Dot 9079 DS1
Dot 9079 DS1
Dot 9079 DS1
AD-A238 812
John Zaniewski
June 1991
S D I
JUL29GETE
1981 -
I UTION 3TATKI A
Dbstzutihm UUimbted w
U.S DepartmenI of Transportaton
rc .1u Aviutfon .dm!n'stra! ,,
91-05839
91l!lllilllll8
NOTICE
NOTICE
The United States Government does not endorse
products or manufacturers. Trade or manufacturers'
names appear herein solely because they are considered
essential to the object of this report
DOT/FAA/RD-91/15, I
June 1991
Unified Methodology for Airport Pavement Analysis and
6. Performing Organization Code
Design
Vol. I State of the Art DTS-77
8. Performing Organization Report No.
7. Author s) DOT-VNTSC-FAlJl-91-7
John P. Zaniewski*
9. Performing Organization Name and Address 10. Work Unit No. (TRAIS)
U.S. Department of Transportation FAIJl/A1063
,John A. Volpe National Transportation Systems Center 1. ContractorGrantNo.
Research and Special Programs Administration
Cambridge, MA 02142 13. Type of Report and Period Covered
I.. Spornboring Ae..cy N.,me on; ',dress Final Report
U.S. Department of Transportation January 1989 - January 1990
Federal Aviation Administration
_ Research and Development Service 14. Sponsoring Agency Code
Washington, DC 20591 ARD-240
15. Supplementary Notes
19. Se ry Closf.(of rh-Isreport) 20, Security Clossil. (of this page) 21. No. of Pages 22. Pr..c
Uo c i7ssifi
it-d Unclassified 14)
iii
specific characteristics of pavement materials, i.e., the
conqtitutive equations of the materials, with respect to elastic,
plastic, and viscous behavior, properly represented. The task is
a formidable one; but with recent advances in fracture mechanics
and numerical analysis, and particularly with the advances in
computational capabilities of personal computers, it is a task
that can be successfully performed over the next several years.
Accession For
NTIS GRA&l
DTIC TAB
Unannounced 0
justificat0io
By
- ~ DI!,tribuItiouf
" st siaf1c o or
iv
METRIC/ ENGLISH CONVERSION FACTORS
CENT. £'-,S 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
25 1
Fcr- ,e e) l a J(4 " .cC, .o n f42 0 s, ssEc ,ES e aneous Pub lCzt;cn 2S6. U rts cf V ' g ,s and
ce'sc
%'tI SL" ;-:,e S2 . SD C2- No. C 13 1 2 6.
17
TABLE OF CONTENTS
Section Page
3. ANALYSIS MODELS......................47
4. DISTRESS MODELS......................79
4.1 Fracture.........................79
Vii
TABLE OF CONTENTS (cont.)
Section Page
viii
LIST OF ILLUSTRATIONS
Figure Page
1-1 PAVEMENT TYPES: (A) FLEXIBLF AND (B) RIGID ... ...... 4
Jx
LIST OF ILLUSTRATIONS
Figure Page
x
LIST OF TABLES
Table Page
xi
1. THE AIRPORT PAVEMENT PROBLEM
Paivements are one of the most difficul ' design problems faced
by civ4l engineers. Pavements are constructed with low-cost
materials whose properties are highly variable and dependent on
environmental and load conditions. Traffic loadings are difficrnl
to f, 2cast as air traffic growth frequently exceeds expectatic.1s
and as new aircraft are introduced. Environmental conditions can
be evaluated on a probabilistic basis f:om historical tiends;
however, the specific environment at a particular point in time
can have a dramatic effect on the perfoimanc- of the pavement.
I
characteristics of the pavement surface, primarily with respect to
ride quality. There is not a comparable definition of failure for
airport pavements. In airport pavements, the accumulated effect
of different types of distress are of major concern to the
pavement engineer. Toward this end, the pavement evaluation
procedures and subsequent computation of a pavement condition
index developed for the PAVER pavement management system " give
guidance for determining when pavement rehabilitation and
reconstruction is needed. The need for these treatments is
indicative of pavement failure. The PAVER method is widely used
for the evaluation of military airfields and civilian airports
and, therefore, limits of the PAVER defined pavement condition
index may become a de facto standard for the definition of airport
pavement failure.
2
This chapter presents a review of the factors affecting
airport pavement performance as a msans of establishing the
foundation for the review of pdvement analysis models. Included
is a discussion of airport pavement types, pavement material
characteristics, aircraft characteristics, functional areas of
airport pavements, and environmental effects. While the
complexity of these individual factors makes the development of
pavement analysis models a difficult task, the interaction of
these factors makes the modeling task even more challenging.
3
Binder Surface I Seal coat
coat
cors coreTc
coure~\coure i i /Tac ,-Prime
I Base course
Subbase course
Compacted subgrade
L----------------------------------------------
Natural subgrade
(a)I
Prlnd-cement concrete
IBas coure mayor may not be used
------------- L-------------------
(b)I
4
cracks can be controlled through the use of joints, reinforcement,
or combinations of such.
5
stresses in the pavement due to prestressing are
cumulative, with the flexural strength to produce an
increase in the stress range in the flexural zone3 . Due
to the greater range of allowable stresses in the flexural
zone, PCP can be thinner than other concrete pavements.
The prestressing is generally accomplished by
post-tensioning steel strands. The size of slabs of
prestressed pavements is limited by the ability to
post-tension the steel strands.
6
There have been several reviews of the literature on material
properties and the use of these characteristics for pavement
design. Research specifically on airport pavement design has been
performed at the Waterways Experiment Station under contract to
4
the Federal Aviation Administration. In particular, Chou
compiled material characteristics data for bituminous concrete,
portland cement concrete, granular materials, stabilized soil and
cohesive subgrade soils.
7
Linear response indicates that the deformation, or strain, is
proportional to the load or stress level (e.g., Houbling the
stress doubles the strain). Conversely, a nonlinear material
would not demonstrate a proportional relationship between stress
and strain.
9
value of the ratio of transverse to longitudinal strains for a
sample measured in uniaxial test condition:
E = ale
p = let/le
J(t) = e(t)/a 0
G(t) = a(t)/E 0
10
define the shear creep function, J.(t). Superposition can be used
to define the cumulative effect of these two creep functions,
i.e., the total strain in the three-dimensional case is the
addition of the strain due to volume and shear strains.
11
the material behavior. Generally, the aging and temperature
effects are considered in an incremental manner in mechanistic
analysis. The following discussion will address these
characteristics for each of the materials. Discussion of the
mechanistic models in subsequent chapters will address how these
characteristics can be included in an incremental manner for
estimating the performance of the pavements.
1.2.2 Steel
Steel is used for pavement reinforcement and load transfer
devices. Steel has a lattice structure composed of iron and
carbon atoms with other alloying agents used to impart special
properties. Compared to other materials used in pavements, the
properties of steel are relatively easy to quantify. For the
conditions encountered in pavement performance, steel is a linear
elastic material. Structural design of the steel in pavements is
relatively straightforward as Hooke's law applies and the modulus
of elasticity and Poisson's ratio are well-defined.
12
factors are equal, the strength of the concrete will be determined
by the water-to-cement ratio.
13
ultimate compressive strength. (Shear stress in the sample as a
result of the compression loading is actually responsible for the
failure of the material at the conclusion of the test.) However,
concrete is much weaker in tension than in compressicn. Since
pavements carry traftic loads in a flexural mode, the
tensile-flexural test is usually used for the design and quality
control of concrete used in pavements.
14
cement whose chemical compositions are carefully controlled. In
fact, ASTM specifications for asphalt cement are written around
physical tests of characteristics rather than chemical
composition. The gradc of the asphalt is evaluated with viscosity
tests.
1. mixture stiffness;
3. durability;
4. fatigue resistance;
6. permeability.
15
With the exception of durability and permeability, Monismith,
Epps and Finn recommend measuring the properties of the mixes in a
form which permits mechanistic analysis. For example, they
recommend measuring the stiffness as:
Smix(tT) = o/E
where:
16
10'
101
-30' C
d 10' .20* C
0 ,
10 10
* 10-aC
20* C
0 250 C
2 10-2 30
~.10-3 400 C
Estimated using shell procedure;
10-4
- ~ Asphalt property:
Pen. @ 25*C =50 dmm 0C
50ayC
1mmin 1hr. lSk
1da 0C
TR &B= T80pn.=54C
*~
10-5
16 1
~lweeks
10'
10-7 10- 10-6 10-4 10-3 10-2 10-1 10 10, 102 1023 10 4 10 3 10'6 O 101
Time, sec.
-E 100
(D
*0
> =c
so
'U0
S60
_
040__
00
2 10i0516 0
Asphalt Concrete Stiffness, psi
17
The behavior of asphalt concrete is very complex and
difficult to quantify. In the conclusion to an in-depth
literature survey of asphalt material properties, Deacon 11 states:
E* = osin(tat)/{Eosin(t -
where:
E = Complex modulus
18
removed, most of the strain will be recovered (after sufficient
conditioning), as shown on Figure 1-5. The resilient modulus is
the ratio of repeated stress to the corresponding recoverable or
resilient strain. The ASTM procedure for measuring resilient
modulus uses a diametral test mode. Researchers have also used
uniaxial and triaxial test modes.
19
U) 0 C/= CFOSIN ((o t)
U)
TIME
IE'I=r o o
FIGURE 1-4. TYPICAL PLOT OF STRESS AND STRAIN VERSUS TIME DURING
THE COMPLEX (DYNAMIC) MODULUS TEST (MAMLOUK AND SAROFIM)
TO7ALI
INSTANTANEOUS
LOAD
it I
DEFORMATION
20
portion of the time. Many pavement design procedures recognize
the problem with saturated bases, subbases and subgrades, and
require material characterization be performed on saturated
samples. Cedergren13 developed a damage factor for comparing the
damage to pavements with saturated bases and subgrades to well-
drained pavements. The damage factors ranged from 10 to 70,000.
Cedergren also demonstrated that pavements can remain saturated
for up to 20 days following a rainstorm.
21
TABLE 1-1. EFFECT OF INCREASING VARIABLES ON THE STIFFNESS
OF COHESIVE SOILS (DEACON)
Loading
Number of cycles Decrease
Minimum Minimum at 1 to 5000
cycles
Incremental strain Decrease Rate of decrease
amplitude depends on maximum
stiffness and shear
stress
Effective mean Increase Effect depends on stress
initial principal or strain amplitude
stress
Transverse stress no effect
Initial octahedral effect negligible after
shear stress 10 cycles
Frequency of loading Increase Effect minor above
10 cps
Strain rate Increase
Overconsolidation Increase Any effect can be
ratio explained on basis of
effective pressure and
void ratio
Stress path Large dependency
Mixture
Soil disturbance Decrease
Void ratio Decrease Maximum effect at low
confining pressure
Dispersion Decrease At small strains
Structure Little effect on max.
shear modulus
Degree of saturation Decrease Modulus of resilient
at compaction deformation
Plasticity Decrease
Compaction Energy Maximum Impact compaction
Environmental
Aging Increase
Degree of saturation Decrease
Time (thixotropy) Increase Recovery after high amp-
litude cyclic loading
or many load cycles
Densification Increase
Time (during sec- Increase Bentonite
ondary compression)
22
However, the response to an increase in the number of cycles is
different for cohesive and cohesionless soils. Table 1-2 is a
summary of how the various factors affect the stiffness of
cohesionless soils11 . Many investigators relate the stiffness of
cohesionless soils to the mean effective stress c o as:
S = K aon
MR = Ka 3n
and
MR = Kaon
4. decreasing fines.
23
TABLE 1-2. EFFECT OF INCREASING VARIABLES ON THE STIFFNESS
OF COHESIONLESS SOILS (DEACON)
Loading
amplitude
Mixture
Environmental
24
TABLE 1-3. EFFECT OF INCREASING VARIABLES ON THE STIFFNESS
OF UNTREATED GRANULAR AGGREGATES (DEACON)
Loading
Number of cycles Constant After 50 to 100 cycles
Initial confining Increase Triaxial compression
pressure
Initial effective Increase
mean principal stress
Incremental stress Constant to Differences in liter-
level Increase ature, large effect if
shear failure
Load duration Constant 0.1 to 0.25 sec.
Load rate or Increase Small increase
frequency
Drainage Constant
Mixture
Void ratio Decrease At low moisture
contents
Increase At high moisture
contents
Angularity and Increase
surface roughness
Fines Decrease Minor effect
Compaction water Decrease
content
Environmental
Degree of saturation Decrease
25
1.3 AIRCRAFT CHARACTERISTICS
One of the primary differences between highway and airport
pavements is the nature of the traffic loadings. Essentially,
commercial airport pavements are designed for a fewer number of
repetitions of heavier loads than highway pavements. However, the
concept of fewer number of repetitions is a relative term. The
1974 FAA advisotv circuLar for pavement design and evaluation (AC
150/5320-6B) had design curves for pavcments receiving up to
2!,000 annual departure,. These curves were extrapolated in the
1978 report to cover up to 200,000 annual departures based on
accelerated test data performed by the U.S. Corps of Fngineers.
However, Kohri ajQ Bentsen 15 report these test data were based on
the equivalent of 17,400 passes of a dual-tandem gear. The Dalla.-
- Ft. Worth Airport averaged 314,000 annual operations from 1974
to 1983 and Atlanta's Hartsfield Airport averaaed 577,000 annual
operations from 1984 to 1986.
8. numler of repetitions.
26
Aircraft characteristics are available in several references
and are only briefly summ.arized here. As shown in Figure 1-6,
there are three basic types of gear assemblies for civilian
aircraft: single tricycle, single bicycle-tricycle combination,
and double tricycle. The Boeinlg 727, McDonnell-Douglas DC-10 anl
the Boeing 747 respectively, are examples of each of these types
of assemblies. There are several types of tire assemblies as
shown in Figuie 1-7. Nose gea- assemblies are predominantly twin
tire assemblies. The main truck assembly of the heavier aircraft
are predominantly twin-twin tandems, e.g., B 747, B 707, DC 8 and
some models of the L 1011 (model 1). The DC 10 has a twin-twin
assembly under the wings and a twin gear in the center. Th ain
truck of the lighter commercial jets, B 727, B 737 and DC 9 are
twin tire assemblies.
27
/ \
/ \
/
/\
TMTM' TT
TNN
Single Tricycle
Tm T
/I' / I '
/ I \\/ I ' '
a Tb b, T M/l
T"O/ TO \T
•0 0 00 0
Twin: Triple Tandem Twin - Twin: Trlp!e Tandem
28
aircraft ti -es has the shape of an ellipse, as shown jn Figure
i-a. The total contact area is determined as a function of the
total weiqht on the tLre and the tire pressure. cenerally, the
conac2 pressure between the tire and pavement is assumed to be
,niforn. Research h:s shown that contact pressure between truck
:_ires and pavements is not uniform ' 1 71 8 "' '. The tire pressure of
commercial ajr' ;aft tires is in the range of 100 to 180 psi.
C
29
03Li 0L 0.31L
l
..............
L.
Oscillation Takeoff -
elDynamic
50 MPH 100 MPH 150 MPH o
Rejected Takeoff
Ree(Emergency)
] U'l
Parking
(Static)
30
TABLE 1-4. ACTIVITIES OF AN AIRCRAFT
IN ITS AIRPORT OPERATIONAL CYCLE (WIGNOT)
Departure Arrival
31
Due to the width of airport pavements, edge loading is of
less concern than it is on highway pavements. The outer wheels
are usually more than 15 ft. from the pavement edge 3
32
4. High horizontal loads during turns produce responses that
are temperature and rate of loading dependent.
1. concrete swell;
33
Low temperatures promote the following effects:
34
Methods are being developed that can be applied to the
analysis of these complex interactions. Figures 1-10 and
i-1i demonstrate the results of pavement fatigue analysis that was
25
developed at the University of Tokyo
35
-8.8
8, -8.9
6 -9.1
E -9.2
u! -9.3
0 c
,, -9.4
g -9.5
0E
-l -11.0
36
2. A GENERAL FRAMEWORK FOR THE DESIGN OF AIRPORT
PAVEMENTS
37
PLANNING D N V
ACTIVITIES DESIGN ACTIVITIES
* Assess Network Input Information on Materials
Deficiencies Traffic, Climate, Costs, etc.
" Establish Priorities Alternative Design Strategies
" Program and 0
Budget Analysis
Economic Evaluation
Optimization
Il .
r(CONSTRUCTIO ACTIVITIES
(1MAINTENANCE ACTIVITES
_S
RESEARCH-DATAPAVEMENT EVALUATION
AcTIITIES BAN
38
The main focus of this project is a review of pavement design
and analysis methods. There are several pavement design methods
available ranging from empirical procedures and guidelines to
sophisticated analytical models. However, as shown on Figure 2-1,
the design activities can basically be divided into three levels:
(Combined Subjective
0 0/Analytical
00 0 0 l Link)
O 4l O 0 O 0 tm
O O 0
O0
Z
IMPLEMENTATION
40
INPUTS Feedback
[i.Pavement
onstruc on
atriaRESPONSE
C FrMdlDeft
Structure
~ion
StanDistortio
Permanent
(Distress)
Disint rain
Weighting
Functions
W~
WOu
SYSTEM OUTPUT FUNCTION
Performance
nivironmentps
Atualy Iransporled--------- ----- -~*FLoad Apps
TRAFFIC ch
z
CLIMATE PAVEME RESPONSES P
I--
z
FUNCTIONS
TRANSFER a
DISTRESS /
PERFORMANCE
FINAL
DESIGN
42
Solar Wind Air Thermal
Radiation Velocity T/ature Prepert
Transverse
i Pavement IMix Stiffness Multi-Layered
Wheel Position Structure Modulus Elastic Theory
I I
Stress, Strain ,
Phase Angle I
Dissipated
Energy
Estimation of Unit
Fatigue Damage
Accumulation of
Unit Fatigue Damage
Final ]
Fatigue Life
43
While empirical pavement design methods were useful in their
time, the dynamics of modern pavement design, including new
materials, changing load conditions, and the need for greater
reliability in the pavement design, have limited the utility of
empirical methods. Thus, the focus of this project is on the
development of mechanistic design methods. The primary
distinction between the empirical and mechanistic methods is the
use of traditional engineering analysis methods for estimating the
performance of the pavements. The key feature of the mechanistic
approach, as shown in the upper left portion of Figure 2-3, is the
use of structural models of the pavement to predict the response
of the pavement to traffic and environmental loading.
44
resembles a standardized test29 . Of particular concern is the
long-term interaction between traffic and environmental loads and
the characteristics of the materials. While the challenge is
great, the application of advanced mechanistic models and test
methods can be used to improve the state of the art in pavement
design and yield greater reliability in the analysis.
45
3. ANALYSIS MODELS
47
initial conditions. Nair describes the relationship of these
components as:
48
Methods for the analysis of the pavement due to load will be
presented first followed by a discussion of the models for
analyzing environmentally induced stresses. The analytical
solutions are the first load response models addressed including
layer theories for elastic and viscoelastic materials and plate
theory. This will be followed by a discussion of the finite
differences and finite element numerical analysis methods. Models
of environmentally induced stresses are then presented. The
theories discussed in this chapter are for predicting the primary
response (i.e , stresses, strains and displacements), of the
pavements. Con;epts for the analysis of the limiting response of
the pavements (e.g., cracking) are discussed in the following
chapter.
49
P
Interface 1
h2 , E2 , /.L2 A1 Bi
Interface 2
Interface n - 1
'Itt
50
5. no shear stresses at the surface; and
51
Lister and Jones 34 studied the effect of nonuniform,
noncircular loads and concluded that the net effect of these two
assumptions resulted in an error of less than two percent under a
whole range of realistic tire and load conditions. Gross
overloading of a tire results in an error of about seven percent.
Saraf et al. 35 concluded elastic layer theory overestimates the
tensile strain at the bottom of the asphalt layer compared to a
finite element analysis.
52
ELT analysis. Engineers that employ ELT for pavement analysis
rely on measuring the material properties under simulated field
conditions. Different test procedures will produce different
measures of the material properties. This will result in
different primary responses estimated by ELT and, therefore, when
the primary responses are used in the limiting response functions,
the prediction of distress may be in error due to the formulation
of the analysis problem.
53
are some problems near the pavement surface directly under the
load. With respect to ELSYM5 Ahlborn states,
The CHEV5L and ELSYM5 programs are widely available and are
in the public domain. The Federal Highway Administration
sponsored the modification of the program to operate on a
microcomputer with a full screen editor for inputting the data.
This program is available from the McTRANS 42 Center at the
University of Florida.
54
Zaniewski 46 compared the output of the three programs. As
shown in Table 3-1, the surface deflection and horizontal strain
at the bottom of the surface are equal for ELSYM5 and CHEV5L and
are similar to the BISAR output. There are considerable
differences in the computed strains at the top of the subgrade for
the three different programs. Parker et al. also found
significant differences in the computed deflections near the load
although the stresses and strains were not very different between
the BISAR and CHEV5L programs.
55
Q. W) I r'.Q ~ .
Vu 4 -7i J-4C .1 0 *
(n r w r' ~) Y. N-
I
-4-4
I I
P
41~
P.4
0 0 4 Az C
Cd) N -4 -4 000u~ O 1 E 2 ~ 3%A T
:3 0 0i ~ 0 0 C% 0 N VIA .7
M. '.I .4 61
'A 0
(n -4-
7,' ;- Go- 0 0% 0
o4
04 a o 0 0
n - n.. 4- VI N -
>4 - -443
>4 -4 C 04
-4 0 0 0 13 N - VI
Q rj
IV ?A r 0j
M-
F,- . -4tv w 0
E-4- 1) C67 V. .0 0 >7
>' F3
.11~~ z -w4-
oo UO -4
O 56 '0
Maxwell materials. In 1962, Dister and W.entman 0 used a
three-element model to characterize the hhavior of a
beam on a Winkler foundation and analyzed this for a
moving point load. Ishihara presented a solution in
1962 for a two-layer viscoolastic system in which he
assumed that the layers were characterized as
incompressible Voight and Maxwell models. However, he
examined the behavior only at zero and infinite times.
Kraft presented an analysis of the deflection of a
two-layered system in 1965 in which the layers were each
composed of three-element models, and the volumetric
.behavior was assumed to be elastic.
iD- EGI(e)t6i , j = 1, 2, 3,
where:
Dj = the creep compliance function
e = natural base
t = time interval
G1 = coefficient of the Dirichlet series
5, - exponent of the Dirichlet series
b/*
G i and 5i require experimental determination of the complex
modulus and the time temperature shift function of the asphalt
concrete. Rauhut et al. 55 demonstrated that rutting predictions
with the VESYS program developed for the Federal Highway
Administration were very sensitive to these parameters. There
have been several revisions to the VEoiS model to refine
capabilities. Hufferd and Lai 56 expanded the capabilities of the
original program to include "N-layers" and reformulated the rut
prediction procedure to reduce the complexity of the mater-Lal
characterization.
58
Khosla concluded the structural subsystem of VESYS IIIA
predicts pavement performance accurately and that the triaxial
compression test can be used for easuring the material
properties. In another paper presented at the same conference as
the Khosla paper, Beckedahl et al.5 8 criticized the ability of the
VESYS model, particularly its characterization of the material
properties. Beckedahl et al. prop)sed several improvements to the
model including the development of procedures for capturing
fluctuations in the material properties over the life of the
pavement.
59
Single degree of freedom models developed by Weiss6,65 use a
combination of masses, springs and dashpots to represent the
pavement, as shown in Figure 3-2. The spring represents the
stiffness of the pavement, the dashpot represents damping of the
pavement materials and the mass represents the weight of the
pavement. When a force is applied to the model, the mass deflects
and may oscillate before coming to rest. Oscillations are a
function of the relative magnitudes of the mass, spring stiffness,
and dashpot dampening coefficient. If there was no mass in the
system, there would be no vibration and, therefore, no dynamic
response and the problem could be analyzed with static models.
Since pavements have mass, there is an expectation of a difference
in the dynamic and static response of the pavement. Although the
single degree of freedom model considers the dynamics of pavement
response, it is limited to vertical loads and responses. Thus,
there is no consideration of the propagation of the response
laterally through the pavement. Deflections at points away from
the load cannot be modeled.
60
MASS
SPRINGI DASHPOT
FIGURE 3-2. SPRING-DASHPOT SINGLE DEGREE OF FREEDOM PAVEMENT
MODEL (MAMLOUK)
APPLIED FORCE
S2 ! R2
61
The single degree of freedom model considers the dynamics of
pavement response, it is limited to vertical loads and responses.
Thus, there is no consideration of the propagation of the response
laterally through the pavement. Deflections at points away from
the load cannot be modeled.
2
(A)uil1 + (E + W)u 1 ij +P& Ui = 0
where:
P= mass density
62
The Kausel and Peek program computes displacements. This
program was modified by Sebaaly 69 to include the calculation of
stresses and strains.
63
5. The load at the interior and at the corner of the
slab is distributed uniformly over a circular contact
area; for the corner loading the circumference of
this circular area is tangential to the edge of the
slab.
3. 5 NUMERICAL METHObS
There are two basic numerical techniques that can be applied
to the analysis of pavement structures: finite differences and
finite elements. The application of finite difference methods for
the analysis of rigid pavementz was dcveloped at the University of
Texas in the 1960s and several successful computer programs were
64
produced . However, advances in the finite element method (FEM),
along with the development of more powerful computers, have led
researchers to concentrate on the development and application of
the finite element method in preference to the finite difference
method. This fulfills the prediction of Nair who stated in 1971,
65
shown it is difficult to be an effective user of an FEM program
without understanding some of the basic concepts and mathematical
techniques employed by the method. In this vein, the Burnet's
description of the salient features of every FEM is presented
followed by a description of various applications to pavement
analysis.
66
governing equations. Algebraic equations are much easier to work
w~th than calculus equations. Derivation of the element equations
is a theoretical procedure performed by the analyst or program
developer. The element equations are algebraically identical for
all elements of the same type. Consequently, element equations
usually need to be derived for only one or two typical eiements,
not every element in the mesh. In addition, since the element
geometry is simple, derivation of the equations is usually
straightforward. The analytical effort for the entire problem has
been reduced to deriving a few algebraic equations for usually
only one or two small elements.
67
Much of the work described above needs to be performed only
once, when developing the computer program. Application of an
existing program requires supplying specific data on the
constituent constants, mesh generation commands, and output
specifications. Due to the popularity of FEM, there are many
commercially available programs that provide a wide variety of
capabilities.
68
I 200"
PRigid Plate (10" X 10") P/ 1500 b
*Granular Base-- -- - 20
Subgrade ----- -
180".
69
3. limits on the principal stresses in granular and
fine-grain soils so that they do not exceed the
Mohr-Coulomb theory of failure.
All the joints on the top surface are free to move. Since
the sides and approached end cross section are often within
the deflection bowl, the theory of linear elasticity is used
to designate boundary deflections. Provision is made for
lateral and longitudinal continuity for each layer with
plastic hystereses on the sides. The main conditions to be
satisfied for the passed end cross section is that the
residual longitudinal strain must equal zero whilst still
allowing flow; also the vertical and lateral strain must tend
not to vary at points that vary in the longitudinal
coordinate only.
70
Sargious3 and Wang, Sargious, and Cheung 8 presented a general
description of FEM for the analysis of concrete pavements. The
foundation for the slab is modeled as either an elastic continuum
or as a Winkler foundation. The differences in these
foundation models are shown in Figure 3-5. The Winkler
foundation, as also used in the Westergaard solution, consists of
a series of springs having a constant modulus of reaction, K. The
reaction of subgrade per unit area at any point is proportional tc
the vertical deflection at that point, but independent of the
vertical deflection at any other point. The elastic continuum
foundation is considered as an idealized half space. The
flexibility matrix for the foundation is obtained by determining
the deflections at all points for each location of a unit vertic7i
point load.
71
pm
177-7777 7 7 7 7 -Z 77-7 -
r P
(a) (b)
72
V
73
modeled by shear springs along the edges. The concrete is modeled
as linearly elastic and isotropic. Force vectors due to the
weight of the slab, thermal gradient and loads are applied
incrementally. The stiffness matrix is adjusted at the end on
each increment according to the new subgrade support conditions.
74
causes the development of stresses in the surface. When these
stresses exceed the strength of the material, transverse cracks
develop. This basic mechanism is the reason conventional concrete
pavements are constructed with joints. The need for design models
for the selection of joint spacings in concrete pavements has led
to the development of models of the environmental stresses in
concrete slabs. Although the mechanism also affects flexible
pavements, there has been relatively little research into
quantifying the effect.
cc = WLf/(24h)
where:
Uc = "friction stress"
L = length of slab, ft
75
L/2
I IA
00
.Friction
(a)
Fully mobilied
resistance f
!= 1000
L/2 -
(b)
76
crack spacing as a function of temperature drop, dryinv-, shrinkage,
moisture change, wheel load, and coefficient of subgrade
resistance.
edge stress
a = CEat/2
interior stress
where:
L = Po;-son's ratio
E = elastic modulus
77
4. DISTRESS MODELS
4.1 FRACTURE
Fatigue is generally considered to be the fracture of the
pavement surface due to the repeated application of traffic loads.
Both asphalt and concrete pavements are subject to fatigue
failure. Two basic approaches have been taken to the modeling of
fatigue behavior: phenomenological and power law. In the
phenomenological approacn, the number of applications a pavement
can carry prior to failure is estimated directly as a function o!
the stress or strain levels generated by the traffic loadings.
The power law approach uses concepts developed in fracture
mechanics to estimate the growth of a crack through the pavement
layer.
79
The phenomenological approach has been widely appiied in the
analysis of pavement fatigue life. Although both asphalt and
concrete pavements will fail in fatigue, the behavior of these
materials with respect to repeated loads is very different in that
concrete appears to have a "fatigue limit." It is generally
assumed that concrete will not fail in fatigue if the stresses in
the slab are kept below 50% of the modulus of rupture, as measured
with the flexural test.
N = a(l/E)b
where:
8o
estimates are expressed in terms of a shift factor. This shift
factor is frequently in the range of 20 to 25, but factors as
large as 1000 have been reported.
N = 259(1/6)3.16(l/E) 1.4
where:
' 5
N = 479(1/E) (I/E) -
81
This equation was developed from an analysis of experimental
pavement sections designed and tested to simulate airfield
pavements 98
While the WES and the Kelly-Thompson equations have the same
form, the coefficients are considerably different. For a strain
of 0.0005 in/in and a modulus of 500,000 psi, the WES equation
estimates approximately 10,000 repetitions can be applied to the
pavement while the Kelly-Thomas equation estimates in excess of
73,000 applications. This demonstrates the problem with the
phenomological approach to fatigue analysis. Prediction models
developed by different researchers appear to be more a function of
the analysis procedure used in the development than determined
from the basic properties of the materials.
n
dc/dN = Ak
where:
82
Determination of the stress intensity factor requires analysis of
the discontinuity of the pavement material in the area of the
crack. Currently, this requires the application of FEM to model
both the material behavior and the stress intensity.
4.2 DEFORMATIONS
Plastic and viscous deformations of the pavement materials
result in permanent deformations of the surface. Channelized
traffic generates an accumulation of deformations in the wheel
83
paths of the vehicle or rutting. Three approaches have been
defined for relating the primary response of pavements to rutting:
1 08
The limiting strain approach was first presented by Dorman
in 1962 and Klomp and Dorman1 0 in 1964. It is currently used in
the Asphalt Institute airfield pavement design method110 and the
Joint Department of Army and Air Force elastic layer theory method
for the design of flexible pavements. The basic hypothesis of
this approach is that if the maximum compressive strain at the
surface of the subgrade is less than a critical value, then
excessive rutting will not occur for a specified number of
repetitions. These relationships were developed based on analysis
of the Corps of Engineers pavement design procedures.
The Army and Air Force elastic layer theory procedure for the
design of flexible pavements uses limiting subgrade criteria for
estimating the number of applications a pavement can withstand
before excessive permanent deformation occurs. The criteria are
specified by the equation:
N = 10,000(A/Ss)
where:
84
Relating permanent deformation to elastic stresses has been
proposed by several researchers. Some have assumed a fundamental
deformation jaw exists through which permanent strains can be
predicted based on the stress state of the material. Other
authors have used statistically formulated equations for relating
the permanent and elastic strains. Neither method has been
particularly successful since the concept, from a mechanistic
aspect, is not fundamentally sound and there has not been an
adequate database for developing the statistical models.
85
5. AIRPORT PAVEMENT DESIGN METHODS
where:
88
3. The number of equivalent departures is summed to determine
the total number of departures for use in design.
T = f\ZBcBR, W NE)
TBmin = f(SGBR,TP)
Ts = max(TrinTsb)
TSB =Tp - T s - TB 6
90
5.2 ASPHALT INSTITUTE
The pavement design charts for the Asphalt Institute110 design
procedure for full-depth asphalt concrete pavements were developed
based on elastic layer theory analysis and two distress types,
fatigue and permanent deformation. The fatigue criteria are based
on limiting the horizontal tensile strain at the bottom of the
surface layer. The permanent deformation criteria is based on
limiting the vertical compressive stress at the top of the
subgrade. All aircraft classes are converted to the "standard
aircraft"; the DC-8-63F was "arbitrarily" selected to be the
standard. Taxiways are considered to be the critical portion of
the airfield pavement. The design process is summarized in Figure
5-1.
The inputs to the design process are the mean annual air
temperature, the design subgrade modulus, and the expected number
of repetitions of each aircraft type. The design subgrade modulus
requires performing a series of soils tests and selecting the 85
percentile value, e.g., only 15% of the subgrade tests have a
lower value than the design value. The manual recommends
laboratory evaluation of the modulus using the triaxial test
procedure specified in the manual. Approximations are available
for estimating the modulus from the CBR test, plate bearing test
and the FAA Soil Classification. The manual provides
specifications for the asphalt concrete but the properties of the
asphalt concrete are not a direct input to the design process.
91
EE
LL.)
CNL-
4) =
06 H
2 W2
- uJi
.= 51 C
0 1A c H
C#2=
W=Z.
co U.2
(U C.5 TO)
Eo m31 W E-4
, %- - L ----
-cm
921
The traffic equivalency curves are then used to convert the
number of aircraft loadings into equivalent loadings.
Equivalencies are determined for four wheel paths and four
different pavement thicknesses. For each assumed thickness, the
critical wheel path is determined as the one with the greatest
number of total equivalencies. The number of equivalencies and
corresponding thicknesses are plotted on the design graph to
establish the predicted traffic curve.
93
the design modulus of rupture of the concrete to the
working stress that will be used for design. The safety
factor ranges from 1.7 to 2.0 for the critical areas,
and 1.4 to 1.7 for noncritical areas. The critical
areas are aprons, taxiways, hard stands, runway ends and
hangar floors. The noncritical areas are the central
portion of the runway and some high-speed exit taxiways.
The selection of a specific safety factor for an
aircraft depends on the expected number of loadings.
The PCA states that the fatigue method of design applies to:
94
The basic st'eps in the fatigue analysis method are:
95
The PCA manual addresses the design of continuously
reinforced concrete pavements, but recommends that the thickness
design be the same as for plain concrete pavements. This manual
states that reducing the thickness can increase deflection and
promote spalling of the joints.
96
WARM-UP APRON TYPE B
LEGEND
WASHRACK
APRON,
97
LEGEND_
LADDER TAXIWAY /
WASH RACK
98
LEGEND
______TYPEABTRAFFICAREA
TYPEPDDTRAFFIC AREA
Ur
- LDDE TAXIWAY
~U E. -~.~1WASH ACK
zz
~~t-~e r-
~~~ tr4r--C.~ m-o
0ot~
0~0
< - C-',
C <
E-4 -a z w -M
m -n
Cl l C Cl6
U) () Qn ) k o
U) 0~ cz 1)CI Zq
0T 0
oc 0 '0T : 0 r00
zn en It0 IT0 000 I
U ClO 0 Cl~~c~Cl~~e~"~ ~z
U) C
HC
kn kn
Hn OCtm -
- U
I~ :D
100
Design of full-depth bituminous concrete pavements is beyond the
scope of this report. The design sequence is:
10]
5.4.2.1 Plain Concrete Airfield Pavements - The edge stresses are
reduced by 25% to account for load transfer afforded by the joint
designs required. The flexural modulus of elasticity of the
concrete is assumed to be 4,000,000 psi and Poisson's ratio is
assumed to be 0.15. The design curves are available for light,
medium, heavy and modified-heavy load pavements. Select pavement
thickness is a function of the flexural strength of the concrete,
modulus of subgrade reaction, and type of traffic area. The
design curve for short-field pavements uses the gross weight of
the aircraft and number of aircraft passes instead of the type of
traffic area. The design thickness is rounded to the nearest half
inch.
where:
102
There is an interdependence between the amount of steel required,
the size of the slab and the thickness, so designing the pavement
requires assuming either a percent steel or reduced slab thickness
and solving for the other value.
Design curves are presented for each of the classes and types
of Army and Air Force airfields. The Army design curves consider:
103
5.4.2.4 Prestressed Pavement Design - The design of prestressed
concrete pavements requires balancing the level of prestressing
with the thickness of the slab to obtain an economical design.
The design equation is:
ds= (6PNB/(w(hp)P)) - R + r. +
where:
104
for other gear types. The design curves are limited to modulus of
subgrade reactions of 100 and 500 pci and interpolation is used
for other k values.
105
SITUMIROUSNRET AR OT IONETTOOLSFO
PARAMETERS. SRAFAIC %
LIHITIN
STRACOMPTERE
ETRI
lAECUREMDUU EIN DLETERMINLE ~
RUTODLUORS INGfRMESIALN
UETR EI.LRCE NOCTS HIKEST
MODULU
SURRAS BONMAT
-a ROUITEDCKCNEES
I
START5 NEUS
USDNoH EISO 05TP PPVMN
TIF DRIINED
E0 ONE AYeST
STRAINS CONST
FR0SABIIZE BASESIMM
cONIDRAIO
EG N AF
FO SEEC Rlil106NA
STARTWITH:
BITUMINOUS CONCRETEMODULUS
FOREACHMONTH BASE COURSE DETERMINED
MODULUS INFLEXURE
EIf
SUBGRADE MODULUS FOR ANDTHAW
NORMAL COURSE
PERIODSBASE MODULUS FROM
DETERMINED EQUIVALENTCRACKED EBc
SECTION
TRAFFIC PARAMETERS COURSE
SUBBASE MODULUS INFLEXURE
DETERMINED ES
LIMITING STRAIN
CRITERIA SUBBASE
COURSE DETERMINED
MODULUS FROMEQUIVALENTCRACKED
SECTION
Ek
ESTIMATED INITIAL
THdICKNESS
INCEASE
THI
KNESS
I OPTs h sN
107KES T
As demonstrated with the design procedure flow charts, many
steps are required for designing a flexible pavement with this
procedure. However, the procedure can be summarized in five
steps.
108
Strains in the pavement structure due to the design aircraft
wheel loading are computed with an elastic layer theory program.
These strains are input to the criteria equations for fatigue and
subgrade strain to determine the allowable number of strain
repetitions.
109
6. DISCUSSION AND RECOMMENDATIONS
111
Generally, maintenance can be divided into two
categories: preventive and responsive. Since preventive
maintenance is performed to reduce the occurrence of
distress, it is not an indicator of pavement failure.
Responsive maintenance, on the other hand, corrects distress
conditions. Taken to an extreme, responsive maintenance can
become cost prohibitive and rehabilitation or reconstruction
may be required. If airport pavement maintenance is used as
a failure criteria and if maintenance of the pavement is
performed to correct the occurrence of distresses, then the
prediction of pavement life is directly related to the
prediction of pavement distress.
112
relationships developed primarily from laboratory and highway
experience with limited calibration from airport pavement
performance data. Thus, the state of the art of airport
pavement design methodology falls far short of a true
mechanistic analysis of pavement behavior and performance.
113
conditions, pavement geometry and the interaction of these
factors.
114
of preservation and restoration of existing pavements. In
many ways, pavement preservation and restoration is more
difficult than the design of new pavements. For the design
of pavement preservation and restoration, the structural
capacity and condition of the existing pavements must be
evaluated. Deflection testing is the current practice fet
evaluating existing pavements. However, analysis of
deflection data suffers from the same variances between
pavement behavior and mechanistic models as is encountered
for the design of new pavements (e.g., the discrepancies
between the true behavior of the materials and the
assumptions used in the analytical models, and the effects of
environmental conditions on the behavior of the pavement).
115
all pavement materials under all types of traffic loadings
and environmental conditions. Even the most precise models
fail to recognize the full range of behavior with respect to
pavement conditions, material properties and environmental
conditions.
116
environmental conditions of the stress state of the materials
and their properties. While three-dimensional finite element
analysis is firmly entrenched in other areas of engineering,
due to its complexity this technology has not been extended
to the analysis of pavements. The intricacy of pavement
materials, loading conditions and environmental effects makes
pavements among the most complex structures designed and
analyzed by engineers.
117
In addition, the existing shortage of quality virgin
materials will lead to a greater emphasis being placed on the
use of new materials, recycling and the use of marginal
materials. The existing models and laboratory procedures are
not adequate to determine the future response of new
materials. This is a critical area demanding the development
of the unified pavement theory. While field verification of
material performance is always desirable, it can also impede
the development and introduction of new materials. Normally,
a pavement should perform for 20 years. Thus, it would take
a minimum of 20 years to prove the value of a new material.
A unified theory for the analysis of airport pavements,
supported by appropriate laboratory testing, can provide a
vehicle for the proof and acceptance of new materials in a
timely manner.
118
LIST OF SYMBOLS
119
Smix(tT) asphalt mixture stiffness at time, t, and
temperature, T
Ss vertical compressive strain at top of
subgrade
SGcBR subgrade California Bearing Ratio (CBR)
t time, (as subscript) time, tangential or
transverse
TpTB, Ts,TsB thickness of, respectively, pavement,
base, surface, and subbase. Total
pavement thickness Tp = T s + T B + TSB
Ai ith component of the displacement vector
in Cartesian coordinates
w ratio of multiple wheel gear load to
single wheel gear load
wi wheel load of the ith whee
W, W D weight, weight of design aircraft
a coefficient of thermal expansion
6i exponent of the Dirichlet series
6 strain
Ce longitudinal strain
6o initial and constant strains
E(t) strain as a function of time
£ Lamds constant
IPoisson's ratio, Lames constant
phase difference between stress and
strain
mass density
a stress
cc "frictional" stress
o hydrostatic stress
a3 transverse principal stress
T shear stress
(angular frequency
120
REFERENCES
10. Monismith, C., Epps, J.A., and Finn, F.N. (1985) "Improved
Asphalt Concrete Mix Design," American Association of
Asphalt Paving Technologist, Proceedings, Volume 54
R-1
14. Hicks, R.G., (19 Factot nfluencing the Resilient
Properties of Grai.ular Mat <iais, Dissertation Series,
University of California, Berkeley
16. Saraf, C., Marshek, K., Chen, H., and Hudson, W.R., (1987)
"The Effect of Truck Tire Contact Pressure Distribution of
the Design of Flexible Pavenents," Proceedings, rhe Sixth
International Conference Structural Design of Asphalt
Pavements, Ann Arbor, Michigan
20. Wignot, J.E., Durup, P.C., Wittlin, G., Scott, R.B., and
Gamom, M.A. (1970) "Aircraft Dynamic Wheel Load Effects on
Airport Pavements," Federal Aviation Administration, Report
No. FAA-RD-70-19, Washington, DC
R-2
25. Himeno, K., Watanabe, T., and Maruyama, T. (1987)
"Estimation of Fatigue Life of Asphalt Pavement," p. 272,
Proceedings, The Sixth International Conference Structural
Design of Asphalt Pavements, Ann Arbor, Michigan
35. Saraf, C., Marshek, K., Chen, H., and Hudson, W.R., (1987)
"The Effect of Truck Tire Contact Pressure Distribution of
the Design of Flexible Pavements," Proceedings, The Sixth
International Conference Structural Design of Asphalt
Pavements, Ann Arbor, Michigan
36. Ahlvin, R.G., Chou, Y.T., and Hutchinson, R.L. (1973), "The
Principle of Superposition in Pavement Analysis," Highway
Research Board, Highway Research Record 466, p. 153,
Washington, DC
R-3
37. McCullough, B.F. (1969) A Pavement Overlay Design System
Considering Wheel Loads, Temperature Changes and
Performance, Doctoral Dissertation, University of
California, Berkeley
44. Parker, F., Barker, W.R., Gumkel, R.C., and Odom E.C.
(1979) "Development of a Structural Design Procedure for
Rigid Airport Pavements," Federal Aviation Administration
Report No. FAA-RD-77-81, Washington, DC
R-4
49. Pister, K.S. (1961), "Viscoelastic Plates on Viscoelastie
Foundations," Journal of the Engineering Mechanics
Division, ASCE, p. 43
53. Hopkins, I.L. and Hamming, R.W. (1957), "On Creep and
Relaxation," p. 906, Journal of Applied Physics, Vol. 28
R-5
60. Bastiani, A. (1962), "The Explicit Solution Cf the
Eq]atioiis of the Elastic Deformdtion= for d Stratified Road
Under Given Stresses in the Dynamic Case," p. 351,
Proceedings, International Conference on the Structural
Design of Asphalt Pavements, Ann Arbor, Michigan
R-6
72. Pickett, G., and Ray, G.K. (1951), "Influence Charts for
Concrete Pavements," Transactions, ASCE
75. Stelzer, C.F., and Hudson, W.R. (1967), "A Direct Computer
Solution for Plates and Pavement Slabs," Research Report
569, Center for Highway Research, University of Texas,
Austin
82. Smith, R.B., and Yandell, W.O. (1987), "Predicted and Field
Performance of a Then Full-Depth Asphalt Pavement Placed
Over a Weak Subgrade," p. 443, Proceedings, Sixth
International Conference on the Structural Design of
Asphalt Concrete Pavements, Ann Arbor, Michigan
R-7
85. Chou, Y.T., (1981) "Structural Analysis Computer Programs
for Rigid Multicomponent Pavement Structures with
Discontinuities, WESLIQID and WESLAYER," U.S. Army
Engincering Waterways Experiment Station, Technical Report,
GL-81-6, Vicksburg, Mississippi
88. Tia, M., Arnaghani, J.M., Wu, C.L., Lei, S., and Toye, K.L.
(1987), "FEACONS III Computer Program for an Analysis of
Jointed Concrete Pavements," p. 12, Transportation Research
Board, Transportation Research Record 1136, Washington, DC
R -8
96. "The AA:YK Read Test. Report 5 - Pavement Researt;<,'
Highway Research Board, Special Report 61E, 1962
R-10