulcanisation est assuree par mise en etuve so us pression (autoclave),
realisation du mandrin par apport de sable et de liant dans des moules ~quipes des protections thermiques
protegees. Apr~s tassage du sable, ce1ui-ci est durci par mise des mou1es en etuve,
assemblage des diff~rentes parties ainsi constituees afin d'obtenir 1e mandrin pr~t au bobinage,
fixation du mandrin sur la machine a bobiner et bobinage,
po1ymerisation du composite en etuve (la realisation des jupettes peut se faire pendant la phase de bobinage ou apr~s 1a premiere po1ymerisation, dans ce dernier cas, un deuxi~me cycle de polym~risation est
nicessaire)
extraction du sable par adjonction d'eau ti~de provoquant la desagregation et l'evacuation du sable,
usinage des jupettes, fixation des cadres de jupettes et operation de finition,
controles geometriques et de zones critiques telles que 1es liaisons en particu1ier. Une phase importante
du contro1e est le timbrage de la structure a une pression interne ega1e a 1.15 fois la pression maxima1e
de fonctionnement.
Enfin intervient la livraison de 1a structure.
6.3 MISE EN OEUVRE DU COMPOSITE
Deux methodes sont uti1is~es. El1es concernent la maniere d'impregner les m~ches de fils avec la
resine.
La premi~re (methode W) consiste a impregner 1es m~ches pendant la phase de bobinage immediatement
avant 1e depot des fils sur 1e mandrin. Nous l'appel1erons technique WET.
La deuxi~me (methode P) consiste a impregner les m~ches avant la phase de bobinage (parfois p1usieurs
mois), soumettre les fils impregnes a un champ thermique pour provoquer un debut de po1ymerisation et ensuite stocker ce produit sur des bobines en congelateurs a - 18 0 C. Nous l ' ape11erons bobinage Prepreg .
Ces deux methodes presentent naturel1ement des avantages et des inconvenients.

5-41
a - Mthode It
Le procd est moins cher que la mthode P. Par contre la stabilit du fil pendant la phase de bobinage
est parfois faible.
b - Methode P
Ce procede permet d'obtenir une meilleure stabilit pendant la phase de bobinage. Par contre, il entralne une augmentation du COllt de la structure du fait du temps supplmentaire ncessaire pour raliser Ie
Prepreg.
Une autre source de surcout li A l'amlioration de la qualite est Ie contrale du produit avant son
utilisation.
Enfin, il faut noter (paragraphe 3.2.4) que, dans certaines configurations Ie Prepreg est necessaire
pour pouvoir bobiner la structure.
7. QUALIFICATION EXPERIMENTALE

En dehors d'actions exprimentales destines A caracteriser les materiaux utiliss pour raliser la
structure, une nouvelle structure fait l'objet d'un essai destructif permettant de connaltre la tenue ultime de l'objet.
Les jupettes sont soumises A des efforts mecaniques dtermins afin de representer au mieux l'tat de
contrainte Ie plus severe.
Le reste de la structure est soumis A un essai destructif, par application d'une pression interne gneree par apport d'eau dans la capacite bobinee.
De plus, un effort mcanique applique sur la liaison arriere represente Ie dlestage de poussee. La jupette avant est donc soumise A un effort de compression.
Ce type d'essai do it demontrer que la pression de rupture PR est telle que
PR;' K x

~F

(PMF = pression maximale de fonctionnement)

La structure est place en position verticale dans une fosse. De nombreuses mesures sont effectuees
aHn de :
- permettre d'expliquer comment la rupture a te initiee,
permettre au groupe calcul de comparer la prvision theorique A la ralit,
- donner au res pons able du chargement de la poudre les conditions aux limites concernant la dformation de
l'enveloppe bobine.
Ces mesures sont realisees A l'aide de capteurs de deplacements et de jauges de deformations.
Les zones particulierement instrumentes sont :
- Ie milieu de la partie cylindrique pour determiner la sollicitation dans l'enroulement circonferentiel,
- les liaisons jupettes/cylindre qui sont des zones de flexion,
- les zones du composite proche des embases polaires (zones de flexion),
- l'embase polaire arriere mtallique toujours optimise car lourde du fait de la grande ouverture.
Ges essais sont toujours longs du fait d'une instrumentation dense (250 voies de mesures pour les essais importants) et de la mise en oeuvre de moyens importants :
- moyens d'enregistrements ncessitant de raccorder chaque voie de mesure A un systeme d'acquisition de
donnees,
- moyens mecaniques lourds afin de garantir Ie respect des regles de scurit.
Enfin des enregistrements video et par camras rapides (~ 3000 images/secondes) sont systematiquement
faits, en couvrant au mieux l'integralit de la structure, pour permettre apres essai de revoir Ie droulement de la rupture.
La Fosse d'essai a une hauteur d'environ 8 metres et un diametre d'environ 7 metres.
La figure ci-dessous permet de preciser les notions rapidement evoques dans ce paragraphe.

Figure 15

5-42
8. RESULTATS
Nous prsenterons dans ce paragraphe quelques'rsultats permettant d'valuer
- les performances obtenues par bobinage compares a quelques mtaux,
l'volution des performances des produits bobins,
la performance spcifiquement attachee a la notion de structure de propulseur.
8.1 COMPARAISON COMPOSITE/METAL

Nous excluerons ici les performances pouvant etre indiques par les fournisseurs pour tablir ce type
de comparaison. En effet, les performances dpendent videmment des qualits intrinseques des matriaux,
mais tout aussi fortement de l'objet (gometrie, sollicitations) et de la qualit de la mise en oeuvre dtermine par la mise au point de bonnes resines, associes a de bons cycles thermiques, la qua1it du bobinage etc
De ce fait, Ie composite bobine valu ici, l'aura te a l'aide de contraintes a rupture obtenues sur
des prouvettes connues et dont il a te discut paragraphe S.
De plus, les contraintes prsentees sont obtenues en prenant en compte, nOn pas 1a section des fibres
seches, mais 1a section totale (fibre seche + resine d'imprgnation).
Enfin, Ie parametre retenu est cr /d (d = densite du composite).
Ainsi on obtient a ce jour
MATERIAUX

cr /d

Composite verre

"

Composite Kevlar

" 1400

900

" 1600 a 1800

Composite Carbone (IM6 Hercules)

Acier

"

200

Alliage lger

"

170

THane

"

220

cr

exprime en MPa.

L'int~ret du composite est tres important 51 l'on compare les valeurs du tableau ci-dessus.
De plus, on cons tate que les fibres successivement utilises : verre, kevlar et carbone ont apporte des
augmentations de performances sensibles. Sur ce point, il convient de faire apparaltre la notion d'effet
d'echelle.
On ne peut, l'exprience l'a montre, juger la performance uniquement sur les rsultats obtenus sur
prouvettes de petites dimensions (diametre" 300 mm).
En effet, il a et constat avec l'utilisation du verre et du kev1ar une diminution significative des
performances lorsque les dimensions de la structure augmentaient. Ce fait exprimental non expliqu par nos
services Etudes est mis en evidence sur Ie diagramme ci-dessous :

UR(MPa)

3000

...--...,..---r----,r----,
Y.

K- --

y.
K

2000

y.

(-11

K'

(- 23%)

1-----+---+--+----/
---1----1-----1----

1000

%)

----

Y : Yerre
K : Kevlar

1----1---+---+--;

L...--5.J0"':'0--1"':'oo.l:-O:---:1~50':':0::-~2~000

Figure 16

!i?(mm)

5-43
II est cons tate done respectivement
- 23 % de diminution (~ = 300 mm a q = 2000 mm) pour Ie kevlar,
11 % de diminution (~ = 300 mm a ~ = 2000 mm) pour Ie verre.
Ce phenom~fie a rendu beaucoup plus difficile l'obtention de definition respectant les specifications
lorsque ces materiaux etaient utilises.
De plus, les performances constatees lors des essais preliminaires d'evaluation a petites echelles
n'ayant pu etre obtenues sur l'objet reel, la masse inerte de la structure a ete augmentee.
En ce qui concerne la fibre carbone, ce phenom~ne semble ne pas exister aujourd'hui, ce qui augmente
encore l'interet de ce nouveau materiau.
8.2 FACTEUR DE PERFORMANCE DES STRUCTURES DE PRDPULSEURS
La grandeur suivante est frequemment utilisee dans"le monde des industriels realisant de telles structures ; elle est definie comme suit :
q

=-..EY.mg

p
V
m
g

Pression de rupture.
Volume interne.
Masse de la capacite bobinee.
Acceleration terrestre.
Ce facteur a la dimension d'une longueur. Pour les pays utilisant Ie syst~me metrique, elle est couramment exprimee en kilom~tres.
Ce facteur est effectivement assez representatif des criteres que souhaitent respecter les fabricants.
En effet, q est d'autant plus grand que
- p est grand (sans commentaire),
- Vest grand, ce qui correspond au souhait de pouvoir transporter la masse de poudre la plus grande
possible,
- m est petit, ce qui correspond a une masse inerte la plus petite possible.
L'augmentation de q represente done bien Ie souci d'apporter, pour un etage, sa meilleure contribution
a la portee du vecteur.
Toutefois, il convient de noter que la comparaison d'une structure a l'autre doit etre faite en prenant
garde au fait que la longueur- totale est imposee au fabricant de structure.
Or, a qualite egale de realisation une analyse rapide montre que, tout restant inchange, augmenter la
longueur entralne une augmentation du facteur de performance.
Ces remarques etant exprimees, nous presentons toutefois, a titre indicatif, les meilleures performances obtenues par notre societe en ce domaine.

(mm)

Diam~tre

Materiaux

1500
Verre

q (km)

2000

1150

Kevlar

Carbone

31

41

20

Le domaine des longueurs etant assez resserre, on peut done clairement constater les progres apportes
par l'evolution des fibres commercialisees.
8.3 INFORMATIONS SUR L'EVOLUTION FUTURE
En ce qui concerne la fibre carbone, deux grands fabricants proposent des fibres de plus en plus performantes : l'Americain HERCULES et Ie Japonais TORAYCA.
En ce qui concerne HERCULES, apr~s les fibres AS4 et AS6 cette societe commercia lise la fibre IM6
annoncee a :
crR = 5000 MPa (essai

m~che)

Une autre fibre existe et est encore peu commercialisee l'I}!7 annoncee a
crR
D'autres fibres au
En ce qui concerne
T800 et T800H qui sont
Plus recemment, ce

5500 MPa (essai

m~che)

stade laboratoire laissent esperer des 'niveaux plus eleves.

TORAYCA, apr~s la fibre T300 (classe AS4), cette societe commercialise les fibres
respectivement comparables aux fibres IM6 et IM7.
fabricant a annonce la fibre T1000 a
crR

7000 MPa (essai

m~che)

On cons tate done une evolution rapide des materiaux mis a la disposition des fabricants de structures
bobinees. Ce fait rend d'autant plus difficiles certaines decisions, que, lorsqu'un programme commence, il
faut toujours avoir a sa disposition les elements qui permettent l'utilisation des meilleures fibres,
celles-ci devant etre constamment evaluees et qualifiees pour application industrielle.

5-44

9. CONCLUSION
L'exp~rience des annes pass~es, en mati~re de conception et de justification, a t dtai11e dans ce
document (paragraphes 4 et 5). 11 est important de noter que ces mthodes ont t d~ve1oppes pour r~pondre
a des basoins bien precis concernant les structures de propu1seurs, dont 1es grandes 1ignes ont ete pr~sen
tees dans 1e paragraphs 2.
Nous insistons, car ceci est l'image d'un effort permanent de notre groupe Etude, sur 1e fait que nous
avons toujours vou1u pr~server 1a rapidite dans 1e domaine de 1a pure conception et 1a pr~cision dans 1a
justification.
11 convient aussi de faire apparaltre 1e souci constant d'imaginer des conceptions permettant a notre
groupe Production de fabriquer des structures de qua1it~ a moindre couto Les commentaires portant sur 1e
choix WET ou Prepreg et sur 1es prob1~mes de stabi1it~ de bobinage avaient pour but de mettre en 1umi~re de
te11es considerations.
Si 1es resu1tats presentes montrent de mani~re evidente l'interet du composite bobine par rapport aux
mat~riaux meta11iques, ceux-ci mettent aussi en e~idence 1es progr~s faits par 1es fabricants de fibres.
Actue11ement, 1es industrie1s doivent en permanence prendre en compte une evolution rap ide des performances
du carbone.
Enfin, la conception assiste par ordinateur, non traitee dans ce document car c'est un domaine dans
lequel 1es moyens se met tent progressivement en place, est un sujet tres important pour notre
Etablissement. II permet aux ex~cutants, d'etre decharges de travaux repetitifs pour mieux se consacrer a
l'optimisation de 1a structure. 11 faut signaler de plus que des moyens permettant Ie transfert automatique
d'~lements de dfinition a la production commencent a se mettre en place. En particulier, des 1a phase de
conception 1es parametres de bobinage sont stock~s sur des supports magnetiques qui sont ensuite directement uti1isab1es aux postes de commande des machines a bobiner. Ceci a pour effet de diminuer 1a maind'oeuvre done Ie cout, ainsi que les risques d'erreurs.
Actue11ement, nous considerons donc que des progres doivent etre faits dans l'automatisation pendant la
phase de conception - justification et pendant la phase de fabrication.
En conclusion, nous esperons avoir fait c1airement apparaltre que notre souci constant a t, en matiere de conception, d'imaginer des solutions prenant en compte non seulement la performance du produit mais
tout autant 1a qualite et l'efficacite pendant la phase de fabrication.

Rl!fl!rences :
[1] TIMOSIlENKO AND WOINOWSKI-KRIEGER - Theory of plates and shells.
(Second Edition) Mc Graw Hill Company - New York - 1959 - Page 435.
[2] HILDEBRAND - Introduction to numerical Analysis
Me Graw Hill Company - New York - 1956 - Pages 236 - 237.
[3] J.P. DENOST - New Design concepts for filament wound pressure vessel with unequal openings.
Societe Nationale Industrielle Aerospatia1e - 1982 - AlAA (Cleveland).

6-1

CONSID8RATIONS FOR DESIGNERS OF CASES

FOR SMALL SOLID PROP8LLANT ROCKET MOTORS.
by:

H. Badham Manager Rocket Motor Design & Development

G.P. Thorp Projects Manager
R.O. SUMMERFIELD
Kidderminster DY11 7RZ
Great Britain
(Operated by IMI Kynoch Limited
on behalf of Royal Ordnance pIc).

Summary
When considering the initial design of the case or load bearing components for a solid propellant rocket
motor a number of possible solutions may be apparent and many factors need consideration before the design
of the individual components can be finalised.
The process of optimisation involves consideration of material properties, methods of manufacture. inspection
and proof as well as interactions with other rocket motor and missile components. The various factors
are considered and indications are given of the interactions to be taken into account.
This paper concentrates primarily on the design of metallic motor cases. Both homogeneous and nonhomogeneous body structures are considered, the latter being fibre overwound metallic and strip laminate.
A relatively recent requirement is that for insensitive munitions and this factor is also discussed relative
to body construction.

1.

Introduction

The design of the hardware, that is the load bearing structure. of solid propellant rocket motors is dependent
on many interacting factors arising from the requirements and the overall motor design. Consequently it
is only possible to give concepts and guide lines and not precise design details. Only metallic structures
are considered in thiS paper, composite structures being the subject of a separate paper.
In hardware design an important aid for the design of the more complex components is finite element stress
analysis, a powerful tool allowing more optimum designs to be achieved and a reduction in testing. Comments
on the use of this technique is included.
The main factors to be considered when designing hardware are the requirements. and the material and manufacturing
methods: separate sections are developed to these aspects.
Rocket motor cases may be homogeneous or non-homogeneous structures and these are dealt with independently.
Other components which are common to both types of cases are considered under the former. Two non-homogeneous
case constructions are detailed, a composite overwound metaliic and strip laminate the latter being an
adhesively bonded structure.
A relatively recent requirement is for insensitive munitions and the fInal section of this paper is devoted
to this topic.
2.

Finite Element Stress Analysis

Finite element stress analysis is a powerfuL aid used extensively in rocket motor design for evaluating
stresses and straIns allowing the design of more optimised components which could require less testing.
The structure or continuum to be analysed is modelled as a set of elements of finite size connected
only at points called nodes. Each node has specified freedoms of translation and rotation in response
to the neighbouring elements or boundary conditions. The stresses are obtained from the field of
nodal displacements together with the appropriate constitutIve equations for the material properties.
Significant advances have been and are still being made In the basic computer stress analysis programmes,
some programmes being very general and others having been developed for a specific industry. Two
significant advances that have simplified stress analysis are programmes for element generation and
the display of results.
However sophisticated these suites of programmes, in general they can only analyse a structure; they
are design aids and cannot replace designers and stress analysts.
To optimise a component a design is first evolved usually based on a classical text book approach
coupled with general design expertese. The stress analyst will then define the elements which from
experience he considers best fit the component. The resuitant analysis will show areas of high or
low stress or strain. If it is considered that changes are necessary to the design then the designer
and stress analyst wIll decide from previous experience the modifications to be made prior to carrying
out a further analysis.

As programmes become simpler to use there is the temptation to design the ultimate component. When
deciding whether a design is satisfactory the limitations of the programme, limitations imposed by
machining accuracies and material properties and the possible advantages particularly in terms of
mass must all be taken into account.
More specific comments on finite element stress analysis are included in the following sections.
Programmes
Many programmes exist for finite element stress analysis and it is essential to attempt to evaluate
these to find the ones most suitable for the work required.
The use of any finite element programme requires a detailed knowledge of the capabilities of that
programme and it is usually best to obtaIn a detailed understanding of one programme. However If,
as is usual, a number of people in one organIsation are involved in stress analysis dffferent people
may be involved with different programmes, thus allowing the 'best' one to be used for any particular
application.
programmes for rocket motor design should be able
almost any geometry, complete anisotropy, various
differing materials in the same body and thermal,
Both two and three dimensional problems should be
to small strains or small deflections.

to deal with almost any stress problem featuring

material constitutive formulations, regions of
pressure, acceleration or point loadings in combination.
handled and the analysis should not be limited

For general rocket motor analysis it could be advantageous for the programme to deal with isotropic
materials haVing a Poisson's ratIo of 0.5. Many programmes do not have this capability, the results
becoming increasingly inaccurate as the value of 0.5 is approached. This limitation is probably because
a formulation of the governing equatIons is reqUired to deal with this special case.
A major finite element programme will offer about a hundred different element types ranging from
those with very general applications such as brick elements to special elements exemplified by those
that simulate a crack tip. The analysis tends to yield relatively inaccurate results if elements
exhibit long, thin aspect ratios or corner angles near to zero or 180. Usually these situations
can be avoided but difficulties can be experienced with long, thin, adhesive layers in bonded structures.
The two elements most frequently used for the analysis of components are the quadrilateral with its
degenerate form the triangle and the hexahedron with its degenerate form the triangular prism or
tetrahedron.
The quadrilateral is used as a plane element in either plane strain or plane stress or as an axisymmetric
element under aXisymmetric loading. For many applications an adequate initial solution may be obtained
by making an analysis using slight modifications to the geometry to bring it within the scope of
this element. Applications include motor bodies, end rings and reasonably axisymmetric end closures
in the aXisymmetric mode.
The hexahedron is used as a solid brick element In a three dimensional analysis. A particular application
would be a closure for a multi-nozzle motor or with non-symmetric features, and launch hooks.
Problem Idealisation
Idealisation is the process of representing a real body or structure in terms of its geometry, material
properties, loading and constraint by a mathematIcal model that is amenable to analysis by the finite-element
method. Very often, an insignificant modification to geometry, loading or support will result in a very
great reduction in the computing and data preparation times. The term "insignificant" depends on behaViour
under load and also on which aspect of the result is under investigation. On the other hand, the model
must possess adequate freedom or flexibility for the results to reflect the essentIal features of the real
situation. The idealisation process therefore requires judgement and experience. Application of the finite-element
method is therefore not an exact science.
Finite element analysis should never be used uncritically or without insight. It Is Vital to ensure that
the elements chosen do not lack a freedom that it is essential to take into account and that they are suffIciently
numerous for the model not to be significantly too stiff. Finite-element models in general err on the side
of being stiffer than the real item. Most finite-element analyses used in rocket motor design are applied
to a geometry of a type that has been analysed preViously and the expected behaViour is known. In the case
of a problem that is new to the user, it is good policy to do an initial run with a coarse division into
eiements to determine the general behaViour. The solution shOUld be examined with a view to understanding
the deformation, shown by plotting the deformed shape, and the stress output. It is possible for a soiution
to appear feasible and yet be quite wrong, so further checks shOUld be made. The solution should reflect
the loads and constraints that were intended in the lnput and, in the case of a compound body, the material
properties should be seen to be distributed as intended. It is usually a simple matter to check the load
balance in at least one direction by ensurIng that the reactions given in the solution are equal and opposite
to the total load that was intended to be applied in that direction. This check should never be omitted.
If further verification is desired, it may be possible to make a very rough approximation to the actual
problem and apply simple engineering formUlae. Having determined that the coarse solution is behaving correctly,
a finer diVision into elements may be made, using the results of the coarse solution to concentrate the
elements in regions in which any stress is changing relatively rapidly with position. The fineness of division
into elements depends upon the accuracy required and is a matter of experience backed, if necessary, by
trial.

6-3

The parameters required to specify the component are:

(i)
(ii)

(iii)

( iv)
(v)

Node numbers and co-ordinates. It is necessary to set up an ordered set of nodes before defining
elements.
Element types and topology. A set of nodes alone does not determine elements. It is necessary
to specify which element types are in use and in what way they connect with the nodes to form
the required model.
Material properties. Every element must be given at least two material properties.
Suppressions and constraints. It is necessary to stipulate which nodes have any freedom of movement
suppressed and in what way. Nodes may be constrained to move by a particular amount in a particular
direction or to relate their movements to those of other nodes. These constraints must be specified.
Load data. Elements that experience applied loads must be identified and the load types specified.
Load types include body forces, differential thermal expansion, pressure applied to element faces
and point forces applied to Individual nodes.

Programmes vary in their output. but stresses and nodal displacements are the fundamental quantities. At
present, strains are not usually given, but programmes are continually being updated. Graphical output
from which deformed plots may be obtained is virtually essential. Plots of stress contours are desirable,
but not essential. Some programmes give stresses only at interior points of the elements, whereas others
extrapolate out to the nodes, if reqUired, and output nodal stresses. Since stresses are frequently at
their maxima on boundaries, nodal stresses are reqUired.
3.

System Requirements

The main requirement in the design of any rocket motor is minimum mass and size consistent with the
design requirements coupled usually with minimum cost, requirements which are not necessarily comoatible.
The factors to be taken into account include,
(i)
(ii)
(iii)

(iv)
(v)

(vi)
(vii)

(Viii)

The internal conditions of pressure, temperature, gas velocity and time,

type of motor; integral, discrete or jettisonable,
mechanical environments, bump, shock, vibration and acceleration, bending moments, shear
forces and couples,
natural environments, which may range from tropical to arctic, dessert to saline, or high
altitude to underwater,
induced environments, particularly aerodynamic heating during air carriage, if relevant,
and flight,
interface with the missile,
insensitive munitions,
safety factors.

The internal conditions in a rocket motor vary considerably from design to design, an indication of
the range of the various parameters being as fol1ows:(i)
(ii)
(iii)
(iv)

pressure
temperature
velocity
time

2 to
2500
o to
1 to

25 MPa
to 3600 0 K
2500 mls
60 s

30 to 3750 psi
4500 to 6500 OR
o to 8000 ftls

In general rocket motor hardware is insulated from the hot gases to limit the temperature rise and
ensure that the ambient temperature properties are Virtually unaffected. Thus for the purposes of
hardware design internal temperature and velocity can generally be Ignored leaving pressure as the
only relevant parameter and allOWing components to be designed on their ambient temperature properties.
Occasionally, however uninsulated components are used in which case heat transfer calculations have
to be carrIed out and the properties evaluated at high temperatures.
The different types of motor are
(i)

(ii)

(iii)

integral, where the motor forms part of the load bearing structure, the case usually forming
part of the outer skin,
discrete, where the motor is housed within the missile and does not form part of the load
bearing structure,
jettisonable, where the motor is attached to the rear of the missile 'in tandem' or mounted
at the side 'wrap round' and is jettisoned when burning is complete.

Integral motors are subjected to the fUll rigours of the mechanical, natural and induced environments
while for the discrete motor many of the environments are alleviated. Tandem boosts form part of the
missile structure being cantilevered from the dart while wrap round motors have mounting points on
the side of the missile. The only alleViation for these motors compared to integral motors is in aerodynamic
heating during flight. Most current motors are either integral or jettisonable.
While the type of motor will influence the environment to which it is subjected, for example, the
vibration spectrum resulting from road transport wili be different for the different types of motors,
the design principles will be the same.

6-4

Other factors that will influence the environments are whether the missile is for use by the Army,
Navy or Air Force.
The Technical Requirement should specify the "full range of environments for the motor. The mechanical
environments should be directly related to the motor and should require no interpretation. With the
natural and induced environments it is essential to ensure that each is relevant, in some instances
interpretation is necessary: this is particularly relevant to the effect of soiar radiation and aerodynamic
heating.
The interfaces with the remainder of the missile includes the motor attachment points and fittings
for wings, launch hooks and cable conduits. While these are the missile contractor's responsibility
to define they can have a significant effect on the design and these attachment features should be
negotiable during development.
The requirement for insensitive munitions is becoming more and more important and is discussed in
section 7.
The safety factors used for rocket motor structural components are iow and can vary according to the
type of load.
Different factors and definit10ns are used in different countries. The minimum values recommended
in the UK are
(i)

(ii)
(iii)

1.5 for the repeated loads such as bending moments and shear forces arising from vibration,
shock and carried flight loads.
1.33 for pressure loads when failure would endanger personnel or the launch platforms,
1.25 for pressure loads in other instances:

for (ii) and (iii) the maximum pressure is defined as that pressure that will not be exceeded on
more than 1 firing in 1000 with 95% confidence at the upper firing temperature. In ali instances
the weakest case 1s considered, that is, minimum thickness and minimum strength.
Limited experience with requirements for contractors outside the UK indicates that similar factors
apply although definitions may vary.
To minimise the mass of the hardware components necessitates working as closely as possible to the
appropriate safety factors.
4.

Materials & Methods of Manufacture

Materials

The materials used for motor body manufacture can be divided into two groups, metallic and non-metallic.
The metallic materials used are generally either steel or aluminium alloys while non-metallic materials
take the form of plastic reinforced by fibres such as glass, carbon and aramid.
Table 1 gives typical properties for metallic and non-metallic materials with weight comparison factors
for pressure and for simple beam bending. Dealing first with metallic materials, figures are given
for two typical steel and two aluminium alloys plus one titanium alloy. Mass comparison factors for
pressure loading show that similar masses are possible with steel and aluminium alloy using stress
as the design criterion but lower masses are feasible with aluminium alloy when strain is the criterion.
The factors also show a significant advantage for Titanium using stress as the criterion but on the
basis of strain masses would be higher than for a steel component.
Now considering hoop bending as the criterion which is applicable to end closures then aluminium
alloys are superior to steel or titanium whether the design criterion is stress or strain.
For motor bodies where the criterion is usually stress then similar masses can be achieved with light
alloy or steel and other criteria have to be taken into account such as
(i)

(ii)
(iii)
(iv)
(v)

(Vi)

the economic manufacturing thickness,

internal volume
the thickness to diameter ratio where a value less than 0.005 is considered undesirable
on a damage criterion,
aerodynamic heating or subsequent high temperature processing which could affect the strength
of aluminium alloys,
possible stress corrosion with aluminium alloys "hich can be virtuaily eliminated by carefui
design in areas of changing section, for example, by the provision of large fillet radii,
external attachments.

When considering aerodynamic heating It ls necessary to consideT the possible effect of this on the
propeliant as well as the case. If the conditions are such that slgnificant propeliant heating occurs
either during carried flight on an aircraft or during missile flight, then this may be unacceptable
as it could lead to high pressures necessitating a thicker case. Repeated excursions to high temperatures,
as could occur during carried flight, could also significantly reduce propellant life. In either
case it may be necessary to increase insulation to limit the temperature rise in the propellant.
This could be done by external insulation which would also limit the temperature of the case.

6-5
Also included in table 1 are properties for fibre composite bodies these being based on information
for filament wound structures. The figures for stress and moduius are hoop composite values for a
balanced structure. Corresponding figures for an uniaxial structure would be approximately 50% higher.
The weight comparison factors for the pressure loading case clearly show the mass advantages possible
in the section of the body designed on pressure alone when using these materials. However many other
factors need to be considered in particular the loss in internal volume; as composite structure design
is the subject of a separate paper no further discussion of these structures will be attempted in
this paper.
Two materials which are showing improved properties are (i) aluminium lithium alloys which are stated
to have a 15% higher modulus and an 8% lower density making them particularly usefuL for comp0gents
where stiffness is required. (ii) fibre reinforced aluminium which has increased strength and ~tiffness.
Both materials require evaluation, the second in particular is expensive and is only likely to be
useful for very special applications.
Manufacture
Steel motor cases have been made by the following processes
(i)
(ii)
(iii)

(iv)
(v)

(vi)

wrapping and welding sheet of the desired thickness.

machining from forgings or extrusions,
deep drawing to the required thickness and machining the ends,
helically winding strip of the required thickness around a mandrel and welding aiong the
helical butt joint,
flow forming,
the strip laminate process,

but
method (i) is no longer used,
method (ii) can be costly in terms of wasted materials unless close tolerance forgings or extrusions
are available; in addition, if a large amount of material has to be removed distortion may occur
when machining,
method (iii) is well established although is more applicable to bodies with a low length to diameter
ratio,
method (iv) suffers from the disadvantage that the poisson effect on winding a relatively thick strip
causes bending of the strip across its width which can lead to difficulties when lining, while the
external surface may not be aerodynamically smooth,
method (v) is now well developed and is being used for a range of motor bodies,
method (vi) is also well proven and is particularly applicable where the body length to diameter
ratio exceeds four.
The strip laminate body is a bonded structure with additional design features to those for a homogeneous
case. The construction and design are given in section 6. This method is particularly useful where
insensitive munitions are required, this is discussed in section 7.
Aluminium alloy bodies have been made by processes (ii). (iii) and (v). As the strength of aluminium
alloys is greater in the direction of grain flow, flow forming gives the higher strength in the hoop
direction which is advantageous for a pressure vessel whereas other methods of manufacture give the
higher strength in the longitudinal direction.
A further method of manufacture that has been used in the UK is a fibre reinforced metal case, this
design is outlined in section 6.
The finai method chosen will be influenced by external attachments and loading conditions. As an
example wing roots could be welded to a simple body or the body, including wing roots, could be machined
from a extrusion. Where attachments are required over a short length of the body then this section
could be machined from an extrusion and welded to a simple tube prior to heat treatment. Whereas
if small or lightly loaded attachments are required then these could be bonded to a simple tube.
Only detailed discussions and costing will allow the most effective method of manufacture to be decermined.
For end closures and blast tubes the most common manufacturing method is forging and machining to
obtain good grain flow and optimum properties. Precision casting, although not as good on grain flow.
offers advantages in the reduced amount of machining and hence cost but addicional inspection is
necessary to ensure the high reliability necessary.
As closures and blast tubes have to be insulated the method of application of the insulation has
to be considered when deciding the best manufacturing route. Rigid insulants are invariably used,
these being filled, cured, resin systems. These insulants may be made separately and bonded into
the component or may be moulded in situ. The former method involves close tolerance machining of
both components prior to bonding under the application of heat and pressure. In this instance the
hardware components can be finish machined prior to application of the insulation and a gap filling
resin has to be used.

6~

When the insulation is moulded in situ the metal blank should be as simple as possible, to allow
it ~o seat in the mould tool, and final machining is carried out after moulding. With this method
it is essential to ensure that there is no significant strain on the components during the moulding
operation. This process is more applicable to aluminium than to steel components as the thermal coefficient
of expansion of aluminium is greater than that of the insulant whereas that of steel tends to be
lower. Thus on cool-down from the moulding temperature the aluminium tends to put the Liner in compression
whereas with steel there is a tendency for the liner to pull away although it is unlikely that a
gap will develop.
5.

Body Design - Homogeneous Metal Cases

For design purposes the motor body may be split into a number of sections as shown in fig leach
of which reacts with adjacent parts or components.
The design of the individual parts are considered in the following sub sections:
Parallel section
The five factors that have to be taken into account when designing the parallel sections of the motor
Cases are

(i)

(ii)
(iii)
(iv)
(v)

internal pressure,
external pressure,
attachments,
loads imparted to body by the missile i.e. bending moments, shear forces and attachment
loads,
stiffness requirements.

Excluding loads arising from attachments such as launcher hooks and wings the over-riding consideration
is usually the internal pressure and for an initial estimate of thickness the simple formula for
hoop stress in a thin walled pressure vessel is used. This wi1i indicate the materials that should
be considered for example for a small diameter or a low pressure motor an economic manufacturing
thickness for steel may not be possible leading to a detailed consideration of light alloy.
It is then necessary to carry out a detailed check on the maximum principal stresses taking into
account all loads and possible materials before the thickness can be finalised. The body is usually
an aXially symmetric structure and thus the principal stresses are in the hoop or axial direction
depending on the load combination. As already stated the hoop stress when pressurised is often the
design criterion; two other conditions that have occasionally governed the thickness are bending
moments when unpressurised, which can result in buckling, and stiffness. The thickness determined
for the body is usually consistent with that described as a thin wall structure. Occasionally this
is not the case and the appropriate formulae then have to be applied.
For short length to diameter bodies theories suggest that the thickened end sections interact to
reduce the maximum hoop stress, thus alloWing thinner sections to be used. However for bodies with
length to diameter ratios as low as 1 : 1 practical testing, inciuding burst tests, does not indicate
any significant effect from the end rings.
A partIcular case that requires more detailed investigation is that of underwater operation where
the external pressure in the unpressurised condition could result in buckling of a body designed
on the basis of internal pressure loads. The simplest approach is to design the thickness for the
external pressure. But as buckling can De prevented by internal circumferential ribs this offers
an alternative which can have a lower mass and provide a greater internal volume. This would lead
to complications in lining and propellant filling but provided these can be overcome a lighter structure
will result.

Attachment features such as launcher hooks and wing fixings require detailed consideration and could
influence the choice of material. for example, alluminium alloy could be acceptable for wing fixings
but steel could be necessary for launch hooks to withstand the mechanical environments. These features
may be incorporated into the body in a number of ways, examples being,
(i)

(ii)
(iii)
(iv)

the body including the attachment feature can be machined from a forging or extrusion,
the section of the body incorporating the attachment features can be machined from a forgIng
or extrusion which is then welded to the remainder of the tube. see fig 2a,
the features can be incorporated into separate rings which can be bonded to or clamped
onto a simple body as shown in figs 2b, 2c or 2d,.
individuai features can be welded or bonded to a simple body.

Method (i) would be applicable to a body incorporating say wing ring fixings which extend over a
high percentage of the length of the body and where the loads are too high for method (iv) to be
used.
Method (ii) could be used for shorter attachment features; it has been used for wing fIxings and
launcher hooks.
Method (iii) is a simple method of attaching relatively short features but does result in local increases
in diameter. To assemble the simple components, shown in fig 2b and 2c. necessitates the rings being
passed along the tube. If this is not possible then the assembly shown in 2d could be used.
Method (iv), like method (iii), can result in considerable simplification of the body tube but bonding,
in particular, is only suitable for iightly loaded components.

6-7
When launcher hooks or wing fixings are incorporated into a body as in methods (i) or (ii) local
thickening of the tube is usually necessary to withstand the additional loads imposed. The optimum
structure. that is one in which the loads result 'in a constant maximum stress in any radial location
will vary in thickness. The thickness will be a maximum in the neighbourhood of the load and a minimum
at a point farthest from the applied load taking into account the degree of machining and mass. In
deriving the optimum thicknesses throughout the ring initial estimates are carried out based on the
theory of curved beams followed by detailed stressing and modifications to the design until an acceptable
solution is obtained.
End rings
The main function of the end rings is to prOVide attachments for the motor end closure and for the
missile. The attachment features ineVitably require these end rings to be thicker than the main body
tube which. when a flush outside diameter is reqUired. necessitates the inside diameter being smaller
than that of the main tube.
The minimum inside diameter at one end of the tube at least will be governed by the charge manufacturing
method. The two basic types of charges are cartridge loaded and case bonded. With a cartridge loaded
charge the outside diameter must be slightly smaller than the inside diameter of the end ring hence
the inside diameter should be similar to the inside diameter of the body lining to ensure minimum
motor length. With a case bonded charge no such restriction exists; in this instance the diameter
would be governed by manufacturing considerations such as core extraction and ease of propellant
filling. In this instance ease of body lining is also an important criterion.
In addition to the pressure and missile loads the end rings must also withstand the bending moments
imparted by the end closures so there is an interaction between the design of these components. The
method of missile attachment reqUired is also an important factor in the design of the end ring.
Other factors such as access holes for igniter cables. safety and arming mechanisms have also to
be taken into account.
Again it is necessary to carry out an initial design based on classical methods and experience followed
by a detailed stress analysis and modification as necessary.
End closures
Three basic forms of end closure are possible
(i)

(ii)
(iii)

integral: where the closure and end closure/body are a continuous structure. In this instance
there may be a small central hole to house an igniter or to ease some aspect of motor manufacture,
see fig 3a.
part integral: in this instance the outer part of the closure is integral with the end
closure-body but the closure has a relatively large central hole. see fig 3b.
separate: where the closure is mechanically attached to the end ring, as shown in fig 3c.

An integral closure will generally be the lighter for any given material and hence with back extrusions,
forgings or for bodies with welded ends this type of closure can be used.
Part integral closures may be for nozzle or blast pipe attachment and are particularly relevant when
no missile attachment is necessary.
However important criteria for the choice of closure will be body lining and charge manufacture as
(i)
(ii)
(iii)

the lining of relatively long motors with sheet material is usually simplified with access
from both ends leading to the use of separate or part-integral closures.
a cartrIdge loaded charge requires a separate closure at one end for loading.
manufacture of a case bonded charge will necessitate a relatively large opening at one
end for ease of propellant filling and core removal so that a part integral or separate
closure is reqUired at one end: at the other end an integral closure may be used for an
HTPB charge but aCcess greatly simplifies the manufacture of CDB charges so that part integral
or separate closures are required.

For small diameter motors, that is up to about 150 mm, there will be little difference mass wise
between separate and part integral closures and body manufacturing considerations become the criteria.
When using a cartridge loaded charge with an integral closure this closure could be at either end
of the motor. However lining of the aft closure is a very critical operation and is more difficult
if this is the integral closure at the end of a long tube. For this reason it is recommended that
the forward closure should be the integral closure.
In general closures are designed so that motor pressure is on the concave side this generally resulting
in a lighter closure and lower bending moments on the end ring. But single chamber designs have been
successfully used with motor pressure on the convex side. This design concept generally arises from
interface considerations.
The operation of a dual chamber motor with a single internal closure may require the closure to be
designed to Withstand the pressure in the two chambers independently. If there is a significant difference
between the two operating pressures then consideration should be given to designing the closure such
that the higher pressure operates on the Concave side to reduce the effect of buckling.

6-8
The ideal shape for an end closure Eor a pressure vessel is a hemisphere but for maximum utilisation of
a volume a Elat plate is the optimum. However the latter Ls much heavier and imparts a significant bending
moment into the end ring. In practise a compromise between these two extremes is sought. From experience
it has been found that an end closure In the shape of an ellipsoid with a ratio of minor to major semi
axes of 0.4 gives a good compromise between mass and volume.
Due to the deformation of an end closure under pressure loads and the bending moments imparted to the body
the over-riding design criterion is usually strain rather than stress. As stiffness is proportional to
thickness cubed a light alloy closure will generaliy result in a lower mass structure and consequently
aluminIum alloys are often used when separate closures are required.
Closures for mUlti-nozzle side vented motors usually require to be very stiff and require special consideration.
It may be necessary to add ribs to the design to ensure the required stiffness.
Closure attachment
Many methods may be used for attaching end closures including
(i)
(ii)
(iii)

(Iv)
(v)

screwing the closure into the body, fig 4a.

locating the closure in the body and holding it with a
locating the closure in the body and holding It with a
on the outside of the snap ring which is bolted to the
and traps the snap ring as shown in fig 4c,
a lock wire in which a wire of suitable cross sectIon,
a slot in the end ring into matIng grooves in the end
bolting the closure to the body, as shown in fig 4e.

screwed ring, as in fig 4b,

snap ring. An additional ring is located
closure: this holds the closure in place
ductility and strength is passed through
ring and closure; fig 4d,

Interface conditions may influence the decision, for example if an internal screw thread is required for
attachment of the rest of the missile this could also be used for closure attachment (i) and (ii).
Method (i) is not recommended when '0' seals are used as the '0' seal has to rotate in relation to the
end ring: with this method it is also very difficult to align the end closure and body radially as is often
necessary.
With methods (i) and (ii) it is necessary to stop the screwed component from rotatIng under vibration,
while with methods (ii), (iii) and (iv) a location feature is necessary if the end closure has to be located
radially; method (v) gives a positive radial location.
For the lock wire method the wires deform as they are inserted. The wires have to be sized to ensure that
any deformation does not result in binding making for difficulty in assembly or scrapping of components.
A further design criterion is buckling of the wire as it is pushed into the slot. Withdrawal of the wire
requires special consideration. For development motors this is possible by making the lock wire longer
than necessary so that it protrudes from the motor allowing it to be withdrawn. An alternative method of
withdrawing the wire would be to make the slot in the end ring wide enough to allow the insertion of a
tool to push out the wire. This method of assembly, because of possible binding and pick up makes it more
SUitable for steel components than for light alloy although it should be applicable to a steel end ring
and a light alloy closure.
Method (v) requires a thick flange on the body and is also more applicable to large diameter motors.
For a given outside diameter of end ring methods (iii) and (iv) will give the maximum Inside diameter,
while method (v) will give the smallest. Taking the different factors into account then method (iii) is
that used mostly with aluminium alloys, while (iii) or (iv) are equaliy applicable Eor steel components.
1llast tubes
Blast tubes may be of two types
(i)
(ii)

supersonic: where the choke is at the forward end so that the gas flow is supersonic throughout
the tube while the pressure is relatively low
subsonic: where the choke is at the aft end so that the gas flow is subsonic throughout the tube,
but the pressure is relatively high.

Blast tubes require insulating, rigid insulants, that is, fibre filled resins, being the usual materials
employed. The insulating liner may be manufactured separately, machined and then bonded into the structural
component using a gap fiiling resin or it may be moulded dIrectly into the structural component. In either
Case the maximum strain aliowable in the liner has to be considered as well as the maximum stress in the
structural component before the thickness of the latter can be determined.
For the supersonic tube with its lower internal pressure then the maximum stress in the structural component
is likely to be the design criterion but for the subsonic tube with its higher internal pressure then strain
is more likely to be the over-riding consideration. In either case, however, on small diameter tubes a
thickness less than the minimum economic manufacturing thickness may be indicated in which case the lowest
density material available should be used.
Blast tubes may be desinged as separate components which are assembled to the end closure after manufacture
or they may be integral with the closure. The latter allows a one piece end closure/blast tube liner which
avoids a joint at the critical entry to the blast tube. This one piece construction is more applicable
to small diameter motors and relatively short blast tubes because of the problems involved in the manufacture
and handling of large components. Oniy a detailed consideration of manufacturing routes will aliow the
best solution to be determined for any particular appiication.

Nozzles
The majority of fixed nozzles, in use on current solid propellant rocket motors for tactical weapons,
have three main components, these being:
(i)

(ii)
(iii)

the outer load bearing shell,

the insulatIng lIner
the throat insert

Figure 5 shows two typical nozzle configuratIons.

It should be noted that under certaIn conditions it is possible to use a simplified structure, for
example a single-piece steel nozzle or a throat insert in a steel shell without an insulating layer.
Only a detailed knowledge of the motor and prope~lant parameters will enable the designer to assess
the appropriate design.
The Internal design of nozzles is dealt with in detail elsewhere and hence comments are made here
in relation to the materials and method of manufacture of the outer load bearing shell only. The
design, and materials used for the insulation and the throat sectIon would result from knowledge
of the propellant and ballistic requirements.
The outer load bearing shell is usually made of light alloy or steel but other materials such as
fibre-reinforced plastics may be used. The shell can be designed as an integral part of either the
motor end closure or the blast pipe or as an Independent unit attached to the closure or blast pipe
with appropriate sealing at the joint. Methods of attachment similar to those described for the closure
to body attachments can be employed.
In addition to the pressure loads from the motor, the outer shell will have to withstand any loads
specIfied in the Technical Requirement and possibly the loads imposed during manufacture, ego pressure
moulding of the insulatIon. Whenever possible the shell should be made in one piece although the
expansion cone may be attached separately.
As with blast pipes, the insulation, and in some instances the throat sectIon, can be mOUlded
into the shell or bonded in separately. When 'in situ' moulding is used, the internal profile
the shell is designed to facilitate retention of the liner during flring. If a pre-fabricated
is to be bonded into the shell, then a mechanical retention device will be required to ensure
the liner does not experience high tensional loads.

directly
of
liner
that

The high pressures associated with moulding 'in situ' will usually dictate the use of an over thick
shell which can be subsequently machined to the sizes requIred by motor and missile loads. Care must
be taken to ensure that the material does not self-relieve after machining, because as with the blast
pipe, the strain capability of the material may well be the over-riding criterion.
If a reinforced plastic is used for the load bearing member, it is usually in the form of an over-wind
onto the insulating layer, or in conjunction with a steel or aluminium alloy where the fibres can
be used as a hoop over-wind in the expansion cone to reduce structural mass.
Sealing of joints
Pressure seals are necessary between mating surfaces of the various components of the pressure vessel.
A common method is an '0' seal which is usually located in a groove on an external diameter although
where other considerations dictate a thick body attachment it may be placed in the inside diameter.
Face seals have also been used.
With screwed joints It is possible to rely on some thread sealant but the '0' seal is a more positive
method with a proven life in excess of 15 years.
The materials used for '0' seals are eiastomeric in nature. The material must be chosen carefully
to withstand the temperature extremes: the most SUitable materials to withstand the low temperature
for Army and Air Force use are silicone rubbers which remain flexible at low temperature. The precIse
size and grade depend on the diameter of the mating parts and the operating pressure of the motor.
Manufacturers recommendations for these and the tolerances on '0' ring grooves and the mating components
should be followed.
All joints should be leak tested at a low pressure, say between 0.2 and 0.7 MPa, 30 and 100 psi.
If an '0' seal holds this pressure It will seal against normal motor operating pressure. One method
of testing is to pressurise the motor and check for leaks. An alternatIve method which is used extensively
is to use double '0' seals, as shown in fig 6, and to leak test by applying pressure between them
and checking for a pressure drop over a period of about two minutes. This is a severe test which
detects very minute leaks because of the small volume of gas between the seals. After a successful
test the leak test hole is plugged and sealed to give added protectlon should the inner seal fail
although this plug itself cannot be leak tested. This technique is more expensive than a single '0'
seal as two grooves require machining and it is necessary to machine holes for the leak test Itself.
In addition each joint has to be tested individually but the ease of testing, as no account has to
be taken of the effect of pressure on the internal components of the motors, is considered to warrant
the additional expense.
The use of '0' seals with a threaded component has the disadvantage that when assembling there Is
movement between the component and the '0' seal and there is more llkellhood of damage on assembly.
When designIng the matIng surfaces it Is essential to ensure that there are no sharp corners whIch
can damage the '0' rings, in addition the minimum Length of mating surface is desirable to assist

6-10
assembly and to

ensu~e

minimum movement of the '0' seal in the

comp~essed

state.

Surface treatment
The majo~ity of metallic materials need to he p~otected f~om the environmental conditions which couid
be encountered during serVice life. Although most external su~faces are painted, this is not usually enough
to combat the effects of the elements such as salt sp~ay. acid ~ain and prolonged exposure to hot-wet
conditions as well as mechanical ab~asions resulting from handling. Hence a number of piating and deposition
procedures have been used on steel and aluminium components.
It is not the aim of this paper to discuss the details of individual p~ocesses, but more to highlight
the need for surface protection to the design engineer and the conside~ations to be used in deciding upon
the appropriate treatment.
For a number of years cadmium plating has b~~n the most commonly used protection for steel su~faces and
indeed this electroplating process is ~elatively simple. and hence low cost; it is also very effective
in te~ms of corrosion resistance. Howeve~. the process with high strength steel can prolnote hydrogen embrittiement.
it has a limited salt spray life and recent medical evidence has shown that cadmium is as toxic as mercury;
thus its use is likely to be severely restricted in the futu~e.
An alternative process is to employ an insoluble phosphate coating which protects by exclusion. whereas
cadmium protects sacrificially. Phosphating needs to be supplemented by paint or an o~ganic finlsh. but
can be applied in very thin coating layers and hence is advantageous in regions of high dimensional accuracy.
Other options include ion vapour deposition of aluminium. mechaniCal plating by dipping using combinations
of tin and zinc or cladding using aluminium and electroplating using tln and zinc. Short term protection
of aluminium alloys is provided by anodising and sealing.
Some materials, in particular stainless steels and titanium can be left unprotected, although as part
of the missile it is likely that they wili be painted. Testing of unpainted bodies in stainless steel.
which have been subjected to normal atmosphe~ic conditions in the UK for up to 3 years has shown no detrimentai
effects.
Testing of
Ali

structu~ai

Component~

COITlpOnents should undergo testing during both the development and

p~oduction

stages.

During development ali components shOUld be subjected to a routine pressu~e test. The component shouLd
be supported in a similar manne~ to that used in the final assembly so that loads are as ~epresentative
as possible. Fo~ example. the body should have end closu~es reproducing as closeiy as possible those of
the finished assembly whiie end closu~es should be held by simila~ retention devices to those used in
the moto~ assembly. The p~essu~e used should be higher than the maximum expected dU~ing moto~ firing:
in the UK a value 12.1/2% above the maximum pressu~e as defined in section 3 is suggested and this p~essure
is usually held fo~ 2 minutes.
In addition to routine testing it is also necessary to carry out additionai limited testing to simulate
other load conditions. using as app~opriate the p~essu~ised and unp~essurised conditions, and tests to
destruction to check that the app~opriate safety factors a~e achieved.
Production components should be subjected to routine p~essu~e testing in a similar manner to those for
development: in addition an occasional test to destruction may also be ca~ried out.
Durlng assembly all joints should be leak tested using a pressure of between 0.2 and 0.7 MPa that is 30
and 100 psi for two minutes or longer depending upon the volume being pressurised. As mentioned in the
section on sealing of joints the use of double '0' seals allows a very searching test to be carried out.
If it is not possible to use double seals it is then necessary to pressurise the motors and account must
be taken of othe~ components such as igniters and nozzle seals when fiXing the leak test p~essure.
6.

Non Homogeneous Cases

Two non homogeneous body constructions have been developed in the UK these being fibre reinforced metallic
and strip laminate. These are described in the following sub-sections:
Fibre-~einforced

metalilc rocket motor cases

In a rocket mutor body the hoop loads are usually higher than the axial r.he ratio generally being 2:1.
If the basic metal tube is designed r.o meer. the axial loads and a material with higher specific strength
is used to r.ake the additional hoop loads, significant mass savings can be effected. Such a material must
be easily applied to the external surface of the hasic metal tube such that the resultant mass saving
does nor. lncur a cost penalty.
Wires and fibres can easily be hoop wound onto a cylindricai tube. but mer.al wi~es are unlikely to pruvide
the additional specific strength and have hence been discounted. Fibres with high speciflc strength are
glass, carbon and aramid. With the glass and carbun fibre. a resin mar.rix is reqUired because of the reiatively
poor abrasion resistance and the need to t~ansfer some stress between fibres. The use of a resin system,
reduces winding speeds and inc~eases the mass of the overwind. Aramid fibres. such as Kevlar, which is
produced by Dupont. have greater abrasion resistance and nominal strength which is less dependent on fibre
length. thus allOWing the possibility of dry winding with the advantages of increased winding speeds and
reduced mass.

6-11
For each design, the metal thickness, winding tension and fibre thickness need to be evaluated. The first
of these is readily determined from the end loads and the material properties. The hoop loading will then
determine the reqUired overwinding by summing the capabilities of the two materials. The tension in the
f.ibre during winding will induce a compressive stress in the metal tube without problems associated with
fibre creep straLn relaxation if it is maintaLned below 30% of the ultimate stress.
This winding tension can be particularly beneficial when an aluminium alloy tube is used; the reason being
that the compressive strain accepted by the metal case raises the pressure at which permanent tensile hoop
straLn occurs.
More stringent design programmes have been developed, and analytical stress analyses for this type of case
construction have been developed by Royal Ordnance Research and Development Centre.
Unlike other methods of case construction it has been found advantageous to ensure that failure occurs
as a result of axial loads. This involves winding additional hoop layers which has little effect on mass
due to the low density of the fibre. If there are significant bending loads, then the thickness required
in the metal case may obviate the gains made by overWinding.
Typical masses for a 120 mm diameter cylindrical Case with a length of 1 m and designed for a burst pressure
of 40 MPa, without ends but with hoop: axial load ratio of 2:1 are:
Aluminium Alioy
Maraging Steel

2
2
(0.055GN/m , 80 000 lbf/in )
2
2
(2.0GN/m , 290 000 lbf/in )

Ai. Alloy overwound

4.33 kg
3.88 kg
2.74 kg

The above figures indic2te that a dry oVirwound aluminium alloy case prOVides a combined hoop stress of
approximately 0.760GN/m (110 000 lbf/in ) and a weight comparison figure of 0.50 for the simple stress
consideration for the pressure loading case' (Ref. Table 1).
Royal Ordnance have employed the system of fibre overwinding on an aluminium alloy case on one shoulderlaunched rocket motor currently in service and a second which is about to go into production.
The combination of a fibre overwind on an aluminium alloy case has led to considerable mass and cost savings
over the more conventional types of case manufacture. Where a maraging steel tube with a separate aluminium
alloy nozzle shell was replaced by a one piece aluminium alloy case with an integral nozzle shell and a
dry overwind there was a 30% mass saving and a 60% reduction in cost.
The work carried out to date has indicated that this method of construction can offer significant mass
savings without incurring a cost penalty. The plant required is relatively simple and hence initial tooling
costs are also kept to a minimum with the advantage that lightweight bodies can be produced very early
during development.
Strip laminate motor bodies
The strip laminate body is a bonded structure manufactured from metallic strip. It was designed and developed
at Royal Ordnance Summerfield and has been used for over 30 years for the manufacture of high strength
steel bodies for solid propellant rocket motors.
Motors using this type of body have shown specific advantages in the field of insensitive munitions: this
is duscussed in section 7.
The basic manufacture of a rocket motor body by the strip Laminate process is fairly simple, the operations
being outlined below.
Steel strip is degreased, shot blasted and degreased again. It is then coated with a suitable adhesive
which is gelled and the strip is coiled. The coated strip has a life of at least six months.
The steel strip is helLcally wound onto a heated mandrel so that there is a small gap between successive
turns of the heliX. Successive layers of strip are added until the desired thickness is obtained. Each
layer is wound in the same direction but the helices are staggered.
After winding is complete the adhesive is partially cured and the tube is removed from the mandrel and
cut to length. End fittings, and any other fittings which may be required, are bonded in place using a
jig to ensure accuracy and the assembly is fully cured.
The finished motor body may be lined, painted and filled in exactly the same way as a body made by any
other process.
Basically the equipment reqUired consists of a strip preparation and coating plant, a winding machine which
may be a converted lathe, a curing oven, and an assembly jig. The end fittings are assembled to the motor
tube in the flnish machined state and this applies also to any other fittings. This gives two of the advantages
of the process, namely, a wide range of supply sources for the body fittings and a minimum of manufacturing
scrap.
The process is capable of producing round stE~ight tubes with combined bow and ovality of iess than 10
on diameter.
on length and end squareness of less than 10
Providing strip is available then development bodies can be produced within 4 months.

-4

6-12

The principles of the construction are

i)
ii)
iii)

the hoop loads are taken by the layers of steel strip,

the longitudinal loads in the tube are transmitted by shear in the resin layers,
loads are transmitted from the tube to the end ring by shear in the resin layer.

The cross section of a three layer strip laminate tube and end rings including shear paths is shown in
fig. 7.
Design considerations for the body are given below: the comments made in section 5 re end closures and
closure attachments are equally relevant to strip laminate bodies.
Body tube
In most respects the mechanical properties of tubes made py the strip laminate process are identical to
those of homogeneous tubes.
For the strength and rigidity of the tube the usual principles of design are applied, based on the total
metal thickness of the tube wall, provided that the strip overlap pattern has been chosen so as to give
adequate longitudinal shear strength between the layers. This would only be a problem if narrow strip was
being considered. In general, the length of shear path in the tube is about one and a half times the length
of resin shear path in the end ring to tube bond, this being a consequence of the construction rather than
a design criterion.
The number of layers used should not be less than three, to avoid distortion under pressure at adjacent
strip edges and to obtain the maximum material strength, but may exceed this by any desired amount. However,
it is advantageous to make the number of layers as small as possible, consistent with reasonable strip
thickness, as the winding time is thus reduced.
The overlap pattern chosen depends on the number of layers of strip. For a three layer tube the pitch is
simply one third of the helix. For a four layer tube a pitch of one quarter has been use-d successfully
as have alternative pitches of half and five-eighths. An overlap pattern with alternate pitches of one-half
and three-quarter strip widths has been successfully used on a thirteen-layer tube. Experience indicates
that where the thickness demands a'large number of layers of strip, then there should be at least four
layers before the pattern repeats.
End rings
The end fittings, which may be in the form of attachment rings or complete closures, are provided with
cylindrical extensions to provide the necessary bonding surface. These extensions may be designed to fit
the strip-laminate tube internally or externally. Internal bonding has the advantages that a flush external
diameter can be provided. If a flush internal diameter is required, however, external bonding must be used
with a consequent local increase in external diameter. Centre rings or closures can also be provided to
join two tubes of the same or differing diameters or thicknesses together. Fig. 7 shows exampies of internally
and externally bonded end fittings.
Care must be taken when designing an end joint to ensure that stress concentrations at the end of the bond
are avoided so that no "peeling" effect is produced; this is particularly important with internally bonded
fittings. Attention must also be paid to the effect of any bending moments which may be present in the
end region; this effect normally takes the form of an additional shear stress, positive or negative according
to the position in the end bond.
In general the end ring bond is designed to be stronger than the tube so that the body wiii fail in hoop.
Attachments
The end rings normally provide attachment locations for other missile items, forward and aft of the motor.
Launcher feet or fin attachments can be incorporated into the end ring design or alternatively bonded directly
onto the motor body. They may be attached as individual components honded directly onto the tube or for
more highly stressed components attached by means of a complete ring which can be bonded onto the body
in the desired position, as shown in Fig 2b. Welding to strip laminate bodies is not possible without affecting
the strength.
In the finished condition the strip laminate body can be painted to suit the particular missile requirements,
in the same way as a conventional motor body.
Electrical continuity of the strip laminate motor body can be provided via appropriate application of silver
loaded resin or by a variety of mechanical links.
Materials
Strip
Most of the experience gained to date with this technique has been based on t2e use of carbon sZeel strip,
rolled and polish ground, having a design ultimate tensile strength of 2 GN/m , 290 000 lb.f/in . Recently
stainless steel strip of equivalent strength has also been quallfied, which obviously offers the additional
advantage of increased resistance to corrosion.

6-13
2
2
Aluminium alloy strip,heat-treated to give an ultimate tensile strength of 0.54 GN/m , 78 000 Ib.f/in ,
has also been successfully used and there is no theoretical limitation on the strip material provided that
it possesses sufficient ductility to enable it to be wound successfuily. Titanium alloys have also been
employed. The material comparison shown in table 1 is applicabie to strip laminate body tubes.
For cheapness it is desirable to use strip as wide and thick as possible as this reduces winding time.
However if excessively wide strip is used difficulties due to lateral curvature and gap variations between
successive turns of the helix arise. From practical experience the strip width should be less than the
tube diameter. The ease of winding a tube of a given diameter will also influence the maximum thickness
of the strip although in practise the design thickness is the over-riding criterion on smaller diameter
tubes.
Many of the bodies produced have been between 125 and 200 mm diameter, 5 and 8 in, with wall thicknesses
of 0.75 to 1 mm, 0.030 to 0.040 in. For these bodies strip 100 mm wide and 0.25 mm thick, 4 in wide and
0.010 in thick, has been used and has become a standard stock material. This strip has been used successfuiiy
for bodies up to 650 mm diameter with a thickness of 3.25 mm, 26 in diameter and 0.13 in thick, but wider
thicker strip could be used on a production motor of this size to reduce winding time and cost. For motors
significantly less than 125 mm diameter narrower and thinner strip would be used. However where the standard
strip can be used it leads to rapid manufactur.e of bodies for initiai development.
Adhesives
The adhesives used in strip laminate are based on epoxide resin chemistry this being the strongest practical
system available for production use. Different epoxide systems are used for strip and end rings2because
2
of the different methods of application but in either case resin strengths in excess of 30 MN/m , 5000 ibf/in
are achieved.

The strip laminate process is applicable to a wide range of sizes of rocket motor tubes. The main characteristics
of the tubes are:
2
2
high strength steeis are used, typically having a UTS of 2 GN/m , 290 000 ibf/in ,
the mechanical properties are identical to those for a homogeneous steel tube in the
same material,
-4
bow and ovality less than 12
on length,
3
end squareness less than 10
on diameter.

(1)
(ii)

(iii)

(iv)
In addition
(i)
(ii)
(111)

(iv)
7

changes to motor length can be easily made,

prototype cases can be produced very qUickly using stock size material,
,relatively simple manufacturing equipment is required,
the ability of the tube to rapidly delaminate under conditions of thermo-mechanical attack
attributes particularly favourable characteristics to this form of construction.

Insensitive Munitions

A relatively recent requirement for solid propellant rocket motors is that for insensitive munitions that
is, ideally, no reaction when subjected to fragment attack and fuel fire.
Whilst the rocket motor is not the only sensitive component of a missile it does represent the largest
single explosive component which makes it the most vulnerable particularly to fragment attack. Trials have
been carried out in the UK in which rocket motors have been subjected to both environments. The results
of these trials are given in the following sub-sections.
Fragment

attac~

Trials have been carried out on a range of propellants, extruded double base, (EDB), cast double base (CDB) ,
composite modified CDB (CMCDB), elastomer modified CDB (EMCDB) and hydroxy terminated polybutadiene (HTPB)
and case constructions, aluminium alloy and steel homogeneous constructIon, strip laminate (SL) and filament
wound carbon fibre reinforced plastic (CFRP). A number of different trials have been carried out including
attacks from 'in-service' rounds of ammunition which provide a fragment with a controlied veiocity and
direction and attacks using 105 mm pack howitzer shells electrically detonated at set distances from motors.
Not all of the combinations were tested but in general there was sufficient read across amongst trials
to show that the major Influences which affect the way in which a solid propeliant rocket motor will respond
to fragment attack are:i)
ii)
iii)

the propellant
the motor temperature and
the motor case material and constructIon

6-14
propellant
The physical properties of the' propellant allied to its latent energy level dictate how it will respond
when attacked.
A brittle propellant of the EOB and to some extent the CMCDB type is extensively damaged when struck by
a fragment. The propellant breaks into pieces which are easily ignited producing a large burning area which
leads to a pressure build up and inevitably an explosion when retained in a strong case.
A more flexible propellant of the COB, EMCOB and HTPB types generally remain intact on fragment attack.
A hole is made in the propellant which ignites and, having a relatively small burning area resuits in a
slower build up of pressure.
Motor temperature
The properties of some solid propellants vary with temperature, becomIng more brittle on thermal cooidown
and conversely more flexible as the temperature rises. Consequently, some propellants which are flexible
at ambient can react lIke brittle propellants at low temperature and will be more vulnerable to fragment
attack.
Motor case material and

~nstruction

Rocket motors using homogeneous cases whether of aluminium alloy or steel tend to explosion or pressure
burst when subjected to fragment attack, although with lower velocity fragments the reaction of aluminium
alloy cases may be reduced to burning only.
However the use of a strip laminate case reduces the effect signif.icantly, in most instances only burning,
or at low temperatures mild explosions, occurring.
Limited work with CFRP cases shows a similar reactlon to strip laminate.
Fuel fire
-----When subjected to fuel fire or external heating the main parameter which affects the response of a rocket
motor is the case. The auto-ignition temperature may affect the time at which the propeilant ignites but
the resultant effect once ignition occurs is similar.
The standard fuel fire test Is to suspend the motor over a bath of burning fuel having a flame temperature
of up to lOOOC. This test has been carried out on both homogenous, strip laminate and CFRP motors.
Homogeneous structures invariably exploded when the propellant ignited while with strip laminate bodies
the result was either propeliant burning or at worst a mild pressure burst. The limited trials with a CFRP
body tended to show marginally worse results than those with strip laminate bodies tending more towards
the mild pressure burst.
Mechanlsm of motor vulnerability under fragment attack and fuei fire
Fragment attack
When a fragment penetrates a rocket motor Case it usually punches a hole at the entry point and a somewhat
larger hole at the exit.
If the fragment ignites and breaks up the propellant the burning area increases and the pressure within
the motor will increase rapidly. Normally the nozzle, fragment entry and exit areas are insufficientiy
large to cater for this increased pressure, and high pressure failure, or even an explosion, will result.
With a strip laminate case, the area of damage at the fragment exit point is more extensive than it would
be in a conventional motor case, due to the strip delamination effect. When the propellant ignites and
the pressure increases the delamination of the strip continues, the pressure rise is restricted and the
motor either burns out or at worst a mild explosion results, the latter usually being restricted to low
temperature conditions with COB propellants.
With a CFRP case extensive damage results to the fibre construction significantly weakening the body and
allowing further delamination of ' the fibres as the pressure builds up again restricting the motor to burning
or a mild explosion, again the latter appears to be restricted to low temperature conditions with COB propellants.
Fuel fire attack
-------When a solid propellant rocket motor is heated externaliy, the propellant eventually reaches its self ignition
temperature and normally an explosion results. This is due to the uncontrolled nature of the burning producing
quantities of gas beyond that which the nozzle alone was designed to vent coupled with the fact that the
case strength is almost unaffected at this time.
Whilst thIs is the mechanism of vulnerability associated with homogeneous motor cases which retain their
strength above the self ignition temperature of the propellant, this is n~t so with the strip laminate
and composite cases. When subjected to heat the resin used for bonding begIns to lose strength at temperatures
above 130C, 270F, and at 2S0C, 480C, it has virtualiy no strength remaining. Most CDB propellants ignice
at approximately l80C but, as the propellant temperature wili lag behind that of the case, the resin temperature
will be higher and its strength will be significantly reduced. At the instant of propellant ignition only
a very low pressure is required to disrupC the motor case thus preventing a rapid pressure rise Limiting
the effect to one of a mild pressure burst.

6-15
Although the auto-ignition temperature for HTPS is higher than CDS the difference is not significant and
similar results are obtained.
Summary
The trials carried out on existing propellants and cases indicates that the degree of insensitivity for
existing rocket motor systems is dependent on both propellant type and case construction. A qualitative
assessment of both factors shows that existing systems may be classified as follows in their reaction to
fragment attack:
Propellant

Case

Most sensitive

EDB
CDS

Steel - homogeneous
AI. alloy - homogeneous

Least sensitive

EMCDB & HTPB

&

CMCDB

SL & CFRP

When considering fuel fire only the case appears to have a significant effect. the classification falls
generally into two groups

8.

Most sensitive

Homogeneous cases

Least sensitive

Strip Laminate & CFRP

Discussion

The optimisation of the design of the hardware for a solid propellant rocket motor requires consideration
of the technical requirement, mechanical and thermal environments and interfaces not only with the missile
but with other components of the rocket motor.
For metallic structures while a number of materials could be used. cost alone has generally restricted
those used to aluminium alloy or steel. The properties possible with these two materials are such that
similar masses can be obtained for components designed on stress alone and other considerations such as
economic manufacturing thickness, internal volume and method of manufacture become relevant in the choice.
In general for low pressure and small diameter motors aluminium alloys would be used for the motor case
while steel would be used for higher pressure and larger diameter motors. For components where strain could
be the design parameter, such as end closures, then aluminium alloy is the preferred material as this results
in a lighter structure.
The method of manufacture used must take into account the fact that a high reliability is required as component
failure could result in injury to personnel or the launch platform. This necessitates a high standard of
specification for the materiai and its manufacture. However improvements in both materials and manufacturing
methods are continuously being made which can improve the choice available to the designer. As examples,
castings have not been favoured for components other than those which are not highly stressed but improvements
in techniques are making casting increasingly attractive; the use of flow forming has allowed high strength,
maraging, steels to be used for body tubes and has resulted in a reduction in the economic manufacturing
thickness and improvements in welding techniques have further simplified manufacture.
For body tubes two non-homogeneous constructions have been outllned. The fibre over-wrapped metal tube
offers considerable potential for mass savings on long tubes when the main component of stress in the longitudinai
direction results from pressure. However as the ratio of longitudinal to hoop stress increases due to missile
loads the advantages of this system reduce. The second method of construction, strip laminate, has advantages
in terms of simplicity and ease of manufacture and is also particularly applicable to longer tubes. The
main advantage of this latter tube is for use in insensitive munitions.
The requirement for insensitive munitions could drive the design of future systems in terms of both propellant
and case. The best metallic case construction demonstrated to date is the strip laminate which reduces
the reaction of a motor under fragment attack and fuel fire to at worst a mild explosion. Fibre composite
cases also significantly attenuate the effect being very similar to strip laminate. No tests have been
carried out on the fibre over-wrapped case but this should be better than a homogeneous case although not
necessarily as good as strip laminate and fibre composite.

6-16

Weight comparisons

Rocket Motor Case.Materials

Pressure loading

Material

Density

e2

Kg/m

Stress

Modulus

f 2

GN/m

E 2

GN/m

~
\~t)/
Stress

f If

Strain
{' IE

Hoop bending

L--------___.7!

'1-----

Stress

el.,/t

Strain
~/.yE

Steel

7850

1. 40

200

1.00

1.00

1. 00

Steel

7850

2.00

200

0.70

1.00

0.84

1.00

Al. alloy

2700

0.45

80

1.07

0.86

0.61

0.47

Al. alloy

2700

0.65

80

0.74

0.86

0.50

0.47

Titanium

4600

1. 30

110

0.63

1.07

0.61

0.72

Glass '5'

2040

1.04

35

0.35

1.49

Carbon

1600

0.85

102

0.35

0.40

Aramid

1440

0.95

58

0.24

0.64

Table 1

Forward Closure

i'l~h","e

\ _Riuu

Body Tube

-Insulation

Figure 1

ROCKET

MOTOR.

1.00

6-17

Tube

I~
I

,t\___

--

=-= 1'==,

-~

--------

Attachments
<

---

- ----

--

...,. .... ~L,

'\weldS/
Fig 2a

Machined & Welded

=~--. ~-----r------

Gille

Line

Tu e

==_~_ ,,'-------'
-------1- --

Fig 2c

Fig 2b
Bonded

END

Fig 3a

Integral

Fig 2 d

Clamped Attachments

CLOSURE

Fig 3b
. Part Integral

Fig 3 c

Separate

6-18

''-Closure
Fig 4 IJ
Threaded Lock Ring

Fig 4a
Thread on Closure

CLOSURE

ATTACHMENT.

, - - Wire Assemhled
~-JhroI Closure

A-

''"--

-B

~-I--~--'

~" }-...,...__~;b,~~*~LL

Section B-B

Section A - A
Fig 4e
Bolts

Fig 4d
Lock Wire
CLOSURE

ATTACHMENT.

6-19

Bonded-in
Throat Assembly

Outer Load
Bearing Shell

-.~-__

Moulded in-Situ
Throat Assembly

Figure 5 TYPICAL NOZZLE CONFIGURATIONS .

./

'O/Seals ~

Figure 6

DOUBLE '0/

SEALS.

6-20

Adhesive Bonding Layer

--'

l:~:~~;~~"'~/H~
Internally Bonded

Externally Bonded
End Ring

End Ring

Molor Axill

---,--

Finll~.

STRIP LAMINATE BODY - SCHEMATIC

ARRANGEMENT.

7-1

PROPELLANT GRAIN DESIGN

S. Scippa
SNIA - BPD
Defense and Space Division
R&D Directorate
Colleferro, Italy

SUMMARY
This lecture reports a general overview of the current methodologies used at SNIA BPD in
designing sol id rocket motor propellant grain.
The Internal Bal I istics aspect in grain design is emphasized.
An overvi.ew of what are the general requirements (Thrust vs time curve, volumetric constraints,etc .. ), the avai lable technologies (propellant types, manufacturing processes, etc .. )
and the methodology for the selection of the most appropriate one is presented.
Then, the methods used for grain design are summarized.
The lecture deals, in particular, with the methods for propellant formulation choice, design of i nterna I geometry, prope I Iant character i zat ion (performances, i nterna I ba I list i cs,
etc ... ) .
An analysis of the predictabi I ity factors (various mechanisms that can affect the thrust
curve, i.e. :erosive burning, hump factor, etc ... ), and how they can be taken into account
for contingency planning is reported.
The experimenthal tests commonly used for grain design val idation are summarized.
GRAIN DESIGN OVERALL PROCESS
The main aim of the sol id propellant grain designer is to real ize a propellant grain that
wi I I evolve consistent with the thrust-time curve required for the mission, taking into account other specified constraints (envelope, weight, etc ... ).
These requ i rements prov i de the bas i s for estab I ish i ng the range of var i ab iii ty of the ma i n
parameters app I i cab Ie to the gra i n des i gn.
The main steps necessary for grain design are:
- evaluation of requirements
- prel iminary design optimization
- propellant selection
- selection and design of grain configuration
- prope I Iant character i zat i on and ta i lor i ng
- analysis of the design
A typical flow-chart for propellant grain design IS reported in fig.
1. Basically the
approach is that suggested in ref. 1.
The gra in des i gn is not rea I i zed i ndependen I y of other rocket motor components design: there is
a necessary inter~ction between grain design and other design areas (fig. 2).
The above mentioned steps in the grain design process are described more in detai I in the
following.
Evaluation of requirements
The requirements which define the basis for selection of grain geometry and propellant
type originate from motor performances specification. They include:
- Bal I istic performance Characteristics(Thrust, time, Total Impulse);
- Mission Requirements (Envelope, Weight, Environment-Rei iabi I ity).
The specified bal I istic performances shal I be evaluated for selecting appl icable grain
configuations.
The envelope and weight constraints define the geometrical motor design parameters
(length, diameter, chamber volume avai lable for the propellant, nozzle length and exit
diameter).
Also some environmental requirements (temperature range, ambient pressure, vibrations,
etc ... ) must be taken into account, because they may strongly influence the grain design.
Prel iminary design optimization
Some "dependent parameters" must be defined before the grain design process begins.
The most importants are:
- Average Operating Bressure (P), and the related MEOP
- Nozzle Throat Area (At) and Expansion Ratio (E)
- Volumetric Loading (VI)

7-2

- Web Fraction (Wf)

- Port-to-Throat Area Ratio (Ap/At)
- Length-to-Diameter Ratio (L/D)
In the prel iminary design phase it is necessary to quickly and accurately evaluate the
effects of these variables.
This study can be accompl ished by sol id rocket motor optimization computer programs
(Ref. 2).
These programs receive inputs from the mission and bal I istic requirements, with a fixed
design criterion (maximum total impulse, minimum propellant weight or length).
AI I the des i gns that sat i sfy the requ i rement s are compared and the motor that meets the
optimization criterion is selected.
Propellant selection

During the prel iminary design optimization,the required propellant properties are defined.
Although the ball istic(burning rate (r), temperature sensitivity (11)and performance
properties (Specific Impulse (Isp), Density (y) ) of the propellant usually are dominant
in the propellant selection, also the other characteristics (Mechanical Properties, Thermal
and Storage Stabi I ity, Hazard Properties, Exhaust Products, Cost) must be taken into account.
Once the propellant requirements have been establ iShed, the I ist of the candidate formulations can be reduced to those that require a minimum tai loring.
- Severa I fam iii es of sol i d prope II ants have been deve loped, character i zed and ut iii zed
at SNIA-SPD.
Among others they include:
Aluminized CTPS composites,
. Aluminized HTPS composites,
. Extruded double base.
For CTPS and HTPS propellants the Aluminum content ranges from 0.5 to 18 per cent and the
sol id loading (Aluminum plus Ammonium Perchlorate) reaches values up to 88 per cent.
Cast double base propellants were also studied and motor tested.
The deve lopment of new prope II ants is st i II under progress, ma i n Iyin the fie Id of a I um inized composites to obtain higher performances, burning rates over the actual range, reduced smoke compositions.
Selection and design of grain configuration
Configuration Selection.The grain configuration types most commonly used are:
- internal/external burning tube
- star
- wagon wheel
- dendrite
- slotted tube
- conocyl
- finocyl
- multiperforated
Combinations of these shapes can be used to satisfy some particular requirements (i .e.
dua I thrust Ieve Is) .
The configuration can be selected on the basis of:
- output of preliminary design optimization analysis - (Ballistic constraints: Wf, VI, L/D)
- thrust vs time shape
- processing technique
- structura I i ntegr i ty
The growth potential of the candidate solutions must be assessed.
Detailed grain design.Prel iminary calculations and parametric grain geometry studies are
sufficient to pre-dimension the grain.
Subsequent iterations typically are necessary for adjusting the grain dimensions to provide the required bal I istic performances with an adequate margin of safety for the propellant structural integrity.
A variety of Computer Programs are routinely uti I ized for geometrical trade-offs (for example between sliver fraction and burning neutrality) and preliminary structural analysis,
to ensure the designer of an optimum design.
Obviously, at the beginning, simplified schemes of the configuration are utilized in the
analysis.
These Programs are described in the paragraph deal ing with the analysis of the design.
In th i s phase of gra i n des i gn, Water tab Ie exper i ments can be ut iii zed for se Iect i on of gra i n
profi Ie.
A Water table Facility exists at SNIA-BPD to analize the internal Flow field in a RocKet
Motor' Chamber'.
Basically, the \Vater table simulates a 2-D Flow of an inviscid gas with '1=2, by a free sur-

7-3

face flow over a horizontal surface.

An hydraul ic analogy exists between the water depth and the gas pressure, so that pressure field in the gas flow can be determined by water depth measurements.
Pictures of the exper imenta I fac iii ty are sho"wn in fig. 20-2l.
"I i pyca I app I i cations are for ana liz i ng:
- the effects caused by aerodynamic interaction between the axial port flow and radial
slot flow;
- the f I ow pattern in the aft dome reg i on for different prope I Iant gra i n conf i gurat ions;
the effects of port sudden enlargement, with formation of recirculation zones and vortex
shedding, for selecting the grain configuration that promotes smooth gas flow in any transition region with consequent reduced probabi I ity of stagnation pressure loss and vortex
dr i ven pressure osc i I I at ions.
Propellant characterization and tai loring
Hav i ng se I ected the prope I I ant, it must be characterized completely, in order to define
theoret i ca I I y and/or exper i menta I I y its main properties (performance, internal ball istics,
mechanical properties).
Then. these bas i c propert i es can be adjusted or ta i lored ina var i ety of ways to sat i sfy
the specified requirements.
I n the fo I low i ng, a short descr i pt i on of techn i ques ut iii zed for Prope I I ant Performance
and I nterna I Ba I list i c Character i zat ion is reported.
Propellant performance characterization,The propellant theoretical Specific Impulse, under
the anticipated motor functioning condition (P, E), is evualuated by the NASA-Lewis EquiI ibrium Chemistry Computer Program.
The estimate of Del ivered Isp is based on a theoretical Impulse reduced by efficiency and
heat losses.
The methods uti I ized at SNIA BPD for predicting Del ivered Specific Impulse are described
in ref 3.
The exper i menta I va I ue of the" De livered Spec i fie I mpu I se can be obta i ned on I y from the analysis
of motor static firing tests results.
Scale testing of propellant by 2" motors shown that the measured Specific Impulse depends
upon the motor size, and the difference between values measured in smal I and large motor
is influenced also by aluminum content.
For this reason the smal I scale measured Isp is used only for correlations and propellants
comparison, not for performance prediction.
Propellant Density can be calculated from the density of the ingredients, and generally
coincides with valuesmeasured.
It is important, however, to calculate the density at the grain casting temperature in
dimensioning the casting mandrel, otherwise a deficiency in propellant weight can result.
Internal bal I istics characterization,Burning rate vs pressure and propellant temperature
measurements are real ized by:
- Crawford strand burners (useful for prel iminary screening and process control)
- 2"/6" standard ba I list i c test motors
Standard Motors firing tests are also uti ized to evaluate the propellant burning rate dispersion.
Scale-up of burning rate data for use in large motor analysis is based on correlations of
simi lar motors data.
Analysis of the design
The above described steps of grain design procedure provide the starting points for detailed design analysis.
In this phase of grain design, detai led calculations are performed, with grain dimensions
being subject to subsequent refinement.
Three main areas of analysis are involved:
- Combustion Geometry
- Internal Ball istics
- structural analysis
Combustion geometry.Computer Codes are uti I ized to calculate burning and wetted perimeters,
surfaces and port areas, scarting from grain initial profile.
These programs ~re either 2-D, 3-D or axisymmetric codes, and can be selected depending
on,the particular grain geometry.
The 2-D Computer Program (f i g. 3) is an in-house deve loped code. It can ana I yze star or
wagon wheel shapes, which are the most common 2-D cross-sections.
They can be described by a series of I ines and arcs.
The cross-sectional perimeter at various stations along the grain IS input, using the parameters defined in fig.4.
The module assumes a I inear change in perimeter and port aerea between stations analyzed
( fig S)and calculates surfaces and port areas history for selected increments of web burnback' and lengths along the grain.

7-4

An axisymmetric module (fig.3) developed by SNIA BPO, can be used for any grain which IS
symmetrical about the motor centerl ine.
The input is extremely sim~le.
A typical axisymmetric grain is shown in fig. 6.
The grain is described by inputting the R-Z coordinates shown in the figure, plus informations as to whether a particular segment is a stright I ine or arc.
A serie of R-Z coordinates is used to describe the motor case.
Surfaces and port areas are calculated using closed form analytical relationships.
The Program outputs the burning surface and port area vs web.
In theOriginal Version of the SPP (ref.4) the 3-D grain analysis is accomplished by a general ized 3-D grain design module (GO 3D).
An extensively revised version of this Program (fig.3) is at present uti I ized at SNIA BPO
FOR 3-D combustion geometry analysis.
Some of the features of this code are described below.
Basically, the code simulates drafting techniques to obtain propellant surface and port
areas vs web burn-back distance for virtually any grain design.
The motor case is assumed ful I of propellant initially.
Voids in the case are simulated by four basic figures (cones, cyl inders, prisms, or spheres) that may overlay and/or protrude outside the case if needed; they are used to simulate the void as it is initially. The figures may be normal outburning figures, grain fi lIed inburning figures, or nonburning figures, and may have rounded corners and edges if
needed. The order of input can be important. Each figure may be placed in any orientation
anywhere in space, ins i de or outs i de of the gra in.
Ordinari Iy, a sol id propellant is symmetric about the motor axis so that only one sector
of the symmetry is analyzed.
The answers are then automatically adjusted for the whole motor.
Examples of the four sol id figures used in the analysis are reported in fig. 7.
An example of the plotted output obtainable for a star-shaped grain is shown in fig. 8~13.
Segmented grains, such as are commonly used in large sol id rocket boosters, require special
attention to provide for fully or partially inhibited vertical slots. The nomenclature associated with a typical slotted grain design is shown in Fig. 14.
A slot face that is part i a I I y i nh i b i ted can be spec i f i ed by super i mpos i ng two cy I i ndr i ca I
disks that are identical except for radius. Each disk is an outward burning void, but the
larger disk has a fu I I y i nh i b i ted face. The growth of a vo i d can be contro I Ied by input
of a bound. Either or both faces can be partially inhibited. An example of a slot with a
partially inhibited aft face is shown in fig.14.
I nterna I ba II i st i cs I n the Interna I Ba I list i c Computer Program deve loped by SN I A BPD,
pressure and thrust histories are calculated considering the effect of mass addition,port
area change, and burn rate variation under quasi-steady state flow conditions.
The burn rate theory used is a mod i f i cat i on of the Leno i r and Rob i I I ard theory. Eros i ve
burning is considered to account for the significant effect of gas velocity and burn rate.
The program has two eros i ve burn i ng opt ions: Mod i f i ed and Standard Leno i r-Rob i I Iard Mode I .
The program can calculate stagnation pressure losses from sudden contractions or expansion
such as occur in segmented motors, uti I izing standard relationships, with the correction
for 20 and compressible flow effects.
Burning rate modifier tables may also be input and used in any element desired.
The governing gas dynamic differential equations are solved by a finite difference technique. The grain may, in accordance with the technique, be considered to be comprised of
nodes or elements of finite length. Pressure, mass flow rate, Mach number, velocity, temperature, and web burned are detemined at each node along the grain length for each time
increment. The resulting thrust, pressure and thrust impulse, throat diameter, exit diameter, amount of propellant burned, thrust coefficient, and del ivered specific impulse are
determined for each time increment. Inputs to the code include: propellant properties;
nozzle throat and exit diameter versus time; perimeter versus web; port area versus web
and wetted perimeter versus web, which may be varied independently for each node. Inputs
and outputs are i I Iustrated in fig. 15.
Outputs include the fundamental quantities of pressure, thrust, and time. Appropriate quantities are integrated with time; these are thrust, pressure, and propellant weight.
Burn rate and other quant it i es usefu lin ba I list i c ana I ys i s are a I so inc I uded in the output.
Motor weight and center of gravity with time are optional output.
A b Iock-d i agram of the over a I I methodo logy for Ba I list i c Performance Prev i s i on is shown
in fig. 16.
Oetai Is o~ the methodology are reported in ref. 3.
This methodology can be uti I ized for calculation of:
- nominal performance
- performance d i spers ion (30"')
The dispersion analysis must be real ized, taking into account:
-propellant characteristies dispersion (density, temperature, burning rate,characteristic
velocity)
- Thrust coefFicient dispersion

7-5

Throat erosion low variabi I ity

- mandre I mi sa I i gnement
- case aval ity and' other tolerances.
This, in order to verify in the worst case assumption (36"), the compl iance with motor
technical specifications.
Structura I ana I ys is. The pr i mary fin ite element code ut iii zed at SN I A BPD is an Ax i symmetric and Planar Structural Analysis code.
It is uti I ized with orthotropic, temperature dependent material properties.
I nput to the program cons i sts pr i c i pa I I y of:
- material properties (Modulus,Poisson's ratio, Coefficient of thermal expansion Vs temperature) .
- geometric configuration, in the form of a finite element mesh.
- loading conditions (thermal, pressure, force, acceleration).
Program output consists of:
- Nodal points displacements
Hoop stress/strain
- Radial stress/strain
- Axial stress/strain
- Shear stress/strain
- Max principal stress/strain
- Min stress/strain
- Max shear stress
- Angle of Max stress
In addition to the printed output, grid plots and isoplots are avai lable.

PREDI CTABI LI TV LI MITS

The basic predictabi I ity I imits have to be analyzed for contingency planning in the mandrel design and to generate the performance envelope.
One of the most widely recognized factors that is not entirely predictable is the erosive
burning.
Its effect on initial chamber pressure depends upon design characteristics, such as:
- Mass flux
- Port-to-throat area ratio
- Reference burning rate
- Complexity of grain geometry
- Binder system
- Chamber pressure
An ana I ys i s that wi I I accurate I y pred i ct eros I ve burn i ng is fundamenta I to succesfu I des i gn
analysis.
Such estimates are establ ished from correlations of actual performance data of simi lar motors.
Examination of large motor data, indicated that in these motors erosive burning did not
significatly affect the bal I istic performance, I imiting the port mass velocity.
For smaller motors, data have been analyzed to establ ish the nature of the erosive burning
exponent (~), used in the Leno i r and Rob i I I ard burn i ng rate theory.
These data indicated that the main parameters affecting pare:
- propellant formulation
- grain internal configuration
Correlations are used as a first estimate of the ~ to be used for a new motor design.
The SNIA BPD Internal Bal I istic Computer Program has two erosive burning options:
- LENOIR-ROBILLARD MODEL
r = apn+

0(

exp (- (3 p f'/G)

Dh 0,2
MODIFIED LENOIR-ROBILLARD MODEL
r = ap n + 0( GO, 8 exp (~
- I~ P riG)
f(Dh)
where f(Dh)= 0,9 + 0,189 Dh( 1+0,043 (Dh) ( 1+0,023 Dh)).
The hydraul ic diameter function attemps to compensate the above mentioned effect of motor
scale on erosive burning.
In many motors, both large and small, the actual trace shapes are initially lower that
theoretical, higher in the middle of burn, and lower again near web time.
At this phenomenon was given the name of BARF (~urning Anomaly Eate factor) or HUMP.
HUMP was found in the 1 m diameter Strap-on-Booster motors for ARIANE 3 and ARIANE 4
launchers (fig. 17-18).
It was also found in Thiokol, Aerojet, LPC motors (fig.19).
The method of quantifying the HUMP phenomenon is to Jetermine the correction factor that

7-6

must be appl ied to the burning rate low as a function of the web to correct the theoretical performance cO the measured value.
The mechanisms that can cause the anomaly are not wei I understood. One possible cause is
the motor f i I ling process.
The anomaly occurs in a quite repeatable manner for a given motor-propellant combination.
ILS effect can be eximated based on correlations of simi lar motors data.
Based upon the frequency of accurence of HUMP, particularly in
large motors, the mandrel
design should include the flexibi I ity to counter it.
Other parameters which could affect mandrel design are:
-predictabilityof Isp
- predi'ctabi I ity of nozzle throat erosion
-predictabi I ity of stagnation pressure drop, due to sudden contractions or expansion In
the port.
Thepredictabilityof Ispis about; 1%
The accurancy of pred i ctab iii ty of throat eros I on I s about 20%.
The impact of stagnat i on pressure losses due to any f Iow anoma I y I S genera I I y lim itated
to the first burning seconds with consequent not serious effects on performances.
Analytical grain design modifications can be undertaken to counteract the effects of the
assumed HUMP or Isp prediction inaccuracy and quantify the potential mandrel changes.
These changes wi I I be contained into the contingency planning.

MOTOR FIRING TEST

The experimental test commonly used for grain design val idation and margins verification
is the motor static firing test.
It can be rea I i zed ut iii zing the motor case rea I conf i gurat i on or an heavy motor case.
Furthermore, it can be or not "dureis" (i .e.: the pressure, can be increased to verify the
propellant integrity at the ignition).
Motor testing is done on static benchs both in atmosphere and in vacuum when is needed.
The fo I low i ng fac iii ties are ava i I ab Ie:
Static atmospheric benchs, of usual type with head chamber pressure and thrust versus
time staRdard measurements;
Six components static bench, for booster motors up to 9.5 propellant tons;
I SA vacuum test fac iii ty, whose character i st i cs are descr i bed in ref. 5.
A vert i ca I stat i c bench for Ar i ane boosters has been bu i Id up and successfu I I y tested.
From the ana I ys i s of the recorded pressure and thrust curves vs time, ut iii zing a computerized procedure set-up at SNIA-BPD (fig22), it is possible to obtain all the ballistic
paremeters, as:
- Burning Rate Scale Factor
- Hump Factor vs Web
- Erosive Burning Coefficient
- Average Characteristic Velocity
Throat Area vs Time
The experimental traces, t03ether with these data, can be compared with the theoretical
va lues, ca I cu I-ated dur i ng the des i gn phase, in order to:
- ver i fy the comp Ii ance wit:, motor tech inca I spec i f i cat ions;
- ver i fy the marg ins, tak i ng into account the performance d i spers i on (36");
- i dent i fy if prope I I ant gra i n des i gn changes are necessary.

COMPARISON WITH EXEPERIMENTAL RESULTS

The discussed gra i n des i gn and i nterna I ba I list i cs ana I ys i s procedure has been app lied to:
- sol id rocket motors for mi I itary appl ication
- sol id rocket boost motors
The main characteristics of several SNIA BPD motors are reported In table 1. The numbers
after the propellant binder identification indicate in the order the Aluminum and binder
per cent.
Motors of very different 'propellant Aluminum content, dimensions and nozzle expansion ratios are included and constitut3 an acceptable base for testing a grain design and a performance prediction methodology.
The good agreement between predicted and del ivered performance is confirmed by internal
ba I list i c ana Iys i s resu Its shown i n JI'~.23-37 where computed traces are compared with exper imental traces envelopes' (e~, 6-1).

CONCLUDING REMARKS
An attempt has been made in this paper to report a general review of the current methodolog i es used at SN I A BPD in the area of des i gn so lid rocket motor prope I I ant gra i ns.
Examples have been used to demostrate the existence of a consistent data base for grain
design (and, more generally, performance prediction) methodology val idation.

7-7
At present, several improvements in the described methodology are in progress. ParticuI ar I y, efforts are devoted in the area of .:.onlputer .1utomated .Qes i gn (CAD) to aid the analyst to procedure and analyze virtually any grain geometry with high accuracy and low
time expense.

REFERENCES
1. Brooks, W.T. : "Sol id Propellant Grain Design and Internal Ball istics", Nasa SP-8076,
Lewis Research Center, Cleveland, Ohio, March 1972.
2. Haymes, W.G., et al.:"Sol id Rocket Motor Design Automation Technology", AGARD
July 1979.
3. Var i ous, "Performance of Rocket Motors with meta I I i zed Prope I I ants",
September 1986.

CP-259,

AGARD AR-230,

4. Nickerson, G.R., et al.: "A Computer Program for the Prediction of Sol id Propellant
Rocket Motor Performance" .AFRPL-TR75-36. July 1975.

5. Solfanelli, G, et al.: "A New Simulated Altitude FacilJtyfor Space Motor Tests", Paper
No 82-11G9, AIAA/SAE/ASME 18th Joint Propulsion Conference, Cleveland, Ohio, June 1982.
6. De Amicis, R., et al.: "Performance Prediction of Apogee Boost Motors and Correlation
with Vacuum Firing Tests and Flight Results", Paper No. 79-1296, AIAA/SAE/ASME 15th
Joint Propulsion Conference, Las Vegas, Nevada, June 1979.
7. Mura, M. and Vari, E.: "Ariane 3 European Launcher Strap-on Booster Development, Quail
fication and FI ight", AIAA/ASME/SAE/ASEE 21st Joint Propulsion Conference, Monterey,
California, July 1985.

7-8
SOLID PROPEllANT GI~AIN OESIGN PROCESS

FIG.l -

.,
I

I I

II
I I

I
I

I I I

I I

I I I
_J

-- I
I

I
I

I
I
I
_____ .--J

-.fItoo NORMAL flOW

___ ~ ITERATION

-,

I
I
L

I
~

I
1

I
I
I
I

I
I
I

_ _ _ _ _ _ _ _ _ _ _ _ _ -lI

FIG.2

SOLID PROPELLANT MOTOR llESIGN ANn

DEVELOPMENT PI1DCESS

7-9

1',; I AltiCe J o!cro:! J

ANI\lYflCAl

CRAIN INITIAL
GEO.4ETRY

1--.

..

DENS J TV

OlJTPlJ

STAll/GelD

I NI'IlT$

'

rECIlNIQlJE

FOI~

GRA 1N

REGRESS I ON CALCULATION
STAI~

rs

p,

PO

SHAPED ANI) AXISYMMETRICAl GRAINS

"

WEB

"ARJAAL[ CIIOSS SECTION

PIWPElLANT

OUTPUTS

GEOJ

INPUTS

THREE DIMENSIONAL GRAIN

[VOLUTION OF smWEO PROPELlANT

VOLUMES

INPur

Ph

Pw

WEB

FIGURE~

F1G.4 - GllAIN CIIOSS-SECTIONAL

INPUT PAnAMUERS

(2-0 COD)

STAR SlIhrE

!l
I

I ,J!

,;

7/'<"

wE IGilT

A"

7-10

~~I)\Nf'IIT~(lI!On

ftC

CI10SS

SECT10HS f\-,\

ANO I"l-Il

(DESCJUBED \."ITII LINES Mn "'~CS)

B-B

8-0

(;RAIN DEscnI3ED

/1,",[T1l

/;

Z.

CIRCLE SPECIFIED

R POINTS

BETWEEN

/;

2 POINTS

i---T---i-/----!--..\

:J

'--

-..J

.. z
FIGURE 6. 1\;( 15YMMfTR I C I NPIIT METHOD

FIG. 1 - SOLID FIGURES

{3 -

0 CODEl

c..m..",

~.In,jer",

(] ..........

(J)

,,'.

"

o::==::=:J

tr::==

""1'"

',-',

0>
0=:::=-

~,

!!.!!!.!'

'",,",
"""',

"""',

5p~ ..U:

-I' fl' ~l

7-11

FIG. '3
fiG.

_ OUTER GRAIN SURFACE

CROSS SECT[ ON , X =

...

'?-eoo.oo

~oo

.00

600.00

FIG. 11

,
~

7~.*'::::OO-=-.OO=--.";'40::::' -;;;.OO=--.i:,;;;-,.-;;,o~o;i:;.O;;;-O ---',;i;;oou.o;oo---'.<\;;00;:;.0;00--;;lao!! .00

Y-AXIS

,-:-I-60-0.-00-~--".------.;:::::-;;;=--;:r;;;;----=~---.=;;;---;j600 00

7-12

FIG, 13

FIG. 12

CROSS SECTION' X

~r--,C"R,-,O",5"5--,5~E,,,C,,-T-,-1 O"'N'---.:'..!cX--=-..--"6,".3-"12s..'-,-_ _, -_ _-,

6275.

o
o

,.-t,:::oo".o::o-~."o::-o.-:o::-o-"2:::"".'::0--10"",0"'0--,-i-,,-,-OO-~<O-O-'O-O

60000

-stO' 00

Y-AX [5

FIG, Ilj -

SEG~\ENTED

MOTUR:; IiOMJ:NCLIHURE

(""1-0 CODE)

Aft race

Forward face

forward

end

XMIN::99

I~'-

~rlblf~geport10n

c-::_

of

"'_=.:.-__--r-I.,....7?<""=-_~::.nrRePOrt;10n
,

,,,r-----!'f-~

I'

port

es':h-e-:-d----"'O"L.~C::...:~:;

by cylinders andl

or cones

of

7-13

FIG. 15
INPUTS

I NTERNAl BAlli ST I CANAL YS I S COMPUTEll PROGRAM

OUTPUTS

INT. BALLISTICS CODE

QUAS I-STEAOY STATE FLOW CONO I TI ON

MASS AOD I T I ON

FLOW AREA CHANGE

P, T, m, M,
Wb

TIME

BURN I NG RATE VAR I ATI ON

FIN I TE DIFFERENCES APPROX I MATI ON

PRESSURE, THRUST
MODIFIED LENOIR - ROBILLARD BURNING

IMPULSE

RATE RELATIONSHIP

a, n, it

BURNED PROPELLANT WE I GHT

HUMP vs WEB
H (Pa) vs TIME

I GN I TER MASS
FI.OW RATE
PROPELLANT
BURN I NG RATE

FIG. 16 - METHODOLOGY FOR BALLISTIC PERFORMANCE PREVISION

7-14

FIG. 17 - ARIANE 3 STRAP-aN-BOOSTER AVERAGE HUMP FACTOR

1.1

1.05

95

240

120

60

WEB (",,1

FIG. 18 - ARIANE 3 STRAP - ON - BOOSTER BALLISTIC PERFORMANCES

----

1000

---- ----

800

600

,..-..------

~-/

I\..,..-'

--

-V--

L--o----

I---.~

II

~ h

Ii

400

..

I
I

I~ n

thrust
(KN)

1\

200

10

30

25

15

time (sec)"

35

A MAX THRUST ENVELOPE

DESIGN NOMINAL THRUST

X DESIGN PREVISION

US I NG HUMP FACTOR

H~bHL EXPERIMENTAL

1.04,.----------

1.02

..~

1.00

~ 0.98
0.96

0.94

- - - nlIOKOl. 4 MOTOR AVERAGE (65 IN. AND 156 IN4 OrA)

- - - - AEROJET GENERAL 3 MOTOR AVERAGB (260 IN. DIAl

0.92

_ ..

LPC 4 MOTOR AVERAGE (156 IN. OIA)

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

WEB FRACTION

FIG. 19 -

I NDIJSTRV EXPER I ENCE WITH LARGE SEGMENTED MOTORS

715

FIG.20 -

s[~VOlk

IUrl'LY lUll \

WATER TABLE FACILITY

FIG.21
- ETRIC CUT AWAY)
( 150M

7-16

FIG.22 - ANALYSIS OF MOTOR STATIC FIRING TEST RESULTS

Texp

vs time

Pexp

vs time
Wp
Pa

At (t=O)
At (t=tf)
Aexit
Pteo vs time

prope II ant burn i"9 rate

Data

..
....

Analysis

Codes

-L i-near variation of nozzle effi

ciency during burning

Isp, CF'; At vs time

-Characteristic velocity efficiency

-...

constant

Hump

vs web

-Correct. ion of measured fwd. - end

pressure
sure

to

based

nozzle stagnation pres_

upon constant port area

-*
C ,{3
burn i"9 rate scale

relationship

factor

MOTOR
fRIANE III

PROPELLANT

PROPELLANT
WT (K g)

(m)

(m)

CTPB

from 11 to 14

0,3

0,23

5-14

ULLAGES

CTPB 16-14

M1

CTPB or HTPB

NOZZLE
EXP.
RATIO

STATE

APPLICATION

Production

Stage Separat i on

56

1,3

0,20

5,8

Product i on

Mi I itary

176

1,29

0,493

35

Product i on

Apogee

1,13

0,684

37

Production

Apogee

1,29

0,766

65

Production

Apogee

1,3

60

Static

Perigee

SIRIO

CYPB 10-14

GEOS

CTPB 10-14

269

MAGE FAMILY

CTPB 16-12

from

IRIS

HTPB 18- 13

1575

250 to 490

Tested
ARIANEIII
BOOSTER

CTPB 16-13

7350

6,5

1,08

Production

Strap-on
Booster

ARIANE IV
BOOSTER

HTPB 10-14

9500

8,5

1,08

Qua Ii fied

Strap-on
Booster

AL SR

CTPB 10-14

550

1,3

0,63

178

2,6

0,28

M2

CTPB 15-13

Technological
Static

Mi I itary

Tested
143

1044

DOUBLE, BASE
"

21

2,0

.12

5,2

Product i on

Military

1.

.08

7.8

Product ion

Mi I itary

MAIN CHARACTERISTIC OF SEVERAL SNIA-BPD MOTORS

TABLE 1

7-17

MICROROCKET IGNITER

AISI 4130 STEEL CASE ROLL AND WELD,

THREE WELDED SEGMENTS

ASBESTOS FILLED RUBBER

THERMAL INSULATION

CARBOXY TERMINATED
POLYBUTADIENE (CTPB) PROPELLANT

STANDARD CARBON CLOTH

PHENOLIC NOZZLE
fiG. 23 -

ARIANE 3 STRAP-ON-BOOSTER
MOTOR LAYOUT

flight experimental pressure

Prevision
Booster PM n. 1
- - - - BoosterPMn.2

press!MPa)

r----"'T""----,-----..----....,-----,-----..,

6 ..

b----:~./4-------=::=:::=1:=:=~~~~==~t:-=-::"...~,;;;l.:J~,-+__~
~
li\
1 +----1
41---_-1+
-I-1
5

i\\\\1

31---+----+---f----+---\ll.t~:\+----1
1\
2 f---+---+----+----l----i\II\r+----1

t.

o L-

-L

-:-

.L.

10

15

-L
20

l-_.....:.::::::....l.
~
25

30

time!sec)
fiG. 24 - AR lANE 3 STRAP-ON-BOOSTER
fL I GHT PRESSURE 0 I AGRAM

fIG. 25 -

ARIANE 4 STRAP-ON-BOOSTER
GROUND PRESSURE 0 I AGRAM

50

40

30

20

10

10
BOOSTER 02-EXPER I MENTAL
_ _ POST-fiRING PREDICTION

20

7-18 .

THERMAL
INSULATION

PROPELLANT
GRAIN

766
1292
Fig. 26 - SIRIO A.a.M. Layout

IGNITER

-.-__

652

1130
Fig. 27-GEOS A.8.M. Layout

IGNITER

______

.nQ2.50

----

FIg. 28 - MAGEOS A.a.M. Layout

IGNITER

Fig. 29 - MAGE 1 A.8.M. Layout

NOZZLE

7-19

50,.-----.,.----,-----,-----.,.---.....,
EXPERIMENTAL
THEORETICAL

o
40

50
TIME (a)

FIG. 30- SIRIO A.B.M. THRUST DIAGRAM

-,"r .. '.

..

_ _EXPERIMENTAL
_THEORETICAL

'"

..
..

"'"
'"
'"

o ~0,----";'1*0----...".20'::----,-3:f:0:-~.....-,-4':0,----.."J50
TIME

(8)

FIG.31 -SIRIO A. B.M. PRESSURE DIAGRAM

50

."
'.".

EXPERIMEN":'AL

THEORETICAL

... 40

...~

30

20

10

O':-

-,l;:

--:~---_+::---~,.__-l.-~

TIME
FIG.32- GEOS A.B.M.

THRUST DIAGRAM

EXPERIMENTAL
THEORETICAL

~ 4

:;"'

."::l

'"

o ~---,-,'-:0:-----2.J.0-----3LO-----l40--...l..l'--5-10
TIME
FHL 33 -

GEOS

A.B.M.

PRESSURE DrAG!lAM

(:'I)

7-20

..
.
."

;;50

VI

_ _ _ EXPERIMENTAL

_ _ _ EXPERIMENTAL

_ _ THEORETICAL

_ _ THEORETICAL

~40

30

20~~

~.

.. ~.,..""

10

,~

~.~

~~

o0~---~'0;;-----2~0:-----=3'::-0----.7:0:-----5"~""'.J
TIME
fIG. 34 -

MAGI'!;OS A.B.M.

10

20

FIG. 35 -

THRUST DIAGRAM

HAGEOS A. B. M.

PRESSURE

(5)

DIAGRAM

___ EXPERIMENTAL

_ _ _ EXPERIMENT At

:::

50
TIME

50

.0

30

(a)

_ _ THEORETICAL

_ _ THEORETICAL
.0

30

.. ~

,
--~
",

............

10 :'

0
0

10

20

30

.0

50

10

20

30

itHE( s)

FIG.36- MAGK 1 A.B. M.

50
(s)

THRUST DIAGRAM

fIG. 37 -

MAGE 1 A.8.M.

PRESSURE DIAGRAM

81

SOLID PROPELLANT GRAIN DESIGN

Bernard ZELLER
SNPE - Direction Technique Autopropulsion
Centre de Recherches du Bouchet, BP N 2, 91710 VERT LE PETIT - FRANCE
ABSTRACT
To design a solid propellant grain is to conceive and to define a grain which satisfies various
requirements. This paper describes the methods and procedures which are used today in France to design
propellant grains. It describes and analyses
- the various types of grain and the various families of propellant which are available and used
today,
- the detailed requirements that a solid propellant grain must satisfy,
- the methods which are used to precisely define the propellant, the architecture and the configuration
of the grain, and more specifically the methods used in order to ensure required ballistic performances
though maintaining structural integrity of the grain (which is submitted to mechanical loads all along
its life,)
- an overview on a method of solid propellant grain reliability assessment.
Lastly, the improvements needed in the area of grain design analysis and the related technical
breakthroughs are briefly presented.
1. INTRODUCTION
During the last twenty years, requirements on performances, reliability and cost of solid
propellant rocket motors (and also on schedules and cost of developments) have become more and more
stringent. Consequently it had a direct effect on solid propellant grain design methods and procedures
and on development program contents.
The need for improved performances is the consequence of the need for longer ranges, higher
velocities and more powerful payloads. The improvement of reliability originates from the need for
higher availabili ty of weapon systems, for lower malfunction probabili ty and fo r longer service life. A
decrease of duration and cost of development program directly reduces program total cost.
During the same period of time, energetic, kinetic, mechanical and aging propellant properties
have also been largely improved. Furthermore, the power of scientific computers has greatly increased
and the use of micro-computers has widely spread within the project managers community.
Due to the pressure of competition (tactical missiles, space launchers) or to
technical/political reasons (strategic missiles), the time assigned to designers for performing grain
preliminary design * has considerably decreased.
It seems appropriate, to present a synthesis of the various methods used today for designing
solid propellant grains, within the larger frame of solid propellant rocket motors design.
Design of propellant grains involves vast knowledges and numerous techniques. It is due to the
nature of propellants, the geometry and architecture of propellant grains and to their operation modes
in rocket motors.

Result of preliminary design is a first propellant grain definition which generally demonstrates how
initial requirements may almost totally be satisfied. Additional modifications of the grain, involving
often the use of large computer codes, are needed in order to establish final design.

82
Grains are made of solid propellant put into a given configuration during manufacture; their
surface is generally locally restricted or inhibited (to prevent ignition and combustion) by a flame
resistant, adhesive material. Other parts of the grain may be bonded by a liner to the motor case
(case-bonded grains).
Weights of propellant grains range from just a few grams to several metric tons, chamber
pressures from a few tenths to more than thirty MegaPascal (MPa), operating times from a few
milliseconds to a few minutes.
Manufacture, fielding, storage and operation of a propellant grain (within a rocket motor)
involve numerous phenomena related to chemistry, thermodynamics, geometry, combustion, fluid dynamics,
mechanics of continuous media, etc ... In the present paper, it is not possible to comprehensively
analyse all the aspects of grain design which precisely defines a propellant grain which can be
industrially manufactured and which must satisfy requirements on storage and operation in various
conditions. A selection has therefore been made in the potential content of this paper. It is assumed
that the reader is familiar with the basic knowledge of solid propulsion (see (1) and (2. Since a
wide variety of solid propellant grain configurations is available, it is not possible to analyse them
in details. Consequently, this paper is mainly focused on design of case-bonded solid propellant grains
used for tactical and strategic missiles. There is no in depth discussion on free standing grains
(although much of the material developed hereafter could be used for such grains), on nozzleless grains
mainly used for inttial boost of ramrockets, on grains containing imbedded metal wires for local
increase of propellant burning rate, and also on clusters of stick propellant used in very short action
time rockets.
Main points discussed include :
- a description of various types of grain and associated propellants (including french terminology),
- an analysis of requirements for solid propellant grains,
- a review of mechanical and ballistic design methods used today in France (the description of related
three-dimensional computer codes and of a rapid grain preliminary design method are emphasized),
- a method of assessing propellant grain reliability and a description of associated tests.
In the conclusion, one tries to point out the main areas wherein improvements on the present
state of the art are needed either in the knowledge of the phenomena or in the analytical and numerical
tools which are necessary to design satisfactory high performance/low cost propellant grains.
2. DESCRIPTION OF GRAIN GEOMETRIES AND ASSOCIATED PROPELLANTS
In this section, various types of grain configurations and of propellants are presented, and
also general principles on configuration and propellant selection.
21. Grain configurations
There are two main types of grain architecture : free standing grains and case-bonded grains.
Grains of the first type are introduced into rocket motor cases (cartridge-loaded) after manufacture.
Grains of the second type are bonded to the motor case during the casting (or injection) and curing
steps of the propellant grain manufacturing process (figure 1)

case
thermal

~~~~~Ii~]inSulation
liner

free standing grain

case-bonded grain
Figure

There is not a single, well defined. procedure for selecting a free standing grain architecture or a
case-bonded grain architecture for a given rocket motor, except when one of these two architectures is
obviously most appropriate for a specific reason. Nevertheless, case-bonded grains generally give
higher performances than free standing grains for equal available volumes. However free standing grains
are largely Wide-spread because this type of archi tecture may present significant advantages, for
instance from the point of view of cost and of overall industrial management. Today, one can yet notice
a trend towards case-bonded architectures due to the demand for higher performances.

When propellant grains have an outer diameter larger than 500 mm or a weight of more than
300 kg, they are almost always case-bonded. High performance, middle-sized (outer diameter between 100
mm and 500 mm, weight between 10 and 300 kg) are case-bonded, but free standing middle-sized grains are
very common. For small rocket motors, free standing grains are generally used.

8-3
Case-bonded grains generally have a central port ; the outer surface of the grain is bonded by
a liner (and a thermal insulation) to the motor case. During firing, combustion of the propellant is
initiated on the internal surface of the central port and proceeds radially towards the case (and to a
certain extent longitudinally depending on the exact geometry). Exact grain geometry is obtained
during manufacture of the grain either by direct casting in the case around the mandrel or by
machining the port after casting and curing have been completed.
- !X:!s,rm.!!!e.!r,!c_c.n.!.i.ll.u.!.a.!i.n.!!.
AXIL Axisymmetric grain with annular slots. The
slots are circular : their axis is the same as the grain
axis. They are located all along the central port
(Figure 2).

I.

_h.ead-e,!1d
figure 2

~.
Figure 3

AXAR configuration

AXAR Axisymmetric grain with annular slots.

This configuration is similar to AXIL, except that
the slots are located near the aft-end of the
central port (Figure 3).

CONOCYL (Contraction of cone and cylinder).

Axisymmetric grain with annular slots. The tips of
the annular slots are inclined towards grain head-end
so that a part of the grain is cone-shaped
(Figure 4).

~~,~
:~~~_

head-end
--

---------.

Figure 5

Star configuration

-----

~ft-end __

AXIL configuration

__

h'"d_~"ft-'"d
figure 4

CONOCYL configuration

- .Y1:.i~d.!.i.a1:. .o~f,!g~r1!t.!.o~s_
STAR
The cross section of the central port
has the shape of a [1 points star. The contour of
the star is constant along the axis (In some cases
it may be slightly evolving for manufacture
practicali ty) (Figure 5).

WAGON WHEEL The cross section of the central

port looks like a wagon wheel (Figure 6). Numerous
parent configurations exist like dendrite, anchor
and dog bone configurations.

figure 6

Wagon wheel

Other configurations may be obtained by combination of some of the above described configurations.
For instance bipropellant star configuration (to eliminate sliver), or AXAR configuration having a
stress-relieving annular slot in the head-end area. Full head-end web grains are also used. Simpler
configurations such as internal-burning tube are commonly used ; the ends are usually unrestricted to
function as a burnIng-surface control; they may also be partially restricted.

8-4

Above described configurations are considered as one or two-dimensional, though of course being
actually three-dimensional. They are either axisymmetric or cylindrical, with, often, an order n
symmetry. It is therefore not too difficult to calculate burning area versus web burned or
stress-strain field. Today, three-dimensional configurations are getting more and more popular among
the designers community; they are also much more difficult to design. Most of these configurations are
referred to as "finocyl", which is a contraction of fin and cylinder. The fins may be located either at
head-end or at aft-end of the grain (and sometimes at both ends) ; they merge into a central
cylindrical port. They may have the shape of slots, which simplifies the geometry (Figure 7).

head-end

~~~-_-.L~
Figure 7 : FINOCYL configuration
(slots and tube)

Often, for stress relieving, there are annular slots. These configurations requi re three- dimensional
analysis for calculating burning area versus web burned as well as stress-strain field or gas
flow inside the central port.

End-burning configuration is not well adapted to case-bonded architecture because of problems of

structural integrity. However it is possible to manufacture such case-bonded grains using
stress-relieving grain support and retention systems which allow thennal shrinkage due to propellant
cooling after curing though permitting pressure to equilibrate during firing.

Free standing grains are generally smaller than case-bonded grains. Because they are not
bonded to the-case wall, except sometimes locally, they allow configurations ,ffiich cannot be obtained
wi th case-bonded grains (for instance internal-external burning tube).
Final checking of the grains is easier than in the case of case-bonded grains. They are
loaded into the motor case during final assembly of the rocket motor. Various support systems may be
used to ensure proper operation during firing. During missile service life, it is often possible, if
necessary, to replace the grain independantly of other motor components.

Star, wagon wheel, tube configurations similar to those above described may be found for free standing
grains. Grain ends are generally simpler: they are plane and may be restricted or not.

To reach high volume loading fractions for free standing grains a configuration was developed : the
cross section of the central port is right circular in the forward section and becomes progressively
star-shaped in the aft section of the grain (Figure 8). In France, this configuration is referred to as
"trompette" (trumpet), though it has not much of the shape of a trumpet.

Figure 8

TRUMPET configuration

The orientation of burning is totally in the longitudinal direction. This configuration is wide-spread
because gas generation rate is almost constant, volumetric loading fraction is high and grain
manufacture is easy. Side and head faces are restricted. Burning times are long and thrust levels are
low or moderate. Thermal insulation and inhibitor play an important role respectively to protect the
chamber walls from the continuous exposure to hot gas and to restrict the canbustion to the desired
area. They also generate pyrolysis gaseous products during firing which must be taken into account in
the total amount of gas generated by the grain. They are used mainly for the sustaining phase of the
flight of some missiles.
213 .e!!e!.al.E.rln..si.E.l~s_f.2.r_s~l~c.i.!~.n_o! ..ar~i!! ..so!!flg~r~tlo!!
A practical procedure for selecting the couple configuration/propellant is discussed at section 4.
Hereafter only basic principles are discussed.

8-5
For selection of grain configuration, the main factors which are taken into account are
- volume available for the propellant grain,
grain length to diameter ratio (LID),
- grain diameter to web thickness ratio (Die),
thrust versus time curve: it gives a good idea of what should be the burning area versus web burned
curve (neutral, regressive, progressive, dual-level),
volumetric loading fraction. It can be estimated from required total impulse and actual specific
impulse of available propellants,
- critical loads (thermal cycles, pressure rise at ignition, acceleration, internal flow)
manufacture practicality, which depends on case geometry (some grain configurations are more or less
easy to obtain)
fabrication cost: it can be the critical factor for selecting a given configuration.
There is no definite procedure to select grain configuration in order to satisfy a set of
requirements, because there are often several technical solutions to the propulsion problem.
Practically, there are some general trends in selecting configurations, based on the shape of
the burning area versus web burned curve (which is qualitatively close to the thrust versus time
curve). Table 1 summarizes these trends.

Burning area
neutrality

Grain configuration

.;

1/AGON 1/HEEL
DENDRITE

Sho rt burning times (

TRUMPET

A constant burning area versus web burned is obtained by progressive cylinder

zone and degressive slots or trumpet zone

TUBE AND SLOTS

3 5), lo"W volumetric. loading fraction

.: 1
for third stage of strategic missiles

AXIL
GOOD
NEUTRALITY
LESS THAN 15 %
RELATIVE CHANGE
IN BURNING
AREA)

Remclrks and comments

Case-bonded grain, L/O

FINOCYL
AXAR
CDNOCYL
TUBE
(end faces may be

Case-bonded grains, 24 LID ,:; 4

for first and second stages of strategic missiles

Volumetric loading fraction less toan 0.8

restricted)

BI PROPELLANT
STAR

END-BURNING

Two different propellants, Internal propellant has a higher burning rate than external
propellant by a factor of about 2. It is a costly configuration requiring long manufacture
cycles (two curing stages). But neutrality is kept, even for high volumetric fraction, am
sliver are eliminated.

Long burning times. low or moderate thrust

Remark : the lower the volumetric. loading fraction, the better the neutrality. A.constant burning area C'.ao b~
accurately obtained~ by adding axisymmetric
slots in the central port or hy restricting speci fic propellants
surfaces. It raises grain cost because it requires additional phases during grain manufacture.

TRUMPET

DUAL LEVEL

PROGRESSIVE

SLOTTED TUBE
AXAR
BI PROPELLANT STAR
END-BURNING
(with annular slots
in the aft-end face)

TUBE
(right circular port
section)
STAR

The ratio of the two levels can be adjusted by varying the geome try of the aft-en<i sect ion
Volumetric fraction may reach 0.88. LID may reach 10
Adjusted by number and geometry of the annular slots
Adjusted by the geometry of the stars and propellant burning rates
Adjusted by the geometry of the annular slot( s)
(boost with radial burning, sustain with end-burning)

The most conmon is the grain with restricted faces

With restricted end faces ; slivers induce long taU-off

TUBE WITH INTERNAL

EXTEHNAL BURNING

REGRESSIVE

STAR. hlITH INTERNALEXTERNAL BURNING

Unrestrl~t.Q<l

END-BURNING HITH
TAPERED AFT-END

Long burnIng tiMe, low to moderate thrust

Table 1

en" f'lces .1nd low to moderate LID ratio

burning area neutrality versus grain configuration

8-6

CONFIGURATION

VOLUMETKIC
LOAOING
FRACTION

wEB
THICKNESS

II

BURNING
BURrfING

AREA

AREA
NEUTRALITY

SLIVER
FRACTION

good

5-10 %

WEB
FRACTION

COMMENTS

Case-bonded

STAR

0.75-0.90

intermediate intermediate

TRUMPET

0.88-0.95

large

intermediate

excellent

SLOTTED
TUBE

0.75-0.85

large

large

good

\lAGON WHEEL

0.5-0.7

SID all

FINOCYL

0.85-0.95

large

3.5-5.5

'>

and free
standing grains
Free standing
grains
Case-bonded

'" 3

grains

Free stand 1 ng
very large

excellent

5-10 %

6-12

and case-bonded

grains
large

good to

2-3

~xcellent

AXIL

0.88-0.93

intermediate

large

excellent

5 %

AXAR

0.88-0.95

large

large

good to
excellent

0.75-0.85

inteno.ediate

1nteMledlattt

good

Mainly case-bonded
grains
high LID

Case-homed
grains
LID ~ 1

2-3

Case-bonded
grains
high LID

5-10 %

Case-borned

STAR

(with full
head-end web)

BI -PROPELLANT
STAR

grains

0.9

large

intermediate

excellent

<5

2.5-3

Case-bonded

grains

END-BURNING
GRAINS

Table 2

0.98-1

very large

small

excellent

Low thrust

main characteristics of common grain configurations

Table 2 presents main characteristics of commonly encountered grain configurations. It allows to aid to
select a configuration.
22. Propellant selection
There are several solid propellant families which differ from their ingredients, their
manufacturing processes and their ability to be processed into certain configurations.

In France, five families of propellant are commonly manufactured and used :

solventless 'extruded double-base propellants (SD), the main ingredients of which are ni trocellulose
and nitroglycerine. Configuration is obtained by extrusion through a die having the desired shape.
Outer diameter is limited to about 300 mm. Additional grain machininE may be performed.
- cast double base propellants (EPICTETE)* ; the ingredients are similar or parents to those of SD
propellants ; they are obtained by casting a mixture of nitroglycerine and triacetin into a mold
containing nitrocellulose based casting powder.
- composite cast double base propellants, which are derived from Epictete propellants by addition of
hexogen, octogen, or ammonium perchlorate and possibly nitroglycerin, in the casting powder (3,4).
- composite propellants based on a non energetic polymeric binder and on ammonium perchlorate, which
may also contain aluminum powder.
- high enprsy propellants based on an energetic binder highly plasticized by a liquid nitric ester, On
he~ogen or octogen~ which may contain ammonium perchlorate and aluminum (3,4).

trade mark

8-7

There is a terminology commonly used in France for the three preceding propellant families.
It is based on the following principles :
the name of a propellant is made of a prefix, one consonant, and a suffix.
The prefix gives some information on the binder :
NITRA
BUTA
ISO

energetic binder (usually containing nitric esters)

binder based on carboxy or hydroxy terminated polybutadiene
binder based on polyurethane
Central letter indicates the nature of the main energetic filler
ammonium perchlorate
octogen (HMX) or hexogen (RDX)
potassium perchlorate

L
M

Suffix indicates the nature of the metallic fuel

ANE
ABE
AZE
ITE

aluminum
beryllium
zirconium
no metal added.
The three above mentioned propellant families include the following main propellants

- Nitramite* E

Isolite*
Isolane*
Butalite*
Butalane*

nitrocellulose/nitroglycerine binder filled with RDX ou HMX. E reminds that this family
of propellants is obtained through a process very similar to the one used for
manufacturing Epictete propellants.
polyurethane binder and ammonium perchlorate
polyurethane binder, ammonium perchlorate and aluminum
polybutadiene binder and ammonium perchlorate
polybutadiene binder, ammonium perchlorate and aluminum

- Nitramite* G

elastomeric binder, plasticized wi th a mixture of liquid nitric esters, and filled wi th

RDX or ill1X and possibly some ammonium perchlorate. Letter G reminds that the
manufacturing process is the slurry cast (global) process.

- Ni tralane*

elastomeric binder plasticized with a liquid nitric ester, and filled with HMX,
ammonium perchlorate and aluminum.

Besides main ingredients, propellants may contain several other ingredients, generally at low
contents, used as stabilizers, afterburning suppressants, combustion instabilities suppressants, burning
rate modifiers. One of the important task of propellant designers is to find a practical way (fillers
particle size, burning rate modifier, ) to control burning rate, which is a key factor in designing
solid propellant grains.

Selection of a propellant for designing a given grain is based on numerous criteria and, here
again, there is no strict procedure for selecting a given composition. The type of architecture
(case-bonded or free standing), energy and burning rate criteria, structural integrity considerations,
smokelessness and safety conSiderations, may lead towards a given propellant family. Each of the
propellant families covers a certain range of properties, and it is necessary that the properties of
the selected propellant allow to design and manufacture a grain satisfyip~ all the requirements. Table 3
summarizes some properties of the main propellant families. The information presented is very succinct
and would need more thorough development. However, it allows, in combination with tables 1 and 2, a
first approach in the selection of the couple configuration/propellant which is detailed at section 43.

* trade mark

8-8
llaximum

delivered
specific
Propellant

impulse in
standard

conditions
( 70/1)

Range
Haximum

density

(kg/dm3)

of burning
rates at
7 HPa
(or at the
plateau)

225

Temperatur
coefficient

Arehit~ture

,,0

1.60

very
low

free
standing

low

ree
standing
and

(plateau)

3 - 28

Cast cOluposlte
modified
double base
NITRA1:lITE E

230 s

1. 70
( plateau)

Non aluminiZed
composite
po lybutad ie-ne
propellant
8UTALITF.

240 5

Aluminized
composite
polybutad iene
propellant

245 s

0-0.2

case-bonded

1.73
0.3-0.5

4 - 60

1.86
5.5 - 80

primary
and

secondary

4 - 22
215 s

EPIGTETE

0.2-0.5

low
to
moderate

free
standing
and
case-booded

low
to
moderate

free
standing
.nd
case-bonded

pr fmsty
and
secondar

245 s
(with API

primary

smoky

primary
1. 79

10 - 25

(and

0.45-0.6
235
(Wi

moderate

case-bomed

1. 75

thout AP)

5 - 10

secondar
without
AP)

Aluminized
cross-linked
double base
XLDB or NEPE

254 s

0.5-0.7

1.86

moderate

case-bonded

according to French regulations

Table 3

t.

(r1UIIIberlJf

diac:hat1Jlt

MlIn"r..:lurinq

,."t

smoky

Notes

AP

Ingredient 8
cost

no

low

low

no

high

low

no

high

moderate

110

100

150

;/
~
;X

(ROX)

no
moderate

low

often

moderate

low

no

moderate

moderate
(ROX)

180

fairly high
(HMX)

no

moderate

180

9 - 25

NITRALANE

SeolliU'IU.y

electtost.tic

;%
1)/

primary
and
secondary

BUTALANE

Non aluminized
cross-linked
double base
NITRAMITE G

IZ
Card

Gll;:Jtellt.

cards)

free
standing

( plateau)

Cast
double base

,~ry

I
very
low

"0

1.65

SO

Pri_ry
-*:ele8sne811

(mm/a)

5 - 40

Extruded
Double base

chsaific:ti:Ol1

Pressure
exponent

ammonium perchlorate

fairly
high
(HMX)

Card Gap Test is French Gap Test

main characteristics of common propellants

3. REQUIREMENTS ON SOLID PROPELLANT GRAINS

This section addresses technical requirements propellant grains must meet. Requirements are
settled as the consequence of an agreement between rocket motor designer and propellant grain designer.
They must be clear, complete and consistent, so that the propellant grain designer may precisely define
the grain and eventually build the corresponding engineering development program.
Requirements are divided into requirements related to functional specifications, requirements
related to operational specifications and interface requirements. They are detailed hereafte~.
31. Requirements related to functional specifications

Average, minimum and peak values of chamber pressure, thrust, total impulse, burning times,
must be specified within the full operating temperature range. Envelopes of thrust versus time or mass
flow rate versus time curves may also be specified.

Other requirements are necessary to the designer in order to define a satisfactory propellant
grain:
- maximum weight of propellant grain
maximum weight of total inert (thermal insulation, liner and restrictor)
maximum axial and transverse acceleration undergone by the propellant grain during operation of the
rocket motor
- rocket spin rate (for instance for unguided rockets)
- dispersions on pressure, thrust, total impulse, burning time have to be specified. Depending on the
corresponding requirements, manUfacturing process and control operations may be strongly affected and
thus the cost of the grain as well.
- plume characteristics (emission and transmission in the visible, infrared, electromagnetic
wave1enghts range).

8-9
32. Requirements related to operational specifications
Depending on environmental conditions, definition of the propellant grain may be significantly
affected. Such conditions must therefore be well defined in order to be correctly taken into account
during grain structural design phase.

Desired maximum shelf-life, related temperature cycles and storage conditions must be defined.
Particular conditions (relative humidity, salty atmospheres ) which could directly affect propellant
grain behavior must be specified.

The nature and number of thermal cycles undergone by missiles (for instance during operational
flights for airborne missiles) must be defined. Generally they are the limiting factors for structural
grain design because very low temperatures may be encountered.

- Acceleration before and during rocket motor operation

longitudinal acceleration undergone by the rocket motor must be specified, as well as radial
acceleration due to rocket spin.
- Handling and transportation : dynamic loadings such as shocks and vibrations encountered during
handling (drops) and transportation must also be specified.

A level of reliability is more and more commonly required. It is essential to define in which
conditions it has to be satisfied. The principle of a method of reliability assessment is discussed at
section 5.

The content and the planning of missiles surveillance, inspection, and maintainance must be defined, as
far as they may have an effect on rocket motor environmental conditions.

These requirements are related to safe ty and survivability of persons and materials. They are not ye t
often taken directly into account during grain design analysis. They may induce an a priori selection
of a type of propellant ( e.g. a non detonable propellant or a propellant having a large critical
detonation diameter) or, during engineering development, the performance of safety and vulnerability
tests.
33. Interface specifications
Close envirorunent plays an important role on grain behavior during its life and operation. It
is often prescribed by the rocket motor designer. The grain designer must take special care that its
definition is complete.

A blue print of the case, or at least, its geometry (length, diameter, configuration of head- and
aft-ends) are mandatory in order to perform grain preliminary design analysis. Physical and mechanical
characteristics of the case have a direct effect on structural and ballistic design:
- type of case (metal, filament winding/resin, )
- thermal expansion coefficient,
- hoop and longitudinal strains as function of internal pressure,
- maximum allowable peak pressure (depending on ultimate elastic elongation of case material),
- maximum temperature allowable at case wall at the end of motor firing.

Nature and geometry of thermal insulations (specially for case-bonded grains) must be known in order to
settle grain definition, either on a ballistic point of view (case wall surfaces subjected to high
temperature combustion products), or on a structural point of view (configuration of stress relieving
flaps and boots). Thermal diffusivity, specific heat capacity and mechanical properties data must also
be available.

8-10

In the case of free standing grains, the support elements ensure that combustion gas may flow between
the grain and the case wall during pressurization due to ignition. The support system must be well
determined so that' prediction of grain operation may be possible at any temperature.
334. Nozzle
The characteristics of the nozzle have a dramatic effect on practical ballistic performance of a rocket
motor. The following characteristics are of particular interest to the grain designer :
- number and orientation of the nozzles (the angle between nozzle center line and rocket motor center
line must be known),
- degree of nozzle submergence,
- erosion of the nozzle (diameter evolution) versus operation time at throat and exi t planes,
angle of the exit" cone (or a dimensioned sketch, in the case of a contoured nozzle),
- failure pressure of the frangible closure disk (it allows to define ignition system and to control
pressurization at ignition),
- dimensions of the blast pipe (between chamber and nozzle), when existing; it affects rocket motor
efficiency.

The conditions of propellant grain ignition depend on its configuration (location, volume, design),
- pressure at the end of ignition,
- pressurization rate (Which affects structural integrity during firing).
Minimum and maximum values of delivered pressure and pressurization rate must be accurately known
be~ause they are important factors governing grain structural integrity. An envelope of ignition
pressure versus time is of interest for this task.
4. BALLISTIC AND STRUCTURAL GRAIN DESIGN METHODS
41. Input
In order to design a propellant grain, two types of data are needed
- technical specifications
the preceding section gives an almost complete list of these
specifications. They are the reduction of functional, operational and interface requirements that must
be satisfied in order that the rocket motor fulfill its assigned mission.
- a data bank on propellants, liners, inhibitors and thermal insulations: it allows the grain designer
to have at his disposal, quickly and with a low probability of error, chemical, physical, kinetical,
mechanical, thermodynamical, etc characteristics of the various candidate materials which may be
used in a rocket motor. The values of these characteristics will be used as input data in analytical
and computational design tools.
42. Proced ure
When performing a solid propellant grain design analysis, two levels of design accuracy has to
be distinguished :
- IiEsl le~el, it is the level of preliminary design anaiysis. The tools used at this level must be
simple and friendly enough to be operated by propellant grain project managers themselves. They usually
are small computer codes based on analytical models, or even graphs which give very simply the first
results.
In any case, the method involves four main stages :
selection of a couple propellant/configuration,
definition of grain geometry satisfying internal ballistic and structural integrity (versus
temperature cycles related loads) requirements,
approximate assessment of erosive burning and potential combustion instabilities,
assessment of grain structural integrity during pressure rise at ignition.
The method is iterative: depending on the results obtained at the third or fourth stage, it allows to
start again at second stage or even at first stage if it appears that the first definition needs strong
modifications.
For a few years, grain designers have been requested to quickly provide fairly precise
preliminary design analysis for a given project. In order to satisfy this request a computer aided
grain preliminary design analysis method (M.I.D.A.P*) has been developed in France. This method is
discussed in details at section 45.
- Second level, it is the level of grain final design. The tools required for this task are more
sophisti~ated. They are operated by grain design experts. They are mainly finite differences or finite
elements computer codes based on two or three-dimensional models of physical phencrnena related to
internal ballistics, fluid dynamics, continuous media structural analysis, etc They allow acc urate
calculations and therefore optimization of the grain final definition.
* M.LD.A.P.

Methode Informatisee de Definition des

~vant-lrojets

8-11
The principle of the method is parent to the one developped for preliminary design analysis, but it
starts from the final result of this analysis, that is to say : the geometry and the propellant
selected at the end of the preliminary design analysis.
Starting from this geometry, the evolution of grain burning surface area versus wP-b burned is
accurately calculated. Taking into account propellant properties, one obtains the evolution of chamber
pressure versus time P(t), and thrust versus time F(t). If necessary, the effect of erosive burning has
to be taken into consideration at this stage. The results must then be compared with corresponding
requirements (maximum pressure, combustion time, total impulse, ). Afterwards structural safety
factor (related to thermal cycles and pressure rise loads) has to be assessed with the aid of advanced
structural analysis computer codes.
If the results are satisfactory, design is correct and the propellant grain definition is
accepted for starting engineering development. If they are .not, grain definition must be modified so as
to increase safety factor in grain critical area. Additional structural analysis must be performed in
order to check the benefits of geometry modification. Evolution of burning area versus web burned,
pressure, and thrust versus time must also be checked so that the ballistic requirements remain
satisfied. It may happen that, after these modifications, some of the requirements no longer be
satisfied. In this event, selection of the couple propellant/geometry has to be changed, or, if there
is no other possibility, modification of some requirements has to be considered, in connection with the
rocket motor designer.
43. Ballistic design analysis

Basic equation of solid propellant rocket motor internal ballistics are

P=

p,S.V c

P
(1)

CD' At

f'

S
Vc

Vc = f(P) , often apn

(II)

F = P. C . At
F

(III)

a
n
F

q:-

chamber pressure
propellant mass density
propellant grain burning area
propellant burning rate
propellant discharge coefficient
nozzle throat area
burning rate coefficient
burning rate pressure exponent
motor thrust (specific itnpulse multiplied by pro pel Ian t
weight flow rate)
nozzle thrust coefficient.

A quick examination of the basic solid propulsion equations indicates the effects of various parameters
on motor operation and therefore on motor and propellant grain design
evolution of burning area versus web burned is directly connected to pre'ssure evolution versus time,
sensitivity of burning rate to factors like propellant initial temperature, rocket motor acceleration,
chamber pressure, gas flow, will have an effect on motor operation,
? and <b which are specific of a propellant may be considered for propellant selection,
initial values, and possible evolutions during firing, of At and CF , which are directly related to
nozzle definition (and also, regarding CF ' to propellant nature), must be accurately known.
In the following sections, the series of stages encountered in ballistic design analysis is described.

This important part of design work has been approached at section 2 but only through a semi-quantitative
analysis. In the present section, it is quantitatively treated using a simple method which yet preserves
designer's judgment.
Selection is performed with the aid of charts and graphs like the one presented at figure 9. The exemple
of this figure illustrates the logical method used,which permanently takes into account technical
requirements, properties of actual propellants and characteristics.of wide-spread actual grain
configurations.

8-12

c"

(;run)

"

Figure 9 : Graph
for aiding in initial
selection of couple
propellant/geometry

Mp (kg)

The steps are

- calculation of propellant mass (Mp), given total impulse (1ft) and standard delivered specific impulse
(Isms) (for an expansion ratio of 70/1 and an optimum expansion ratio nozzle) measured for the propellant
likely to be selected. This first calculation is iterative, for,the value of Isms has to be corrected so
as to be representative of the average conditions of motor operation :
average chamber pressure (Pc) estimated from the specified maximum pressure
nozzle expansion ratio depending on maximum allowable nozzle exit cone diameter
As is limited by the specification on maximum diameter of nozzle exit cone

Mp

At equals

, where t c is speci fied

- assessment of volumetric loading fraction (C R ) required to obtain specified total impulse, given the
mass density of the propellant likely to be selected and the volume available for the propellant grain.
- selection of grain configurations. For each family of grain configuration, an empirical maximum
volumetric loading fraction has been determined. Thus, given the volumetric fraction required, one or
several configurations can be selected. Other criteria, like processing praticali ty, difficulty of
structural analysis, propellant web thickness, have also to be taken into consideration.
- definition of propellant burning rate

Vc

8b

Vc =

t;;-

- verification of consistency between specific impulse, density, and burning rate (at the average chamber
pressure)
This approach must be completed by an accurate calculation of nozzle throat diameter generating a maximum
pressure lower than that required by the specifications. This step requires a precise definition of
grain geometry in order to calculate burning area evolution which is needed for the determination of At

s.

Vc

On figure 9, the various steps of the method can be represented by the path from A to B, then to C and D,
or to C' and D'.

Accurate prediction of chamber pressure evolution versus time depends on accurate calculation of
propellant burning area versus web burned. Computational tools which are commonly used belong to two
families: one for "two-dimensional" configurations, the other for three-dimensional configurations.
Actual grain configurations are three-dimensional, but in numerous cases their geometry is defined by
only two coordinates (r,e) or (r, z) ; in that case, configurations are said to be two-dimensional.

8-13

Figure 10

Figure 12

GEOETOIl code

GEOfAR code

Burning area evolutions

- two-dimensional geometries burning area computer codes
These codes predict the evolution of burning area of propellant grain of following configurations
cylindrical grains (constant port area section), for instance starshaped grains, (GEOETOIL
code, figure 10).
axisymmetric grains (grains presenting a symmetry of revolution with respect to motor
center.line for instance Axil, Axar, Conocyl configurations (GEOAXI code, figure 11).
end-burning grains having one (or several) axisymmetric slots in the aft-end face
(GEOFAR code, figure 12).
- a three-dimensional geometries burning area computer code

.--

,:~

,~S,'-=======-~

The code, named FRONT, allows the calculation

of burning area of finocyl grains having possibly
axisymmetric slots. These grains generally
present an order n symmetry about the motor axis.
Burning area versus web burned is therefo re
calculated only in a 2 'l(/n angle sector.
Performing a calculation requires grid generation
in the grain initial volume. A balance must be
found on the number of elements between accuracy
and computer C.P. U. time. Performing the
calculation on a 2 'IT'ln sector reduces computing
time for equal accuracy. The initial volume grid
of the aft-end (only zone to be three-dimensional)
of a finocyl-shaped grain presenting an order n
symmetry is displayed on figure 13.

Figure 13 : volumetric grid of aft zone of a finocyl

grain (symmetry number = 6) for the calculation of
burning evolution.
Generation of the grid can only be performed by
skilled experts who know where a denser grid is
required for better accuracy. Results are given as
data listings. Visualization of the results is
possible by processing data. Figure 14 shows three
stages of burning area evolution of the grain
partially represented on figure 13 .Complete
performance of such a calculation requires about
eight manhours (mainly devoted to grid generation and
data input). Actual central processor unit computing
time on a VAX 11/750 computer is about ten minutes.

Figure 14 : several stages of a finocyl grain

burning area evolution

8-14
- ~ ~i~p.!.i.fi~d_c2.d.~ .fo!: ~12.t~ ~ni .!:.u.e_c2.n.fi.B.u.E.a.!i~n~
Grain presenting the configuration of slotted tubes are three-dimensional, but a more thorough analysis
of burning area evolution shows that three zones may be distinguished (see figure 15)
. zones A and C are two-dimensional
. zone B is three-dimensional.
Surface area of zone B is less than 10 % of total surface area~ A parametric study previously
performed on that kind of configurations has given results which allow to approximate fairly well the
evolution of zone B burning area. Associated with algorithms of GEOAXI (zone A) and GEOETOIL (zone C),
these results have led to the code GEOFIN which is accurate enough for preliminary design of that type
of grains.

Figure 15 : Partition of a slotted tube into three zones for

a simplified calculation of burning area evolution

Burning rate is one of the major propellant characteristics. It is measured on standard ballistic
evaluation motors and it is stored in the data bank mentioned earlier (section 41). It is sensitive to
several factors :
- pressure. In the pressure range in which rocket motors operate, a de Saint Robert's burning rate law
(V = apll ) is generally preferred. It is also possible to directly use plots of actually measured
burning rates versus pressure. The lower the pressure exponent, the more stable the rocket motor
internal ballistics.
- temperature. Environmental and use conditions of rocket motors may correspond to a wide temperature
range. It is therefore necessary to know burning rate sensitivity to initial propellant temperature. It
is generally expressed at a given burning surface to throat area ratio, K, as a coefficient --r( K
defined by :

<TIK= _1

where

(a..Y...)
e

Vc
d
K
is propellant temperature.

- acceleration. Propellant burning rate is sensitive to acceleration, but it is taken into account only
when it is more than 10 g.
- manufacturing process. "Hump" effect is the result of change in burning rate as a function of web
burned (enhancement of burning rate in radially burning grains in the zone between central port and
motor case walls). It is related to manufacturing process. Empirical correlations, drawn from
experience, are generally applied to take account of this phenomenon in ballistic design.
- internal flow. Combustion products interact with propellant combustion phenomena and may locally
change burning rate law which is no longer the one expected. Because of the significant effect of this
phenomenon, it is discussed in more details in the follOWing section.
Burning rate laws, evolution of burning surface versus web burned, and basic internal
ballistics equations prOVide pressure versus time and thrust versus time evolutions. In the simple case
where internal flow do not significantly interact with burning rate, equations (I) and (II) of section
431, combined with Vc = de/dt, lead to a differential equation which is numerically solved and which
provides web burned versus time e( t), burning area versus time S( t), pressure versus time P( t) and
thrust versus time F(t).

435. Effect of internal flows

------------

It is often assumed that flow velocity in central port exit plane is low enough so that it can be
neglected in internal ballistics analysis. It is then assumed that flow is accelerated only in the
convergence zone of the nozzle so that it reaches sound velocity at nozzle throat. In fact this
assumption is not satisfactory because flow calculations demonstrate that velocities of the order of
100 to 150 m/s are observed in port exit plane after complete ignition and pressurization. Depending on
grain configuration and on propellant properties, two types of phenomenon may be generated :

8-15
- a pressure drop between forward and aft-end of the central port
- a local increase of propellant burning rate due to erosive burning.

When performing a ballistic design analysis, one has to quickly assess the magnitude of the phenomena
connected with internal flow. Table 4 summarizes the knowledge empirically acquired in this field as the
result of numerous solid propellant grain design analysis. This table involves a factor J, which is
defined as :

area of a given cross section of central port

J=2
K

0,25

0,33

0,5

> 0,6
Table 4

propellant burning area up-stream the above cross section

propellant grain burning area

At

nozzle throat area

Nature and intensi ty of phenomena due to internal flow

..:. 0,2
N

5'

no abnormal effect
pressure drop in the central port must be es tima ted
accurately
calculate maximum pressure

in order to

no erosive burning
pressure drop must be taken into account
erosive burning occurs fo r low burning rate (< 10 mml sat 7 MPa)
propellants, or at a very high operating pressure
pressure drop is high and may induce propellant grain failure
erosive burning occurs for all type s of propellant

propellant grains do

not function, except nozzleless grains

Effect of factor J on nature and intensity of phenomena due to internal

flow inside a propellant grain

Pressure drop
Pressure drop is related t~ a decrease of pressure from grain head-end to grain aft-end. It induces an
increase of head-end pressure at the first phase of motor firing, and therefore maximum pressure
generally increases. Pressure drops are generally due :
- to energy losses inside the flow and to phenomena occuring at the interface of flow and propellant
surface or to sharp changes of port section or of flow direction,
- to side injections from burning propellant walls.
One of the critical steps in rocket motor operation therefore occurs just after ignition when port
sections (through which combustion gas must flow) are minimum. Average pressure drops values
encountered are of the order of 0.1 MPa between head and aft-end. In some cases, for special
configurations, pressure drops of more than 1 MPa have been observed.
A gaseous flow 1s fully characterized by the knowledge of local velocities and pressures. Computer
codes have been developed in order to know such characteristics. They are named PROCNE 2 and PROCNE 3
(depending whether geometry is respectively two or three-dimensional). They allow:
- to describe unsteady phases during pressure rise at ignition,
- to know steady flow just after ignition, in the whole cavity and in nozzle convergence section.
In order to use these codes, one has to generate a grid of the combustion chamber. Order n symmetry
(when existing) is taken into account so as to reduce the analysis to a sector of 2 '1T'/n (n = symmetry
number). Figure 16 presents an exemple of grid created inside the cavity of a finocyl propellant grain
having a symmetry number of sixteen. Results may be, presented either as gas velocity or pressure
contours lines in various cavity sections figues 17, 18) or as curves representing for instance gas
velocity as a function of radial distance to central axis (figure 19) in central Port cross section.

8-16

o
Figure 16

Fi9~re

Figure 18

17

Three-dimensional flow inside rocket motors;grid of

central cavity

Flow inside rocket motors

velocity field

Flow inside rocket motors :pressure

field

8-17

,r-L_f\~

Erosive burning

o,'-------------....L--__
Rocket .q'IOtor center line

Enhancement of propellant burning rate

due to tangential gas flow (compared to
propellant burning rate without tangential
flow) is known as erosive burning. It
occurs When propellant burning surface
is subjected to a high velocity combustion
gas flow parallel to it. The phenomenon
is due to an increase of heat transfer
from the flame zone to the propellant surface.
There are numerous physical models to explain
and to quantify this phenomenon (6,7).
Practically, a simple computer code
(COMBEROS), based on a monodimensional flow
model, allows to calculate head-end and aft-end
pressure evolution in a grain experiencing
erosive burning. Erosive burning rate law
selected for the model is :

Figure 19
radial distribution of flow velocities
in plane AA inside grain central cavi ty
Ve = Va [1
Ve
Va
G

CO

is
is
is
is

0( (G - Go)]

burning rate with erosive burning

burning rate wi thout erosive burning
mass flow rate unit in the given port cross section
mass flow rate unit threshold (beyond which erosive burning occurs).

COMBEROS code is used systematically in

preliminary ballistic design analysis.
It implies that grain geometry be
described by the cross section contour
perimeter evolution along grain axis.
Erosive burning is calculated in several
cross sections of the central port
according to local flow characteristics
(static pressure P and local mass flow
rate G) and to above erosive burning law.
Ignition phase is simulated as an unsteady
phenomenon ; time steps range from 1 to
5 ms. A complete motor firing may be
simulated, using a steady state model
and time steps generally ranging from
0.05 to 0.1 s (figure 20). A more
comprehensive investigation of erosive
burning in propellant grains, though
keeping a one dimensional geometry
assumption, may be performed with the
aid of PROC~~ 1 computer code (8).

210

I
I

10'

G-

I
I

erostve b rni"'ll (heatend)

\~ kerO~iYe burning (aft-end

".
..

.,.. I\"::::r:''"'~
~

"'
=

e, I
0
0

0-

I
I

no leroste
I I
i
i

I
I

:
I

0 15

\\

I I

I
0 10

\i

bu ningl

I
o us

.-"" h

(i

20

Z5

HIOAP

'\

'.

10

0, ]S

a_~5

"""~
50

0.5.5

0.60

6S

Ti... (s)

Figure 20 : pressure versus time with, or without,

erosive burning

436 .O!!!b~slin_i~sla.E.i!.ili,!!.s_
Grain design must incorporate an assessment of combustion stabili ty during motor firing. -The
phenomenon of combustion instability may occur when perturbations excite oscillation modes of the
chamber cavity. Interaction with combustion, flow, particles, nozzle, etc, may induce either an
increase or a decrease of the phenomenon. When it increases, pressure vibrations and pressure
increase may consequently be driven to an unacceptable level. In order to assess combustion
stability. a two :;tep:; procedure is follo\led (9).
Pressure inside combustion chamber cavity is assumed to be
~

pI

Wi

'Ii

M
c(

pI

p;;-

i=1

average chamber pressure

instantaneous pressure at point M
pulsation of mode of rank i and of frequency fi
spatial form of mode of rank i
point in grain cavity
damping coefficient (when 0{ i
0 ), or gain factor (when 0{ i >
mode of rank i

<

0) of the

First step of the analysis consists in calculating the various acoustical modes which are specific
of the grain cavity .. A finite element two-dimensional computer code, named VASAX, is used. Example
of two dimensional grid and corresponding results are presented respectively on figures 21 and 22
(the rank of the mode is J).

8-18

Figure 21 ; Combustion instabilities :

grid of a motor cavity for calculation
of acoustical modes

Figure 22 : Combustion instabilities

pressure contour lines

Second step of the analysis consists in calculating the values of 0(. These calculations need not only
the results of the first step but also data describing propellant response to pressure, effect of
condensed particles, etc . . . . Computer code AVER is used.
Depending on the value of ~ (equal to the algebric sum of the various
possible to conclude on the grain propensity to experience combustion
frequency fi , a value ofQi1arger than 0.1 fiindicates that there is
combustion instability may occur. Grain configuration (or propellant)

gain and damping factors), it is

instabilities : for a mode of
a significant probability that
has to be modified.

44. Structural design analysis

Various loads are imposed to propellant grains all along their lifetime, from their manufacture until
motor firing. These loads depend not only on rocket motor own characteristics but also on manufacturing
conditions (temperature), envirornnental and operation condi tions. Various factors affe<:t loads imposed
to a grain (specially a case-bonded grain) :
- curing temperature,
- acceleration of gravity,
- type and number of thermal cycles undergone during storage and transportation (for instance captive
flights for airborne missiles),
- acceleration during boost phase,
- pressurization during grain ignition.
The goal of structural design analysis is to calculate a safety factor defined as

K=

cis

where C is propellant (or bond) structural capability (allowable), and S is a functional related to
stress/strain induced in the propellant grain region undergoing the more severe loads (margin of safety
may be defined as C - S or C/ S - 1). In order to compare them, C and S must be of the same physical
nature.

Safety factor must be and stay, until the end of motor firing, higher than 1 during rocket
motor lifetime. According to this definition, it is assumed that grain cracking or propellant/liner
debonding induce significant modifications of rocket motor internal ballistics haVing consequences
ranging from failure of missile mission to rocket motor explosion. It is assumed that failure at the
most stressed (strained) point does not depend on stress (strain) gradient in the surrounding region.
If the safety factor calculated for a given propellant grain and given imposed loads is lower
than a requi red value, grain has to be redesigned un til a satisfactory safe ty factor is obtained.

8-19
Assessment of capability variations (due to manufacturing process, to material
reproducibility, to mechanical testing, to aging ) and of induced stress/strain variations (due to
uncertainties on boundary conditions, on imposed loads, on stress/strain determination methods) allows
as a result of .8 probabilistic analysis to estimate reliability of a series of propellant grains of a
given definition. This subject is discussed at section 5.
The procedure followed in order to predict safety factors comprises two major aspects : it must
define how to assess propellant and propellant-liner bond structural capabilities on one hand, and how
to determine induced stress/strain in various loading conditions encountered by the grain, on the other
hand (Figure 23).
Propellant and propellant-liner bond capabilities are determined by performing various
mechanical tests and require a failure criterion which is defined as the critical value (at failure) of
a functional related to the state of stress (or strain) of propellant or bond.

MECHANICAL
PROPERTIES
TESTING

Determination of induced stress/

strain involves a structural
analysis requiring input data such
as geometry, boundary conditions
(e.g. case displacement),
propellant and bond mechanical
behavior.
Results are expressed using the
same functional selected for
failure criterion so that they may
be directly compared to propellant
and bond capabilities.
Experimental validation of the
procedure has to be performed,
either on the propellant grain
itself or on subscale analogs,
whenever new elements, such as
uncommon grain configurations, new
propellants, new bonding systems,
have to be considered in safety
factor assessment.

I
I

t
PREDICTION OF
STRUCTURAL
INTEGRITY

Figure 23

VALIDATION
TESTING

procedure for predicting grain structural integrity

Propellant or propellant-liner bond capability is maximum mechanical loading which can be imposed to
the propellant or to the bond until failure occurs. Capability is determined by performing tensile
testing on various specimens. Effect of main parameters is thus obtained (10) :
- loading rate (which are very different when thermal cooling or pressurization at ignition have to
be simulated),
- temperature,
- surrounding pressure (when simulating ignition pressurization).
Evolution of tensile properties (tangential modulus E, propellant maximum stress)tn and corresponding
strain ~, bond normal stress~ and shear stress Q ) versus loading rate and temperature provide
corresponding master curves. These curves can be obtained because the materials behavior is such that
the principle of time-temperature equivalence may apply ; they involve shift factors of W.L.F
theory (11).
Figure 24 summarizes the various steps of the experimental work which has to be performed in order to
eventually provide master curves for a given propellant and bond : evolution of E~ , ~,Gn ,'1;;
versus reduced time t/&r (t is the reciprocal of loading rate - or specimen loading time, a is the
shift factor related to the test temperature T).
When mechanical behavior, for a given loading rate and a given temperature, has to be determined,
reduced time t/aT is calculated Which; using master curves, simply provides the requested data.

8-20
The effect of pressure on propellant capability
has to be known in high loading rate conditions
(which correspond to ignition times in propellant
grains). Pressure main effect is to delay
occurence of vacuum holes (dewettlng) and to limit
their extension in the propellant around energetic
fillers (such as ammonium perchlorate or octogen).
Tensile testing at several constant strain rates
under several pressures allow to quantify this
effect for given propellants. They generally
display an improvement of mechanical properties
(ranging from + 50 % to + 100 % on maximum stress
and corresponding strain,in the case of Butalanes
-CTPB or RTPB aluminized propellants) which is
taken into account in solid propellant grain
structural design when ignition pressurization is
the design limiting loading condi tion.
The whole experimental work performed on this
subject has led to the conclusion that propellants
behavior is (see for instance 12, 13, 14) :
- viscoelastic (experimental evidence originates
from relaxation study),
- non linear (but assumed to be linear for small
deformations) ,
is very
- incompressible (Poisson's ratio
closed to 0.5) until dewetting is significant
enough so that volume variation occurs during
tensile testing.

DETERMINATION
OF l.09 OT
VERSUS T

I'~~

-"'-,------~~ T.

.;....Log
~
~
~.-~1
T
log OT1~-'~
rTo
'---;-'-----+-'- - 4_ _ l.09
0,.

t..

t.,

t.

t..

PLOTTING
MASTER
CURVE
l.09 (t./o,)
l.09

OT

PLOTTING
l.09 OT
VERSUS T

Figure 24 : how to obtain and to use

master curves

The functional which expresses propellant

capabilities is discussed at section 444.

Determination of induced stress/strain field in the propellant grain requires the knowledge of:
- geometry on which loads are imposed,
- boundary conditions which describe imposed loads,
- propellant and propellant-liner bond behavior.
In most of the cases, geometry is three-dimensional, loads are static, dynamic or thermally induced ;
propellant behaviour is viscoelastic, non linear. Loads which are the limiting factors in structural grain
design are generally :
- thermally induced, in the case of grains for tactical missiles (low temperature cycling),
- pressurization induced, in the case of grains for large ballistic missile~ (stored in almost isothermal
conditions)
At the preliminary design phase, expected maximum value of stress/strain induced in the grain is quickly
assessed using analytical expressions. For instance, in the case of fairly simple internally perforated
grains, following expressions are commonly used

0..

AT. K

C. (b/a)2

for a thermally induced strain.

is equivalent strain at grain innerbore surface,

is propellant thermal expansion coefficient (assumed to be at least an order of magnitude higher than
case material thermal expansion coefficient),
~T is difference between cure temperature and temperature at which induced strain has ,to be estimated (AT
may be as large as 100 C),
~ and C are corrective coefficients taking into account respectively central port exact geometry and end
effects,
b and a are respectively grain outer and inner radii
~

In the case of a pressurization induced strain,

E.

= k c.~

is hoop strain of the empty case submitted to ignition maximum pressure,

takes into account case stiffness increase due to propellant grain,
is an empirical coefficient.
These values of

are input data for a first assessment of grain safety factor.

Final design phase involves extensive use of computational methods based on finite element techniques
applied to grain stress/strain field analysis.
The procedure comprises three stages

8-21

The material is assumed to have a linear behavior.

The incompressibility is treated using Hermann's reformulation (15)
The assumption that propellant is incompressible is justified by the fact that, even wTten severe loads
are imposed to the grain, the regions, wherein vacuum holes may occur (because of high propellant
elongation) and therefore wherein propellant is no mOre incompressible, represent only a small
percentage of grain total volume. Stress/strain field therefore results from the incompressible
behavior of almost all the propellant grain. Any structural analysis which does not take into account
Hermann's reformulation and which is based on the assumption that Poisson's ratio is closed to 0.5
(slightly less) may fail, in certain conditions (for instance poor accuracy of computer calculation),
to correctly predict stress/strain field and thus may lead to an erroneous value of the safety factor.
Mechanical load, which determines boundary.conditions, is expressed either as prescribed
displacement or as prescribed strengths at the nodal points of surface elements. Several computer
structural analysis codes may be used for this stage of the analysis : PALEST and PROVISC for
two-dimensional gometries, CAMILLE for three-dimensional geometries. CAMILLE has the main following
characteristics :
- finite element method,
- HERMANN'S reformulation,
- quadratic elements with 20 nodal points,
- quadratic skin pseudo-elements with 8 nodal points, in order to accurately predict stress (or strain)
at the surface of the grain where maximum stress (or strain) very often occurs. The use of skin
pseudo-elements improves accuracy because it reduces the uncertainties connected with extrapolations
which are needed to calculte maximum stress (strain) when there are no skin pseudo-elements.
Predic.tion accuracy depends also on the grid network generated for representing grain volume. The
number of nodal points is limited because of computer power and computer time limitations. A typical
grid network may contain 7000 nodes and 1000 elements.
Exemples of grid networks generated for structural analysis of two-dimensional and
three-dimensional grain geometries are shown on figures 25 and 26.

Figure 25 : Exemple of two-dimensional

grid network

Figure 26 : Exemple of three-dimensional

grid network

- Ko~t=p!.o.e~s.!.n.a .!n.!l.zs.!.s. Propellan t behavio r is assumed to be linear and incanpr essible. Regions of
the grain where stresses (strains) are maximum are identified. Figure 27 gives an exemple of equivalent
stress contour lines (where equivalent stress has a given value) in a three-dimensional grain. In this
exemple, maximum stress occurs at the forward slot-bore junction.

8-22
- e.!.e.!.m.!.n.!!.t.!o~ .f_sy!.s.!.!n_t~e
.!.ei.tl.-e~p!.r.!e~c.!n. ,E.h!. ~o.!t
severe induced loads. Starting
Irom-above-res~lts~stress
calculation is refined in the most
loaded region : a viscoelastic non
linear behavior is assumed in the
case of thermally induced stress
(strain) ; an elastic non linear
behavior is assumed in the case of
ignition pressurization induced
stress

TIME = 0.10000E+01
CONTOURS OF EFF. STRESS (V-M)
MIN- 0.152E+OO IN ELEI"fNT
MAX= 0.186E+02 IN ELEI"fNT

thermally induced stress. The

structural analysis code is
V.E.N.L. (16). It provides"at any
moment of an imposed thermal
cycle, the values of principal
stresses in the propellant
( G""1 th /
th) ()3 th
)
The numerical method is
incremental with respect to the
time. It consists in calculating
stresses at a given time t from
the values observed at time t - At
For this purpose, effect of
thermal cycles are assumed to be
successive infinitesimal strains
with simultaneous relaxation of
the stresses observed at the
preceding time. This code is
commonly used for grain
engineering design analysis and
simultaeously improved by
comparison with actual grain
structural behavior

451
358

CONTOUR VALUES
A = 4.00E+OO
B = 7.00E+OO
C = 1.00E+01
D = 1.30E+01
E = f.60E+01

G2

MAXIMUM

STRESS

Figure 27 : contour lines of equivalent

stress in a finocyl grain

pressurization induced stress. The corresponding structural analysis codes require propellant master
curves and data describing pressurization (pressure rise time, final pressure, temperature). They
provide magnitudes of principal stresses (<5"" 1p ,U-Zp ,U3p) corresponding to maximum pressure.
Often, structural analysis of a propellant grain, which is fired after having experienced
thermal cooldown, has to be performed. Corresponding maximum principal stresses results from the
addition of thermally induced stresses and pressurization induced stresses as far as principal
directions are identical (which is generally the case on grain external surfaces where most stressed
areas are often located).

At the stage of preliminary design analysis, simple analytical formulas provide magnitude of strain
either due to thermal loading or due to pressure rise (see section 443). In both cases ,propellant
capability is obtained from the master maximum strain curves at tla T corresponding to the loading
conditions. So a first assessment of safety factor is

C (due to thermal loading or pressure rise loading)

K

em

(at tla T corresponding to the loading conditions)

Propellant capability (section 442) is related to uniaxial tensile tests ; it is represented by maximum
stess CJ1m) or maximum strain (<m)' Induced stress (strain) (section 443) is the result of a stress
(strain) analysis; i t is expressed as principal stresses (01
,(3) (strains, C
E ,C
3
in the most severely stressed (strained) region of the grain.
1 '
Z

,"2

In order to be abble to directly compare capability and induced stress (strain), failure
criteria are needed (17). They are based on an equivalence between principal stresses and an
equivalent uniaxial stress defined by :
~
2.
2. \~/2.
~/2
- Von Mises criterion
Uo [ (V;;-~) -t-(C2,-CJ?) +(03 -er:;) / 3

:rStassi criterion

0-0

= (cr',;O"";L +~) +l(~

+0-.1.

o~f+ b [(<31 -C>2.)

2.

i(Gi. -63 )<--1(02,

-"';Slr2.}/

a, b, c are coefficients which generally depend on the propellant, but do not depend on strain rate and
temperature. When ()o = , ) m ,the combined STASSI-VON MISES cri terion defines a fail ure surface (of
revolution), in principal reduced stress space ,which is coaxial to the hydrostatic axis ("1 =
=<S})
(figure 28).

az

8-23

HIGH PROBABILITY
OF FAILURE

,62

Depending 01\ the magnitude of ()1

,6"""3 '
STASSI criterion, or VON MISES criterion is
used. STASSI criterion is represented by a
paraboloIdal surface coaxial to the
hydrostatic axis. It is mainly used in the
first octant (1)1 >0,
>0, ~3 )0) which
generally corresponds to thermally induced
stresses.

03 /

8M

OZ

VON HISES criterion is represented by a

right cylinder coaxial to the hydrostatic
axis (and to the STASSI paraboloId). It is
mainly used in the eighth octant (b1.( 0,
6""" 2 <: 0, ~.,<0). This state of stress
corresponds to pressurization at propellant
grain ignition (for an operating pressure of
7 MPa, pressure induces negative stresses in
the propellant, which are much larger than
any other stress). The two surfaces define
Figure 28 : Failure surfaces based on ccrnbined
two regions : internal region represents
STASSI/VON MISES criterion in principal stress space
states of stress which do not involve
propellant grain failure ; surface and
external region represent states of stress involving failure. The parameters defining theses surfaces
(a, b, c) are obtained experimentally, for each kind of propellant, by performing tensile tests under
various pressures and at several temperatures and strain rates.
Propellant grain safety factor is then defined as the ratio of maximum stress obtained in a
uniaxial tensile test (performed at a strain rate and at a temperature identical to those encountered
in the grain) to uniaxial maximum stress (obtained either from STASSI criterion or VON MISES criterion,
depending on the type of stresses encountered in the most stressed region of the grain) equivalent to
the maximum three-dimensional state of stress calculated in the propellant grain :

j
~

m (t/aT)

(Stassi or Von Mises)

Safety factor may also be defined as

Ke. -[; 0

(t/aT)

E.

(Stassi or Von Mises)

is the ratio of uniaxial equivalent stress to modulus,

c;

= ~
E

In the above described definitions of safety factor, it is assumed that the propellant has not
been damaged before experiencing the imposed load. In fact, most of the time, the propellant of a
rocket motor is subjected to a series of mechanical loadings (for instance thermal loads due to
temperature cycles) which damage it. Several investigations have been performed on this subject, e~ther
for general purpose (18), or more specifically on solid propellants (19, 20). The method consists to
define damage D as :

o =

2
i

represents time during which stress i (strain i) is imposed to the propellant, and
represents failure time when stress i (strain i) is imposed.

As defined, and after validation, damage D is equal to

when propellant has not yet been
subjected to loads ; it is equal to 1 when failure occurs. The method therefore consists to decompose
mechanical loads undergone by the propellant during its life into elementary loads and then to canpare
at every moment elementary load to propellant capability. Use of experimental results fran relaxation
tests, creep tests, or various tensile tests allow to assess damage law for propellants.

8-24

G"F

---F . tR 1/ = _1 ,
The simplest f orm of damage law used for propellants ( creep failure law is ~<
stress, t R = time to failure, m = empirical exponent obtained from creep tests) is: Do

= Do~

(Jom

Z"/aT

] 11m

C; is uniaxial equivalent stress based on STASSI-VON MISES failure criterion; m and 00

obtained from experimental tests and a
is the temperature ,shixt factor of time/temperature
equivalence.

are

This cumulative damage method may be applied to the grain entire life including final firing.
It is yet mainly used for grains subjected to thermal cycles before firing, i.e. grains for tactical
missile rocket motors.
Safety factor based on damage is then defined as :
K = 1

Other definitions and prediction methods of safety factors are available. A comprehensive analysis of
most of them was recently performed (21).
In the case of .propellant/liner bonds, the problem is different because of the presence at the
interface of two different materials (propellant and liner). The states of stress/strain are different
on the two sides of the interface. Only the strength applied to the unit interface is continuous. Its
components are: a strength ~ perpendicular to the interface, and a shear strength~. Safety factor
is defined by comparing magnitude of induced interface strength (eon, ~) to magnitude of strength at
failure observed (at same strain rate and temperature) on a propellant/liner bond specimen.
Most of the time propellant/liner bonds are designed so that failure occur in the propellant
near the interface. If, in addition, propellant of this zone has the same properties as bulk
propellant, safety factor is then determined as first described, so that :,

min.

bond
45. Computer aided preliminary design of propellant grains

As mentioned at section 42, there is an increasing

pressure to have, as early as the preliminary design
phase, quick and rather accurate results defining
propellant grain. Moreover, further changes in
technical requirements have to be easily taken into
account. A computer code allowing to satisfy these
needs is now on service. It is named MIDAP (22)
and it involves, today, around 20,000 statements
in its newest version. Figure 29 presents the general
architecture of the code. Each tYPQ of grain
"ontiguration (star-shaped, slotted tube, axisymetric,
finocyl, ) is individually treated inside the code.
The procedure for any of these configurations is
the one which is generally followed to perform
propellant grain preliminary design analysis
(it is described in section 42). The architecture
of the code is modular so that any addition of new
module or any improvement of an existing module
may be very simply worked out

V.2
STAR
SHAPED

Figure 29 : general structure of preliminary

design analysis computer code

8-25
Runs are controlled by the user from the graphic terminal. C.P.U. time is negligible as compared to time
spent by the user performing the design analysis.
The process is iterative and, besides the input of technical specifications, the user has only
to answer yes or no to the option which is proposed on the screen. Results are presented either as
tables or as curves. The block diagram presented in figure 30 enphasizes the role of
propellant/configuration selection, which allows to provide several possibilities,ranked according to a
given set of criteria. The selection of propellant/configuration depends :
- on one hand, on technical requirements (total impulse, burning time, etc ".),
- on the other hand, on semi quantitative requirements related for instance to manufacturing process
praticality, industrial and economical aspects, etc
Due to the dual nature of the criteria, an expert system was selected and implemented for this
critical stage of preliminary design analysis.

All the branches of the code have the

same basic structure. The slotted tube
branch is detailed hereafter

rr======n=="======rr=-=- "- -c..-c..----- .-.>

On flow chart of figure 30, the main

stages of the analysis appear.
Two possibilities are provided to the
user :
- design of a case-bonded (or
free-standing) grain meeting technical
requirements,
- for a given rocket motor, calculation
of motor operation (pressure versus
time, thrust versus time, etc ... ).

DESIGN
PARAMETERS
MODIFICATION

CALCULATION OF HEAD-END

AND AFT-END PRESSURE EVOLUTION

(EROSIVE BURNING)

In the case of slotted tube configurations, geometrical characteristics

which are taken into account for design
analysis are:
- cylindrical motor case, presenting
possibly thermal insulation overthickness in the aft-end zone (slotted
zone)
- plane grain aft-end zone
(slotted zone)
- cylindrical central port,
- slot walls may be parallel or not,
(star-shaped) ,
- possibility of a tapered port in grain
aft zone (in order to limit erosive
burning effects).

~----=====fI=====--liliIl
L . . . . . - - - - - - - - - j f - - - - - - - - - - - I OUTPUT

Figure 30 : detailed flow chart of slotted tube propellant

grain preliminary computational design
The successive steps of the analysis are
!if.s, .iefi~i.i~.n_of .E.h!: ..8.r~i~ ~n.i .h!:l'!!al 2.tf.u.st~r2.l_a~alY2.i2.
This definition meets various requirements, the priorities of which are ranked as follows :
1) structural integrity (for thermally induced strains) corresponding to a safety factor higher than 2,
2) maximum operating pressure,
3) total impulse,
4) evolution of pressure versus time,
S) burning time.
The structural analysis is based on data obtained from regression analysis of stress/strain field
determination results, performed with the aid of three-dimensional computer codes (e.g. CAMILLE,
section 443) on various selected slotted tube geometries
!uf.n.!.n.a 2.r~a_e!.olu!i~~
This task is performed by dividing the grain into three parts as described at section 433. A simplified
version of GEOFIN is used in this preliminary design phase.

8-26
Determination of nozzle throat initial diameter
Ie-is the minimum throat-diameter vaTue consistent with the specification on maximUl1l acceptable
pressure.
Determination of pressure, mass flow rate and thrust evolutions versus time
In-a-first-appro'ach,-erc;sive-burnini 15-oot-taken-into ac~oiint.-The-cal~uTations provide expansion
ratio, and thus nozzle optimum expansion ratio and exit diameter. Nozzle exit diameter is then compared
to corresponding requirement. Afterwards, ratios K (burning area to nozzle throat area) and J (burning
area to central port cross section) are calculated. If needed, a tapered zone is designed in the grain
slots region so as to meet a criterion on J (maximum permitted value). Burning area versus time is then
calculated again.
'E~~~~~~~~~~~~~~~~~~~~~~~~~E~~~~~~
It is based on equal flow rates in any port cross section for actual (three-dimensional) and equivalent
(two-dimensional axisymmetric) contours. It allows a simplified analysis of erosive burning which is
taken into account at the following stage.

Calculation of head-end and aft-end pressure evolution

At-tht'S 'itage erosive burnin&' is-taken-into ac~oiint.-The-module provides pressure evolution inside
grain central port, as well as peak pressure at ignition and pressure rise time.
!r!.d!c.E.i~.n_o!. ~a!.e.YJ!c.E.0E Eela.E.e.!! .E.0_in!t!o~ .E.r!.s~uEi.=.a.E.i~n_
Preliminary structural design analysis is performed, as described at section 433.
5. PROPELLANT GRAIN RELIABILITY

Reliability is probability that a system fullfill a required mission in given conditions and
during a given period of time. Reliability must be considered :
- at design phase; the system must be designed so that its reliability will meet the requirement,
- at realization phase
it must been demonstrated that reliability requirement has been met.
Reliability of a solid propellant rocket motor results from the reliabilities of constitutive
elements, like case, thermal insulation, igniter, nozzle, propellant grain, etc Grain reliability has
several components; but main components are ballistic and structural reliability. In the case of
case-bonded grains, past experience and analysis performed according to F.M.E.C.A (Failure Modes,
Effects, and Criticality Analysis) nave demonstrated that structural reliability is the most important
component of overall grain reliability.
There are two possible approaches in assessing propellant grain reliabili ty
approach and an experimental approach ; both of them are complementary.

an analytical

51. Analytical approach

It is performed according to F.M.E.C.A. method (23). It consists in

- listing the functions the propellant grain must fullfill,

describing failure modes,
assessing probability of failure occurence for each mode,
validating assessments by comparing with overtests and analogs experimental results.
As mentioned above, this method has emphasized structural component of case-bonded grains reliability.
This aspect is therefore discussed hereafter. A comprehensive description of the methods used in
propellant grains structural reliability assessment would need extensive discussion because of the
complexity of phenrnnena and analytical tools which are involved. Consequently, the following section
provides only an idea of the principles governing structural reliability assessment.
Safety factors are considered at section 4 as haVing known values : propellant grain failure occurs
when K = CiS = 1. K is structural safety factor, c; is capability and S is induced stress/strain. In
fact, most of the parameters involved in safety factor prediction are randomly distributed and their

statistical distribution law is not always well known.

These parameters are related to
-

grain geometry
boundary conditions
propellant and bond behavior
capabili ty
failure criterion

which define induced stress/strain field

which may take aging into accoun t

When designing a case-bonded grain, the distribution law of the parameters defining grains and imposed
loads has to be known, so that the distribution law of C and S be known. It is then possible to
determine minimum. safety factor ensuring required reliability. In a second stage, taking into account
variations due to manufacture (and possibly to aging), it is possible to define a mean safety factor
(higher than the preceding one) that must be the objective at design phase. Grains I designed so as to
meet this requirement on K, have the desired reliability at a high confidence level.

8-27
For a given propellant grain, variations of capability and induced stresses (strains) are due
(0 errors in test measurements of propellant and bond mechanical properties,
- to the probabilistic nature of loads imposed to the grain before, and during firing,
- to uncertainties related to structural models and to failure criteria determination.
-

Let C and S be the mean values respectively of capability and of induced equivalent stress (strain)
and CV, and C~ c~r~sponding deviation factors (which are assumed to be independant of the mean
value), then K = cIs, and it can be demonstrated that probabi~y that grain failure does not occur is

Prob

(C)s)

-lff

K -!

(K2 CVC 2 + CV S2) 1/2

is the repart1tion function of normal distribution law ( N (0,1)). C and S are generally
-

where If
assumed not to be correlated (which is not correct but acceptable). It is possible, however, to take a
correlation into account if it is clearly demonstrated.
Ninimum safety factor, K min' ensuring required reliability
value of Kminwhich satisfies the relationship :

If

is then obtained by determining the

~m~~c;

[cK
: CV 2) 1/2] = F
min
S
Taking into account variations due to manufacture, deviation of safety factor is assessed. It is then
possible to calcultate a value of safety factor which is the objective of structural design analysis :
it ensures that grains accordingly designed have a given probability of meeting reliability
requirement.
52. Experimental approach
Safety margin i~ C-S. There is a discrepancy between actual margin of safety (C-S)R and predicted
margin of safety (C-S)C , due to the use of an approximate model. It is possible to write :
(C-S)R = (C-S)c
+ ;- ,where f; is assumed to obey a normal distribution law. ~ mean value,
me ,represents the shift of the model. Deviationui represents variations of this shift. mg and
Oi must be assessed by performing significant experimental tests. There are two possibilities:
either overtests or test on grain analogs.

The first method consists to assume that ~ obey a given normal distribution law and to use overtests
results in order to refine this distribution law: it is the Bayesian method (24). Grain overtests are
tests which have a moderate probability (much higher than in a normal motor firing or temperature
cycling) that failure does occur. Nuch information is thus obtained on grain reliability. Overtests
are defined by changing thermal cycles applied to the grain (colder temperature, larger cycles number)
or firing conditions (reduced nozzle throat diameter) compared to normal conditions.
The most interesting information is obtained when overtests performance does not induce propellant or
bond failure: a more accurate definition of !; distribution law may thus be proposed. Overtests best
definition in however always difficult to establish.

A second method to quantify the shift of the model consists in performing loading tests on analogs
such as the one represented in figure 31 (25)

Case

IW~~
Figure 31 : Analog grain for assessment of propellant grain structural reliability
This analog grain definition has the following characteristics :
- configuration is simple enough so that analogs may be easily manfactured at low cost,
two-dimensional geometry induces low cost computational structural analysis,
ratio ot maximum stress (strain) to mean stress (strain) induced in the analog is of the same order
of magnitude as the one encountered in actual grains,
maximum induced stress (strain) can be easily adjusted by simply modifying analog manufacture
tooling.
The main drawback is that the analog is not .,. the grain itself, which involves that propellant,
liner and bond are not exactly the same, and are not in the same surrounding conditions, as those
constitutive of the actual grain.
Mechanical testing consists in loading a given number of analogs in identical conditions until failure
occur. Analysis of failure results and deviations yieds the shift ~ between actual margin of safety
and predicted margin of safety. It is then assumed that the shift observed on the analogs is equal to
the shift existing in actual grains.

8-28
These set of complementary theoretical and experimental methods, that must still be improved,
allow to assess structural reliability of case-bonded solid propellant grains.

6. CONCLUSIONS
The discussion, outlined in the present paper, on solid propellant grain design methods
presently used in France, emphasizes the significant improvements obtained in this area during the
last twenty years. Phenomena, which were not well understood, or which could not be considered
(because of their complexity) in ballistic and structural description of rocket motor operation, are
now taken into account. It is the result, on one hand, of a better understanding of physical phenomena
or of material behavior, and, on the other hand, of the recent possibility of modelling these complex
phenomena due to the dramatic improvement of both computational methods and computers performances.
More specifically, the three-dimensional aspect of phenomena is more and more commonly treated. These
improvements were naturally transposed into the more accurate design tools. Besides, preliminary
computational design analysis of propellant grain can now be performed, not only by grain design
experts, but also directly by project managers. It involves shorter response times and lower
probabilities of error.
However the situation is far from being fixed. It is continuously evolving because of two
types of factors : needs are still growing, techniques and methods improve. Needs, which appeared
during the last twenty years, are growing : more accurate and quicker responses (though maintaining an
acceptable cost) to propellant grain design requirements are requested. New needs also appear, or will
appear; they are related to requirements which, until now, are not (or are not well) taken into
account during design phase : corresponding physical phenomena are not well understood ; corresponding
eKpression of specifications is difficult or is now considered as being of secondary importance. These
are requirements related to safety, vulnerability, plume characteristics, and grain industrial cost.
During the next ten years, it can be anticipated that grain design methods will therefore
evolve in two directions. A better account of physical phenomena complexity will be taken in the
dimensioning computational tools (three-dimensional aspects, more realistic mechanical behavior of
propellant and propellant-liner bond, two phases and turbulence aspects of flow inside grain central
cavity, etc ). Research work, now performed on better understanding of phenomena involved in safety,
vulnerability and plume characteristics, will affect, on one hand, nature and expression of
requirements in these fields and, on the other hand, the advent of methods for dimensioning propellant
grains with respect to these new requirements.
Performing the entire set of tasks above described requires a close collaboration of
theoreticians and experimenters who must coordinate their energies to consistently and continuously
improve grain design methods.

REFERENCES
(1)

Jean BOISSON. La propulsion par fusee

Ecole Nationale Superieure de Techniques Avancees. tome 1 (1972)

(2)

J. DARDARE, J. MERIGUET, L. VAILHE. Reacteurs-Fusees

Ecole Nationale Superieure de l'Aeronautique et de l'Espace, tome 1 (1981)

(3)

Alain DAVENAS. Amelioration des proprietes balistiques et des proprietes mecaniques tout
temps des propargols sans fumee.
AGARD PEP 53rd meeting, OSLO (Norvege), avril 1979

(4)

Rene COUTURIER. Propergols sans fumee contenant des nitramines

AGARD CP nO 391 (Confidentiel OTAN), janvier 1986

(5)

Guy DELANNOY, Bernard LOUBERE. A physical method for predicting thrust inbalance of solid
rocket motor pairs for a satellites launcher (to be published)
AIAA/ASEE/SAE/ASME 23rd Joint Propulsion Conference, SAN DIEGO (1987)

(6)

Jean-Claude GODON. Modelisation de la combustion normale et erosive des propergols

composites. These Universite de Paris 6 (1983)

~29

(7)

M. K. RAZDAN, K. K. KUO. Erosive burning of solid propellants.

Fundamentals of solid propellant combustion, Progress in Astronautics and Aeronautics,
Vol 90, p 515-598 (1984)

(8)

Guy DELANNOY. Prediction of antitank solid propellant rockets internal ballistics.

AlAA-84-1355
AIAA/SAE/ASME 20th Joint Propulsion Conference, CINCINNATI (1984)

(9)

Alain PHILIPPE, Paul TCHEPIDJIAN. Prediction of longitudinal c~nbustion instabilities in

aXisymmetrical propellant grains. AlAA-84-1358
AIAA/SAE/ASME 20th Joint Propulsion Conference, CINCINNATI (1984)

(10)

Bernard GONDOUIN. Fiabilite des ensembles propulsifs a propergol solide.

Balistiques et Spatiaux a Propergol Solide. A.D.E.R.A (1985)

En&~ns

(11)

M. L. WILLIAMS, R. F. LANDEL, J. D. FERRY. The temperature dependence of relaxation

mechanisms in amorphous polymers and other glass-forming liquid
J.A.C.S., 77, 3701 (1955)

(12)

R. J. FARRIS. Development of solid rocket propellant nonlinear

theory. AFRPL-TR-75-20 (1975)

(13)

E. C. FRANCIS et al. Propellant nonlinear constitutive theory extension.

Preliminary results. AFRPL-TR-83-034 (1983)

(14)

Jean-Noel LHUILLIER et al. Tenue mecanique et fiabilite des chargements

solide. Sciences et Techniques de l'Armement 52, p 11-144 (1978)

(15)

L. R. HERRMANN. Elasticity equations for incompressible and nearly incompressible materials

by variational theorem
AIAA Journal, vol 3, p 1896-1900 (1965)

(16)

G. MElLI, G. DUBROCA, M. PASQUIER, J. THEPENIER. Etude mecanique de chargements moulescolles en propergol double base composite par une methode viscoelastique non-lineaire
Propellants, Explosives, Pyrotechnics, Z, p 78-84 (1982)

(17)

N. W. TSCHOEGL. Failure surfaces in principal stress space

Polymer Science Symposium
11, p 239-267 (1971)

(18)

M. A. MINER. Cumulative damage in fatigue

J. Appl. Mech., Trans. ASME, Series E, Vol 67 (1945)

(19)

K. W. BILLS Jr et Al. A cumulative damage concept for propellant-liner bonds in solid

rocket motors
J. Spacecraft, 1, 3 (1966)

(20)

R. A. SCHAPERY. A micromechanics model for nonlinear viscoelastic behavior of particlereinforced rubber with distributed damage
~lli-4867-86-1, AD-A165728, (1986)

(21)

D. T. WANG, R.N. SHEARLY. A review of solid propellant grain structural margin of safety
prediction methods, AlAA-86-1415
AIAA/ASME/SAE/ASEE 22nd Joint Propulsion Conference, HUNTSVILLE (1986)

(22)

G. URHIG, B. DUCOURNEAU, P. LIE8A. Computer aided design of propellant grains for solid
rocket motors (to be published)
AIAA/ASEE/SAE/ASME 23rd Joint Propulsion Conference, SAN DIEGO (1987)

(23)

L'analyse des modes de defaillance, des effets et des probabilites.

Cahiers de securite de l'Union des Industries Chimiques
Cahier nO 4, Paris (1981)

(24)

}lichel QUlDOT. Methodes d'incorporation de resultats d'essais

Note technique interne nO 98-77-CRB SNPE (1977)

(25)

Jean THEPENIER, Hugues }lliNEZ-COUTENCEAU, Bernard GONDOUIN

Reliability of solid propellant grain :
mechanical analog motor design and testing (to be published)
AIAA/ASEE/8AE/ASME 23rd Joint Propulsion Conference, SAN DIEGO (1987)

viscoelastic constitutive

propergol

la mesure de la fiabilite.

THANKS. The author thanks Bernard GONDOUIN, Roland LUCAS and Bernard PLANTIF who actively
participated in writing this paper of AGARD L8 150 Lecture Series, and Francine VANDEVYVER
who typed it.

9-1

State of the Art of Solid Propellant Rocket Motor Grain Design in the United States
Durwood I. Thrasher
Structural Analyst
United States Air Force Rocket Propulsion Laboratory
Edwards Air Force Base, California, USA
SUMMARY
This paper begins with a brief discussion of the basic ground rules and overall considerations in
solid rocket motor design.
This discussion includes ballistic design considerations, grain features
related to ball istic design and their purposes, primary factors which determine stress and strain
levels, and design approaches for avoiding structural weak points. The major section of the paper deals
with solid rocket motor propellant grain structural integrity assurance, including materials
characterization, structural analysis, and structural capability verification.
The topics addressed
include viscoelastic material behavior and the requisite thermomechanical characterization testing and
anal ysi s approaches; fa il ure criteri a and the appropri ate testing approaches; experimental structural
methodS, including in-situ stress and strain measurement technology; and service life considerations.
This section of the paper also addresses remaining unresolved problems and summarizes some of the more
important ongoing work in the area.
INTRODUCTION
The purpose of a rocket motor or engine is to provide the propulsive force for a missile or other
aerospace vehicle. Solid rocket motors are generally used for applications which require relatively low
cost, long service or shel f 1ife, mechanical simpl icity, and a predetermined thrust level. The sol id
rocket motor is somewhat more 1imited in performance than the 1iquid rocket engine and is less easily
operated in a controllable thrust mode. The typical solid rocket motor has fewer parts (and far fewer
moving parts) than the typical liquid rocket engine, which leads to a fundamental difference between the
design and testing of solid rocket motors and of liquid rocket engines. Liquid rocket engines can be
designed and tested to a much larger extent at the basic component (e.g. injectors, pumps, valves)
level, and the rel iable operation of individual components can be verified prior to their end use by
component testing and by all-up testing of complete engine assemblies. However, a solid rocket motor is
essentially a one-shot proposition. Despite the advent of reusable motor cases, a complete rocket motor
is used only once, and cannot be pre-tested in full operation. As a result, individual rocket motor
rel iabil ity must be assured by assuring the structural integrity of entire popul ations of motors on an
en-masse basis. Heavy reliance on engineering design and design verification processes is unavoidable.
Other consequences of this situation are a strong reliance on past experience and a largely evolutionary
(rather than revolutionary) technology development process.
The design of sol id rocket motor propellant grains is in many ways an exercise in the art of
compromise.
Sol id propellants are developed, selected and employed (in most instances) more on the
basis of intrinsic ball istic performance than for their abil ity to provide structural integrity in a
rocket motor. Solid propellants lack many of the basic desirable properties for a structural material reproducible mechanical behavior, for instance - yet they must withstand severe mechanical loading
without structural failure to actually deliver the desired ballistic performance. The structural design
of rocket motors is necessarily concerned with making such materials perform satisfactorily as
structures, rather than with selecting optimum structural materials. Solid propellant grain design must
be discussed with this necessary process of compromise in mind.
The design process will be discussed in terms of the two basic design areas: ball istic design and
structural design.
BALLISTIC DESIGN
The overall design considerations for the internal ballistic design of a rocket motor includ<: ~h"
physical constraints on the motor, the total impulse requirement, and the environmental limits which
must be met. These requirements are all ultimately defined by the mission (booster, space propulsion,
tactical missile, etc.) requirements. Physical constraints on the motor in~lude length, diameter, total
volume, weight or mass fraction, and center of gravity location. An air-launched motor, for example,
requires a large aspect ratio (typically, LID'" 10) to reduce drag in fl ight by minimizing the frontal
area of the missile, while a motor used for space propulsion will usually have a more spherical shape
(typically, LID'" 1) for better volumetric efficiency.
In the case of multi-stage missiles, the
individual motors' shapes and sizes are dictated by complex tradeoffs between individual stage
performance and overall propulsion system performance.
Most motors must be designed for a nearly
constant center of gravity location. Since motors are always designed as part of a missile or vehicle,
there are typically physical constraints in the form of critical interface dimensions which must be
maintained.
A motor must be designed for storage and operation within specific temperature and humidity limits.
Table 1 outlines the typical environmental requirements for the different types of motors.
Motors may be subject to other environmental factors which could require special gra i n des i gn
features or special design of other components to isolate the grain from the environment. For exampl e,
many motors are des i gned with "weather seals" to protect the propellant gra i n from environmental
conditions such as external moisture, pressure, or vacuum.
The primary motor design requirement, of course, is delivery of a specified total impulse. Usually
other ballistic performance parameters must be met as well.
Typically the designer must meet
constraints on the following ball istic parameters: motor action time, peak pressure, peak-to-average

9-2

pressure ratio, peak ignition-transient pressure, and ignition time.

Action time, peak pressure, and
ignition time are also required to meet maximum variability requirements over the operating temperature
range.
TABLE 1.

TYPICAL ENVIRONMENTAL REQUIREMENTS

Booster

Space

Air Launch

Storage Temperature (F):

Minimum
Maximum

40
90

40
100

-65
140

Operating Temperature (F):

Minimum
Maximum

40
140

40
100

-45
140

Relative Humidity (%) :

Minimum
Maximum

50
85

(Vacuum)
85

20
95

The first design decision which must be made is selection of the propellant.
The primary
considerations here are the required ballistic performance parameters discussed above. The propellant
properties used to select the propellant include specific impulse, burn-rate characteristics, density,
hazards classification (i .e., Class I or Class III), and exhaust product requirements (e.g., minimum
smoke, reduced smoke, or none).
These requirements will guide the preliminary propellant selection.
Typically, there will be no existing propellant formulation which exactly meets all of the ball istic
requirements with the mechanical properties which a new motor design requires.
Usually a sl ight1y
higher performance is required, or the motor design requires a slightly higher structural capability.
One of the early processes in the development of a new motor is a propellant tailoring effort to achieve
the necessary ballistic and mechanical properties. Over the years, continual growth in the capabilities
of solid propellants has occurred through this tailoring to meet motor requirements, as well as through
more intensive development programs to introduce new binders, oxidizers, and energetic ingredients. In
the United States, this growth process has led to the successive replacement of older polyurethane
binders by various new polybutadiene binders in most solid rocket applications; increases in the
propellant sol ids loading from the 80 percent level to 90 percent and greater; the development of
composite-modified double base propellants, and (very recently) the advent of nitrate-ester plasticized
propellants for high-energy applications.
Some of the traditional methods of tailoring propellant to control the
burn-rate characteristics include changing the oxidizer particle size and/or
a burn-rate catalyst or changing the level of catalyst, and changing the
binder.
All of these modifications to an existing formulation have the
propellant's structural properties or affecting its aging stability.

base burning rate and other

the solids loading, adding
heat of combustion of the
potential of modifying the

Once the propellant has been selected, the overall design of the propellant grain begins. A first
consideration is the basic grain geometry.
Figure 1 shows a "generic" solid propellant rocket motor
(While the insulation
which illustrates most of the common geometric features of a propellant grain.
and I iner are parts of the propellant grain, and the motor case has a profound relationship with the
grain structural behavior, discussion of these will be left until later.)
To illustrate some
terminology, the grain configuration shown in Figure 1 is basically a "finocy1" (derived from "fin" and
"cy1 inder") grain with a single radial slot. The primary purpose of the features just mentioned is to
obtain a nearly neutral relationship between burning surface area and burn distance as the propellant
grain burns. In the absence of the other geometric features, the circular cylindrical bore or port of
the motor would rapidly increase in area; in other words, it would have a highly "progressive" burning
characteristic.
In most motor applications, the progressive characteristic of a "pure" internal burning circularport grain would be undesirable for the following reasons:
(1) The efficiency of combustion would be low during most of the burn (efficiency increases with
pressure) .
(2) Nozzle operation would be degraded because the fixed area ratio of the nozzle would result in
overexpansion or underexpanslon over most of the motor duration.

(3) The high pressure at the end of the bu.rn would require a heavy case to provide the necessary
pressure vessel strength.
(4) The resulting progressive thrust history would produce a variable acceleration during the burn,
which is usually undesirable. (In fact, a moderately regressive thrust history would be required for a
constant acceleration, depending on the payload being propelled by the motor.)
A highly regressive grain geometry would be undesirable for the same reasons. In order to achieve
a high volumetric loading in the motor, the designer must therefore incorporate some initial burning
surface in the motor in the early stages of the grain burn back while reducing the net burning surface
later in the burnback.
A feature such as the radial slot in Figure 1 accompl ishes this purpose.
Consider the aft face (i.e., the face on the nozzle side) of the radial slot. Once the bottom of the
slot has "burned out" (reached the case insulation), the area of the slot face decreases with increasing
burn distance. At the same time, the burnback of the slot face toward the nozzle end reduces the length
of the burning surface of the circular bore. The same interaction occurs between the forward face of
the slot and the circular port section forward of the slot. In effect, once the bottom of the slot has
burned to the wall of the case, the two slot faces are equivalent to the burning surfaces at the ends of
a simple uninhibited cy1 indrical grain.
Also note that the shape of the propellant within the dome

9-3

region provides a regressive influence on the surface area evolution. While the fin region in the aft
end of the motor has a much more compl icated geometry, it provides a regressive contribution to the
surface area evolution in the same way as the radial slot. Note that the radial slot has a progressive
area eVolution until the first burnout occurs; this is probably true of the fins as well.
Besides
initial surface-adding features such as slots and fins, the designer frequently applies the inverse
approach: eliminating initial burning surface through the addition of inhibitors or restrictors (pieces
of rubber insulation) to selected surfaces to change the burning pattern of the grain. Depending on its
placement, an inhibitor may influence only the initial burning pattern or produce a drastically
different surface evolution throughout the motor operation.
It may be clear at this point that achieving a smooth surface area evolution and approximate
neutrality in the grain at the same time is very difficult to do; most motor grain designs do have peaks
and valleys in their pressure traces due to local burnout of some grain features and complex
interactions of the surface features as they burn back.

Sutton and Ross(l) have an excellent discussion of grain geometry and burnback. They also cover
the quantitative aspects of rocket motor sizing and the analysis required to calculate internal
pressures based on the usual power law burning rate/pressure relationship and the internal gas dynamics
of the motor chamber.
In practice, the internal ball istics analysis is usually done using computer codes such as the
Solids Performance Program (SPP)(2). This program (discussed more extensively in Ref. 2) has two-and
three-dimensional burnback models coupled with a one-dimensional nonequi1ibrium internal ballistic
model, and is capable of accounting for erosive burning and nozzle throat erosion. The Sol id Rocket
Motor Design and Optimization Program (SRMDOP)(3) uses a pattern-search technique to adjust geometric
parameters of a motor to provide a specified thrust-time trace, to optimize the motor's total impulse to
weight ratio, or to minimize the motor's weight.
The case insulation is typically a filled rubber material; typical fillers are si1 ica and Kevlar~
fibers. (A major development in rubber insulation within the past 1.0 years is the nearly total
disappearance of asbestos-filled rubber because of the health problems produced by aSbestos.)
Particularly in air-launched rocket motors, hard materials such as silica-filled phenolic are sometimes
used as case insulators. Design of the case insulation is primarily based on thermal considerations.
The insulation withstands the high heat fluxes from the chamber gases by charring and ablation. Once
the insulation material has been selected, the required thickness is determined by the time of exposure;
greater thickness may be required in areas of high gas velocity. In the motor of Figure 1, only a very
thin layer of insulation is needed in the mid-grain area, while the insulation must be substantially
thicker under those grain features that burn out early in the motor operation, such as the radial slot
and the aft-end fin tips.
While the aft dome insulation could be "scalloped" to provide thickened
insulation only where it's needed, this is rarely done in practice because of the added manufacturing
complexity.
The above discussion provides a qualitative overview of the ballistic and some other nonstructural
facets of designing sol id propellant grains. We now turn our attention to the structural aspects of
propellant grain design.
STRUCTURAL DESIGN
The structurally important features of a typical case-bonded sol id rocket motor propellant grain
are shown in Figure 1. Structurally, a solid rocket motor consists of:
(1) the solid propellant grain itself;
(2) the liner (or, in some systems, a "powder imbedment layer"), whose primary purpose is to
provide an adhesive bond between the propellant grain and the case insulation;
(3) the case insulation, which provides thermal protection to the case from combustion
products and also structurally supports the propellant grain within the motor case.
Any of these structural elements may become the weakest link in the propellant grain structure.
The primary structural design concern with the insulation is generally the strength of the bond
between the insulation and the case. Generally the propellant does not inherently bond well to the
insulation; this problem is solved by using a 1iner which is formulated to both bond well to the
insulator and to provide a good bond with the propellant when the propellant is cast and cured against
it. Usually the liner uses the same polymer as the propellant itself, and is only partially cured
before casting the propellant, so that a strong chemical bond is obtained upon propellant cure.
The motor case itself, while not considered a part of the propellant grain, structurally interacts
with the propellant grain to various degrees under different loading conditions. In general, composite
motor cases produce substantially larger stresses and strains in rocket motors under pressure loads for
fully case-bonded grains than do metal cases because of the larger dome deflections. While the nozzle
and the igniter are indispensible parts of the rocket motor, they will not be addressed in the present
discussion.
The propellant grain must withstand a series of structural loads throughout the service life of the
rocket motor; normally the final structural load is that imposed by motor operation (firing). The loads
applied to the motor include thermal loads, acceleration loads, vibration loads, and ignition
pressurization.

94
The primary thermal load on the propellant grain occurs under a simple uniform-temperature "thermal
soak."
At any temperature which differs from the motor's natural reference state ("stress free
temperature"), the grain is subjected to stresses and strains produced by differential thermal expansion
of the propellant grain and case materials. Since the propellant grain materials typically have thermal
coefficients of 1inear expansion that are an order of magnitude greater than that of the usual case
materials, relatively large strains are induced in motors at low temperatures. Tensile stresses at the
propellant/liner/insulation bondline can also become quite large. The motor's stress free temperature
corresponds to a temperature near its cure temperature, but is usually somewhat higher due to shrinkage
of the propellant during the cure process. The stress free temperature can be altered by curing the
motor under high pressure.
Additional thermal loads are produced by transient temperature loading conditions such as
aeroheating, aerocool ing, or rapid changes in the environmental storage temperature.
("Rapid" is a
relative term; a large motor may require many days to reach thermal equilibrium.) In aeroheating during
air-carry of an air-launched motor, the case temperature increases rapidly while most of the propellant
grain remains at the original temperature. The temperature gradient produces a sudden thermal expansion
of the case which results in tensile stress at the bondline and a tensile bore strain. These transient
thermal stresses and strains are ordinarily rather small perturbations of the stresses and strains
produced by thermal soak loads.
However, aero heat load i ng of a cold motor can produce
propellant/liner/insulation bond stresses that exceed the"bond strength because of the decrease in bond
strength as the bondline warms up.
Acceleration loads include the long-term gravity load (often referred to as the "slump" loading
condition) during motor storage.
While this load is relatively small, the stresses and strains are
additive to those produced by the thermal storage load. The change in grain shape, or slump, produced
by the gravity load can produce significant changes in internal ballistics.
Additional acceleration loads include
missile motors by aircraft maneuvering, and
flight conditions. Upper stage motors and
during booster operation, and all motors
operation.

the air-carry acceleration loads imposed on air-launched

vibration and shock loads imposed during ground handling and
space propulsion motors are subjected to acceleration loads
are subjected to in-fl ight acceleration during the motor

The ignition pressurization load is the ultimate test of the motor's structural integrity. The
case deformations under pressurization produce distortional strains in the propellant grain, resulting
in
tensile
strains
at
the
motor's
inner bore
surface and
shearing
stresses
at
the
propellant/l iner/insulation bondl ine. The pressurization loading is typically appl ied to a motor that
is already under stress and strain due to thermal load (possibly including aeroheat) and is typically
accompanied by axial acceleration.
The structural response of the propellant grain to all of the loading conditions described above is
highly dependent on the grain configuration.
Two geometric parameters often used to describe the
propellant grain geometry are the web fraction (b-a)/b and the length-to-diameter ratio LlD=L/2b (see
Figure 2). These basic geometric parameters apply specifically to a right-circular-cyl inder geometry
with free surfaces at the ends of the propellant grain, but have at least cQnceptual equivalents for any
motor geometry. Results of closed form solutions for simple geometries(4,5) provide some insights into
the effects of the motor geometric parameters.
Increasing either the web fraction or the L/D will
result in an increase in bond stress and bore strain under any of the usual loading conditions. The
stresses and strains become extremely sensitive to the web fraction for high volumetric loading. More
specifically, bond stresses in an infinitely long grain are approximately proportional to the quantity
~2/(b2_a2ij, while bore strains are approximately proportional to the square of the same quantity. The
stresses and strains are reduced in short grains but are essentially the same as those in an infinitely
long grain once L/D exceeds a value of 10. For a rigid motor case, bonding the grain ends to the case
domes essentially makes the grain length infinite.
In real motors, the highest stresses usually occur at discontinuities such as the grain bondl ine
termination point, and the highest strains occur on the bore surface either in the middle of a long
circular section or at discontinuities in the geometry (such as the forward termination of the fins in
Figure 1). The designer's task is to assure that the stresses and strains at these critical locations
in the motor are within acceptable bounds; meeting this objective constrains the designer's options in
designing the grain geometry to meet ballistic requirements, and sometimes requires special features in
the grain design in addition to those incorporated for purely ball istic reasons. In actual practice,
the ball istic design and the structural design are often done by two (or more) different special ists.
It then becomes the task of the "structural designer" to modify the "ballistic designer's" design to
produce a grain design with structural integrity, without significantly changing the motor's ball istic
characteristics. To accomplish this, the designer concentrates on the critical areas of the motor: the
free surfaces and the bondline.
Consider a motor configuration like that in Figure I, but without the t'adial slot. A high strain
area is located near the center of the long circular portion of the bore; due to strain concentration,
the region where the fin bottoms intersect the bore surface may have an even higher strain. One option
the designer has for reducing these strains is to incorporate a stress relief flap (sometimes called a
boot) as shown in Figure 1. This reduces the effective L/D of the motor, producing lower bore strains
and lower bond stresses in most areas of the motor.
The flap has the additional advantage of
drastically reducing the bond stresses at the bondline termination in that end of the motor, so that the
critically stressed region is moved to the vicinity of the flap termination or "hinge." The crack-l ike
singularity in the flap is buried in the rubber insulation, which is normally far tougher than the
propellant or 1iner.
By thickening the insulation at the flap termination in a smooth contour, the
designer can reduce the stresses at the bondline to an acceptable level. Note, however, that the use of
flaps carries the penalty of an increase in inert weight and a loss of propellant.

9-5

Since the motor of Figure 1 has the radial slot, the flap shown would have little value because the
stresses in that end of the motor are already reduced by the radial slot.
The radial slot also
accomplishes the same effective length reduction as the flap. In this case, the structural penalty is a
highly-strained region at the bottom of the radial slot, and a substantial amount of propellant would be
lost if the designer incorporated the radial slot for stress and strain relief. Because of the profound
effect on the motor ballistics, such a slot would rarely be incorporated for structural reasons.
However, the structural designer might alter the location of the slot (or its width or depth) to gain
some structural advantages at the penalty of slightly altering the pressure trace.
Returning to the aft end of the motor, the designer might find it necessary to increase the width
of the fi ns to reduce the stra i n concentra t i on at the intersect i on with the bore or the stra ina t the
fin tips.
Once again, a penalty of propellant loss would be paid, and the pressure trace would be
altered somewhat.
The substitution of an elliptical or approximately elliptical contour for a
semicircular contour in the fin bottoms can substantially reduce the strains at the critical fin regions
with virtually no propellant loss, but may increase the cost of building the tool ing.
Similar
considerations apply at the bottom of the radial slot. Occasionally, the designer may incorporate a
flap (e.g., in the aft end of the motor of Figure 1) to reduce the strains in a fin region of a motor.
A secondary consideration in the structural design of fins or radial slots is the locally higher bond
stresses which occur when the fin or slot closely approaches the bondline; slight modifications of the
slot or fin geometry may be needed to reduce these stress concentrations.
For long motors, an option that is sometimes used is the stress-relieving liner (SRl). The SRl is
a special built-up insulation assembly which allows the grain to pUll away from the case under thermal
load and is designed to vent chamber pressure into the area between the case and the propellant-to-SRl
bond(6)
The long-term trends in recent years that affect the structural design of propellant grains have
not been trends in basic grain design approaches. Instead, the important trends have been:
(1) the steady growth in propellant strain capability;
composite cases and (3) the increased application of pressure cure.

(2)

the increased application of

STRUCTURAL INTEGRITY ASSURANCE

The above discussion of rocket motor structural design was conducted without specific reference to
details of the mechanical behavior of the materials involved or the methodology required to assess
stresses and strains in
particular motor design. The following discussion introduces these details
and shows how structural integrity assurance is provided for solid rocket motors.

Materials Characterization
The primary materials in a solid propellant grain (propellant, liner, and insulation) are basically
nearly-incompressible, rubber-like materials, and this basic property must be recognized in dealing with
the materials on a structural basis.
All of these materials have bulk moduli of compressibility of
200,000 psi or greater in the virgin (undamaged) state. A typical undamaged solid propellant contains
on the order of one percent or fewer voids on a volume basis. However, the mechanical behavior of these
materials (particularly the solid propellant) is tremendously sensitive to mechanical damage, and the
materials are easily damaged by appl ied loads. The primary result of such damage is the appearance of
porosity and binder-to-particle dewetting, which is manifested as dilatation under tensile or mixed
loading conditions and as initial compressibility under compressive loads.
Mechanically, the propellant and the other grain materials respond as viscoelastic materials, but
show strong strain level dependence and other nonlinear characteristics; hence they are not strictly
treatable as linear viscoelastic materials. Complex mechanical behavior noted in propellants includes a
degree of thermorheological complexity and a nonlinear response to combined straining and cooling or
heating.
Another nonlinear characteristic of propellant is an ability to reheal following damage.
Application of these materials to extreme ranges of temperature and strain rate, as well as strain
levels that range from near-zero to the vicinity ofl00 percent, leads to more than a little difficulty
in characterizing and analyzing them as engineering materials.
The following discussion concentrates on the propellant itsel f; however, most of the information
applies equally to typical insulations and liners. In point of fact, the mechanical characterization of
insulation and liner materials is almost totally ignored in most motor development programs; assumed
properties are usually used for these materials in structural analyses.
Routine laboratory characterization of solid propellants consists of constant strain rate testing,
constant strain level testing, and constant load testing. These are discussed separately below.
Constant strain rate testing is primarily done for the determination of failure properties and for
manufacturi ng quality-control purposes.
The stress-strain curve shown in Fi gure 3 ill ustrates the
mechanical parameters measured in the typical uniaxial constant strain rate laboratory test.
The
initial or tangent modulus of elasticity, Eo, is primarily used as a qual ity-control parameter rather
than as an engineering mechanical response quantity (instead, the relaxation modulus, discussed below,
is used to determine the modulus for stress analysis). The maximum stress, a m, is used to define the
stress allowable for the propellant, while the strain at maximum stress, e: m, is used to define the
strain allowable.
In more recent years, there has been a trend toward using the rupture strain to
determine allowable strain, and to use the "corrected" or "true" stress (stress multiplied by the
quantity (1+e:) to correct for the reduction in area under strain) in calculating am and e:m' Opinion on
the validity of these approaches is divided, however. Typically, the data obtained from constant strain
rate tests is transformed into master curves of am and e:m versus reduced strain rate or reduced time to

9-6

failure using techniques similar to those discussed below for stress relaxation data.
Two types of constant strain level tests are commonly employed to characterize propellants: the
stress relaxation test and the strain endurance test.
The stress relaxation test is the primary source of data used to establ ish the viscoelastic
response (relaxation modulus) of the propellant for structural analys1s.
Typically the stress
relaxation test is run at strain levels of three to five percent. but higher and lower values are
sometimes used. The relaxation test is conducted over a range of temperatures (typically between -65 0 F
and 1100F. although wider ranges are sometimes used). The temperatures and test durations are selected
so that the data from different tests overlap. allowing a master curve to be established for the stress
relaxation modulus as a function of reduced time t/aT.
This procedure. known as "shifting."
simultaneously establishes the time-temperature superposition function. aT. also known as the shift
factor or shi ft function".
A typical propellant master curve is shown in Figure 4; the inset graph shows the variation of aT
with temperature.
When both the relaxation modulus and the reduced time are plotted on log-log
coordinates. as in Figure 4. the shifting can be performed graphically.
Since the shift factor is
conventionally defined as equal to unity at nOF ("room temperature"), the nOF relaxation curve is held
stationary while other curves are shifted to the left (apl) or right (aTd) to obtain a smooth curve.
The resulting shift function is. of course. purely empirical.
Other approaches to shifting the
relaxation data include constraining the aT function to fit the theoretical Wil1iams-Lande1-Ferry (HLF)
relationship.
(1)
where To is the reference temperature (i.e . the temperature for which aT=l).
conveniently transformed version of the more basic HLF relationship.

Actually. Eq (1) is a

(2)
where Ts is a temperature which is approximately 500 C above the glass transition temperature and Cl'.
C2' are the "universal" HLF coefficients(7).
The glass transition temperature determined by shifting
relaxation modulus data usually agrees fairly well with that determined from thermal expansion tests.
(At a fundamental level. the glass transition point and th~ time-temperature superposition principle are
both closely related through the concept of free vo1ume.)ll)
Another relationship sometimes used is the Schapery power law function(5.8),
m
aT = [(To+C)/(T+C)]

(3)

where C and m are constants determined by fitting the data. Generally. the constant C is close to the
glass transition temperature. Several mathematical representations are used for the relaxation modulus
itself. including an exponential series (the Prony series)(8.9) and several variations of the power law
\9,10 .
The strain endurance test is a test to determine the level of strain at which the propellant has
There are several versions of this test in use.
A representative technique
"unl imited" endurance.
involves bo1ding samples at several strain levels for a specified period (typically two weeks), and
reporting the strain endurance value as the strain level at (and below) which all samples survive.
A
sl ightly more sophisticated (and preferable) version of this test requires periodic examination of the
samples so that their failure times can be determined.
Instead of a single "endurance" value, this
procedure provides a curve of strain level versus time to failure such as shown in Figure 5.
The only constant
This test is similar to
(using a simple pendant
a curve of stress level

load test generally used for sol id propellants is the stress endurance test.
the strain endurance test; however, each sample is subjected to a constant force
weight) instead of a constant strain level. The stress endurance test provides
versus time to failure such as shown in Figure 6.

The laboratory tests described above are usually performed under uniaxial loading; however, they
are occasionaliy performed under biaxial loading as well.
In such cases, the tendency is usually to
"cal ibrate" the relationship between biaxial behavior and uniaxial behavior rather than to perform a
complete biaxial characterization. Generally. replicate samples are tested to account for experimental
and material variability.
In addition to the routine testing described above, special tests that are frequently performed on
propellant grain materials include fracture testing (typically using cracked biaxial sheet samples or
notched uniaxial tensile specimens). constant strain rate tests with dilatometry, pressurized constant
strain rate tests, and "similitude tests" consisting of low-rate prestraining of the specimen followed
by a high rate test to failure.
The latter test is intended to simulate the loading history of
propellant in a rocket motor fired under low temperature thermal soak conditions and is used as a source
of failure data for this loading condition.
Occasionally. the "similitude test" is used to establish
modulus values for analysis input(8).
Another frequently-measured special property is the dynamic
Generally. the real and imaginary components (or. equivalently. the real component and the
modulus.
tangent of the phase angle between the excitation and the response) are determined using any of several
test methods and reported in the form of master curves (i .e., as functions of the reduced frequency
waT)
Two additional mechanical response properties needed for completeness of the propellant
characterization are the bulk modulus of compressibility and the thermal coefficient of linear expansion
(TCLE). The bulk modulus is usually measured by applying fluid pressure to a small cube-shaped specimen

9-7

and measuri ng the compress i on wi th 1i near di sp1acement transducers.

There a re several techn i ques for
measuring TCLE; one of the more commonly used methods is the quartz-tube dilatometer test, in which the
sample is heated in a quartz tube and the linear expansion is measured with a displacement transducer.
Bondline strength evaluation entails another battery of laboratory tests.
Several variations of
the peel test (e.g., gO-degree peel, 180-degree peel) are used to qualitatively assess bondline
strength, but are not used to establ ish strength for analysis purposes.
Instead, a multitude of
special-purpose bondline samples are used by various companies and government agencies to determine the
bondline strength.
To obtain uniform bond stress distribution, the bondline specimens typically
incorporate a miniature version of the motor stress relief flap, formed by cutting a slit into the edge
of the insulation portion of the bondl ine specimen.
These samples are often pulled at an angle to
simulate the combined tension/shear stress field in a motor.
High rate bondl ine tests are typically
done under pressure, again to better simulate actual motor conditions.
Structural Analysis Practices
The overall structural analysis process is depicted in Figure 7.
As shown in Figure 7, the
structural analyst must, in principle, make a series of basic decisions about the analysis approach to
use in a particular situation.
The first set of decisions involves translating the overall design considerations into more
concrete analysis input data.
These decisions are fairly straightforward; the important structural
loads are fairly consistent from one motor appl ication to another, and the motor geometry is usually
predefined by the motor designers.
The material properties must be selected based on analysis of
available laboratory data or (often, unfortunately) assumed where actual data are missing. The material
properties consist of the response properties (e.g .. modulus and Poisson's ratio) and the failure
properties (stress and strain allowables). The response properties are used directly in the structural
analysis model, while the failure properties are appl ied when using the structural analysis results to
calculate the margins of safety for various failure modes.
As indicated in Figure 7, all propellant
grain material properties are affected by chemical-structural aging, which must be considered in the
service life determination process.
The selection of response and failure properties is, of course,
interdependent with the selection of analytical treatment and failure theories.
A primary decision the analyst must make is the selection of analytical treatment for the
propellant grain materials. As shown, the analyst may choose to regard the propellant grain materials
as elastic, (either linear or nonlinear), linear viscoelastic, or nonlinear viscoelastic.
While the
propellant grain materials in fact behave as nonlinear viscoelastic materials, in practice they are
rarely modeled in strict accordance with even linear viscoelastic theory in actual structural analyses.
This is the case for a number of reasons, but primarily because no existing theory seems to fully
explain propellant mechanical behavior. The analyst, faced with theoretical shortcomings regardless of
his choice, usually opts for the simplest theoretical treatment, i.e., linear elasticity. In order to
accommodate the viscoelastic behavior of the materials, however, the analyst uses a "quasi-viscoelastic"
approach which involves the use of an effective modulus. The effective elastic modulus is conceptually
based on the linear viscoelastic evolution of stress through the Boltzmann superposition integral (g).
given (for a simple uniaxial strain history) by
(J=

[~[Er(~-~')}d~'

(4)

where [Er(t)] is the relaxation modulus function, ~ is the strain rate as a function of time, and ~ is
the reduced time. For an isothermal loading process, the reduced time is given by
~

where aT is the time-temperature superposition factor discussed earl ier.

history, the Moreland-Lee Hypothesis\g) is generally adopted, so that
~ =

(t

in
_0

(5 )

= t/aT

For a nonisothermal loading

(l/aT)dt

(6)

The instantaneous effective modulus is simply the instantaneous secant modulus,

Eeff(t) = (J(t)/(t)

(7)

where (J (t) is given by the superposition integral, Eq. (4).

It can be shown that for a 1inear
viscoelastic material, the effective modul i for two strain histories that are exactly proportional to
each other are exactly the same.
If the strains everywhere within a given material in the motor are
proportional to a particular normalized strain history (i.e., if the strains are affine), then the
response can be represented by a single effective modulus value throughout the materia.Ta't any instant
of time. The effective modulus is therefore dependent on the "shape" of the strain history, rather than
upon the absolute strain level(5,IO).
Note that these equations imply exact knowledge of the strain history (and in the case of non-isothermal
loading, the temperature history) for the particular rocket motor being analyzed. Generally the analyst
cannot predict the exact thermal or strain history in advance, so rather gross approximations must be
made in the loading history.
In actual practice, the typical method for calculating the effective
modulus is to replace the strain history (t) with a simple step function at t=O and to determine the
effective modulus at a time, t, which is characteristic of the problem.
For motor ignition, for
instance, t would be equal to the duration of the ignition transient. The effective modulus is then
simply equal to the relaxation modulus at time t. While this method is crude, it is often defended on
the basis that the actual load history is not known accurately enough to warrant more exact techniques.

9-8

A frequent practice of structural analysts for treating the effective modulus is to use an orderof-magnitude approximation for the propellant modulus in the structural model, then to scale the
resulting stresses and strains according to the actual modulus later on. Sometimes "unit loads" (e.g., a
pressure of one psi or a temperature change of one degree) are used in such analyses and the results are
sca1ed accord i ng to the actua 1 load.
P,'ovi ded such effects as case-gra i n structural interaction are
effectively constant, this procedure is val id for 1inear analyses. This "unit analysis" approach is a
most useful way of coping with the analyst's typical dilemma; the results are needed "now," but the
propellant is still being tailored.
The structural analysis proper is carried out primarily by finite element computer codes, although
the older approximate techniques based on closed-form elastic solutions (made more practical by the
availability of programmable calculators and desk-top microcomputers) are sometimes used for preliminary
analyses(5).
Two major advancements in elastic finite element analysis for rocket motor pr9pellant
grains were the introduction of the Herrmann reformulation for finite elements in the mid-l960's\11) and
the introduction of isoparametric curved-sided finite elements in the early 1970's(12,13). These two
features are necessary for efficient and accurate modeling of propellant grains.
A typical finite element model of an axisymmetric propellant grain contained in a composite motor
case is shown in Figure 8. Note the concentration of elements in the critical regions near the stress
relief flaps at each end of the motor to obtain precise modeling in these locations. Elements are also
concentrated in the bond1ine region to capture stress gradients and through the thickness of the case to
obtain accurate model ing of the case bending stiffness.
This particular finite element model used
Herrmann-reformulated quadratic-displacement (1 inear strain) isoparametric elements for the propellant,
liner, and insulation; on the order of four times as many elements would be required to achieve the same
accuracy with linear-displacement (constant strain) elements.
The loads are applied to the finite
element model as boundary conditions (e.g., internal pressure, and loads such as nozzle ejection load
due to pressure), or as distributed body forces (e.g., acceleration and thermal loads).
In more comp1 icated grain geometries (such as that of Figure 1) a three-dimensional (3D) finite
element model would be used; however, 3D finite element analysis is avoided whenever possible because of
the tremendous increase in model ing complexity and computer time consumption that results. The recent
introduction of interactive computer-graphics aided finite element modeling codes such as PATRAN-G~ has
greatly increased the efficiency of the motor structural analyst in model generation.
Using an
interactive computer-graphics model ing code greatly reduces geometry and connectivity errors in the
initial modeling and allows rapid "debugging" of a model once it's generated.
The results of the finite element analysis are stresses and strains at the element nodal points or
at "integration points" within the element.
The stress and strain values at the most severely
stressed/strained locations in the motor are used to calculate margins of safety for the motor.
The
procedure of locating the highly stressed/strained areas can be greatly aided by interactive computergraphics postprocessing of the finite element code results.
As shown in Figure 7, the analyst must make yet another basic decision before calculating a margin
of safety for a motor failure mode. This decision is the selection of a failure theory to apply to the
motor failure mode.
In general, three primary propellant grain failure modes should be addressed(9):
fracture (cohesive or adhesive); excessive deformation; and autoignition. These failure modes will be
addressed individually.
Cohesive fracture of propellant grain materials is traditionally considered to fall into two
regions of behavior(9), as shown in Figure 9. The conceptual approach recognizes that propellant and
other materials inherently contain numerous microscopic flaws. The stress required to propagate a flaw
increases as the flaw size decreases; when the flaw size becomes small enough (ccr in Figure 9) the
failure becomes dominated by the ultimate strength of the propellant, and classical failure criteria may
be appl ied over "Region 1." If the initial flaw size is larger than ccr (i .e., in "Region II"), then
fracture mechanics must be applied instead.
For the most part, current analysis practice regards propellant grain failure modes as "Region I"
failure modes, and classical failure theories are applied. No global failure criteria (in the sense of
a failure surface in maximum principal stress or maximum principal strain space) have been successfully
applied to solid propellant grain materials, probably due to the inability to accurately calculate
stress and strain under realistic loading conditions. However, simple failure criteria have been found
to work satisfactorily under specific conditions; for instance, the maximum principal strain failure
criterion is usually successful for defining failure at a free surface of the propellant grain away from
material interfaces. The maximum principal stress failure criterion has been successfully app1 ied to
bond1ines under thermal and gravity loads. Under ignition pressurization, the maximum deviatoric stress
and the maximum octahedral shear stress criteria have been successfully used for bond1ine failure modes.
In applying the above failure criteria, the effects of time and temperature are as important as in
determining the effective modulus. As mentioned earlier, the results of constant-rate laboratory tests
are generally reduced to master ~urves of stress or strain as a function of reduced strain rate, (~aT)'
or reduced time to failure, (Em/EaT). Typical master curves of constant rate uniaxial failure data are
shown schematically in Figure 10.
To determine the appropriate failure stress or strain value, the
analyst would use the appropriate reduced strain rate for the app1 ied load used in the structural
analysis.
For a maximum principal stress or maximum principal strain failure criterion, the data from
the master curves would be used directly; for other criter'ia, the failure stress or strain can be
calculated from the uniaxial values. For instance, the deviatoric stress at failure would be given by
(8)

where 01 is the axial stress and 02. 03 are the transverse stress components (the transverse compOnents

9-9

being zero for uniaxial stress).

For a test done under superimposed pressure the term (-p) would be
added to all the stress components, so that
( af,dev)pressurized

=2

a 1/3 - p.

(9)

Note that a sufficiently large value of a1 will result in a positive (tensile) value of af dey
under pressure, even though the maximum principal stress would be negative. It is this property of ' the
deviatoric stress that makes it a useful failure criterion for propellant grain failure under ignition
pressurization loads.
A different approach to the treatment of constant strain rate failure data is based on crossplotting the stress and strain failure curves, eliminating the reduced strain rate variable. A plot of
log (stress) versus strain, such as Figure 11, is known as a stress-strain failure envelope (often
referred to as the "Smith failure envelope," after its originator, Thor Smith(14).
For some
propellants, the stress-strain failure envelope has been shown to coalesce data from a variety of
laboratory tests, including constant-rate uniaxial and biaxial tests, strain endurance tests, and stress
endurance tests. When such results are obtained, the failure envelope can be very useful as a tool for
selecting failure criteria. For example, a very high loading rate would produce failure near the upper
limits of stress on the failure envelope.
Examination of Figure 11 shows that the failure stress is
relatively constant in this region, though the failure strain varies greatly. One would expect that a
stress-based failure criterion would work best in this situation. On the other hand, an intermediate
loading rate might produce a stress-strain trajectory that crosses the failure boundary at the "knee" of
the curve, and a very low loading rate trajectory would cross the failure boundary near the strain
endurance limit (low-stress end of the curve). These regions of the failure envelope show a much better
definition of failure strain than of failure stress; a strain-based failure criterion would appear to be
the choice for these conditions.
As suggested earlier, superimposed pressure does affect the failure properties (as well as the
response properties) of propellant grain materials. This is usually seen as an increase in the stress
capability of the propellant, while the strain capability may increase or decrease slightly; Figure 12
shows the effects of pressure on constant rate behavior of a typical propellant.
For the failure
envelope, the pressure effects are usually accounted for by considering the deviatoric stress (both
pressurized and unpressurized) as the stress parameter.
When this is done, the pressure effect is
manifested as an upward shift of the failure envelope.
Typical analysis practice is to either use
pressurized failure test data directly or to use a correction factor based on 1imited pressurized
testing. The same procedures are used to account for pressure effects on bondline failure data.
For
consistency, of course, the analyst must account for the pressure effect on the modulus values used in
the finite element analysis. (It should be noted here that this treatment is at variance with the basic
work which has been done with polymers, in which the effect of pressure is treated as a shift in the
time v9riable, similar to that produced by temperature changes and again related to the concept of free
volume(15).
As indicated in Figure 7, another failure theory often used is linear cumulative damage(9). This
failure theory is based on the concept that damage to propellant under constant load is proportional to
the time under load. For a particular sample subjected to a stress a
D = t/tf

(10)

where D is the damage, t is the time under load, and tf is the time to"failure under the stress a. The
damage varies from zero to 1.0; a damage of 1.0 implies failure. The primary characterization data for
cumulative damage is the stress endurance test (Figure 6).
As is evident in Figure 6, considerable
scatter is found in the time to failure at a particular stress; this data scatter must be taken into
account through a probability distribution function.
The relationship used to calculate damage
accumulation for arbitrary stress and temperature histories(9) is
D = [1/(pto(aot-C)B] i t [(at'-C)B/aT]dt'

(11)

where to, aot> C, and B are constants obtained from the stress endurance data, (It', is the stress
history as a function of time, P is the probability distribution function, and aT is the timetemperature superposition function which will be a time-dependent variable unless the temperature is
constant.
For propellants, the stress vs time-to-failure relationship is a simple power law
relationship (a straight line in Figure 6), so that the constant C is typically zero.
Typical practice in the United States for reporting propellant grain structural analysis results is
to report the margin of safety relative to a minimum required factor of safety, based on an allowable
value of the failure criterion parameter (e.g., maximum principal stress); the latter is usually based
on the failure data but is adjusted to reflect aging, data scatter (i.e., pure experimental .error in
laboratory test results), sampling bias (e.g., carton-to-motor bias) and variability (i.e., actual
batch-to-batch or motor-to-motor variation in propellant properties).
Vlhen scatter or variability
adjustments are made to allowables, the typical practice is to use the "lower-three-sigma" (i.e., three
standard deviations below the mean) value. The basis for these adjustments is direct laboratory
measurements where possible, but very often is simply past experience. The typical practice is not to
adjust the modulus for variability or scatter; instead, mean experimental values are used (possibly
adjusted for experimental bias or aging). By definition, then,
MS = [ALLOVIABLE 1 (REQUIREMENT X FS)] -1

(12)

where MS is the margin of safety, FS is the imposed minimum factor of safety, "ALLOWABLE" is the
allowable value of the failure criterion parameter, and "REQI:JIREMENT" is the corresponding value of this
parameter determined from the structural analysis.

9-10
When linear cumulative damage is used, the margin of safety (for a factor of safety of 1.0) is given by
MS

= (1/0)-1

(13)

Fracture mechanics is rarely used in the routine initial design analysis of rocket motors in the
United States.
It is more often app1 ied in analyses done during the investigation of motOt' failut'es.
but even in such applications its use is still rather exceptional. When fracture mechanics is used, the
two classical approaches (strain energy release rate and stress intensity factor) are used with
approx ima te 1y equal frequency.
Some of the reasons for the infrequent use of fracture mechan i cs in
routine propellant grain analysis are the inability to calculate motor stresses with sufficient
accuracy, the rarity of known f1 aws in motors (other than motors rejected for use), and the 1ack of
workable analysis techniques for crack pressurization and propagation in burning propellant.
Excessive grain deflection can be a problem in several different situations.
One of the most
"interesting is the nozz1e1ess rocket motor(16). In the nozz1e1ess rocket, the aft end of the propellant
grain provides a sonic port which effectively replaces the usual hardware nozzle. The nozz1e1ess motor
usually has a drastic pressure drop from the head end of the propellant grain to the aft end. resulting
in large shear stresses and bore deflections that significantly affect the flow field in the motor.
Typically, the grain deflects inward at the aft end. increasing the pressure drop compared to that
produced by an undef1ected geometry.
Since the resulting ball istic/structura1 interaction is highly
nonlinear, the problem has to date only been solved approximately(16). Essentially the same problem can
be encountered in motors with nozzles when the port-to-throat ratio is close to one. Deflections under
pressure loads can have a discernab1e impact on ballistics in most motors(3). Long-term slump response
of a propellant grain under storage loads can also produce excessive deformations which can cause
ballistic problems.
These potential problems should be addressed in motor structural analyses.
In
practice, however, they are relegated to such a low priority that. they are usually not performed for
most motors.
Since the concept of margin of safety does not directly apply to such analyses, margins
are normally not calculated for grain deflections.
Autoignition could result from causes not usually
considered by the structural analyst (e.g., static discharge or high environmental temperatures) but
could also result from internal heating of the propellant under vibration loading.
Autoignition
typically requires an induction time as well as an excitation such as temperature; assessment of the
likelihood of grain failure through autoignition is normally based on empirical treatment of laboratory
data (9).
Structural Capability Verification
It may be obvious from the foregoing discussion that structural analysis results are frequently in
need of verification.
Such verification is the role of the experimental structural methods used in
rocket motor structural integrity evaluation.
The primary experimental structural methods used for
rocket motors are motor instrumentation and motor structural overtest.

Rocket motor instrumentation available for propellant grain structural integrity evaluation
includes case strain gages (which are routinely used in motor firings). internal in-situ stress and
strain gages, and embedded t~ermocoup1es. The use of in-situ instrumentation in rocket motor propellant
grains has a long history(17).
Because of the expense and checkered success record of such
instrumentation, however, its use has been somewhat rare. Recent improvements in transducer stabi1 ity
and accuracy, however, have led to increased use of motor stress and strain instrumentation for motor
structural capability verification.
Figure 13 shows a normal stress transducer designed for use as an embedded gage to measure stress
at the bondline of a rocket motor. I~ use, such transducers are usually embedded in the case insulation
or mounted against the insulation, with a layer of liner between the stress transducer and the
propellant. The usual practice is to select areas where stress gradients are expected to be small.
While it would be more desirable to measure stresses in more critical locations, better results are
obtained in low-gradient locations; also, the chance of inducing spurious motor failures is reduced by
selecting low-gradient locations.
Figure 14 shows a typical clip strain gage, used to measure surface strains in a rocket motor.
Other devices, such as linear displacement transducers, are often used to measure grain deformations.
Dynamic deformations of propellant grains are often measured using accelerometers.
To verify the safety margins of a rocket motor directly, the motor (preferably with in-situ
instrumentation installed) must be subjected to either an overtest to failure or a margin limit test.
The margin 1imit test is defined as a test in which the structural load (usually temperature or
pressure) is app1 ied in excess of the design condition, but at a level lower than that required to
reduce the calculated margin of safety to zero. If a motor successfully passes such a test (statistical
considerations aside), the test provides assurance that the motor has a minimum margin of safety near
the calculated value when subjected to the design load.
The overtest. of course.' provides a more
definite val idation of the margin of safety; it also has the advantage of verifying the critical motor
failure mode.
While either test could conceivably involve a live firing of the rocket; motor. more
information may be gained by using a cold gas or liquid pressurization to simulate a firing(18).
In practice, the margin 1imit test is more commonly used. One reason for this is that a valuable
motor asset is thus saved for other poss i b1e uses.
Another reason is that a conserva t i 'Ie1y-des i gned
motor may be nearly impossible to fail without destroying the validity of the test as a margin of safety
verification. For instance. cool down to failure may require cooling the motor to temperatures below the
glass transition temperature of the propellant.
In any case, drastic departures from the design
condition may result in spurious motor failure modes.
Another useful tool for aiding in verifying the structural capability of motors once the design and

9-11

manufacturing procedures have been established

specimens from dissected motor samples are the
material properties is truly representative of the
motor dissection frequently show the presence
"laboratory" test articles that must be taken
assessment'\ 19).

is motor dissection(18).
Propellant and bond1ine
only direct assurance that the laboratory data on
actual materials in the rocket motor. The results of
of property gradients and other variations from
into account for an accurate structural integrity

structural test vehicles or analog motors (with or without in-situ instrumentation) are sometimes
Structural tests of these devices provide an
used in the structural integrity verification process.
indirect check on material properties and structural analysis procedures since they are more like the
actual motor than are simpler laboratory samples.
Service Life Analysis
The goals of rocket motor design do not end at assuring the initial structural integrity and
performance of the rocket motor. The designer must assure that the motor has the potential to survive
for the intended service life.
Typical service life goals are from 5 to 20 years; furthermore, the
system user requires anywhere from 2 to 5 years' "visibility" (i.e., warning) of the impending end of
the service life of any group of motors in order to prepare for motor replacement.
The techniques used for service life analysis and service life assurance vary, depending largely
upon the cost of individual motors.
For tactical motor systems, there are typically large numbers of
rocket motor assets available for this purpose, and the cost of each unit is relatively small.
Under
these conditions, testing of many individual motors for service life assurance is practical, and the
emphasis in current service life assurance programs for tactical motors is primarily on test firings of
aged (either in the field or in controlled storage) motors. However, the trend of developing technology
is to place more emphasis on temperature-acce1er!!.ted aging and accelerated damage of motors, motor
dissection, and motor structural overtesting(19, 20). A greater emphasis is being placed on the role of
structural analysis and cumulative damage analysis as well.
For larger and more expensive motors with a smaller total population, the emphasis is on analysis,
laboratory testing and analog motor testing.
The basic premise is that the motor service 1ife is
1imited by the degradation in structural integrity with age.
In the ideal service 1 i fe assurance
program, initial structural overtests of full-scale production motors are used to define the actual
motor failure modes. The overtests thus provide the focus for material property testing on aged grain
materials and for structural analysis updates, as well as validation of the original structural analysis
models. Analog test devices designed to duplicate the critical motor regions provide additional failure
data without the cost of testing full-scale motors. Material property tests are performed according to
carefully planned
time-phased, multilevel
test matrices
based on chemical-structural
aging
considerations.
Maximum use is il)ade of temperature-accelerated aging techniques and dissected motor
propellant grain material samples\IB, 21).
In actual programs, however, the full ideal service life
assurance program is often somewhat curtailed.
UNRESOLVED PROBLEMS
The primary unresolved problems in propellant grain design were alluded to in the foregoing
discussion.
The two problems of greatest concern are the current inabil ity to model the nonl inear
aspects of solid propellant behavior in routine stt'uctural analysis, and the inability to successfully
employ fracture mechanics in routine structural analysis. Government-sponsored work is being conducted
in both areas, and in the long term, satisfactory solutions to both should become available.
The area of greatest impact is that of nonlinear propellant behavior. Work on nonlinear mechanical
constitutive theories for prope1laryt grain materials under Air Force sponsorship has proceeded
intermittently for more than 10 years\8, 22, 23). Some results from the most recent completed effort(8)
will provide insight into the magnitude of some aspects of propellant nonlinear behavior.
The linear viscoelastic techniques discussed above were carefully aIJP'lied to a variety of loading
conditions. using a Prony series representation of the relaxation modulus(8). The success of the linear
viscoelastic techniques in predicting the stress response to different constant strain rate loading
histories varied with the strain rate.
When a slightly more complex loading history (a dual-rate
history) was attempted, the results were as shown in Figure 15.
Notice that the stress was
overpredicted by more than 50 percent at the end of the higher-rate segment of the loading history.
Under a more complex loading history with some cyclic content (as well as creep at zero load during some
portions of the strain history), the linear viscoelastic prediction (Figure 16) was in error more than
100 percent over significant portions of the history.
Part of the nonlinear response of the propellant is a manifestation of damage. This is illustrated
in Figures 17 and 18.
These figures are "hysteresis plots," or stress-strain curves for a sample
subjected to a sawtooth strain history consisting of a constant strain rate loading process, followed
immediately by a constant-rate unloading process to the zero-stress point, whereupon the sample was
allowed to creep under zero load. After a period of "reheal ing" time, the entire process was repeated.
Figure 17 shows the results for "zero" reheal ing time.
While the stress response of Figure 17 shows
that the sample had not fully recovered to its original length, the more significant aspect of the
response on the second loading is the characteristic "s-shaped" stress-strain curve. This "s" shape is
considered to be characteristic of damage. A closely-related characteristic is the tendency to return
to the "virgin curve" as the previous maximum strain level is approached(23). 'To further complicate
matters, these damage-induced phenomena tend to slowly disappear if the propellant is allowed to "heal"
in a stress free condition for a time.
As shown in Figure 18, less than a day of "healing" time
produces a significant degree of recovery of the original behavior.
After 30 days of healing time,
essentially all traces of damage typically will have disappeared.
For identical first and second

9-12
loadings, the response characteristics of the "virgin" hysteresis loop appear to be recovered in the
following order: (1) unloading response; (2) strain at zero stress; (3) maximum stress; (4) loading
response.
The first two recovery phenomena may possibly be explainable by linear viscoelasticity;
however, the damage phenomenon itself and the second two recovery phenomena are definitely recognized as
nonlinear behavior.
Another nonlinear phenomenon is exhibited when propellant is subjected to combined straining and
cool ing(9).
As shown in Figure 19, linear viscoelastic analysis drastically underpredicts the stress
response of propellant under these conditions.
The recently completed Air Force sponsored research contract on nonlinear constitutive theories for
solid propellant\8) resulted in significant improvements in stress prediction capability for isothermal
loads; of five candidate constitutive laws investigated, three were developed to the point of being able
to predict stresses for the load histories of Figures 14 and 15 to within 10 percent of the measured
values.
However, combined straining and cooling and response under pressure were not directly
addressed. A follow-on contract now in the early stages will continue the constitutive law development
to cover these loading conditions; in addition, humidity will be incorporated into the constitutive law
as a constitutive variable.
One of the more challenging aspects of the follow-on contract will be
incorporating the constitutive law into a finite element -code so that it can be assessed by comparing
predicted responses with measured responses in instrumented motor tests.
REFERENCES
1. Sutton, G. P. and Ross, D. M., Rocket Propulsion Elements, 4th ed., New York, John Wiley and Sons,
Inc., 1976, pp 354-382.
2. Lamberty, J. To, United Technologies Corp., Chemical Systems Division, "A Report on the Grain Design
and Internal 8all istic Module of the Improved Sol ids Performance Program," 19th AlAA Aerospace Sciences
Meeting, St Louis Mo., Jan. 1981 (AIAA Paper 81-0034).
3. Sforzini, R. H., "Automated Approach to Design of Solid Rockets," J. Spacecraft, Vol. 18, No.3, May
1981, pp 200-205.
4. Williams, M. L., 8latz, P. J., and Schapery, R. A., California Institute of Technology, "Fundamental
Studies Relating to Systems Analysis of Solid Propellants," Final Report, GALCIT 101, Feb. 1961.
5. Leighton, R. A., Air Force Rocket Propulsion Laboratory, "Quick-Look Structural Analysis Techniques
for Solid Rocket Propellant Grains," 1982, AFRPL-TR-81-80.
6. Fullbright, J. L., Rockwell International Corp., Rocketdyne-McGregor Div., "Stress Relieving Liners
for Air-Launched Tactical Rocket Motors," 13th AlAA/SAE Joint Propulsion Conference, Orlando, Fla., Jul.
1977 (AlAA Paper 77-827).
7. Ferry, J. D., Viscoelastic Properties of Polymers, 2nd ed., New York, John Wiley and Sons, Inc.,
1970, p. 318.
8. Francis, E. C. et al, United Technologies Corp., Chemical Systems Division "Solid Propellant
Nonlinear Constitutive Theory Extension," 1984, Air Force Rocket Propulsion Laboratory Report AFRPL-TR-

83-071.
9. Noel, J.S., Rockwell International Corp., Rocketdyne-McGregor
Structural Integrity Analysis," 1973, NASA SP-8073.

Div.,

"Sol id

Propellant

Grain

10. Thrasher, D. 1. and Corbett, E. C., Air Force Rocket Propulsion Laboratory, "An Analysis of a Sol id
Propellant Transient Viscoelastic Response under Motor Ignition Conditions," 1982, AFRPL-TR-81-55.
11. Herrmann, L. R., and Toms, R. M., "A Reformulation of the Elastic Field Equation, in Terms of
Displacements, Valid for all Admissible Values of Poisson's Ratio," J. Applied Mechanics, Trans. ASME,
Vol. 86, pp 140-141, 1964.
12. Zienkiewicz, O. C., The
McGraw-Hill, Inc., 1971.

Finite Element Method in Engineering Sciences, 2nd Ed., New York, NY,

13. Hofmeister, L. D., Isoparametric Finite Element Analysis for Solid Rocket Motors, Computers and
Structures, Vol. 3, Nov. 1973, p. 1369-1376.
14. Smith, To L., "Characterization by a Time and Temperature Independent Failure Envelope, J. Polymer
Science, Part A, Vol. I, No. 12, Dec. 1963, pp 3597-3615.
15. Fillers, R.W., and Tschoegel, N.W., "The Effect of Pressure
Polymers," Trans. Soc. Rheol., Vol. 21, No. I, 1977, pp 51-100.

on

the Mechanical

Properties of

16. Procinsky, I.M., and McHale, C.A., "Nozzleless 800sters for Integral-Rocket-Ramjet Missile Systems,"
J. Spacecraft, Vol. 18, No.3, May 1981, pp 193-213.
17. Danieu, D. J., and Ruggles, V. L., Thiokol Chemical Corp., "Minuteman STV Stress Analysis and
Testing," 1974, Air Force Rocket Propulsion Laboratory Report AFRPL-TR-74-27.

9-13
18. Neely, R. B., TRW Systems Group, and Veit, P. W., Aerojet Solid Propulsion Co., "Solid Rocket Motor
Grain Design Verification Through Cold Gas Pressurization and Motor Dissection," AIAA/SAE Joint
Propulsion Conference, San Diego, Ca., Oct. 1974 (AIAA Paper 74-1202).
19. Thrasher, D. 1., and Hildreth, J. H., "Structural Service Life Estimate for a Reduced Smoke Rocket
Motor," J. Spacecraft, Vol. 19, No.5, Nov. 1982, pp 554-570.
20. Svob, G. J., and Bills, K. W., "Predictive Surveillance Technique for Air-Launched Rocket Motors,"
J. Spacecraft, Vol. 21, No.2, Mar. 1984, pp 152-157.
21. Kelley, F. N., and Trout, J. L., Air Force Rocket Propulsion Laboratory, "Elements of Sol id Rocket
Service Life Prediction," 8th AIAA/SAE Joint Propulsion Conference, New Orleans, La., Nov. 1972 (AIAA
Paper 72-1085).
22. Quinlan, M. H., "Materials with Variable Bonding," Arch. Rational Mech. Anal., Vol. 57, 1978, pp
155-181.
23. Quinlan, M. H., Air Force Rocket Propulsion Laboratory, "An Appl ication of the Theory of Materials
with Variable Bonding to Solid Propellant," 1979, AFRPL-TR-78-37.

9-14

Case
Termination

Liner

jE~~i~~~""'" Insulation

Igniter

Liner
Insulation
"::''-'-'':'''''-'-fII--

Motor Case

Figure 1.

Fin

Generic Solid Propellant Rocket Motor

Ha:Cimum stress am

t ~ ---------

u.

-Hf---- - - - - - -

------

Basic Geometry Parameters

(stress at
break Gb

----=-=J.1__ ...,
Strain
~

at maximum'

~Strain

stress em :

1----- L ------1

Figure 2.

Bo re

break

at
E: b

Strain E: ---..

Figure 3.

Uniaxial Stress-Strain Behaviorat Constant Strain Rate

9-15

200,000
100,000
60,000

Temperature
F

10,000
6,000

RJ:t"

3,000

~ ')~

'Vi
n

r~

.....

-.,
~

1,000

-<

V>

:J

600

'3

E
c
.!'!

200

100

'0
0

n;

a:

60

'r

1
0

i" "

-1

7.52
6.60
5.19
3.84
2.20
0
-123
-2.30

i l ' l r I hl

G(

..
!

~~

L~

)(

-45
-42
-29
-2
6 19
o 76
0118
0152

~~ t@ ("\

log AT

IVr;:;

-11'-'

"Lll

~~
.....
-u U13

........
I".. ~

Y' r-....

-2
-80

-40

40

80

160

120

Tempeuture F

10
10-/

10-2

10-'

10 -1

10 0

10'

10

Reduced lime II/Arl, min

Figure 4.

Stress Relaxation Modulus Master Curve for a Typical Solid Propellant

100

eo
60
40

20

...

".

.S
~

..
.3

10

...

'n;

"

.c

";'

...

"'"

fOO
nme 10 laliure. min

Figure 5.

Strain Endurance Data for a Typical Solid Propellant

9-i6
1,000 , - - - . . , - - - - . . , - - - - . . . , . . . - - - . . . . . , . . . - - - - - - , 100

w
Remarks: All samples good breaks

t----+-----+----+----+---~ 40
~

a.

---=t'_~------:~---I-----I

-; 200 1 - - - -

~
;. 100
~

20

'----..j...:::''D-=:::---+-----+-----+------.:::::::::~

<i
10 ._~
~.

~ 60

~
30

t----+-------+----+----+--=::::".-J

10

!-.

10 0

-.l....L-_ _---l

102

10 1

J.....I..

L_L--_..1 1

103

104

105

Time 10 failure, min

Figure 6.

Stress Endurance Data for a Typical Solid Propellant

~r------~
o
~
~

'--

--.J

STRESS
ANALYSIS

in

6,--------~

~~-~=======---.I

e3'-----_---.J
o

ALTERNATIVE FAILURE THEORIES

Figure 7.

The Structural Analysis Process

9-17

Blowup of Aft Case/Grain Grid

Aluminum boss

Blowup of Forward Case/Grain Grid

Figure 8.

Typical Finite Element Model (Composite-Case Motor)

\
\

..

,,!l3. 0
o

Failure stress \

----I

2.5

- - -_Ultimate strength

Fracture stre,9$
a cr

w
w

,::
en

Failure domina ted by ul tima te strength

Failure dominated
by ini tial fl...

Region I

Region II

Figure 9.

cr

Flaw size c --...

Conceptual Relation Between Flaw Size

and Type of Failure

Figure 10.

Constant-Rate Failure Data

Master Curves (Schematic)

9-18
3.000

1.000

il.

300 _

-~

Temperature F

-30

.10

(j,

20
77

a
;;

.1

[J

140

180

o~

-60

~
:i

100

o P

;t
30

30

20

10

80

50

Strain,%!

Figure 11.

Stress-Strain Failure Envelope for a Typical Propellant

70 F. 1,000 psig
2051n,Jmin
400

'iii

300

c.

~
~

'"'"~

en

200

100

OL-_...J.._ _I.-_...J.._ _J . - _ - I . - _ . . . . I . . . _ - - '

10

20

30

. 40

50

60

70

Strain, 0',

Figure 12.

Effect of Superimposed Pressure on Uniaxial

Sample Stress at Constant Strain Rate

IOfTMtl--IZ'
or
r

HHiI

9-19

MIN

f -$~===aF:edCItEDE::
;:=:~iEl.'lI~~
Zs
Z
10.0"

MIN

.700 MAX

DIA MAX300SERlES CAU TUlIt

DIA MAX SOOStRIn eRts TUBE

5 NO.2I!JlWO TEFLON
INSULATED TYPE E COlOR

COOED WIRE PEA Mtl-W-161870

Figure 13.

Normal Stress Transducer

CLIP STRAIN GAGE

STRUCTURE

MOUNT I NG PADS
(BONDED TO PROPELLANT
SURFACE)

Figure 14.

qO

Clip Strain Gage

TtNPEA"TUA[ MISTDAT, 10
STAn, N HIS lOOT', )0

- "(RSIIO[O STAO IN

3S

1",

30

-...
z

a:
a::

2S
20

IS

V1

10

.-

S
0

~~~_~L~I-N~[h~n:-:'V':'"Is"":c"":o':'"tL-n"":s-:'':'"1C--S-:,"-:E-~-:S---r---""---""'--""''":X:::H::D'":n:::n:':'':'r

I 60

'":D::-.-;I~IN::",::":':,;:'N--:'':0--:':'"'.';;'O'":';';:N;";':";';N:-T
'

'":H::':':":5":"O::O:':T:'"',

<!l - "[AN "[051111(0 SIn(ss

::
V1

120 1 - - - - 1 - - - - - - - - - - 100

80 ~--_j---f--_+_---t---_t--___Ir_--I------'

~~~

V1

IU,

...a::

V1

I~-t_-.---- ----- ---

:: ------- - - - - - - - - - - ----- -----+----r----+-....

.I

2:L::6d:::d::::=t=t:LL-LJ----J
o

10

'5

20

25

30

TI ME. 1M I HI

Figure 15.

35

qO

QS

50

SS

GO

" E I

Linear Viscoelastic Stress Prediction Compared to Measured Stress (Two-Rate Strain History)

9-20

Iq

r:l!I:"'-~"'::ER::":S~U'::RE::":D:""::ST::":R~R~1"~----r-----..,.-----....,r-----"':l~E~Hp::":E:=R'::RT~U:=RE~H~I'::ST:-::O~R'::T1-=1::"~"':Fr------r

12

STRRI" HISTOR1I 12.e.12.~.e.D.~.O

f-----_t_---!!r--t_~dl!r--_t_-----~-----_r-----t_----_f

1-

10

8f-----~---~\IF-----=~----_,w,t_----+_----_t_----_f

a:
a:

VI

f-

t-~+--I--------

-----I---'ll/f'---I---I!\}n---I---nUt---t-----t

OI!P-----I-----l.------O..

....

-4-

+--

80 l"':1!I:-".":'l":'IH~E:-::R~R-:Y:-::I~SC::":O:-::E~lR::":S:-::T":'I[~S":'TR::":E:-::S~S"T""-----..,..-----..,..-----..,..--:X~HD~R~RT::":E~H::":IS:":r'::OR::":Y~,~1
70
(!) - "ERII "ERSURED 5 TRESS

--4

":'I~"/~H:":'I~"-,

60 t - - - - - _ t _ - - . . J ! ! - - - t _ - - - - - j - - - - - - + - - - - - - j - - - - - - t _ - - - - - J
50
qO
VI
VI

a:

f-

t----~'jlf----<\HJt_-tm"'-;iif_lIJ_-+-----t_----+-----t__----_f

30
20

f---;,rfD---_t_---OJIoo--tnl':l=---''\;;,-~~--_"II''__71kr----_t_-_=---t_----_f

VI

10

0f-~~___i-_=t~-F~~~~t!-_1
"__
...I-l__'

-llIf------l--

10

15

TIME (MINI

Figure 16.

20
H

25

30

35

E I

Linear Viscoelastic Stress Prediction Compared to Measured Stress (Complex Strain History)

20 C lesl

1
"

'"'"
'"

V>

Strain ()

Figure 17.

..

Load-Unload Curves with Zero Healing Time

9-21

20 C test

J
"

First and second unload

..
Figure 18.

Load-Unload Curves with 1000 Minutes' Healing Time

.~

plii

60

50

Computed

Experimental data

data

.~

10

.d'
/'

lU-..
/" "-

...8--C1'

o b-o~=.:!=-::.:-=.o-:....J-L-_-_-L_ _--l_...L_L-_ _..l-_ _:J..

+150

Figure 19.

+100
+50
Cooling

-50 -75 -50

Temperature

0
Uesting

+50

of

Stress Response Under Combined Straining and Cooling

10-1

DESIGN AND ANALYSIS OF SOLID ROCKET MOTOR

INTERNAL INSULATION

by

Mr ATruchot
Agence Spatiale Europeenne
Division STS-AR
8-10, rue Mario Nikis
75738 Paris Cedex 15
France
The main purpose of internal insulation is to protect the motor case from the thermal environment of
propellant combustion products. Several secondary objectives must also be met. Internal insulation must
bond to propellant and case, transmit case strain into propellant and sometimes seal the case.
In the first step, the general process of internal insulation design is presented
evaluation of
thermal environment, selection of material, thermal and structural analyses. For each phase the different
analysis methods are described : aerodynamic, thermal, and structural.
To conclude the presentation, the different phases of internal insulation fabrication are
ted : elastomer formulation, fabrication of parts and integration into the motor case.

presen-

1. - !~Q~~!!Q!
The main purpose of internal insulation is to protect the-motor case from the thermal
combustion products. The following objectives must also be met

environment

of

bonding to propellant and case

inhibition of certain propellant grain surfaces
transmission of case strain into propellant
tightness of the case for composite chambers.
The case insulation integrity must be maintained during the life expectancy of the
elastomers used must maintain their properties during aging and environmental conditions.

The
internal
is sized
flaps,
internal

motor.

The

general configuration of a motor case loaded with propellant is presented in figure I.

The
insulation is located between the propellant and the case. The internal profile of the insulation
so that the maximum quantity of propellant can be introduced into the structure. The stress relief
which are represented in the forward and aft domes, are used to prevent case deformations due to
pressure from inducing excess stresses in the internal insulation or propellant.

Propellant grain

Stress relief flap

Fig. 1 -

GENERAL CONFIGURATION OF INTERNAL INSULATION

The general process of the internal insulation design is presented in figure 2.

ves four steps.

This process invol-

In the preliminary phase an evaluation of the thermal environment is required. The design of internal insulation depends on parameters such as combustion pressure, duration, internal flow fields, combustion geometry, and type of propellant.

The selection of the material is based on its thermal and mechanical properties.

This step calculates the thickness of insulation on each point of the internal profile of the case.

10-2

A mechanical analysis of the motor case loaded with propellant,

gradients or internal pressure, is required in order to verify the
insulation. This step allows to refine the stress relief flaps design.

subjected to loads such as thermal

structural behavior of internal

EVALUATION OF THERMAL ENVIRONMENT

-

internal balistic
combustion products characteristics
pressure - duration
grain design
internal flow fleld etc...

,
,

MATERIAL SELECTION
DATA
- characterization
- firing test results

THERMAL ANALYSIS
- erosion rate prediction
- insulation thickness evaluation

STR UCTURAL ANAL YSIS

- design of stress relief flaps

Fig. 2 -

DESIGN PROCESS

The insulation thickness is a function of exposure time of each point of the internal profile of
the case. For this reason the design of internal insulation is greatly dependant on the combustion geometry. Thus in the example presented in figure 3, the insulation thickness varies from 14 to 1 millimeter,
depending on the exposure time of each point.

Fig. 3 -

EFFECT OF GRAIN DESIGN ON INTERNAL INSULATION THICKNESS

10-3
The erosion rate of internal insulation increases with the gas velocity. Thus for external nozzles, the aft dome erosion rate is greater due to the increase of gas velocity near the wall. A decrease in
rubber thickness is possible with submerged nozzles, because the gas velocity near the aft dome is
significantly lower.

EXTERNAL NOZZLE

Low mach numbers

SUBMERGED NOZZLE

Fig. 4 -

EFFECT OF NOZZLE CONFIGURATION ON INTERNAL

INSULATION DESIGN

The design of internal insulation is also a function of propellant grain design. Thus for finocyl grains, the internal flow is greater in front of the slots, creating a non symetric erosion of internal insulation.

Finocyl fins

Nozzle

Internal insulation

~~~-~-~~/

Fig. 5 -

FlNOCYL GRAIN

10-4
These two examples illustrate the importance of an accurate evaluation of the internal flow field
inside the motor case.
For this purpose, two or three dimensionnal analyses are used to calculate gas
velocity in the motor. The results of a typical analysis are presented in figure 6. This analysis uses a
method of singularity assuming an irrotationnal and ideal gas.

----

............

............

"

I
/

Fig. 6 -

"

3D INTERNAL FLOW FIELD ANALYSIS

The selection of rubber is based on thermal and mechanical properties.

Thermal properties are
based on laboratory characterizations and firing tests results. The mechanical properties are a measurement of the material behavior in response to applied tensile and shear forces. Testing of bonding to the
propellant and the case is also required.

The thermal diffusivity of a rubber can be measured by tests performed in laboratory. However
this sort of testing is limited in temperature by the chemical decomposition of rubber, which occures at
approximately 300C.
The response of rubber to a heat flux can be evaluated by tests using oxyacethylene
torch or plasma arc. These tests are not representative of the ablation process by hot gases and firing
test are required to evaluate ablation as a function of propellant gas velocity.
Preliminary firing tests are performed at a reduced scale, in order to investigate a large
number of specimens at a low cost. Later fullscale tests are performed. Two examples of test motor are
presented in figure 7. The first one evaluating specimens with constant gas velocity and the second one
with variable gas velocity.
Propellant grain
Cylindrical specimens

First case .' constant velocity gas

10-5
Propellant

Second case

8 specimens

thermocouples

variable gas velocity

Fig. 7 -

SUBSCALE FIRING TESTS SCREENING OF MATERIALS

These tests obviously present limitations and can not replace fullscale tests,
mit evaluation of many specimens at a low cost, with different test conditions.

but they per-

The mechanical response of an elastomeric material to applied forces is measured by tensile or

Typical specimen geometries and measured stress-strain curves are presented in figure 8
shear specimens.
and 9.

b - STRESS STRAIN CUR VE

Stress C (MPa)

Mean modulus E (MPa)

a - SPECIMEN DESIGN

I
I

C200

I
I
I
I

C 100

I
I Strain (%)

o
Fig. 8 -

TENSILE TEST

100

200

10-6
Strength (daN)

qtI:=~I'~~!I=~!!r~J

1+
Fig. 9 -

SHEAR TEST

A common test for elastomers is the peel test. This test does not give tensile or
but it does verify bonding to the propellant and the case.

shear

data

Strength
F

Bonding

Elastomer

Fig. 10 -

PEEL TEST

The criterion for thermal analysis is the maximum allowable temperature of the motor case. This
temperature is defined by the structural capability of the case and the bonding integrity of internal
insulation. Sometimes some equipment located near the motor case can be a limitation factor.
2.3.1. Ablation process by bot gases
At a specific time of combustion,
ness (see figure 11).

three zones can be considered in the insulation thick-

~ _Motorcase
Zone C

CHAR ZONE
(T> lOOOC)

Zone B

.-

DECOMPOSITION ZONE
( 350C<T<lOOOC)

VIRGIN ZONE
(T<350C)

Zone A

----

Hot gases

Fig. 11 -

Receeding surface
Initial surface

ABLATION PROCESS

10-7

In the virgin zone near the motor case (zone c) the temperature is low and the
decomposition reactions are not initiated. The heat transfer is achieved by a conduction process. In the
decomposition zone (zone B), a great part of the energy is absorbed by the chemical reactions which are in
majority endothermic. In this zone the energy transfer is achieved by a both conduction and mass transfer
proces9= In front of the combustion products,

the rubber is fully decomposed and is at the state of carbon

char.
On this chemical erosion effect is superposed a mechanical erosion due to the friction of
boundary layer. This effect becomes preponderant for gas velocities exceeding 40 m/s.

2.3.2. Analysis method

The ablation process is a very complex phenomenon depending on the combustion products,
the elastomer formulation and the boundary layer configuration. For these reasons, its accurate modelling
for thermal analysis is very difficult. Simplified methods are used in usual practice. One of these methods
is presented in figure 12. This method consists to assume on isothermal ablation of the rubber. The eroded
thickness is supposed to be a result of a constant erosion rate. This erosion rate is estimated with
results obtained in firing test. Some empirical correlations are made in order to take into account
parameters such as exposure time, gas velocity or combustion pressure. In the remaining material, the heat
transfer is achieved by a conduction process with equivalent thermal properties.

Temperature

Va

...-Ta

Motor case
Equivalent thermal
properties (A. P, Cp)
Va : erosion rate
Ta : ablation temperature
Thickness

1-. .1_
. _.I
Fig. 12 -

ANAL YSIS METHOD

All the data needed for this calculation such as erosion rate,
equivalent thermal properties are obtained from firing test measurements.

ablation

The mechanical analysis is the last step of internal insulation design.

variations, flight loads or internal pressure at ignition.
Temperature variations can be induced by processing of propellant,
variations may induce stresses in the internal insulation by the difference of
case and the propellant.

temperature

and

The loads can be thermal

handling or storage.
thermal

expansion

These
of

the

At motor ignition, the internal pressure creates different deformations of the case and propellant. The most severe conditions are obtained for ignition at low temperature. The effects of contraction
of the grain due to temperature and exp~nsion of the case due to internal pressure are directly cumulated.
Finite element analyses are used in order to verify the structural design of the case loaded with
propellant. The results of a typical analysis are presented in figure 13 and 14, taking into account the
cumulative effects of internal pressure at ignition and flight loads.

10-8

Fig. 13 -

Fig. 14 -

MODELiNG

STRAIN CALCULATION

FINITE ELEMENT ANAL YSIS

10-9

The composition and fabrication of an elastomeric internal insulation are very similar to those of a
classical reinforced plastic. The material consists of an elastomer such as EPDM or natural rubber, reinforcing fillers (carbon, silica or Kevlar) and different ingredients such as plasticizer, antioxidant and
vulcanization agents. The fabrication of the part is made by the same processes as those used for
reinforced plastics : molding, winding or lay-up.

Some elastomers used for internal insulation are presented in table 1.

can use a mixing of these different ingredients.

NAME

ASTM ABREVIATION

Natural

NR

0.93

Butyl

IIR

0.93

EPDM

EPDM

0.86

Neoprene or
Polychloroprene

CR

1.23

Sometimes

formulation

DENSITY

----------------- ------------------ --------Fillers have no chemical effect in the formulation of the rubber. They are used as reinforcement
in order to improve thermal and mechanical properties of the product. The main fillers which have been used
for internal insulation are carbon black, asbestos, silica or kevlar. The erosion resistance of the
material is a function of the filler used.
Other ingredients are added to the formulation such as plasticizer, which are used to facilitate
mixing of the ingredients, vulcanization agents, antioxidants, etc

The fabrication of an elastomeric internal insulation involves 4 phases.

Initially a compounding process is used in order to blend the different ingredients. This operation uses internal mixers or rollmills. This step is an important phase of the process and the quality of
the final product is greatly dependant on this operation.
In a second step a calendering process is used in order to give to the
shape of sheets thus allowing the fabrication of the part.

unvulcanized

rubber

the

The fabrication of the part is made by processes such as molding, lay-up or winding. A molding
process (see exemple in figure 15) is well suited for small parts produced in large quantities. However "he
cost of tooling is very high and for reduced quantities, a lay-up process can be used (see figur~ 10). This
type of process is difficult to automate. For this reason, winding process becomes more and more usual.
This process presented in figure 17, leads to a more reproductible quality at a lower ~ost.
The vulcanization i. an operation of curing under pre55ure,
mechanical propertie~ to the part;

For molded part9 y

trical resistances incorporated into the mold.

thus giving the

final

thermal

and

vulcanization can be performed by heating with elec-

10-10

A - B

TRANSFER MOLD

MATERIAL RESERVOIR

PISTON

MATERIAL

TOOL ASSEMBLED

VULCANIZATION

Fig. 15 -

DEMOLDING

TRANSFER MOLDING PROCESS

10-11

ELASTOMER

L;j
WRAPPING

WRAPPING

Cutting

Cutting

Lay-up on mandrel
WRAPPING

Lay-up on mandrel

VULCANIZATION

Cd

VACUUM BAGGING

Transfer of lay-up into the mold

TI
"""

#'

VULCANIZA TION

\/~\

VULCANIZATION

MACHINING

Machining

Vulcanization

INTERNAL
INSULATION
Fig. 16 -

LAY-UP PROCESS

10-12

SPOOL Of CALENDERED RUBBER

WINDING

VACUUM BAGGING

VULCANIZA TION

MACHINING

CYLINDRICAl.
INTERNAL
INSULATION

AFT DOME
INTERNAL
INSULATION
Fig. 17 -

WINDING PROCESS

10-13

The integration of internal insulation into the chamber depends on the technology used for the
case construction. Thus for metallic chambers, the internal insulation is bonded into the case with a
pressure applied by specific tooling. In the case of kevlar or graphite chambers, the filament winding
process can be performed directly over a sand or metallic mandrel equiped with the internal insulation.
Afterwards the mandrel is removed in order to cast the propellant into the chamber.

The most important step of internal insulation design is the thermal analysis. This analysis is always
performed using mainly experiments, correlations and simplified methods. However important improvements
have been performed in the field of elastomer formulation, thus increasing the insulation performance and
decreasing the density of the rubber.

11-1

OVERALL OPTIMIZATION OF SOLID ROCKET MOTOR

by

Mr A.Truchot
Agence Spatiale Europeenne
Division STS-AR
8-10, rue Mario Nikis
75738 Paris Cedex 15
France
ABstRACT
The overall optimization of a solid rocket motor consists of determining the most efficient configuration of the motor in terms of range or payload of the vehicle. After a presentation of the general requirements of a solid rocket motor, the different design parameters are detailled : operating conditions, geometric parameters and technologies used for each component. In order to achieve parametric studies, a computer program has been developed by SEP. The principle and main characteristics of this program are presented
and two examples of application are given in order to illustrate the possibilities of the program.

1. -

INTRODUCTION

Based upon the general requirements of a propulsion system, a great number of motor configurations are
possible ;
thus it is possible to change the operating conditions (pressure, duration), some geometric
parameters (case length, nozzle submergence, etc ) and the technologies used for each component.
All
these parameters are closely linked and influence upon motor efficiency. The main objective of the optimization is to determine the most efficient configuration in terms of range or payload of the vehicle.

2. - REQUIREMENTS - OPTIMIZATION CRITERIA

The overall optimization of a solid rocket motor is an iterative process. The requirements of the propulsion system are a result of a trajectory calculation of the vehicle using the estimated performance of
the motor. These requirements are associated with optimization criteria concerning either the range for a
fixed payload, or the payload for a fixed range. These preliminary data enable the motor designer to determine the most efficient configuration of the motor. Then, a new trajectory calculation is performed, which
gives new requirements and new optimization criteria. Several iterations are often required in order to
obtain the most efficient configuration of the vehicle and its propulsion system.
2.1. Solid rocket motor requirements
The requirements of a solid rocket motor can be very different for ballistic,
applications; however, they can be usually divided into the three following classes :

space or

tactical

Performance requirements :
The performance requirements consist of a total impulse to be delivered by the motor, with certain limits to the burning time. The thrust-vs-time curve can be constant or as constant as possible. Sometimes a specific thrust-vs-time curve can be required ; this type of specification is usual for first stage
of ballistic missile or space launcher, for which a high level of initial thrust enables a decrease of the
gravity losses. Sometimes certain masses (total, inert or propellant) may be required.
Geometric requirements :
The envelope limit is usually a requirement of the vehicle designer. The geometric contraints are
of great importance; often the optimization consists to determine the most efficient configuration in a
limited external volume of the motor; moreover the motor configuration can result directly from these
requirements ; it is the case of submerged nozzles for limited length motors or blast tube nozzles for tactical missile applications.
Other requirements
Other requirements are of great importance,; thus, the objectives of cost, reliability and development time can restrict the use of certain efficient technologies, but not sufficiently demonstrated such
as nozzle or case materials, TVC systems, etc
2.3. Optimization criteria
The optimization is based upon criteria concerning p.ither the range for a fixed payload or the
payload for a fixed range. The optimization criteria are often given by partial derivatives indicating the
influence of design parameters upon the range or payload of the vehicle.

11-2
An example of partial derivatives of range with respect to inert mass, propellant mass,
impulse, and burning time is presented hereafter.

H'
roc
dP
~

dWiP
aP
,)Is
~p

3 km/kg

(partial derivative with the inert mass)

5 km/kg

(partial derivative with the propellant mass)

50 km/ s

(partial derivative with the spec ific impulse)

specific

- 10 km/s (partial derivative with the burning time)

The partial derivatives enable the comparison of a given motor configuration (with parameters
m i , m p , I sv , t cu ) to the reference (with parameters m i (0), m p (0), I sv (0), t cu (0 by calculating the range increment A P.

orIs

(0))

( h'I

If the range increment

the reference.

r - I'Y'r

dE'
.)hu

-1-.)

(0) \

I-CJ.< - h:J.A )

P is positive, the motor configuration is said to be more efficient than

Partial derivatives are in fact influence coefficients, which are only valid for small variations
of motor parameters compared to the reference, which has been taken into account for the'trajectory calculation.
3. - IN1!'LUENCE OF DESIGN PARAMETERS

The design parameters of a solid rocket motor can be divided into three classes : the operating conditions (pressure, burning time), the geometric parameters (case length, nozzle submergence, exit cone half
angle, etc ) and the technologies used for each component ( propellant configuration, case and nozzle
materials, Tve system, extendible exit cone, etc ). The variation of one of these parameters leads to a
new configuration of the motor, requiring an accurate evaluation of its performance.
3.1. Operating conditions
It is often interesting to optimize the operating parameters of the motor; thus, the chamber
pressure has a direct influence upon the throat area, the nozzle dimensions and inert mass, the case inert
mass, etc Moreover these interactions are conflicting; thus, an increase of pressure leads to a higher
inert mass of the case (designed for internal pressure) and a lower one of the nozzle (decrease of the
throat area).
The burning time can also be optimized ; this parameter influences throat area, nozzle dimensions
and inert mass, thermal design of nozzle and internal insulation, etc
In order to determine the optimum operating conditions, it is necessary to select several values
of the maximum pressure and burning time. The performance of the motor will be evaluated for each couple of
pressure and burning time. The results can be represented.by the diagram illustrated in figure 1.

t
1

Range increment (km)

60

Optimum
/

Pressure " 70 MPa

Bum time " 50 s

40

70 s

20

5.0

6.0

7.0

8.0

9,0

Pressure (MPa)
Fig. 1 -

OPTIMIZATION OF PRESSURE AND DURATION

11-3
In this example, the optimum operating conditions are the following
maximum pressure
burn time

7.0 MPa
50 s

3.2. Geometric parameters

The main geometric parameters of a solid rocket motor are presented in figure 2.

Fig. 2

INFLUENCE OF NOZZLE SUBMERGENCE

The motor efficiency is a main function of these parameters. In order to illustrate the importance of the geometric parameters two configurations of a solid motor in a same external volume are represented in figure 3. The only difference between these two configurations is the nozzle submergence.

Fig. 3 -

MOTOR GEOMETRIC PARMETERS

As shown, this variation influences upon the nozzle inert mass, exit cone expansion ratio, propellant mass, case inert mass, etc... The optimal nozzle submergence will be greatly dependant on the partial derivatives with respect to inert mass, specific impulse and propellant mass. Other geometric parameters can be optimize : case length, exit cone expansion ratio and half angle, aft-opening diameter of the
the case, etc

11-4
3.3. Technologies used for each component
Different technologies can be used for each component (propellant grain, case, nozzle, etc ).
The use of a given technology can influence other design parameters. For instance, the optimum operating
pressure is often greater with a solid rocket motor using a graphite filament wound motor case because of
the better mechanical properties of graphite fibers compared to those of kevlar fibers or metal. For the
different components, the technologies which can be used are the following :
a) Motor case :
The motor case can be metallic or composite. The first generations of composite motor case
were made up of glass fibers. The use of kevlar fibers and more recently of graphite fibers has allowed on
increase of performance. The use of a composite motor case leads to a lower inert mass of the motor.
b) Propellant grain design
For a given propellant, the desired thrust-vs-time curve defines the burning area evolution.
This evolution can be achieved by different grain shapes or burning rates, etc The optimization of the
propellant grain must be performed taking into account certain constraints : casting process, erosive burning, combustion instability, mechanical behavior, etc
c) Internal insulation :
The design of internal insulation is a function of the combustion geometry ; the inert mass of
internal insulation is a function of both the propellant grain design and the thermal properties of the
se lee ted rubber.
d) Nozzle and TVC system
The motor performance depends on both nozzle specific impulse efficiency and inert mass. These
two parameters are functions of the nozzle design: aerodynamic contour (expansion ratio, exit cone half
angle deflection angle), nozzle configuration, selected liner and insulator materials (phenolic, carboncarbon, etc ), TVC sys tem, etc
4. - COMPUTER AIDED PRELIMINARY DESIGN
In order to perform quiCkly and accuratly parametric and sensitivity analyses, a computer program has
been developed at SEP. Based upon data given by the user concerning performance or geometric requirements,
the program builds the motor, component by component and estimates its performances (inert mass, propellant
mass, specific impulse, etc ). This program enables the comparison of a great number of motor configurations and the inves~igation of the influence of design parameters. The principle and main characteristics
of the program are presented in reference C 1]. A brief summary is given hereafter and different examples
of application are given in order to illustrate the possibilities of the program.
4.1. Principle
With the data given by the user, the program designs separatly each component. This design phase
uses simple models in order to obtain acceptable calculation time necessary to decrease the optimization
time for parametric study. Afterwards, an iterative method is used to assemble the motor components. The
outputs of the program consist of numerical data (geometric, performance) and graphical displays in order
to verify the coherence of the solution. At the end of the program, a modification procedure is available
in order to define a new configuration by changing only the desired parameters.

11-5
4.2. Characteristics of the program
The characteristics of the program which have been selected to facilitate the utilization are
interactivity
visual graphics
modular

structure

: the program user controls inputs and outputs data on a visual unit

graphics routines are incorporated, allowing a visualization of the calculated motor

configuration
this structure has
gies.

been selected in order to facilitate extension

to new technolo-

4.3. Data input

The data can
each motor component.

be divided into two classes

data relative to requirements and data relative to

a) Geometric and performance requirements

Several options can be selected in the following list if they are not conflicting
fixed
fixed
fixed
fixed
fixed
fixed

diameter
space envelope
motor or propellant weight
total impulse
thrust-vs-time curve
ideal velocity increment.
b) Data relative to motor components :

The program user must input the data relative to each component. For instance, for the motor
case, the user must specify the technology used
metallic or composite, and select the design criterion:
stress or strain limited, imposed stiffness.
4.4. Analysis principle
The central portion of the program consists of the automated design process.
In a preliminary
phase, a series of compatibility tests is performed ; the program verifies that data for each component are
sufficient and not conflicting. Then each component is designed with a specific design routine. This step
uses several data basis containing informations such as propellant characteristics, materials thermal and
mechanical properties. The specific routine used to design each component is based on the technology used,
the function to be achieved by the component and the design criteria. The models which have been developed
are based on experience, state of the art, technology limits, simplified analyses and results from other
computer programs. No use has been made from sophisticated analysis program such as aerodynamics or finite
element codes. The reason for this, was the need of short computational time in order to decrease the duration of optimization time. At the end of the program, all the different components designed-are assembled
together, using an iterative process, taking into account the interactions between each component, the conformance to space envelope, etc
4.5. Analysis results
After the assembly process, the final configuration of the motor is obtained. Two categories of
results are available: firstly numerical results relative to performance or components parameters and
secondly a drawing of the motor configuration with the possibility of zooms. Typical numerical results are
presented in figure 4.

11-6

PROGRAM OUTPUT

1.600
2.400
6,155
5,670
6.7
6.1
53.
180.
16.8
15.8
275.
15.290
0.91
0.62
350.
3.78
10.
2.1
800.
23.5
5.

Motor diameter t m

Overall length, m
Motor weight, kg
Propellant weight, kg
Maximum pressure, MFa
Average pressure, MPa
Burn time, 5
Nozzle throat diameter, rom
Expansion ratio, initial
average
Vaccum specific impulse,s
Total impulse, MN.s
Volumetric loading
Web fraction
Port diameter, mm
Port to throat ratio
Burn rate (7 MPa), rnrnls
Case length, rnrn
Aft boss opening diameter, mrn
Average exit cone angle, degrees
Maximum vector angle, degrees

TABLE 1

---..,

r--e:-+--------+~-

'-----7

/
I

_ _-----'--,__

+1

l\

\\
\

\
\
\
/~---~
'-- -""'=1========-+--=- ----"
Fig. 4 -

MOTOR CONFIGURATION

I
I
I

\
\

\
Fig. 5

ZOOM OF THE NOZZLE

11-7
4.6. Accuracy
A comparison of results obtained by the program and by a more detailled design of the motor is
presented in table 2. The accuracy obtained on inert mass and propellant mass is in the range of 2 to 3 %.

MASS

PROGRAM RESULTS
(kg)

DETAILLED DESIGN RESULTS

(kg)

Case

357

363

Internal insulation

115

104

Nozz le

139

119

Igniter

Miscellaneous

10

618

604

11213

11188

Total inert mass

Propellant mass

TABLE 2

4.7. Applications
Two examples of applications will be presented hereafter. The first application is representative
of a second or third stage of a ballistic missile. The second presents the developments of the program,
which have been achieved in order to optimize the ARIANE V solid rocket boosters.

4.7.1. Optimization of a solid rocket motor in a fixed space envelope

The fictitious solid rocket motor studied is representative of a second or third stage of
a ballistic missile. The design criterion is a fixed space envelope and the optimization criterion a maximum missile range. The motor to be optimized, incorporates the following technologies :
a kevlar filament wound case
an internal insulation made of EPDM rubber
an axisymetric grain with a circular port
an high solid and aluminized propellant
a submerged contoured and movable nozzle using a carbon-carbon integral throat and
phenolic for the insulators and the exit cone
the thrust vector control device is a composite flexible bearing.

entrance

and

carbon

The general configuration of the motor is presented in figure 4. The optimization was conducted with the following parameters :
case length
web fraction
maximum operating pressure
burning time
nozzle submergence
exit cone half angle

L (m)
e (mm)

Pmax (en MP a)
tcu (en s)
x/L
Om (en 0)

The nozzle expansion ratio has not be choosen as an independant parameter; it is deduced
from nozzle throat location and exit cone half angle. The iterative process, which has been used for the
optimization is presented in figure 6.

11-8

INITIAL VALUES OF xiI AND Om

I
I

OTIMIZATION OF Pmax AND tcu

OPTIMIZATION OF xiI AND Om

OPTIMIZATION OF LAND e

I
I

FINAL CONFIGURATION
FIGURE 6
The optimization was based on partial derivatives with resp~ct to burning time, inert
mass, propellant mass and specific impulse. The total process has required the evaluation of two hundred
motor configurations. The work has been done in two days. Typical results of step 1 and 2 are presented in
figure 7 and 8.
The main characteristics of the optimum motor are presented in table 1.
of the motor is presented in figure 4. The optimum design parameters are the following :
case length
web frac t ion

maximum pressure

Pmax
tcu
xiI
Om

burn time
nozzle submergence
exit cone half angle

2.1 m
379 mm
6.7 MPa
53 s
.61
23.4

Range increment (km)

60

40

14~11
l\
\
I.
7

53 s

70 s

6.7 MPa

20

Maximum operating pressure (,VIPa)

Fig. 7 -

OPTIMIZATION OF PRESSURE AND DURATION

The visualisation

11-9

Range increment (km)

100

optimum 850 mm

50
lines of
constant aft
opening diameter
(A.O.D. )

a
0.61
0.58
Nozzle submergence

0.55

Fig. 8 -

0.64

OPTIMIZATION OF NOZZLE SUBMERGENCE

AND EXIT CONE HALF ANGLE

4.7.2. Optimization of ARIANE 5 solid rocket motor

The ARIANE 5 european launcher first stage consists of a central cryogenic
strap-on solid rocket boosters. These boosters incorporate the following technologies:

stage and two

segmented steel motor case

composite propellant : 86 % solid, 18 % aluminium
submerged movable nozzle hinged by a flexible bearing.

Fig. 9 -:- ARIANE 5 BOOSTER CONFIGURATION

The computer program has been adapted in order to achieve the optimization of these boosters. For this purpose, two modules have been developed to design the segmented case and grain.
Segmented case design module
This module has the capability to design three to five segments of either identical or
various cylindrical section lengths. The parameters taken into account are the following: mechanical properties of the selected material, geometric characteristics, forward and aft opening diameters,
maximum
expected operating pressure, safety factor and proof pressure ratio.
Segmented grain design module
The required thrust-vs-time curve presented in figure 10 consists of a high thrust at
lift-off followed by a regressive thrust phase to minimize aerodynamic loads during the transonic phase. At
the end of the burn time, the only limitation is the maximum acceleration of the launcher. The grain configuration consists of center-perforated grain with a forward or aft finocyl star. Because of the dominant
role of the thrust-vs-time requirement, it has been necessary to develop an efficient combustion module to
calculate the burning area evolution. The grain module is based upon 1) internal ballistic aspects in order
to produce the desired thrust-vs-time curve, 2) mechanical aspects, 3) internal flow field in order to
avoid erosive burning.

Thrust

Time

Fig. 10

THRUST VS TIME SHAPE REQUIREMENT

11-10

In addition to these new modules, work has been done 1) on the internal insulation module
to extend its capabilities to segmented grain, 2) on the nozzle module to adapt the model to large nozzles,
3) on the specific impulse prediction model.
This new version of the optimization program is extensively used in parametric studies,
which are conducted by the designers in order to select the basic options, geometric and operating parameters. The main parameters which are studied are:
motor overall length and outside diameter
number of segments
length of cylindrical sections
location of the star section
thrust-vs-time curve parameters
maximum operating pressure
nozzle expansion ratio
Typical results obtained by the program are presented in figures 11 and 12.

...,
I

I
I

...J

Fig. 11 - a
THREE SEGMENTS - STAR SECTION IN THE FORWARD SEGMENT
b - FOUR SEGMENTS - STAR SECTION IN THE AFT SEGMENT

.,

Thrust

--

---~~,---

.-

-~

--~~---r------~-f - - -

-;;;;;-.-

~ ;...--

...-.:::

- - .......~

..

_-

._-

\ Time
Fig. 12 -

THRUST TIME COMPARISON

- - - design requirement
____ actual from the code designed grain

s. -

CONCLUSION

The development of automated design computer programs has allowed great simplifications of solid
rocket motor preliminary design and optimization. The program, which has been presented in this paper, is
an efficient tool Eor parametric studies and gives sufficient data to start the detailled design of each
component. These programs must be continuously extended to new technologies or new applications.

RTD-I
ROUND TABLE DISCUSSION
Rijswijk, Netherlands, 7 April 1987

1.

QUESTIONS CONCERNING NOZZLE PROBLEMS

1.1 Mr G.Kristofersen, NDRE, NO

With regard to nozzle inserts, there has been a trend from metallic (W, Mo) to graphite and to reinforced graphite. How
does silicon carbide (SiC) fit into the picture?
MrTruchot
The main drawback of silicon carbide is the lower melting point (about 2500 'C) than that of carbon-carbon (4000 'C). So
silicon carbide can be used as a throat material only when the burning temperature of the propellant is low.
Mr Kristofersen
I thought that SiC could be used at 3000 'C.
MrTruchot
No, I think it is too high! But an advantage of SiC is an excellent oxidation resistance, much better than that of carboncarbon.
Mr Kristofersen
Is it possible, in order to enhance the oxidation resistance of a graphite, to impregnate it with SiC?
MrTruchot
I think you talk about "overcoating" carbon with SiC? This technology has been used, but the main drawback is that the
very thin layer of SiC is usually really very brittle, and when cracks go through this layer they propagate very quickly in the
carbon material. I think the best solution is really to use a ceramic matrix.
Mr Kristofersen
Is it true even with a small burning time?
MrTruchot
I think that the problem of the cracks in SiC could be critical.
Comment by Mr D.Reydeliet
I think that if you manage to have a "good" layer of SiC and a "good" bonding between carbon and SiC, and if your burning
time is short enough, you could have a good solution. But if the SiC is too brittle, or if the bonding is weak, orifthe burning
time is too long, you will run into difficulties. Anyway, it is better to put SiC ill the carbon material and not only on the
material.
1.2 Mr A.Mason, R.O.Summerfield, UK
Concerning materials for ITE, could the speaker quote any specific values for erosion rates of carbon-carbon throat
inserts, giving details of propellant and motor performance to which they relate?
MrHiidreth
I can't give you precise values for all kinds of propellants, including the whole range of aluminium percentage; I can say
that the carbon-carbon erosion rate has an order of magnitude of 7 milliinches per second to 15 milliinches per second
(about 0.2 mmls to 0.5 mm/s). It is about one third or one fourth of that of a carbon phenolic material.
1.3 Mr Kristofersen, NDRE, NO
a) More nozzles can be substituted for a large one. Is there an optimum for the number of nozzles?
b) An annular nozzle can be looked upon as the limiting case for many small nozzles. Do annular nozzles have any
future?
MrTruchot
I think that the best solution for minimum weight is one single nozzle. But it may happen in some cases that several small
nozzles are useful (for controlling the rolling movement, for instance).
Mr Kristofersen
I thought that one single nozzle was heavier than several small nozzles.
Mr ReydeUet
I don't think so: when you look at the thermal insulation, for instance, you need much more weight for protecting the aft
dome with several nozzles than with one single nozzle. In addition, the internal aerodynamics is very bad!
MrTruchot
Furthermore, with several nozzles the case is generally much heavier, because of the weight of the bonding for the nozzles.
Mr Hildreth
I know that in USA some early booster type motors were using several nozzles, primarily for roll control purposes. Now, I
know that almost all the motors which are designed now are using a single nozzle. I think that the first advantage is
simplicity. My opinion is that the fewer nozzles you have for a large motor the better it is likely to be.

RTD-2

1.4 Mr R.J.Ballinger, British Aerospace, UK

Would anyone like to comment on the potential problem of aluminium deposition on the nozzle throat during the initial
stages of burn? Is it taken into account in the initial design? Do nozzle materials influence the situation?
MrHildreth
For large nozzles, I think this phenomenon is negligible. In smaller designs, it could be more significant. It is generally
ignored, because if there is some aluminium deposition at the very beginning of the firing it is generally removed very
quickly. In RPL we have a test motor which is used for measuring the performances of the propellant formulations. We
have thousands of firings with this motor; it has a bulk graphite throat of about one inch diameter. It seems that there is
some evidence of aluminium deposit during the first second, or one second and a half. After that time, the graphite
temperature seems to be sufficient for eliminating the deposition. The complete firing duration is about 5 to 6 seconds. But
I don't think that this phenomenon was ever considered in the calculations.
Mr Ballinger
My concern was for tactical applications, for boost motors where we need the high practical specific impulse, with
aluminium propellant, for high accelerations with very short burning time (1.5 to 2 seconds).
MrHildreth
I am not really familiar with air launched missiles; my best answer is that there could be

a significant problem for you.

Mr Reydellet
It is basically a problem of wall temperature. The example given by Mr Hildreth shows that you have small throats, short
burning time and a material with high diffusivity. You have a large amount of material to be put to a high temperature very
quickly. So that wall temperature has not enough time to reach the alumina melting point. On the contrary, if you use a
material having low diffusivity (carbon phenolic for instance), the thickness of the material to be heated will be much
thinner. In this case, the equilibrium temperature will be reached for the wall very quickly. I don't guarantee the result, but
it might help you if you have some difficulties.
2.

QUESTIONS CONCERNING INTERNAL THERMAL INSULATION

2.1 Mr Raudsandmoen, Raufoss, NO

As both composite and strip laminate motor cases are very sensitive to leaks of hot gas through the insulation, could you
please comment on how to reduce the risk of insulation leaks; e.g. material choice (type of rubber, fibre-filled, powder
filled), method of application (spray or rotation insulation, insulation vulcanized into the case, bonded into the case, or
winding on to the insulation).

MrTruchot
This question was first put to Mr Denost, Mr Thorp and Dr Evans. But since composite cases cannot alone hold gas
pressure without leaking, I must take this question. For composite cases it is essentially a problem of rubber insulation:
only the rubber can prevent gas from getting out.
MrThorp
For strip laminate I must say that it is exactly the same as for monolithic cases: you have to protect it from the hot gases
(3000 OK, typically). So there is no difference. What I must say is that we have to protect the case from water during the
pressure proof test, because the resin we use is water sensitive. But, besides that exception, there is no difference with
monolithic cases.
Mr Raudsandmoen
It seems to me that the consequences are very significant for composite cases.
MrDenost
For composite cases, it is unreasonable to rely on the resistance of the case against hot gases if, for any reason, there is a
thermal insulation failure or some leakage. The answer seems to be very rude, but it is only reality.
Mr Raudsandmoen
Does that mean that the best way to prevent the case from leaking is to wind on to the insulation?
MrDenost
Not really. The designer must just make sure that the insulation will survive and protect the case. That is what is done
generally.
Dr Evans
In addition, you will have to protect the case by an insulation during the hydro proof test. If you don't the case will likely be
destroyed during the test.
3.

QUESTIONS CONCERNING CASES

3.1 Mr Kristofersen, NDRE, NO

Filament winding is demonstrated for large motors. Is there a recommended lower limit with regard to motor diameter.
Dr Evans
The lower limit is established by the availability of small precise winding machines and somewhat by the size of fibre
bundles commercially available. I have wound a lot of pressure vessels at J inches diameter with absolutely no difficulty on
production size winding machines. Some demonstrations have been made for vessels of 1-:;. inch diameter.

RTD-3

3.2 Mr Monvoisin, DGAIDEn, FR

Is there a rule or a method to define the succession of the hoop and helicoidal layers within the composite'?
Dr Evans
In general it is preferred to alternate layers in a way that produces fibre direction symmetry about the mean radius or mid
thickness ohhe laminate, to minimise warping of some plys under load. However, in practice, considerable liberty may be
taken with this "rule" with little penalty if other behavioural characteristics are desired. For example, if the motor case
structure is severely aeroheated during the post-burn flight period, one can somewhat thermally protect the helix layers,
which provide bonding strength and stiffness, by placing most of them radially inboard and consequently concentrating
the hoop layers to the outside where they serve as an insulator after they have performed their pressure vessel function.
MrDenost
Generally, we alternate layers. The first reason is that we try to protect the layers that have to support higher loads (i.e.
hoop layers). The second reason is that generally the mixing of layers helps to solve the problem of the bonding with the
skirts, which are the cylindrical parts bonding the vessel to the rest of the missile. Since one must stop the hoop winding at
the tangent line (because one cannot put hoop layers on the dome) we try to stop it as smoothly as possible, in order to
avoid bending loads.

3.2 Mr Monvoisin, DGAIDEn, FR

The composite is damaged during the manufacturing process of the filament wound cases. That makes its mechanical
properties decrease in comparison with unidirectional composite. What causes the damage'? Is it possible to minimise this
damage by slight changes in the manufacturing process'?
Can the behaviour of glass, kevlar and graphite fibres be compared from this point of view'?
Dr Evans
In my presentation I oversimplified by stating that the reduction in fibre strength achieved in a pressure vessel with respect
to the inherent strength is due to "damage incurred in manufacture". There are other factors, including non-uniform fibre
collimation and tensioning as a consequence of winding and cure (mandrel expansion before the resin becomes solid
tends to more uniformly tension each fibre). Fibre/resin interaction can also cause reductions in delivered fibre strength.
For example, high modulus resins result in high radial tensile loads being placed into each fibre and can cause severe
strength loss in kevlar which is inherently quite weak in this direction.
Certainly, however, fibre damage during winding is a significant factor and is caused by fibre abrasion as the fibre is pulled
under winding tension through and across the various contact points comprising the resin impregnation and fibre delivery
systems. These kinds of damage are minimised in practice by ensuring that contact points are smooth and oflarge enough
radius to avoid fibre damaging by bending. Even so, fibre damage does occur. Graphite or carbon is more sensitive to this
phenomenon than kevlar or fibreglass.

3.3 Mr Monvoisin, DGAiDEn, FR

The graphite/resin composite is said to be more shock sensitive than the kevlar/resin composite. Do you take this factor
into account when designing or manufacturing the cases'?
Dr Evans
.
Graphite vessels do appear to be more "shock" or "impact" sensitive than those of glass and particularly kevlar. This
tendency leads oftyn to the use of kevlar or glass overwraps for added protection if the impact environment is severe.
3.4 Mr Sauvet, DGAIDEn/CAEPE, FR
During his presentation, Mr Denost spoke about the difficulty of making filament wound cases with openings of different
sizes at the forward and the aft dome. Have you encountered this difficulty'? If yes, how do you manage to solve it?
Dr Evans
Mr Denost correctly presented the typical solution for wound motor cases, where the aft polar opening which houses the
nozzle is signifiqmtly larger than that of the forward opening. In practice, accommodation of such differences in polar
opening is achieved by "deviating" both dome profiles and their winding angles and paths to "share the distress" by
introducing shear stress field into both domes. For most vessels, a combined "deviation" for both domes of 10.15' can be
accommodated. Resin properties of course influence how much deviation can be accommodated. Also, the path is not
geodesic, so there is a tendency for the fibre to slip away from where it is placed.
MrDenost
1agree with all these statements. All these methods are to be judged when coming to the winding machine and when one
sees if it works or not.
3.5 Mr Thepenier and Mr Lucas, SNPE/USM, FR
For the optimisation calculations of the case, do you take into account the influence of the grain? If yes, how?
Dr Evans
In Hercules, we consider the interaction effects of propellant grain strain under pressure and the case stiffness
(particularly hoop stiffness) during the preliminary motor design. It is a design trade. If propellant bore strains are
predicted to be excessive as a result oflower case hoop stiffness than desired, then the hoop stiffness can be increased by
tlie addition of more hoop wraps. The,IM (intermediate modulus) graphite fibre like IM6 and IM7 with fibre modulus of
42 X 10 6 psi provides adequate case stiffness in most instances when the case is designed for pressure.

RTD-4

3.6 M.Vigier, Aerospatiale, FR

As a result of the studies concerning the graphite fibres, have you some results that can investigate the influence of ageing
or of accidental shocks on their performance? Is the behaviour different from that which has been observed for other
fibres (kevlar in particular)?
Dr Evaiis
It is a very difficult subject to be answered in a few words. With respect to the shock induced damage this is, as you know, a
quite complex topic which cannot be easily addressed in abrupt response. But one can say that graphite pressure vessels
subjected to impacts by sharp objects appear to undergo more strength loss than do those of either glass or kevlar.
However in the context of generalized use of composite motor cases in a tactical missile environment, the larger question
is how can one determine when an unacceptable damage level has been incurred after accidental impact, regardless of
fibre choice. One other problem is how to get a criterion by using non-destructive techniques.
With respect to ageing, particularly under stress, I have seen no data to suggest that graphite fibre undergoes ageing losses.
Kevlar and glass both exhibit creep rupture behaviour and consequently do not carry long term loads very well, but this
characteristic is not important for motor cases. Matrix or resin dominated properties like compression and shear strength
and stiffness are of course affected by resin ageing regardless of the fibre.

3.7 Mr Krist~fersen,NDRE, NO
Cartridge loaded grain necessitates large ports, thereby requiring metal end rings. What is a typical propelIant mass
fraction for a small tactical rocket with cartridge loaded grain?
MrThorp
The mass fraction will depend very much on the requirements of the missile. Typical1y, if you have a very simple
cylindrical body, and a long burning (end burning grain), with no requirement for a blast pipe, or attachments, you will
probably have a mass fraction of about 0.8, with a light al10y or steel body. Graphite fibre can also be used, but it depends
very much on the stiffness requirements.
At the other end of the scale, if you have a motor with wing-ring, long blast pipe, long attachments, the mass fraction can go
down to 0.5 or 0.6.
I want to point out that strip laminate cases can be used with case bonded grain as wel1.
3.8 Mr Thepenier, SNPE/USM, FR
What is the factor of safety used for a "strip laminate" body? Is the strain under pressure for such cases similar to others?
MrThorp
The factor of safety used is exactly the same as for a monolithic steel case. If you choose a determined strip thickness, it
may happen that you get a slightly higher margin of safety than desired, because of the nature of the laminate. But strip
thickness can be modified to match exactly the factor of safety that you want.
The strain under pressure is the same as that for a monolithic case. The adhesive layer is capable of taking high shear loads,
thus transmitting al1loads to the strip.
3.9 Mr Holz, Thomson-Brandt, FR
Can a filament wound case for tactical missile be competitive (in terms of ratio propelIant mass/missile mass) against a
maraging steel case, if a first frequency made bigger than 70 Hz is required?
Dr Evans
A graphite motor case designed for pressure only will usual1y have an axial stiffness value about equal to that of aluminium
of the same thickness (but the aluminium one has far less strength). If a greater stiffness is required to meet a given natural
frequency, this may best be achieved by adding either a low angle helix layer or unidirectional layers. For this purpose the
higher modulus fibres like HM (E = 60 X 10 6 psi vs E = 42 X 10 6 psi for 1M) is advisable to minimise the quantity
required. The graphite case approach may wel1 not be competitive if the missile system requires a very high natural
frequency or if there is a volume limited by available envelope.
4.

QUESTIONS CONCERNING GRAIN PROBLEMS

4.1 Mr Calabro, Aerospatiale, FR

In the current propel1ant grains, the finocyl shape is mostly at the aft dome. In the older SRM, this shape (star) was at the
forward dome; what are the criteria for choosing that location? In particular, if the shuttle's boosters were to be
redesigned, would the same solution be chosen (in the forward dome)?
MrThrasher
I was not aware that there was a long term trend for moving the finocyl star from the forward end to the aft end of the
motors; but I can find some reason for that. First, the tooling removal for the star shape is easier through the large port,
which is usual1y located at the aft end. The second reason is that the star in the forward dome creates a big pressure drop
from the forward to the aft dome, because of the big gas flow through the cylindrical bore of the grain. I know that there is
some consideration about instabilities risk. The driving of longitudinal burning instabilities seems to be more effective if
there is a star shape in the forward dome.
As far as the shuttle boosters ar.e concerned, I am only aware of some problem about the star: there are some smal1 cracks
(that can be easily removed) due to thermal cool down. I see no reason to bring the star shape to the aft end. It is probably
easier to manufacture the motor with the star in the forward dome.

RTD-5
4.2 Mr Gondovin, SNPE/CRB, FR
For the purpose of high energy (NEPE type) propellant grains structural analysis, which stress-free temperature do you
assume - cure temperature or another temperature? (It is assumed that the grain is not pressure cured). In other words is
there a significant cure shrinkage for that kind of propellant?
MrThrasher
I am not absolutely sure of the value, but it seems to me that I have seen some structural analysis assuming a cure shrinkage
corresponding to a 10'F elevation of temperature.
MrGondovin
For that type of propellant, we don't have precise values. We only know that for classical propellants (CTPB for instance)
there is no cure shrinkage. So we assume that the stress free temperature is the cure temperature.
Mr Thrasher .
I guess that the common practice in USA for old propellants is to take a slight elevation of temperature.
4.3 Mr Thepenier, SNPEIUSM, FR
In your model for calculating the margin of safety, how do you determine the value of FS (factor of safety)? Do you take
into account for choosing this value the level of knowledge of the propellant characteristics? (Well-known or new
propellant.) Do you need a reliability model?
MrThrasher
In USA, the FS is typically given by the SRM designer. It is a requirement. For boosters, we used a FS of 2 for bond-line
critical points, and a FS of 1.5 for central bore critical point. We don't take into account the fact that, at the beginning, we
don't have a lot of data relating to the propellant. The mechanical characteristics data box has to be developed during the
programme. If the propellant doesn't satisfy the factor of safety, it has to be improved, or the design must be changed.
In addition, it is not a common practice in USA to assess a number for the structural reliability in rocket systems. There
has been some attempt, I think, but the success was poor.
4.4 Mr Thepenier SNPE/USM FR
How do you use propellants which have "non classical" tensile curve (with two maxima for instance). What parameter of
the curve do you take into account for designing the grain?
MrThrasher
It is a very difficult question. At first, if you use corrected stresses, you find the phenomenon much more often than if you
use engineering stresses. My personal opinion is that something bad is happening to the propellant when you reach the
first maximum. This could be a very wise criterion to be chosen. But sometimes, some people think that the second
maximum could be used. Nobody knows who is right.
4.5 Mr Fleming, UK
In the paper of SNIA BPD no mention was made of the spin on erosive burning effect. Can these effects be important?
(30% increase in burning rate.)
MrBiagioni
We don't take into account this effect, but for the size of motor for which I have a personal experience, this effect is not so
high. We consider it quite negligible, but the spin is not very high.
Mr Thepenier, SNPEIUSM, FR
In the grain predesign, are criteria relating to mechanical strength pressure temperature taken in to account? What are the
safety ratios that are used to achieve good operation during the lifetime under operational conditions?
MrBiagioni
I am not a structural specialist. But I know that in my company, structural specialists do use mechanical criteria in the
predesign, based on experimental correlations to take into account the effect of pressure and temperature - I have no
value of factor safety in my mind now.
4.6 Mr Fleming, UK
"Base Bleed" is essentially a rocket propellant burning at low pressure (so acting as a gas generator) in a suitable motor at
the rear of a shell (projectile) to extend range. Since such motors are now being produced in the hundreds of thousand per
year (propellant by SNPE, Raufoss, ERT etc...), when does AGARD plan to study "BB"?
How would the members present here react to an independently organised meeting, perhaps towards the end of 1987, to
consider quality control and testing?
Dr Zeller
At the last PEP meeting, there was a proposal from Mr Cruttenden (UK), about a meeting dealing with that kind of
subject, including "base bleed". But, as far as I know, this subject has not been selected by AGARD for a next specialist
meeting. I personally would be interested by a meeting on this subject.
D.Reydellet
In addition, I think you will find more interested people, if you ask people involved in gun propellant.

RTD-6

4.7 Professor De Luca, Politecnico di Milano, IT

Usually, a rocket motor is designed to be ignited, to give thrust during a certain duration. then extinguished. But what
happens if you want to ignite, extinguish, then reignite, and so on..? And what if you want several levels of thrust? How
much of your method can be used? How much must be changed?
DrZe!!er
It is really a very large question. There are several architectures that are possible: for instance for the short range attack
missile, you have two grains that are separated, and that burn separately. There are some advanced studies using laser
beams to interfere with burning. As a conclusion, one can say that it depends on the kind of architecture you choose: in
certain cases you may have to develop new design methods.

Mr Reydellet
Concernil!-g thrust modulation, I must say it is a very difficult problem. The main question is to find propellants that burn
steadily in a large range of pressures. Even if you find them, you may encounter some difficulties for controlling the
pressure level. As a conclusion, one can say that rocket propulsion has some limitations. Thrust modulation is one of
them, and you must not try to compete in that field with other propulsion techniques that are more flexible.

RTD-7
ROUND TABLE DISCUSSION
Lancaster, California, USA, 23 April 1987

1.

QUESTIONS PUT TO MR HILDRETH

1.1 MrJ.Gross, NWC, China Lake, US

How do you estimate nozzle erosion rates in the design process?
Mr Hildreth
99,99% is empirical based on past experience, subscale, fullscale; the range runs from about 7 to 15 milliinches per
second (0.2 to 0.5 millimetres per second) for carbon-carbon, depending upon whether you have aluminized or nonaluminized propellant.
MrGross
What is the relative cost of a carbon-carbon throat insert for a tactical motor compared to a graphite throat (i.e. ATJ)? or compared to a pyrolytic graphite throat?
MrHiidreth
I can't give you dollar values. I can give you only a general statement. If it is compared piece to piece, I think that the
carbon-carbon piece will be more expensive. Depending on the kind of carbon-carbon you use, it can be moderately more
expensive, or extremely. For instance, in one case, if you can make a log of 3D cylindrical weaved and cut a number of
small throats in it, or if you use a 4D or multi D technology, you can make a huge billet and punch out a large number of
pieces for tactical motors. It would be different for a space boost motor, where you have to weave every piece individually.
In that case, I think there is no question: carbon-carbon is more expensive; however, you also have to consider that in
general carbon-carbon nozzles are probably simpler; you don't need as much back-up materials, as much insulator
materials. The whole nozzle design changes, which directly affects the cost. I don't know if, for tactical motors, a carboncarbon based design is much more expensive than a phenolic or other technology design; I guess it is, probably, but it has
also much higher performances. Concerning the pyrolitic graphite, I have no information on it: this technology appeared
and disappeared in very few years.
1.2 Mr H.Platzek, NWC, China Lake, US
The extendible-exit-cone (BEC) does add a little more impulse to a motor, but it seems as if it's more than off-set by the
increased complexity, increased inert weight. Have any type of quantitative studies been conducted to show how much is
gained and what is the cost?
Mr Hildreth
Again, I don't have any precise number for you. I would only assume that, if there was not a pay-off, they wouldn't be used.
But is it really for gaining performance or reducing the losses? For instance, where you have some geometrical constraint
(vehicle length for submarine launched missiles) you can minimize the loss by using an EEC. Although EECs are complex
and expensive, I know that they have been used in ballistic and space applications. So there must be a gain.
Comment by Mr Truchot
The complexity and the cost ofEEC depends on the technology you use for carbon-carbon parts. Ifyou use a 2D involute
or 3D, large exit cones are very expensive. But if you have a less expensive technology, the cost can decrease. In addition, if
the material allows simplification of the design, you can reduce the complexity and the cost. Anyway, if you compare the
efforts you have to perform for gaining a few seconds of specific impulse by improving the propellant, I think that EECs
are rather cheap (you can gain about 8 seconds in specific applications). Anyway, there is a trade-off to be made in every
case.
Comment by Mr Reydellet
In addition, I can say that EECs are particularly interesting for submarine launched strategic missiles.

2.

QUESTIONS PUT TO MR TRUCHOT

2.1 Mr B.Christensen, Hercules/Bacchus, US

What flow models are used for internal fluid flow calculations?
Are they viscous? Finite elements? Finite difference? Two-dimensional? Three-dimensional?
MrTruchot
Currently in SEP, we have several computer codes. The first one is based on finite difference methods with only 2D
capabilities. This code takes into account the viscous effect. The second code is a 3D code. It uses the methods of
singularities. Finally, we have a third computer code which is still in development. It is a 2D, two-phase flow code.
Comment by Dr Zeller
In SNPE we have a 3D code; Ideal gas, one phase, inviscid, non turbulent. It uses a finite difference, unsteady method
(Godounoff method).
Mr B.Christensen
How are erosion rates for internal insulators determined for new motors (no firing data for that motor available). What is
the overall design strategy for internal insulators in a new motor?

RTD-8

MrTruchot
We use empirical correlations and firing test results. For the first motor, we design some extra thickness. After the first
test, we decrease the rubber thickness. !fyou know another method, you're invited to give more details. We would be very
interested.
Mr Christensen
How do canted, submerged nozzles affect internal insulator char rates?
MrTruchot
I have no personal experience of canted nozzles, but I guess that the canted nozzle is not a good design from the point of
view of thermal insulation erosion in the aft dome.
Comment by Mr Reydellet
I think that the problem is more severe if the canted nozzle is

/lot submerged.

Mr Christensen
What measurements (instrumentation) for internal insulators are taken during firings? Are the same made for flight and
static motors?
MrTruchot
In SEP, we measure the temperatures during the static firing test by putting thermocouples in the rubber.
Comment by Mr Reydellet
Usually we don't put thermocouples in rubber for flying tests, because we don't have a sufficient number of measurements.
In addition, in France, we don't recover the stages after night (even the first stage). Have you in the USA experienced some
differences? If you have, have you some explanation?
Mr B.Christensen
Our experience is that we have seen over twice as much char as compared to static tests. These experiences were at first
with silica-filled EPDM. We wonder if this applies to kevlar or asbestos filled materials. This phenomenon occurs in the
forward dome. At this time, there is no satisfactory explanation; maybe a violent flow bringing hot gases to the forward
dome, perhaps flight acceleration causes more charring in this area.
D.Reydellet
Under axial acceleration, the gap between the flap and the forward dome insulation is bigger than during static tests. So
you may have more circulation in this gap leading to more heating of the flap surface and of the insulation.
Mr B.Christensen
It should be, but we also have this phenomenon with honded forward dome.
2.2 Mr O.Casillas, UTC/CSD, US
What are the historical trends in the internal insulation technology (materials and processes) and what do you see in the
future for insulating methods?
MrTruchot
The first generation of rubber used asbestos fillers (10 years ago). The density of these rubbers was 1.5; the next
generation was EPDM silica rubbers. We decreased the density to 1.1, with the same thermal properties and the same
erosion resistance. Now, the modern compositions allow the density to decrease to 0.85. The trend for the future is to
decrease the thermal diffusivity of the rubber, decrease the density and increase the erosion resistance.
2.3 Mr B.Lichtinger, UTC/CSD, US
What research is being performed to find new and lighter elastomers for insulators in place of EPDM?
MrTruchot
I have no answer to that question. If you have some idea...
Mr B.Lichtinger
What additional losses are involved in the supersonic split line Thrust-Vector-Control (TVC) system over conventional
systems?
MrTruchot
At this time, we have only performed a demonstration test for that technology. The accuracy of the measurement during
this test was not sufficient to detect the loss due to that effect. r guess there is some loss, but r can't give you any numbers.

3.

QUESTIONS PUT TO DR EVANS

3.1 Mr B.Lichtinger, UTC/CSD, US

Could you explain the significance and implication of fibre stress ratio in composite case design?
Dr Evans
Stress ratio is the ratio of the stress developed in the helix layer divided by the stress developed in the hoop layer. Another
way to look at that: the lower the stress ratio, the more high stresses are in the hoop layers. Usually, the design drives to a
failure in hoop layers. because that type of failure is more repeatable,

RTD-9

Mr B.Lichtinger
Could you explain Non Destructive Test (NDT) and inspection methods used to detect tlaws in current composite motor
cases?

DiEvans
That is the subject of about a six months seminar! A lot of work has been done on this subject in this country by all
propulsion companies for composites: ultrasonic, acoustic emission, cat-skin etc...; not only for cases, but for composite
parts in general; as a matter of fact, for cases, only two kinds ofNOC are routinely used: X-RA Yfor obvious unbounds and
cracks, separations between the layers; and proof-test, which is the standard acceptance test criterion: if the case passes the
proof-test, you have a good one. The other techniques are "non-state-of-the-art". The challenge is to develop a "finger"
print" for each unique configuration, so that the comparison of the same configuration with various defects can be done.
Mr Lichtinger
What is your .opinion concerning the future of composite cases in the new HLLV (Heavy Lift Launcher Vehicle) and
future launch vehicles?
Dr Evans
My opinion is that the design of a new heavy launch vehicle is strictly an economic issue: the lowest price per pound in
orbit. The composite cases used for thirty years in strategic system have certainly proved a sufficient reliability. The
appear, in any judgement, to be cost-competitive, for satisfying the cost requirements.
3.2 Mr R.Leighton, USAF Astronautic Lab. Edwards, US
What role does the propellant grain structural integrity play in composite case design considerations? What situations
might warrant consideration of the grain effects (stiffness contributions, inertia effects...)?
Dr Evans
The case and the grain have to be considered interactively, because the grain bore strain during motor pressurization is
directly influenced by case (hoop layer) rigidity. With the strength of materials growing every day, and the case wall
becoming thinner and thinner, one can have to put extra material for having a large web fraction. More generally, the
stiffness is a concern for tactical applications, where natural frequencies of the motor are a design driver; during the
operation time, the mass is decreasing, and the natural frequencies are changing dramatically; in addition, with
aeroheating, the case stiffness is decreasing. This concern has to be taken into account se.riously.
MrLeighton
Likewise, how does the nozzle loading or the case enter into case design requirements? Also, how critical is the boss blowout failure mode in case design requirement?
Dr Evans
The nozzle thrust force "unloads" the aft dome. This fact is routinely considered for the design of the polar reinforcement
of the dome. In this area, the simplified netting analysis is not convenient. Furthermore, the problem is more critical when
the opening becomes larger. We use the most powerful finite element code we have and, even so, we have encountered
some surprises. If you need it (if the shear level is high) you can incorporate some helical reinforcement between layers.

4.

QUESTIONS PUT TO MR DENOST

4.1 Mr D.Thrasher, USAF Astronautics Lab., Edwards, US

Both speakers addressed netting analyses which define mandrel contours for geodesic and modified geodesic domes.
Apparently, these analyses ignore the thickness build-up near the polar openings. How is the discrepancy addressed in
practise?
MrDenost
Knowledge of the precise thickness in every point of the dome is very important for the design. It has been necessary to
develop a specific code for evaluating the thickness and validate it by experimental data. In particular, it is necessary to
take into account the bandwidth of the filaments, and the winding angle (which varies very rapidly in this area). We have
also to determine how to place the successive layers very precisely. Sometimes one can use local reinforcements, wafers
for instance; but I think that since the quality of the theoretical predictions is not very good in this area, the wafers are often
used as a security, but I don't think they are always necessary.
4.2 Mr Carrier, DREV, CA
Do you know of any public domain "Design handbook for composite cases for SRM"?
MrDenost
I'm not aware of any publication, beside the two papers of this L.S., of course.
Mr Reydellet
There are a few specific courses in the French school, "Ecole Nationale Superieure de l'Aeronautique et de l'Espace"
(Toulouse, France), but they are in French.

5.

QUESTIONS PUT TO MR THORP

5.1 Mr Platzeck, NWC, China Lake, US

We heard of the strengths and advantages of strip laminate and filament cases that have been around some time, but still in
limited use. What are some of the weaknesses of disadvantages of these types of cases?

RTD-IO
MrThorp
I will speak only of strip laminate. At first, I must say that the "Rapier" system, which is one of the best missile systems in
UK uses a solid rocket motor with a strip laminate case. The second generation of the Rapier will use also a strip laminate
case. A number of other systems currently in production in UK use also strip laminate cases. I agree that all around the
world it has a more limited use, but we will try to do something for that. This technology is applicable to high tensile steels,
but not to light alloys. Particularly high length to diameter ratios are required, because if the ratio is smaller than 4, the size
of the end rings becomes a problem, so that you are probably better off having an homogenous monolithic case. Another
disadvantage is the aerodynamic heating that you may have with missiles fired from high speed aircraft, because at some
temperature the resin starts to lose its strength. This phenomenon occurs from lOOC or 130C. The strength decreases
continuously, and at 200C, you can run into trouble. The answer to this is to apply a thermal protection to the case: A lot
of materials are available for that use. Another case where protection will be necessary is the underwater use; the
protection will be necessary to protect the resin against humidity.
Dr Evans
I shall answer for the filament wound cases. Primarily, there is a problem of initial cost (due to materials and processes)
that often excludes these technologies from tactical applications. The other problem is that composite cases are said to be
more subject to damage under severe environments (aeroheating, impact...)
But requirements for tactical applications are becoming more and more ambitious every day. Composite cases could be a
possible answer to the search for performance. But there could be some limitations with the complexity of design for some
tactical motor cases. However, the data base for designing and manufacturing composite cases becomes greater and
.greater every day, so we can hope that we will be able in the future to solve technical and economical problems as well,
specifically for mass production.
MrPlatzeck
I heard about bonding and rings to the strip laminate cases and bonding the strip laminate body, but no indication of the
adhesive used, cleaning preparation, nor inspection procedure to ensure that the bond is acceptable.
MrThorp
Firstly, the adhesives we use are basically epoxide resins. We have two different adhesives, one for the strips and one for
the end rings, because of the different requirements, but they are very similar; the strip adhesive is a high molecular weight
resin; the end ring adhesive has a lower molecular weight. I have some very long chemical formulas down here. I don't
intend to read them, but you are welcome to come and take a look at them.
Concerning cleaning and preparation, we take great care to degrease the strip when we receive it, then shot blast it; then we
degrease it again before we put resin on it; when resin is applied, it is semi-cured and then it can be held for six to nine
months before we use it.
Concerning inspection, for the strip and the resin, we use stringent methods (detailed chemical characterization for the
resin, for instance). For the motor body, the ultimate inspection is the pressure proof test. Every case is proofed to a
pressure well above the level of the maximum operation pressure. The body is measured very precisely before and after
the test, to make sure that there was no movement of the end rings during that test. We have just tried to use some acoustic
emission analysis during the pressure test, because the cases groan and moan a little bit. But it was just research. In
addition, we have never used ultrasonic inspection methods for the case, but only for chccking the propellant grain
integrity.
5.2 J.Lu, USAF Aeronautics Lab, Edwards, US
I would like to know your opinion on the incorporation of fracture mechanics in motor structure design and service life
prediction, and, also, the current status on fracture mechanics research in France and in UK.
MrThorp
Fracture Mechanics is not in my line; I think Mr Reydellet is more equipped to answer this question, for France and UK,
and he has promised to do so.

Mr Reydellet
The fracture mechanics can be used for propellants, but I never heard that it can be used for composite cases. It is now very
classical to use it for metallic alloys; but the advances on that subject don't come from rocket propulsion, but from aircraft
applications. However, fracture mechanics can be used for metallic cases in SRM for different purposes.
The first is to assess the adequacy of non-destructive inspections for metallic cases: by measuring the K! C of the materials,
it is possible to calculate the size of the critical defect that will lead to failure under operating pressure, and compare this
size to the smallest tlaw that can be seen by non destructive tests. It is /101 advisable to have a very small critical size (1 mm
is bad: 5mm is better!)
The second concern is the stress corrosion problem; if you have permanent stresses, you have to take into account the
K1SCC of your alloy ("K 1 seuil de corrosion sous contrainte" in French), that is the value of K 1 under which you cannot
have stress corrosion.The last concern could be a fatigue problem. if you have cyclic loads; in this case, you have to use another criterion, based
on another threshold for K!.
As a conclusion, all these methods are not quantitative design methods, but rather qualitative criteria (determined by
experience), to be used in order to make sure that one will not be running into difficulties with a fracture mechanics
problem.

RTD-ll

Dr Zeller
I would like to point out that some work is being performed currently in France dealing with the kinetics of crack
propagation in propellant grain. The AIAA paper of Mr Nottin which was given at the Joint Propulsion Conference last
year can give you a good idea of what is done in France in that field. But it is at this time only research; it is not used as a
routine acceptance criterion for current production.

6.

QUESTIONS PUT TO DR ZELLER

6.1 MrLichtinger, UTC/CSD, US

What work is France doing in the area of 1.1 class solid propellants?
Dr Zeller
The threshold between 1.1 and 1.3 propellants is not the same in USA and in France; in France, the 1.1 propellants begin
at 240 cards of card-gap test, but the cards used have not the same thickness as in USA. In USA, 1.1 propellants have a
NOL card gap test greater than 70 cards. This is equivalent to 90 French cards (according to an experimental program
that we have performed). As a consequence, 240 French cards are equivalent to about 190 US cards. So, in France almost
all cured propellants are 1.3, but some of them could be 1.1 in USA!
In the 1.1 US class category, we find extruded propellants (which are still very useful), cast double base or composite
modified cast double base, and lastly the new high energy cross link double base or NEPE 'type propellants. All these
propellants are 1.1 class in USA, and we routinely manufacture any kind of these propellants.
Mr Lichtinger
Can you provide me with a summary of your processing capabilities? How many mixers? What size? Pressure cure
capabilities? Curing ovens? etc...
Dr Zeller
I shall not enter into details. For extruded and cast double base, we have manufactured so many grains every year, that it is
an evidence that we have all necessary facilities. For NEPE and cross-link double base, we have all the facilities starting
from the one pint mixer (for research centre) to the 420 gallons mixer (industrial production).
6.2 M.Carrier, DREV, CA
Can you tell use what is being done in Europe to experimentally validate the grain structural analysis codes that are used in
all European projects?
Dr Zeller
In France, we use in-house developed 3D codes. We use these codes every time we develop a new case bonded grain.
Since we have already successfully developed a lot of grains, I think that it is a kind of validation. In addition, during every
engineering development, we usually perform "overtests", in order to assess the actual margin of safety of the grain (if
necessary by obtaining mechanical failure). We can say that, roughly speaking, we have a good agreement with the
calculations. We can also check the validity of the codes by using analog motors; a paper will be presented on the next
AIAA propulsion meeting in San Diego on that subject. On the other hand I must say that we dOIl't use transducers for
measuring stresses in the propellant.
6.3 Mr Thrasher, USAF Astronautics Lab, Edwards, US
What sort of structural modelling capability is used within your grain design code? (Closed form, finite element or
something else?)
Dr Zeller
At the very beginning of a project, we use closed form equations including the central port geometry and end effects. For
special geometries, such as slotted tubes, we use the results of a parametric study using 3D finite element code
calculations; we use them on a data base.
MrThresher
Are structural deformations considered in your grain burnback analysis? If so, how?
Dr Zeller
No, we don't take into account the deformations of the grain for ballistic predictions.

7.

QUESTIONS PUT TO MR THRASHER

7.1 Mr Platzeck, NWC, China Lake, US

Some grain designs, especially radial slots, look impossible to inspect. How's it done?
MrThrasher
Most of the inspections are made visually; by using an endoscope, borescope or something similar.
Another technique is X-Ray, which can be useful to detect cracks (even radial cracks).
Computed tomography can be also effective, but it is for the future.
7.2 MrJones, Bristol Aerospace, CA
As a result of casting now effects, the mechanical properties and burning rate of composite propellants vary with position
and orientation in the motor. In the future. is it going to be necessary to allow for these effects in the design stage and how
may they be predicted?

RTD-12

MrThrasher
This is definitely a real-world problem, because it really does happen; another problem is the effect of chemical migration
and reaction which leads to mechanical properties gradient. Before looking if it will be "necessary", I would like to look if it
will be "possible" to predict these phenomena: In fact, the flow of uncured propellant is a very complex rheological
unstable phenomenon. I don't think it will be possible to obtain an advanced prediction tool. However, I think that the
phenomenon is rather repeatable, when the distribution of aluminium and oxidisers are determined. It could be possible
to dissect a motor and to look at the ballistical and mechanical gradients and anisotropy; these data could be taken into
account in the analysis from a ballistical and structural point of view.
Dr Zeller

We have developed an internal ballistic code which can take into account gradients of burning rate with the location and
orientation in the grain. The problem is that we have at this time no data to run this code!

8.

QUESTIONS PUT TO ALL SPEAKERS

8.1 Mr B.Geisler, USAF Astronautics Lab, US

In your experience, which factors do you believe have caus~d the majority of motor failures?
a) poor design
b) poor manufacture
c) ageing and environmental factors
d) other factors?
MrHiidreth
I can speak of four examples: two involved spare boost motors and two others launch vehicle boosters. The failures of
space boost motors were nozzle failure. In one case, it was obviously a bad design, not necessarily at the time of the design,
but subsequently with the involute exit cone we had to improve the design and the technology, because we couldn't
manufacture well enough.
The second one was a critical tolerances problem.
The fiist failure of launch vehicles was an excessive erosion problem in nozzle. This failure was the result of changes in
material, over a period. It was neither a design problem, nor a manufacturing problem, but changes were made since the
first design.
The second one is the really catastrophic accident that everybody knows. This failure occurred after many successful
flights, so one cannot say that there was a catastrophic flaw in the design. Something particular must have happened to that
particular segment, maybe in environmental conditions. So one has to put this failure in another category.
As a conclusion, I could say that the major problem we have is not only to have a good design, but also to manufacture it
over a long period of time.
MrTruchot
We only experienced failures during the development of the first generation of our strategic missiles. In one case, it was
poor mechanical and thermal design of the nozzle (specifically with pyrolithic graphite throats).
Dr Evans
I would say at first that, relating to problems that appeared in the propulsion community in this country, I'm convinced
that our customers have a certain responsibility. Rocket motors are basically very simple devices. They have a relative
technological maturity; in the past ten years, we have seen a big pressure for decreasing the amount of money and work
devoted to the effort of understanding what could be the necessary margins of safety. In addition, competition between
manufacturers leads to propose fewer and fewer test articles in an engineering development. Furthermore, the search for
performances has never been so great. This is, I think, what has pushed us in that direction.

MrDenost
I have no personal experience of major failure in composite vessels. But I must say that the designer has to take into
account a sufficient margin of safety to accept the variations that may occur during fabrication. He has to check that there
is no potential danger: for instance critical areas where slight changes during the production phase may lead to a poor
margin of safety. So, if in the future we meet some major failure, I think we shall have to look at first at the design.
MrThorp
The example I have now in my mind concerns a tactical missile with four SRM boosters, which must peel away at the same
time. After some time in service, it appeared they were not peeling away at the same time. As a matter of fact, it was a
problem of ageing of the propellant. Is it a design or an environmental problem? It is probably a combination of the two
causes, plus the fact that insufficient testing was done in the development phase to prove the service life.
Dr Zeller
The most recent failure I have in mind occurred in an end-burning free-standing grain. The cause is probably a mixing
between poor design and ageing. The failure was due to migration of nitroglycerine in the inhibiter; the result was an
ignition in an area which was not predicted by ballistic codes, then overpressure and exlosion of the case. During the
engineering development of this motor not enough work was performed concerning the ageing, not only on the propellant
by itself, but also ageing of the whole propellant grain in the real environment during its life. So I think the most important
thing is to have a sufficient number of good ageing tests.

RTD-13

MrThrasher
At first, I would like to say that, in my opinion, a designer must apply all the knowledge that is available in a reasonable way
(structural, ballistic...) In that context, I have only one example in my mind: that was really a design error. This concerns a
free-standing grain, for a nozzle-less motor. The designer failed to consider that there was a gradient of pressure along the
grain, and a very rapid peak of pressure grew up at the nozzle end of the grain; the grain collapsed, and there was a very
effective explosion! Anyway, every other failure has been basically due to problems in designing, specifically because of
the lack of realistic data for the analysis: the materials in the motors have often not the same properties as the properties
used for the analysis. The lack of data in a sufficient spectrum of loading conditions causes major failures.

Additional Comment by Mr B.Geisler

Previously, we have put a lot of money into studying the ageing of the propellant. But, with the experience, we discovered
that it was not the cause of failure: we'd better put more money in design and manufacture practices.
The other trend I see is that we are cutting the margin of safety continuously. We think we are smarter; we make the parts
thinner and thinner, lighter weight, and when our understanding lacks a little bit, we have a failure.
The other frequent problem is to get out of the original design by slight modifications, after slight modifications... and so
on... and what was once a good design no longer works.

BIBLIOGRAPHY
This Bibliography was compiled by CEDOCAR (CEntre de DOCumentation de 1'ARmement) in consultation
with the Director of the Lecture Series, Ingenieur en Chef de l'Armement D.Reydellet.

B-1

BIBLIOGRAPHY OF LECTURE SERIES N 150

"DESIGN METHODS FOR SOLID ROCKET MOTORS"

I. - General Optimization

1.1

24063 C.CEDOCAR
REFERENCE QUESTE
titre ang.
Auteurs
Type de doc.
Langue
No rapport
Source
resume

Classification
Descripteurs

1.2

24062 C CEDOCAR
REFERENCE QUESTE
titre ang.
Auteurs
Type de doc.
Langue
No rapport
Source
resume

Classification
Descripteurs

AE-024064
User's manual for solid propulsion optimization code
(spoc) volume I - technical description.
ROYS G. P.
RAPPORT
ENG
NTIS AD A 108224
NP. 240; 91 FIG. ; 24 REF. ; DP. 08/81; MICROFICHE
Code de calculs pour optimiser les recherches
concernant la propulsion de moteur fusee a propergol
solide. Le spoc adaptable sur ibm 4341 et cdc 6600
elabore un avant projet compte tenu des exigences des
performances, des contraintes de conception et des
limites d'operations. Determination des principaux
composants moteur et de leurs possibilites en utilisant
les dimensions et caracteristiques de base. Modele de
schema d'optimisation non lineaire a partir de
l'algorithme hooke et jeeves. Formulation du propergol,
tuyere combustion, dimensionnements. Manuel I de
l'utilisateur.
014
Moteur fusee ; propergol solide; Modele mathematique ;
Non linearite ; Calculateur numerique ; Optimisation ;
Code ; Avant-projet ; Conception; Analyse structurale
Simulation analogique ; Combustion ; Rentabilite ;
Fortran.

AE-024063
User's manual for solid propulsion optimization.code
(spoc). Volume 2. User's code.
ROYS G. P.
RAPPORT
ENG.
NTIS AD A 108225
NP 217; 67 FIG. ; DP. 08/81 ; MICROFICHE
Volume 2 du manuel d'utilisateur d'un code pour
ordinateur qui execute les avant-projets de detail
de moteurs-fusee a combustion solide. Tous les
composants principaux et les performances du moteur
sont mathematiquement determines a partir des
dimensions et des caracteristiques de base. Une
methode d'optimisation.en recherche non lineaire
basee sur l'algorithme de hooke et jeeves est
utili see pour etablir les caracteristiques du moteur .
et optimiser l'un quelconque des parametres. Les
variables sont les proprietes du combustible, les
dimensions du nez et des canaux.
014
Propulseur; Moteur fusee ; Produit combustion ;
Propergol solide ; Conception; Projet ; Projectile
Ordinateur ; Logiciel ; Optimisation ; Algorithme ;
Mise en oeuvre.

B-2

1.3

24064 C.CEDOCAR
REFERENCE QUESTE
titre ang.
Auteurs
Type de doc.
Langue
No rapport
Source
resume

Classification
Descripteurs

1.4

155533 C.CEDOCAR
NUMERO
titre fro
titre ang.
Auteurs
Affiliation
type de doc.
Langue
Titre publi.
Source
CODEN
localisa tion
Classif.
resume

source analyse
Code Cosati
Descripteurs

AE-024065
User's manual for solid propulsion optimization code
(spoc) volume 3. Programm description.
ROYS G. P.
RAPPORT
ENG
NTIS AD A 108226
NP. 24; 8 FIG. ; DP. 08/81; MICROFICHE
Volume 3 du manuel d'utilisateur d'un code pour
ordinateur qui determine les avant-projets de detail
de moteurs-fusee a combustible solide. Tous les
composants principaux et les performances du moteur
sont calcules a partir des dimensions et
caracteristiques de base. Une methode d'optimisation
en recherche non lineaire basee sur l'algorithme de
hooke et jeeves est utilisee pour etablir les
caracteristiques du moteur et optimiser l'un quelconque
des parametres. Les variables sont les proprietes du
combustible, les dimensions du nez et des canaux.
014
Propulseur; Moteur fusee; Produit combustion; Propergol
solide; Conception; Projet; Projectile; Ordinateur;
Logiciel; Optimisation; Algorithme; Mise en oeuvre;
Programmation.

C-84-003334
(Conception assistee par ordinateur des moteurs fusees
a propergols solides).
Computer aided preliminary design of solid rocket motors
JACQUES L. ; ROUX J.
(I) Societe Europeenne de Propulsion, Saint Medard
en Jalles (France)
Publication en Serie
ENG
AlAA Paper (US)
NO 83-1254 (06/83); 8 P; 13 fig. ; 1 tabl.
AlAA/SAE/ASME 19th Joint Propulsion Conference,
Seattle, WA. ; 27-29/06/83
AAPRAQ
05; Me 300-1
2108
Description du programme de calcul de la configuration
d'un moteur en fonction des criteres appliques a chacun
de ses composants. Exemple d'application a un etage de
missile balistique de portee maximum.
INFO/SJ
21 08
- Moteur fusei propergol solide
Conception ~ssistee
par calcul
; Etude conception moteur fusee ,

*;

- Missile balistique ; Performance systeme propulsion

Configuration systeme propulsion ; Structure moteur
Progr&~me

1.5

1216 C.CEDOCAR
REFERENCE QUESTE
titre ang.
Auteurs
Type de doc.
Langue
No rapport
Source
resume

Classification
Descripteurs

calculateur .

AE-001217
Advanced air launched missile rotor design methods.
ROYS G. P.
RAPPORT
ENG
NTIS AD A 108634
NP. 42 ; 10 FIG. ; 13 REF. ; DP. 09/81 ; MICROFICHE
Rapport sur un projet en 3 phases pour la mise au point
d'un programme d'ordinateur pour calculer l'avant-projet
de moteurs-fusee a propergol solide. Tous les composants
principaux et les caracteristiques sont determines numeriquement.
Une optimisation basee sur l'algorithme de hooke et de jeeves
sert a determiner le compromis ideal pour des parametres. Les
contraintes choisies sont les parametres precites, les formes. Les
variables sont la formulation du propergol, Ie rendement de
combustion, les dimensions de la tuyere, etc.
014; 023
Propulseur; Moteur fusee; Methode; Calcul numerique; Logiciel;
Propulsion chimique; Propergol solide; Optimisation; Combustion;
Caracteristique aerodynamique; Missile.

B-3

I. 6 98253 C. CEDOCAR
NUMERO
titre fro
autre titre
Auteurs
type de doc.
Langue
editeur
Source
localisation
Classif .
resume

source analyse
Code Cosati
Descripteurs

Identificateurs
1.7

124194 C.CEDOCAR
NUMERO
titre fro
autre titre
Auteurs
type de doc.
Langue
editeur
Source
localisation
classif .
resume

source analyse
Code Cosati
Descripteurs

Identificateurs
1.8

8631 C.CEDOCAR
NUMERO
titre fro
titre ang.
Auteurs
Auteur coll.
type de doc.
Langue
Titre publi.
Source
CODEN
localisation
classif.
resume
source analyse
Code Cosati
Descripteurs

C-82-002087
(Les moteurs - fusees a combustible solide) .
Rakeinye divigateli tvierdogo topliva
FAKHROUTDINOV I. K.
Ouvrage
RUS
Machinostroenie (Moscou)
(1981), 223 P. ; 87 ref.
101 fig.
10 tabl.
3 photo
05 ; 440-132
2108
Theorie et pratique de la fabrication des moteurs fusees a combustible solide. Calcul des parametres
questions d'organisation et questions economiques
concernant la fabrication et la conception de ces
moteurs.
CAEN/DI.EX/MP
21 08
- Moteur fusee propergol solide *; Etude conception x;
- Parametre ; Protection thermique; Charge propulsive ;
fiabilite; Technologie; Etude theorique; Etude technique
Monographie; Ouvrlge;
Combustible solide ; Publication sovietique.

C-82-018365
(Bases theoriques sur la conception des moteurs-fusees
a combustible solide).
Teoretitcheskie osnoy proekirovaniia RDTT
EROKHIN B. T.
Ouvrage
RUS
Machinostroenie, Moscou
(1982), 206 P. ; 81 ref.
38 fig.
8 tabl.
05; Numero CEDOCAR/440-482
2108
Modeles physique et mathematique pour la conception
des moteurs-fusees a combustible solide. Methode
mathematique d'optimisation des caracteristiques geometriques
des moteurs-fusees a combustible solide.
CAEN/MP
21 08
x
Etude c~nception moteur fusie ; Moteur fusee propergol
solide ; Etude conception ;
- Modele mathematique; Tuyere moteur fusee; Chambre
combustion; Etude theorique; Ouvrage:
Publication sovietique; Optimisation.

C-79-001619
(Etude thermique, analyse et performances d'un systeme
rotatif a propergol solide).
Thermal design, analysis, and performance of a solid
propellant spin system
HWANGBO H. ; EBY R. D. ; SUMMER P. R.
Agence Spatiale Europeenne
Publication en Serie
ENG
ESA Symposium "SP"
NO 139 (11/78); PP. 365-373
9 ref.
17 fig. Munich
(10-12/10/78)
ESFPA4
OS: M.3091-4
2108
Modele de calcul des caracteristiques balistiques ,
thermiques et d'ecoulement de fluide. Prediction des
performances en vol.
CAEN/DA
21 08
- Moteur fusee iropergol solide ; Etude conce~tion
moteur-fusee ; Commande attitude satellite ;
- stabilite attitude ; Analyse thermique ; Balistique ;
Ecoulement fluide ; Modele mathematique ; Corps rotation;
Generateur gaz.

B-4

2. - Case Design
2.1

51378 C.CEDOCAR
NUMERO
titre fro
Auteurs
Affiliation
type de doc.
Langue
Titre publi.
Source
CODEN
Isbn
localisation
Classif.
resume

source analyse
Code Cosati
Descripteurs

Identificateurs

C-80-F04021
Enroulement filamentaire de fibres non metalliques.
Conception et realisation de structures de propulseurs
a poudre.
LAMALLE J. ; CABANEL D.
(1,2) SNIAS, St Medard en Jalles
Publication en Serie
FRE
Institut National des Sciences et Techniques Nucleaires
Saclay (FR)
NO 22 (1980), PP 355-382; 25 fig. (22e Colloque de
Metallurgie de Saclay, 19-20 et 21/6/79)
PCMLA2
2-225-66132-4
05; M.5570
1104
Conception et dimensionnement de structures de propulseurs a
poudre realisees par erroulement filamentaire de fibres nonmetalliques. Les differents materiaux utilisables ; les
procedes de fabrication et de contr6le; les moyens mis en
oeuvre; des realisations de la SNIAS.
Auteur-ATRE/PZ
11 04
- Composite matrice ep~xy;: Construction filament erroule ;:
Reservoir propergol ;
- Plastique renforce fibre verre ; Propulseur; Fusee
Resistance structure ; Calcul structure : Bobinage
Memoire congre~ ;
Kevlar materiau ; Materiau preimpregne : Sac lay congres
1979 annee ;

B-5

3. - Propellant Grain Design

3.1

1480 C.CEDOCAR
NUMERO
titre fro
titre ang.
Auteurs
type de doc.
Langue
Titre publi.
Source
CODEN
Issn
localisation
Classif .
resume

source analyse
Code Cosati
Descripteurs
Identificateurs
3.2

147252 C.CEDOCAR
NUMERO
titre fro
titre ang.
Auteurs
Affiliation
type de doc.
Langue
Titre publi.
Source
CODEN
Issn
Isbn
localisation
classif .
resume

source analyse
Code Cosati
Descripteurs
Identificateurs

C-79-000221
(Optimisation des performances des propergo Is
solides a additifs metalliques)
Performance optimization of metallized solid
propellants
SWAMINATHAN V. ; RAJAGOPALAN S.
Publication en Serie
ENG
Propellants and Explosives ; (DE)
VOL.3, NO 5 (10/78), PP. ISO-ISS ; IS ref.
3 tabl.
PREX2S
0340-7462
05
1901
On cherche a ~eterminer la composition optimale
d'un type de propergol solide contenant un additif
metallique en vue d'obtenir l'impul~ion specifique
maximale. On a utilise la methode de projection des
gradients, proposee par Rosen. On presente et discute
les resultats.
ATRE/PO

..

19 01
.
~
- Propergol sol~de ; Impuls~on spec~f~que
- Liant propergol solide ; Chaleur formation
Poids moleculaire
Modele mathematique ;
Propergol metallise

C-83-015378
(Analyse dynamique de structures viscoelastique
utilisant la methode des elements finis incrementale).
Dynamic analysis of viscoelastic structures using
incremental finite element method
CHEN W. H. ; LIN T. C.
(I) National Tsing-Hua University, Hsinchu, Taiwan.
Publication en Serie
ENG
Univelt, Advances in Astronautical Sciences (US)
VOL 50 (1983), PP. 710-722 ; 23 ref. ; 6 fig.
NCKU/AAS. International Symposium Tainan-Taiwan,
29-31/12/81 (Part II)
ADASA9
0065-3438
0-87703-166-51
05; Me 920
2011
Presentation d'une methode simple et efficace, qui
ne comprend pas de transformations integrales basee
sur Ie principe variationnel d'Hamilton, pour etudier
des structures viscoelastiques de geometrie complexe,
soumises a des charges dynamiques.
Modelisation du comportement constitutif, technique de
resolution. Deux exemplea illustratifs montrent la
precision de la methode. Etude des effets transitoires
des structures viscoelastiques soumises a des charges
instationnaires.
INFO/SN
20 11
- Viscoelasticite*; Analyse dynamique structure *
- Propergol fusee solide ; Aube turbine gaz ; Methode
element fini ; Dependance du temps ;
Charge variable

B-6

3.3

174955 C.CEDOCAR
NUMERO
titre fro
titre ang.
Titre conf.
LIEU DE CONF.
DATE CONF.
Auteurs
Affiliation
Auteur colI.
type de doc.
Langue
Titre publi.
Source
CODEN
Issn
localisation
resume

source analyse
Code Cosati
Descripteurs

Identificateurs

C-84-015323
Influence directe des contraintes sur la vitesse de
combustion des propergols solides composites.
The direct effects of strain on burning rates of
composite solid propellants.
20th Joint Propulsion conference
Cincinnati (US)
1984/06/11-1984/06/13
LANGHENRY M.T.
Auburn Univ. AL (US)
American Inst. of Aeronaut. and Astronaut.
SAE ASME
Memoire Congres
ENG
AIAA Papers (US)
NO 84-1436; 8 p.; 12 ref.
9 fig.
I Tabl.
DP.1984

AAPRAQ
0146-3705
05; ME.300-1
Presentation d'un modele mathematique permettant de
prevoir l'augmentation de la vites~e de combustion due
aux contraintes mecaniques dans un propergol composite
solide. Ce modele attribue cette augmentation a la
capacite des flammes a penetrer dans les petites fissures
et les vides qui se forment lorsque Ie propergoI se deforme.
Le nombre et la dimension des fissures sont obtenus en
appliquant une analyse de propagation de defaut a partir d'une
distribution initiale aleatoire due a la decohesion sous
contraintes des particules de liant et d'oxydant.
INFO/CR
19 01 ; 20 II
- Propergol composite x; Combustion propergol ;
Relation deformation contrainte
- Performance systeme propulsion ; Vitesse combustion
Moteur fusee propergol solide ; Propagation fissure
Modele mathematique ; Thermodynamique statistique ;
Propagation flamme
Facteur intensite contrainte

*;

B-7

4. - Nozzle Design
4.1

53667 C.CEDOCAR
NUMERO
titre fro
Auteurs
Auteur colI.
type de doc.
Langue
Titre publi.
Source
CODEN
Isbn
localisation
classif .
resume

source analyse
Code Cosati
Descripteurs

Identificateurs
4.2

180094 C.CEDOCAR
NUMERO
titre fro
Titre conf.
LIEU DE CONF.
DATE CONF.
Auteurs
Affiliation
type de doc.
Langue
Ti tre pub li .
Source
CODEN
Issn
Isbn
localisation
resume

C-80-F04680
Prediction du comportement des materiaux phenoliques
ablatifs.
BONNET C.
Advisory Group for Aerospace Research and Development
Publication en Serie
FRE
AGARD Conference Proceeding (FR)
VOL. CP-259, NO 30 (7/79), PP. 30 1-30. 16
7 ref.
25 fig.
AGCPAV
92-835-0243-4
05 ; Me 372-15
2108
Expose destine a montrer les efforts accomplis pour ameliorer
les methodes de fabrication des materiaux composites ablatifs
et des pieces,mettre au point de nouvelles methodes d'analyse
et des programmes de calcul permettant de predire Ie comportement mecanique et thermique des pieces de tuyere en tir,
caracteriser ces materiaux aux temperatures d'utilisation.
CAEN/MS
21 08
~ properg~I so l'd
- .
.
*,
- Moteur f usee
~ e *; M
ater~au
compos~te
tuyere moteur fusee ; materiau ablation
- resine phenolique ; resistance mecanique ; Resistance
thermique; Procede fabrication
Comportement materiau .

C-85-F00952
Etudes aerodynamiques liees au developpement des divergents
deployables.
21e colloque d'aerodynamique appliquee.
Ecully (FR)
1984/11/07-1984/11/09
VENABLES A. : LARUELLE G.
ONERA (FR) ; ONERA (FR)
Memoire Congres
FRE
AAAF Notes Techniques WR)
,NO 18 ; 71 p. ; 16 Ref.; 38 Fig.
ONERA TP 1984-149
DP. 1984/4T
AAAN2Y
0243-0177
2-717-00794-6
05 ; M 1180
Dans Ie but d'augmenter la portee d'un missile balistique
a encombrement donne, on etudie l'adaptation de divergents
deployables pour les etages superieurs. Description des
diverses configurations de cas divergents, expose des

source analyse
Code Cosati
Descripteurs

Identificateurs

methodes de calcul utilisees pour l'estimation des efforts,

essais effectues pour verifier les resultats.
INFO/MS
21 08 ; 12 01
- Tuyere moteur fusee *; Calcul mathematique *; Tuyere
divergente *;
- Etude conception tuyere ; Missile balistique ; Mesure
pression ; Essai soufflerie ; Moteur fusee propergol solide
Moteur missile
Methode calcul ; Mesure ecoulement.

*;

B-8

4.3

202235 C.CEDOCAR
NUMERO
titre fro
titre ang.
Titre conf.
LIEU DE CONF.
DATE CONF.
Auteurs
Affiliation
Auteur colI.
type de doc.
Langue
Titre publi.
Source
CODEN
Issn
localisation
resume

source analyse
Code Cosati
Descripteurs

Identificateurs

C-86-002072
Conception et optimisation de systemes de propulsion
utilisant des tuyeres chanfreinees.
The design and optimization of propulsion systems
employing scarfed nozzles.
21st AIAA/SAE/ASME/ASEE Joint Propulsion Conference
Monterey US
1985/07/08-1985/07/10
LILLEY J.S.
US Army Missile Command, Redstone Arsenal, US
AIAA/SAE/ASME/ASEE
Memoire Congres
ENG
AIAA Paper (US)
NO AIAA851308
24 p.
5 ref.
20 fig.
6 tabl.;
DP 1985
AAPRAQ

0146-3705
05 ; ME 300-1
Decription d'une etude comportant deux phases:
utilisation d'un modele mathematique de performances
d'un systeme de propulsion pour evaluer l'influence de
la goemetrie des tuyeres sur leurs performances.
Utilisation des resultats de cette premiere phase pour
la conception et l'optimisation de tuyeres, devoyees et
chanfreinees, en fonction de la mission du missile qu'elles
equipent.
INFO/CR
21 08
- Moteur fusee propergol solide
- Performance systeme propulsion ; Etude conception moteur
fusee ; Ecoulement symetrie axiale ; Tuyere devoyee ;
Etude conception tuyere
Missile tactique.

*;

REPORT DOCUMENTATION PAGE

1. Recipient's Reference

2. Originator's Reference 3. Further Reference

AGARD-LS-150
(Revised)
5. Originator

ISBN 92-835-0454-2

4. Security Classification
of Document

UNCLASSIFIED

Advisory Group for Aerospace Research and Development

North Atlantic Treaty Organization
7 rue Ancelle, 92200 Neuilly sur Seine, France

6. Title

DESIGN METHODS IN SOLID ROCKET MOTORS

7. Presented at

8. Author(s)/Editor(s)

9. Date

Various

April 1988
11. Pages

10. Author's/Editor's Address

Various
12. Distribution Statement

248

This document is distributed in accordance with AGARD

policies and regulations, which are outlined on the
Outside Back Covers of all AGARD publications.

13. Keywords/Descriptors

Solid propellant rocket engines

Solid rocket propellants

Rocket engine components

Design

14. Abstract

This Lecture Series will try to summarize the current state-of-the-art in designing solid rocket
motors and their components. The aim is to collect the experience of several countries in using new
technologies and new methods which have been developed over the past ten years.

Specific sessions \vill deal \vith propellant grain, cases, nozzle, internal thermal insulations; the
question of the general optimization of a solid rocket motor will be emphasized.
The material in this revised publication was assembled to support a Lecture Series under the
sponsorship of the Propulsion and Energetics Panel and the Consultant and Exchange Programme
of AGARD presented on 18-19 April 1988 in London, United Kingdom, 21-22 April 1988 in
Saint-Aubin de Medoc, France, 25-26 April 1988 in Neubiberg, Germany and on 28-29 April
1988 in Rome, Italy.

AGARD Lecture Series No.150 (Revised)

Advisory Group for Aerospace Research and
Development, NATO
DESIGN METHODS IN SOLID ROCKET MOTORS
Published April 1988
248 pages

AGARD-LS-150
(Revised)
Solid propellant rocket
engines
Solid rocket propellants
Rocket engine components
Design

This Lecture Series will try to summarize the current stateof-the-art in designing solid rocket motors and their
components. The aim is to collect the experience of several
countries in using new technologies and new methods
which have been developed over the past ten years.

I AGARD Lecture Series No.150 (Revised)

Advisory Group for Aerospace Research and
Development, NATO
DESIGN METHODS IN SOLID ROCKET MOTORS
Published April 1988
248 pages

Specific sessions will deal with propellant grain, cases,

nozzle, internal thermal insulations; the question of the
general optimization of a solid rocket motor will be

P.T.O

P.T.O

This Lecture Series will try to summarize the current stateof-the-art in designing solid rocket motors and their
components. The aim is to collect the experience of several
countries in using new technologies and new methods
which have been developed over the past ten years.

AGARD-LS-150
(Revised)
Solid propellant rocket
engines
Solid rocket propellants
Rocket engine components
Design

Lecture Series No.150 (Revised)

I AGARD
Advisory Group for Aerospace Research

and

Development, NATO
DESIGN METHODS IN SOLID ROCKET MOTORS
Published April 1988
248 pages
This Lecture Series will try to summarize the current stateof-the-art in designing solid rocket motors and their
components. The aim is to collect the experience of several
countries in using new technologies and new methods
which have been developed over the past ten years.

Specific sessions will deal with propellant grain, cases,

nozzle, internal thermal insulations; the question of the
general optimization of a solid rocket motor w:ill be

Specific sessions will deal with propellant grain, cases,

nozzle, internal thermal insulations; the question of the
general optimization of a solid rocket motor will be

P.T.O

P.T.O

AGARD-LS-150
(Revised)

Solid propellant rocket

engines
Solid rocket propellants
Rocket engine components
Design

This Lecture Series will try to summarize the current stateof-the-art in designing solid rocket motors and their
components. The aim is to collect the experience of several
countries in using new technologies and new methods
which have been developed over the past ten years.

Specific sessions will deal with propellant grain, cases,

nozzle, internal thermal insulations; the question of the
general optimization of a solid rocket motor will be

AGARD Lecture Series No.150 (Revised)

Advisory Group for Aerospace Research and
Development, NATO
DESIGN METHODS IN SOLID ROCKET MOTORS
Published April 1988
248 pages

AGARD-LS-150
(Revised)
Solid propellant rocket
engines
Solid rocket propellants
Rocket engine components
Design

emphasized.

emphasized.

The material in this revised publication was assembled to support a Lecture Series under
the sponsorship of the Propulsion and Energetics Panel and the Consultant and Exchange
Programme of AGARD presented on 18-19 April 1988 in London, United Kingdom,
21-22 April 1988 in Saint-Aubin de Medoc, France, 25-26 April 1988 in Neubiberg,
Germany and on 28-29 April 1988 in Rome, Italy.

The material in this revised publication was assembled to support a Lecture Series under
the sponsorship of the Propulsion and Energetics Panel and the Consultant and Exchange
Programme of AGARD presented on 18-19 April 1988 in London, United Kingdom,
21-22 April 1988 in Saint-Aubin de Medoc, France, 25-26 April 1988 in Neubiberg,
Germany and on 28-29 April 1988 in Rome, Italy.

ISBN 92-835-0454-2

ISBN 92-835-0454-2

emphasized.

emphasized.

The material in this revised publication was assembled to support a Lecture Series under
the sponsorship of the Propulsion and Energetics Panel and the Consultant and Exchange
Programme of AGARD presented on 18-19 April 1988 in London, United Kingdom,
21-22 April 1988 in Saint-Aubin de Medoc, France, 25-26 April 1988 in Neubiberg,
Germany and on 28-29 April 1988 in Rome, Italy.

The material in this revised publication was assembled to support a Lecture Series under
the sponsorship of the Propulsion and Energetics Panel and the Consultant and Exchange
Programme of AGARD presented on 18-19 April 1988 in London, United Kingdom,
21-22 April 1988 in Saint-Aubin de Medoc, France, 25-26 April 1988 in Neubiberg,
Germany and on 28-29 April 1988 in Rome, Italy.

ISBN 92-835-0454-2

ISBN 92-835-0454-2

~@rn,[Q)

-~-

NATO

OTAN

mSTIURUTION OF UNCLASSIFIED
AGARD PUBLICATIONS

7 rue Ancelle 92200'NEUILLY-SUR-SEINE

FRANCE
Telephone (1)47.38.57.00 Telex 610 176

AGARD does NOT hold stocks of AGARD publications at the above address for general distribution. Initial distribution of AGARD .
publications is made to AGARD Member Nations through the following National Distribution Centres.Further copies are sometimes
available from these Centres, but if not may be purchased in Microfiche or Photocopy form from the Purchase Agencies listed below.
NATIONAL DISTRIBUTION CENTRES
ITALY
Aeronautica Militare
Ufficio del Delegato Nazionale all'AGARD
3 Piazzale Adenauer
00144 Roma/EUR

BELGIUM
Coordonnateur AGARD - VSL
Etat-Major de la Force Aerienne
Quartier Reine Elisabeth
Rue d'Evere, 1140 Bruxelles

LUXEMBOURG
ee Belgium .

CANADA
Director Scientific Information Services
Dept of..Mo';_-_1 ,,-~-- Ottawr
DENMARE
Danis.
Ved Id
2100 (
FRANCE
O.N.E.
29 Ave
92320
GERMANY
Fachini
Physik,
Karlsru
D-7514

postage and Fees Paid

National-Aeronautics and
Space Administration

A~ 1\

, \;;ill ,
National Aeronautics and
Space Administration
Washington, D.C.

NASA-451
Offlcla~Bu.lne ..

00
~

Panalty for Privata Usa $300

SPECIAL fOURTH CLASS MAIL

BOOK

20546

ishment

~ .:.~F.~,~_~.'0i...~IFli.-.~.~.r'!'"; ,.~.:.I~ .:.:.~_~,~.r"IF ~. I !,~,f_~'~';'}'~':It.~~.'.;.~,'1~8S000i6iD

..

",!.'i\II ,\

1'1

,.

f...:II:_:NTE!~~ (i~lF SC)

uE,,:. ... '.JlI .1,1. 1,. I

\GARD

'l~]:~~~~y;ri6cUMENTS
...

.....

us.0l ..

_.

ARNOLD AF STA IN

"1 .. t-~"")

~{~d~

GREECE
Helleni~

Aircraft
Departn
Holargo
KINGDOM
Defence Research Information Centre
Kentigern House
65 Brown Street
Glasgow G2 8EX

Ul'lllbU

ICELAND
Director of Aviation
c/o Flugrad
Reyjavik

UNITED STATES
National Aeronautics and Space Administration (NASA)
Langley Research Center
MIS 180
Hampton, Virginia 23665
THE UNITED STATES NATIONAL DISTRIBUTION CENTRE (NASA) DOES NOT HOLD
STOCKS OF AGARD PUBLICATIONS, AND APPLICATIONS FOR COPIES SHOULD BE MADE
DIRECT TO THE NATIONAL TECHNICAL INFORMATION SERVICE (NTIS) AT THE ADDRESS BELOW.
PURCHASE AGENCIES
National Technical
Information Service (NTIS)
5285 Port Royal Road
Springfield
Virginia 22161, USA

ESNlnformation Retrieval Service

European Space Agency
10, rue Mario Nikis
75015 Paris, France

The British Library

Document Supply Division
Boston Spa, Wetherby
West Yorkshire LS23 7BQ
England

Requests for microfiche or photocopies of AGARD documents should include the AGARD serial number, title, author or editor, and
publication date. Requests to NTIS should include the NASA accession report number. Full bibliographical references and abstracts of
AGARD publications are given in the following journals:
Scientific and Technical Aerospace Reports (STAR)
published by NASA Scientific and Technical
Information B r a n c h '
NASA Headquarters (NIT-40)
Washington D.C. 20546, USA

Government Reports Announcements (GRA)

published by the National Technical
Information Services, Springfield
Virginia 22161, USA

Printed by Specialised Printing Sen1ices Limited

40 Chigwell Lane, Laughton, Essex lGlO 3TZ
ISBN 92-835-0454-2