j} TJ c

- -- . ~

-
KAPL-3158
AEC Researclt and
Development Report

KNOLLS
ATOMIC POWER
LABORATORY

\
Fatigue Design Data
-
. for the
--
Titanium Alloy, Ti-6Al .:6 V-2Sn

D.F. Mowbray

May 20, 1966

Operated for the
United States Atomic
--
Energy Commission by

GENERALfJ ELECTRIC -
-
PATENT REVIEW APPROV
GOVERNING PATENT REV~SWRELEASE. PROCF.DURES
FILE IN RECEIVING SECTIQ['(. AND R~l,,I~AS~ A~ QN
 DISCLAIMER

This report was prepared as an account of work sponsored by an

agency of the United States Government. Neither the United States
Government nor any agency Thereof, nor any of their employees,
makes any warranty, express or implied, or assumes any legal
liability or responsibility for the accuracy, completeness, or
usefulness of any information, apparatus, product, or process
disclosed, or represents that its use would not infringe privately
owned rights. Reference herein to any specific commercial product,
process, or service by trade name, trademark, manufacturer, or
otherwise does not necessarily constitute or imply its endorsement,
recommendation, or favoring by the United States Government or any
agency thereof. The views and opinions of authors expressed herein
do not necessarily state or reflect those of the United States
Government or any agency thereof.
 DISCLAIMER

Portions of this document may be illegible in

electronic image products. Images are produced
from the best available original document.
 H.c..·$~,; MN ,
UNCLASSIFIED - _>'0

KAPL-3158
UC-25, Metals, Ceramics, and
· Materiais ··
(TID-4500, 49~h Edition)

._ .
'··
FATIGUE ·DESIGN DATA FOR THE' TITANIUM A.J.;UJY, Ti-6Al-6V-2Sn

D. F. Mowbray

May 20, 1966

\ __I_:_E_:_E_:_:_.:_.:_R_F_:R_c_I ~_ :_,
:_:_U_:"":"':_:_:_:_:_:_r_s_\

r;;=;~-:.:::~::.~M;~~~ ;1~~~~,- ··~· ~ ,.~

A. Makes any warranty or representation, expredsed or tm I c ommioalon:
racy, completeness, or usefulness of the infor U 1 pi ed, with respect to the accu-
of nny information apparatus method mn o~ contained In this report, or that the use
privately owned rl~hts· or • . • or process ~Bclosed in thJs report may not infringe
B. Assumes any tiabillttes wtth respect tO the u~e of 0 f, •
U90 Qf P.riy lnfflr-!!l.AU-:"n, 3Pf'(\!'P.h1i:t mnthM nr rrnn~g!g> tU I r ~rt da~ges resulting from the
I

As used in the above " . ' .. -~· ··• . . :- . _lln_n"~- n t hi _rQ(V'rt,

ployee or contraCtor of th~ c:;~~:s~~:'!::!:!::1:~fth:uc~ : ~!::!on" includes any em-
0
such employee or contractor of the Commiesio ~
0
tor' to the extent that
disseminates, or provides access to, any lnfor:~u:t employee of such contractor prepares,
With the Commission, or his employment with such cP:::~:~ to hJs employment or contract

••
~uthorized Classifier
.1,M o?~ Date
/f le?
General Electric Company
KNOLLS ATOMIC POWER LABORATORY
Schenectady, New York
Operated f.or the
United States Atomic Energy Commission
Contract No. W-31-109 Eng-52

UNCLASSIFIED
 UNClASSIFIED ·

t.EGAt NOTICE

This report was prepared as an account of Government-

sponsored work. Neither the United States, nor the Commission,
.nor any person acting on behalf of the Commission:
A. Makes any warranty or representation, expressed or
implied, with respect to the accuracy, completeness,
or usefulness of the information contained in this·
report, or that the use of any information, apparatus,
method, or process disclosed in this report may not ·
infringe privately owned rights; or
B. Assumes aey liabilities with respect to the use of, or .
tor damages resulting fran the use of any infonnation,
. Apparatus, method, or process diaulosed in this report.
As used in the above, "person acting on behalf of the
Commission" includes an.v employee or nnnt.rMtor of the Commiaaion,
or emplqyee of such contractor, to the extent that such empl~~
or contractor of the Commission, or employee of such contractor
prepares, disseminates, or provides access to, any infor.mation
pursuant to his emplo.yment or contract with the Commission, or
his emplo,y.ment with such contractor.

Printed in USA. Price $2.00. Available f~om the

Cl~aringhouse for Federal Scienti'fic and Techni.cal
Information, National Bureau of Standards, U. s. •
Department of Commerce, Springfield, Virginia 22151.

UNCLASSIFIED KA.PL-.3158
ii
 UNCLASSIFIED

KAPL-.3158
UC-25, Metals, Ceramics, and
Materials .
(TID-4500 1 49th Edition)
(Nonstandard)

DISTRIBUTIO~

No. of popies
AEC 1 SNR
Cramer, CC J
Baldwin, EE 2
Div:J.sion of Ter.hni t>,Al Tnfnrmet.i.nn F.xt.P.md.nn J
· Document Library 4.'
Jekkals 1 I lr
Ferril_, D 1.
G.lasser, TH 1
Kelleman 1 RW 1
Miles, A 1
McCalley, RB 1
Mehringer~ FJ 1
Miller, DR 1
Mowbray 1 DF . . 5
Naval Reactors Library 2
Shriver,. PR. 1
Sokol, GJ '1
Technical ~~blications/JG Shaw 1 ·.-- .
TIG File/CJ Schmidt 5
Total
-·
J5

UNCLASSIFIED
iii
 'i

THIS PAGE
WAS INTENTIONALLY
LEFT BLANK
 CONTENTS

~
ABS'rRACT • ix
NOONQLATURE . xi
INTRODUCTION 1
TEST DESCRIPI'ION · • 1
Material • .. 1
Sped.mens • . ~ 2
Test Equipment and ProcedUre 5
· TEST .RESULTS . .. 6
..DISCUSSION OF TEST RESULTS • 10
. Unnotched Fatigue Strength • 10
Effect or Mean Stress on Unnotched Fatigue Strength • 14
Notch-Sensitivity • 17
SUMMARY AND.CONC~USIONS. .. 18
REFERENCES 19
APPENDIX .A. MAXIMUM EFFECT OF MEAN STRESS BASED ON A PARABOLIC
DAMAGE LAW •. 21

APPENDIX B, EQTJATIONS FOR USE WITH ·THE 8-CONCEPT • 23

v KAPL-.Jl5G
 THIS PAGE
WAS INTENTIONALLY
LEFT BLANK .
 ILLUSTRATIONS

No.
1·
Title
(0002) Pole Figure for the ~Phase of Rolled 1.0-in.-diam.
---
Page

Bar (KS-62295, Unclassified) • • .•

2 Unnotched Cylindrical Fatigue Specimen (KS-62296,
Unclassified) •
3 Notched Cylindrical Fatigue Specimens (KS-62297,
Unclassified) • 4
4 Monotonic and Cyclic Stress-sttain Respou~e·at noom
Temperature and at 600F (KS-62298, Unclassified) 7
5 . Unnotched Fatigue Data (KS-62299, Unclassified) 8
6 Notched Fatigue Data (KS-62300, Unclassified) 12
7 Variation of Notch-sensitivity Factor (q) with Notc.h-Root
Radius (R) (KS-62301 1 Unclassified) · ·• •
8 Comparison·of TitaniumStrain-Cycled Fatigue Strength with
Other Materials (KS-62302, Unclassified) • 14
9 ·Fatigue Damage Laws for Considering MeS:nStress (KS-6230),
Unclassified) 16
. ,-

vii KAPL-3158'
 'j

THIS PAGE
WAS INTENTIONALLY
LEFT BLANK
 ABSTRACT

An investigation was conducted to determine some of the basic

fatigue properties of the titanium base alloy (Ti-6Al~6V-2Sn) in
the annealed bar stock torm. The basic S-N curves in.the cyclic·
. 2 6 .
lifo .region from approximate~ 10- to 10 oyoloo, tho offoot of
mean stress at lives from 10 4 to 106 _ cycles' and the effect of
notches on fatigue strength at 106 cycles w:ere e~tab:J..i~l)~d at
room temperature and at 600F. ·
The test results indicate that thi.s material possesses goqd
unnotched fatigue strength, but has a rather high sensitivity
to ·mean stress and notches •. These latter re.sult_s are as ex ...
pected because of the material '.s high ratio of yield to :ultimate
tensile strength.
The test results were analyzed in terms of current de~ign
methods. It was found that the effect of mean stress on the
unnqtched fatigue strength is best described by a parabolic
. damage law. Values of 8 are established for estimating fatigue
. notch factors for members containing cr&ck-like defects.

ix
 THIS PAGE
WAS INTENTIONALLY
LEFT BLANK
 NOMENClATURE

A Minimum cross-sectional area of fatigue specimens, in~ 2

8 · Distance from surface of notch root (see ·Reference 10), in.
D Minimum: section diameter of notched fatigue specimens, in.
Cyclic st~ain range in'unnotched fatigue tests, %
Young's modulus, psi
Poi$son's ratio (O.J is asslimed)
Ratio of local tensile stress to nominal net-section stress (a/an)
Fatigue-notch factor for axial load* (defined as the ratio.of
fatigue strength of unnotched specimen to fatigue strength.-of
notched speciulen at a given number of cycles to failure).·
Theoretical elastic stress concentration·factor for normal stress
(am8JC/an) .
.·Kt .. C.ombined theoretical elastic stress concentration factor taking
··.into account normal stress concentration factor and Mises
··.criterion ·of failure (see Reference. 2)
Fatigue":"'notch factor predicted by the 8-concept.

Theoretical elastic stress concentration·factor at a distance 8

from the notch-root surface for the normal stress in a notched
cylindrical specimen
Theoretical elastic stress concentration factor at a distance 8
from the notch-root surface for the tangential stress in a notched
cylindrical specimen
Kf - 1
q Fatigue-n<?tch sensitivity faator,
Kt, -.1

R Notch-root radius, in.

a weal tensile stress, psi
a Maximum elastically calculated tensile stress at notch root, ps;l.
max
an Average net-$eation stress at the minimum section of a notched bar,
psi

*Used in t.his :r;-eport only for axial load.

xi KAPL-3158
 CompletelY reversed (zero mean stress) stress ~plitude* that pro-
duces failure in a given number or _cycles (E x €a/2) 1 psi
s'a Stress amplitude* that, combined with the maximum possible mean
stress, produces failure in the same number of' cycles as a com-
pletely reversed stress amPlitude Sa, psi
Limit or elastic behavior, psi
.Mean stress, Sb·- S~, psi
Su Ultimate tensile strength, psi
T ~otch depth, in.
y Coordinate (se.e F:igure B.·l), in.

*Fictitious stress.
xii KAPL-.3158
 ..-

FATIGUE DESIGN DATA FOR THE TITANIUM AIJJ)Y, Ti-6Al-6V-2Sn

D. F. Mowbray

INTRODUCTION
The application of high-strength titanilDil alloy, ~i-6Al-6V-2Sn, is
being increased in the nuclear industry where fatigue loadings are com-
monly encount~red in the low- and intermediate-cyclic-life regions over
the ;ra:nge of .temperature from 75 to 600F.. T9 assist in the design of
components fabricated from bar stock of this alloy, an investigation was
•Jndertaken to (letermine sam~ of the baai.c fatigue properties' at the
cyclic lives and temperatures of interest. Displacement-controlled tes-ts
were conducted on unnotched specimens to establish fatigue c'urves and
the.so called "maxim\Dil effect of mean stress" over the cyclic life region
of -10 2 to 106 .cycles;, load-controlled tests were conducted on notched
spec~ens to determine fatigue notch factors for several different
geometries at 106 .cycles. An equal number of tests were conducted at
both 75 and 600F.

TEST DESCRIPTION
Material
The material used in this investigation was the titanium base alloy
designated.Ti:-6Al-6V-2Sn.l. This titanium alloy is double-vacuum...melted
by consumable electrode·techniques. The material was obtained as 1.0-in.-
diameter hot-rolled bar stock in the ful~y annealed condition. The anneal-
ing treatment consisted of heating in vacuum at 1300F for 2-1/4 hr,
followed by an air cool.
· ·The chemical composition was as follows:

Element wt %
Aluminum 5.6
Va.nadium 5.5
Tin 2.0
Iron o. 76
Copper 0.71
Oxygen 0.155
Carbon 0.023
Nitrogen 0.018
Hydrogen. 0.0053
Titanium Remainder
1 KAPL-3158
2

The conventional tensile properties (in the rolling direction) at roam.

temperature and at 600F are given in Table 1. Note the high-yield-to-
ultimate-tensile-strength ratio exhibited by this material.

TABLE 1. CONVENI'IONAL TENSILE PROPERTIES OF TEST MATERIAL*

UTS, 0.2% YS, 0.2% YS Elongation

Temperature, F psi psi · UTS in 2 in., % RA 1 %
75 159,000 155,000 0.98 18.5 J5.0
75 161,100 151,000 0.94 18.5 J6.2
600 127,000 105,200 O.SJ 19-5 50.0
600 128,000 104,500 0.82 18.; 50.8

*The tensile and fatigue specimens were vacuum-annealed for 2 hr

at lJOOF following machining.

UTS Ultimate tensile strength

YS Yield strength
RA Reduction in area

In light of the fact that preferred orientations are known to

develop in rolled titanium products, a (0002) pole figure was determined
for the a-phase (HCP structure) of the material used in this investigation.
The pole figure, shown in Figure 1 1 indicates a very definite,preferred
orientation with a preponderance of basal planes perpendicular to the
rolling (and loading) axis.
Specimens
Axial specimens with their longitudinal axes in the rolling direction
were used for both the ·unnotched and the notched tests. The· unnotched
specimen,. shown in Figure 2 1 had a 0.160-in. diameter and was subjected
to a constant-displacement cycle. Notched specimens i~corporating 60-deg
V-grooves were tested with two minimum-section diameters: 0.160 and
0.480 in. These specimens are shown in A and B1 respectively of Figure J.
The 0.160-in·. minimum-section-diameter specimens were tested with a single
notch-root radius (0.001 in.). However, the 0.480-in. minimum-section
diameter specimens were tested with notch~oot radii of 0.0015, 0.005,
and 0.015-in. It was originally intended t~at the larger diameter speci-
men be tested with a hotch-root radius of 0.001-in. the same as the
smaller diameter specimens, but a 0.0015-in.-notch-root radius was. the
smallest that could be achieved. All the notched specimens were 'subjected
to a constant-load cycle.
KAPL-3158
 3

BAR AXIS

_I_
I

NOTE: AREA DESIGNATED I

CONTAINS SPORADIC
AREAS OF 1/2 AND 2

FIGURE 1. (0002) Pole Figure for the a-Phase

of Rolled, 1.0-in.-diam Bar.'
KS-62295 Unclassified

i ln.- 20NF THREADS

t----2.900---~

DIMENSIONS IN INCHES

FIGURE 2. Unnotched Cylindrical

.... •Fatigue Specimen. ·
KS-62296 Unclassified

KAPL-3158
4

t-------- 4t ------..-1

0235.
•.0233

~I::=::.=:::ttJ,._-Ju
0-001 in. ROOT
radius

A. Small Diameter.

~
~~------~-------- 6-------------~

~-----~
-------~----fl~tln_l~_. . ,: -~-
-=-=-=-
F"==""=-- - -

I· in-12 UNF-2A
THREADS

NOTCH DETAIL

B. Large Diameter.

(All Dimensions in Inches)

FIGURE 3. Notched Cylindrical Fatigue Specimens.

KS-62297 Unclassified

KAPir3158
,,,
 All specimens were machined from the fulzy annealed bar stock. The
machining was performed carefully to prevent _excessive residual stresses
on the specimen surface. After machining, the test-section surfaces of
the unnotched specimens were polished with 600-grit paper in the longi-
tudinal direction. The grooves in the notched specimens were finish-
machined by a lapping operation which minimized the presence of machining
scratches transverse to the test,direction. Following final.machiniog,
the specimens were given an additional 2-hr vacuum-anneal at.l.)OOF (vacuum
was maintained·at 10-5 mm Hg).
Test Equipment and Procedure
The test work in this investigation was conducted by Lesseils and
Associates~ Inc.~ Waltham, Mass. The description of the ·test equipment
and proced~e to follow is abstracted from their data report.
The displacement-controlled tests were conducted in a test machine
which has a rigid frame containing the specimen grips, the lower grip
being driven by a rotating eccentric which m~ be set for any desired
displacement amplitude. The upper specimen grip consists of a cylindrical
rod threaded at its lower end to accept the specimen. It is free to slide
vertically within a hole in the upper frame, except when clamped by screws
on the split, upper crossbar of the frame. The lower grip slides in the
lower .crossbar of the frame, being guided by a very close fit with a
cylindrical impregnated bearing which is pressed into the cross member.
Attached to the bottom of the lower grip is a flex plate, below which is
a push bar instrumented with strain gages for load measurement. The
machine cycling rate is. variable. from 18 to 600 cpm. A cycling rate
of 120 cpm was used for the tests at the lower strain amplitudes and a
rate of 18 cpm at the higher strain amplitudes.
· Specimen strain was me<:u:~w•ed by one of several methods. . In tests
conducted at room temperature at strains <1.9% range., strain. gages were
attached directly to the specimen test sections. Values of strain re-
ported for these tests were the first-cycle strain-gage readings (i.e.,
no variation of strain range ·was noted during the tests). In four tests
conducted at each of the two test temperatures at strains ?4·0% range,
strain was measured by means of a diameter transducer gage, incorporating
an LVDT. Values of strain reported for these tests are transducer
readings at shakedown (the first cycle at room temperature and approxi-
mately the lOOth cycle at 600F). This diameter-transducer gage was also
·used to measure strain in the remaining 600F tests (<1. 5% range), but
the gage was not considered by the author to yield sufficiently accurate
measures of the strain. Consequently, the 600F strains reported at <1.5%
range were determined by dividing the P/A stress by the elastic modulus.
In all of these tests, the material appeared to remain elastic.
6

Signals from the load transducer, specimen strain gages, and diameter
tr~nsducer were amplified and record.edon a Honeywell Visicorder. Records
w.ere made during the first twenty cycles and periodically thereafter.
A small split-shell resistance furnace with appropriate thermocouples
and an indicating pyromet~r-controller were used for the elevated-
temperature tests. At 600F, the temperature gradient along the specimen
test section was < 5F.
The load-controlled tests were conducted on a Schenk six-ton vertical
fatigue machine. This machine incorporates automatic control of the
alternating load, and the temperature in a three-zone f~nace when test-
ing. at elevated temperatures. At the 600F operating level, temperature
level and test-section _temperature gradient were within ±2F.
The purpose of the notched tests was to evaluate Kf only at 106
cycles. To accomplish this, six to nine specimens pf each not'ch geometry
were tested at each of the two test temperatures. Attempts were made in
testing each group of' specimens to fail some qf the specimens below 106
cycles, some above 106 cycles, and some as near 106 cycles as possible.-
Yalues of Kf were then evaluated by plotting the six data points on an
vs cycles-to-failure coordinates, joining a mean line through the points,
determining the value of on corresponding to the mean line at 10 6 · cycles,
and dividing the unnotched value of Sa at 106 cycles by an at 106 cycles.
Specimen failure was considered to be complete specimen separation.
Use of a dispJ,.acement-controlled unnotched test result and a load.::.
controlled notched test result for determining Kf is justified in this
case.-by virtue of the fact that the fatigue stresses at 106 cycles to·
failure were strictly elastic. This can be observed in Figure 4 where
the cyclic and monotonic (static) stress-strain response at the two test
·temperature are plotted. The values of elastic modUli used for con-
structing the elastic portions of the curves were 15.0 x 106 psi at room
temperature and 1,3.6 x 106 psi at 600F. These values are in ·agreement with
~namic test results obtained for the test material, dynamic values
reported in Reference 1, and static stress and strain measurements made
on the unnotched fatigue specimens at room temperature.

TEST RESULTS
The results of the room temperature .and 600F unnotched fatigue tests
are listed in Table 2, and plotted in A and B of Figure 5 in terms of
fictitious stress amplit"1,lde (E€a/2) and cycles to failure. Part A of
Figure 5 presents the room teJ!iperature data and B of Figure 5 the cor-
responding data at 600F. Best-fit curves have been faired (by eye)
through the data points for the fulJ.i-reversed cycle tests (zero pre-
strain) -to establish S-N curves over· the range of cycles from "'10 2 to 106 •.

. KAPL-.3158
 7

MONOTONIC AND CYCLIC 0

RESPONSE AT 80f 0
0

-----
MONOTONIC
RESPONSE AT 600f

(J)
(J)
w
a::
f-
(J)

CYCLIC· STRESS- STRAIN DATA

0 - ROOM TEMPERATURE (75Fl
c - 600f

AMPLITUDE , %

FIGURE 4. Monotonic and Cyclic Stress-Strain

Response at Room Temperature and at GOOF.
KS-62298 Unclassified

Data points for tests in which a 1.0% prestrain was superimposed are
plotted in A and B of Figure 5 as solid points. The effect of the tensile
prestrain is to reduce the cyclic strain"required for failure in a given
number of. cycles at lives >5 x 10 3 • These reductions can be considered
appreci.able, particularly at room temperature. For materials Uke the
present one, which possess high ratios of yield-to-ultimate tensile
strength, this result should be anticipated.
In Table 2, the results discus~ed .above are ~lso reported ·in 'tenns
of strain range, which was the controlled test variable. Va;t.ues of
eias"Lic moduli uiled for converting strain-range to fictitious stress-
amplitude are those quoted in the previous section.

KAPL-.3158
8

o ZERO PRESTRAIN
e 1.0% PRESTRAIN
- INDICATES NO FAILURE
I06 r-~r-TO-rrnTr---r-r-rrrnTr-~r-TO-rrnTr--,--,_,rn~--~~~~~

',.
''<'
---·--·
.
CORRECTION FOR _/"
MEAN STRESS (EQU. A.5) ',_
_________ ,a:__ _
..._..__

A. At Room Temperature.

''....:
CORRECTION FOR )',
MEAN STRESS (EQU. A.5) .. , ee e
,_
---- -----
/
......
.
....----~-­

104~--~--~~~~~--~~~~~~~-L~~~--~~~wu~~~~~-L~~
~2 ~ ~
CYCLES TO FAILURE
B. At GOOF.

FIGURE 5. Unnotched Fatigue Data.

KS-62299 Unclassified

KAP!r3158
 UNCLASSIFIED

QQNPlB'SNl'IAL • 9

TABLE 2. UNNOTCHED FATIGUE TEST RESULTS

Test Alternating Cyo1es.to

·Temperature, Al ternatiog Fictitious Stress Failure
3
F Strain Ra~e, ~ Amplitude, psi Preatrain£ ~ 'X 10- Failure
75 5.,30. 397,000 0 0~151
4·70 352,000 0-400
4.08 306,000 0.573
4-04 303,000 0.619
1.75 131,000 2.82
1.47 110,000 4~57
1.2J n;soo 10·4
·1.09 81,800 14.7
0.98 73,500 26.8
0.85 63,000 58.0
. 0.78 58,500 78.0
0.78 58,500 85.5
0.75 56,200 1420~0 None
0.72 54,000 0 1110.0 None
1.00 75,000 1.0 8.37
0.99 74,200 8.69
0.38 28,500 33.3
0.36 27,000 66.9
0.36 27,000 50.8
0.30 22,500 1240.0 None
0.26 19,500 1030.0 None
.. 75 o.zo 15,000 1.0 1710.0 None
600 5~44 372,000 0 0.402
4-?9 328,0QO 0.414
4-30. 294,000 0 •. 617
4·00 :).74,000 o:619
1.50 102,800 5.'00
.1.38 94,500 '7.06
1.10 74,500. 10.;9
0.82 55,700 20.4
0.65 44,200 34-7
0.59 40,100 51.4
0.59 40,100 1000.0 None
0.57 39,100 10.30.0 None
0.50 34,300 0- 1000.0 None
0.96 65,700 1.0 16.3
0.96 65,700 18.5
0.94 64,400 11.4
0.50 34,300 17.5

600
0.50
0.44
0.42
34,300
30,100
28,800
1
1.0
28.4
36.9
310.0

.UNCLASSIFIED
- KAPI.r-3158
10

The results of the notched specimen tests are listed in Table 3 and
plotted in Parts A and B of Figure 6 in terms of stress amplitude and
cycles to failure. Part A of Figure 6 includes the data at room tempera-
ture and B of Figure 6 the corresponding data at 600F. · Best-fit curves
have been faired (by e,ye) through the data points for each notch geametr,y.
. : The values o~ Kr at 106 cycles derived from the notched and un-
not'ched tests· are listed in. Table 4; Also listed are values of the
notch se~sitivity factor 2 (q) defined as: ·
Kf - 1
q =
K' ~ 1
t
This factor is a commonly used index for.judging the notch sensitivity·of
materials in fatigue. It provides a scale of notch sensitivity that
varies from q = 0 1 or no notch effect, to q = 1 1 or full theoretical
effect. Plots of. q vs R for roam temperature and for 600F are shown in
Figure 7.
Attention is directed again to the stress-strain diagrams plotted in
Figure 4· Note that the cycling did not alter the stress-strain behavior
at roam temperature (i.e., no cyclic hardening or softening), whereas at
600F the cycling caused a slight hardening. All the cyclic s~ress-strain
data points plotted are first-cycle values except for the four at 6dOF
which show a deviation from the monotonic curve. The stress values for
these points correspond to the lOOth cycle, where the peak-load valu,es
generally occurred.

DISCUSSION OF TEST RESULTS

Unnotched Fatigue strength
The unnotched fat,igue test results indicate that this particular
alloy of titanium possesses good unnotched fatigue strength in air rela-
tive to other materials, when evaluated on a strain range basis. This is
illustrated in Figure 8, where the unnotched and zero mean strain test
results have been plotted in terms of strain range and cycles to failure.
Shown for comparison is a scatter band for test results obtained by Manson
and Hirschberg 3 on fourteen different materials, including steels,
aluminums, titaniums, and high-temperature alloys. Their individual data
points for the titanium alloy, Ti-6Al-4V, heat treated to obtain an
ulttmate strength of 179,000 psi, are also shown.

KAPL-3158
 UNCLASSIFIED

G@N~IQENTIAis "* 11

TABLE J. NOTCHF.D FATIGiJ'E TEST RE.'>ULTS

Test .Minimum- Stress Cycles to .

:remperature, Notch-Root Section Amplitude, Failure
Radius, in. poi 3 Failures
F Diameter, in. X 10-:-

. 75. O.OOJ. 0.160 20,000 165.0

I
20,000 1J6.0
17,000 261.0

0.001
1
0.160
17,000
17,00C
17,000
lO,OOO.C
1, 550.(1
282.0
None

0.0015 0.480 JO,OOO 25.3

l
25,000 J9.6
20,000 1JO.O
18,000 UJ.O
15,000. £60.0
J
0.0015 0.480
10,000
10,000
9J2.0
529.6
0.005 0.480 20,000 75.0
15,000 192.0 :.
15,000 400.Q
15,000 1,510.0
i.4,000 285.0 .-
1
0.005
1
0.480
lJ,OOO
12,000
10,000
·8, 140;0
15,110.0
12,020.0
None
· Nc;me

0.015 0.480 JO,OOO '40.1

25,000 147.0.
! .

75
600
1
0.015
0.001
I
0.480
0.160
25,000
22,500
22,500
20,000
,15,000
17,000
149.0
1J,OOO.O
J4J.6
21,960.0
9,.320.0
.340.0
None

None
None.

I
17,000 JlJ.O
17,000 32.8.0,

1
0. (X)l
0.0015
0.160
0.1.80
17,000
20,000
20,000
12,000
7,54().0
1,7.0
104.0
5'iU.'O
None

10,000 .527.0

I
0.0015
U.!JU5
I
0.480
0.480
·10,000
10,000
9,000
.9,000
8,ooo
15,000
551.0
46J.O.
5'n.O
8,000.0
4,490.0
81.0
None

-
I
15,000 186.0
12,000 241.0
12,000 . JJ7.0

1
0.005 0.480
12,000
12,000
12,000
822.0
921.0.
8,220.0 None
o:o15 0.480 22,500 . 85.J
22,,ao 89.5
20,000 149.0
:w,ooo 2,320.0

600
J
0.015
1
0.480
20,000
.20,000.
5;090.0 .
.5,JOO.O
..,

UNClASSIFIED KAPL 3i58

12

20
0=0.160 in.
10 R=QOOI in.
0

30

20 0=0.480 ....
R = 0-015 in.
10

0
0=0.480 in.
R=0.005 in.

30
·;;;
..... 20
ILl
0
:::>
1- 10 0=0480 in.
:J
. [l.
R =o:ool5 in.
·:IE
•<t
0
•, ~

. (/)
"(/)
A. At Room Temperature.
ILl
a:
1-
(/)
30
C)
z 2
~ 0 =0.160 in.
o-
<t
z R =0.001 in.
,,,a:
1-
...J
<t

30
-G)--
20 0

o~o4eo in.
10
R =0.015 in.
0

B. At .600F.

FIGURE 6. Notched Fatigue Data.

KS-62300 Unclassified

KAPL-3158
 13

TABLE 4· FATIGUE NOTCH FACTORS AND NOTCH SENSITIVITY

AT 106 CYCLES .

Notch -Notch
·Radius,· Diameter, . 5
T~perature, F in. · in. •xt
~ ~
Kr Kr
~-
_q

. 75 0.001 0.160 6.9 3-.J 3-4* 0.39

75 -0.0015 0.480- 10.0 6.1 5.9*. 0.57
. 75 0.005 0.480 5.7 . 4·5· 4·6* 0.74
75 0.015 0.480 J.4 2~6 . 3_.2* 0.67
600 O.O~Jl 0.160 6.9 2.4 2.!i*" 0.24
600 0.0015 . 0.480 10.0. 4•3 4·3** ·O.J7
_600. . 0.005 0.480 5.7 3.5 3-9** . 0.53
600 0.015 0.480 J-4 2.0 . 2.9** 0.42 •.
* = 0.0006 in.
·s
**5 :; 0.0012 in •

.. '
1,1

1.0 o ._ROOM TEMPERATURE

o -600F ·
- 0.9
-~
II: 0.8
~
u
~ 0.7
/----
. ----~
.

>-
-~ 0.6
--
i=
~ 0,5
---
Vl
J: 0,4
~
-~ 0.3

0.2

0.1

FIGURE 7. . Vari~t ion of Notch-Sen-·

sitivity Factor (q) with Notch-Root
. . · .. Radius (R) •
KS-62301 Unclassified

KAPL-Jl58
14

0 Ti-6AI-6V·2Sn, 75°F
0 Tf-6Af-6V-2Sn,600"F
6 Tf-6AI-4V, 80°F (REF. 3)

1()3 104
CYCLES TO FAILURE

FIGURE 8. Comparison of Titanium Strain-Cycled

Fatigue Strength with Other Materials.
KS-62302 Unclassified

The following observations are 1nade:

1. The room temperature fatigue strength of the Ti-6Al-6V-2Sn
alloy falls on the upper side of the scatter band, exceed-
ing it in the low-cycle life region.
2. The 600F fatigue strength of the Ti-6Al-6V-2Sn alloy exceeds
the scatter band in the low ..cycl~ lifo region but i'alls to
the lower side of the ~cat·~er hand ;f n thli intermediate-cycle
life region.
J. The room temperature fatigue strength of the Ti-6Al-4V alloy,
heat-treated to obtain an ultimate tensile strength higher
than that of the Ti-6Al-6V-2Sn ~lloy tested i.n this investi- ·
gation (1'79,000 psi vs 160,000 psi), possesses slightly
greater fatigue strength in the high-cycle life region but
slightly less in the low-cycle life region.
Effect of Mean Stress on Unnotched Fatigue Stret¥th
The effect of mean stress on the fatigue strength of the titanium
alloy was studied by applying a prestrain of 1.0% to a number of the
unnotched specimens. The static strain of 1.0% is small·enough so that
the damage imparted to the specimen and the effect observed on the proper-
ties can be considered to be due to mean stress.

KAPL-3158
 15

Because of the large reductions in fatigue str.ength caua_ed bY the

presence of m~an stress, it was thought de~irable, for design purposes,
to describe the data by means. of an empirical damage law. TWo of the
damage laws proposed for predicting the effect of mean stress on fatigue.
strength were examined in ·an attempt to describe the data; these \-rere the
Goodman (linear) and Gerber (parabolic) laws. 4 · These were employed in
· the form in which the Goodman law is presently used for establishing
design curves for material being strain-cycled. s,s The design procedure
· incorporates a method for estimating the maximum effect of mean stress
on the fatigue strength. At present, this method is based on the follow-
ing assumptions:
1. Plastic deformation will limit the maximum cyclic stress
to the "limit of elastic behavior" (~).
2• ·Actual stress-strain.behavior oan be approximated by An
elastic-perfectly plastic stress-strain c:mrve, provided
that the idealized value of yield stress. is crhosen· suffi-
ciently l.arge.
3· A mod.ified Goodman diagram (Part A of Figure 9) accounts
for the fatigue damage du~ to m~an stress.
· 4· The limit of elastic behavior (~) can be estimated· by a .
· 0.2% offset to the monotonic stress-strain curve or- the
cyclic stress amplitude-strain amplitude curve, if the
material's stress-strain response is altered by cycling.
The choice of a 0.2% offset strain is arbitrary.
To det~rmine.the maximum effect of mean st!ess, a yield; line is
superimposed on the modified Goodman diagram for each fatigue life (Part A
· of Figure 9); the largest mean stress that can be mS:intained.· at a give·n .
fatigue life corresponds to the intersection of the Goo~an and yield
linea. .The fatigue curye is then adjusted for the maximum effect of mean
stress.
The above method has been devised assumiQg idealiz~d st~ess-strain
behavior. For actual materials, which always exhibit some strain harden-
ing, this method will be conservative, provided the strain corresponding
to the 'selected value of ~ is ? the maximum strains expected ·in service.
For total strains exceeding the offset strains corresponding to ~~ the
amount of mean stress that can be supported becomes a function of the
strain hardening .and maximum total strain, as well as the cyclic strain.
Despite. the fact·that the above described method is not necessarily con-
.servativefor cases where the maximum strain exceeds the strain corres-
ponding to ~~ as ,:,as the case in this investigation where a value of 1.0·%
mean strain was applied, it has previously been found that this method
. provides a sui table working theory for describing Zircaloy6 , 7 and Inconel 8
test dat~.
KAPL-3158
16

'/YIELD LINE
,.
'
''
z s I -------
C)
~
~
z
a ,., LINE
I '
ffi I '
~ I
I
'
'
~

sm 'sb Su
MEAN STRESS ---...

A. Modified Goodman Diagram.

LINE:

PARABOLIC LINE

I ''
''
''
Sm ' Sb Su
MEAN STRESS --+
B. Gerber Parabolic Diagram.

FIGURE 9. Fatigue Damage Laws for

Considering Mean Stress.
KS-62303 Unclassified

KAPL-3158
 17

Attempts to describe the titanium data with the Goodman law in the
method for estimating the maximum effect of mean stress did not yield
satisfactory results. The correction to the fatigue curve 'fell well
belowthe experimental data. For this reason the Gerber law was employed
with the same method to see if a better description of the data could
be achieved. The results are shown in A and B of Figure 5 by the dotted
lines. One· can observe that the maximum effect of mean stress concept
incorporating the Gerber parabolic dam~e law produces a.fairly good
description of the test. data, although it appears conservative in the
very high cyclic life region. Values of ~ and Su used were 150,000 and
160,000 psi at room temperature and 120,000 and 1.30,000 psi at 600F. An
·equation expressing the maximum effect of mean ·stress in terms of the
Gerber law is derived in Appendix A.
· Notch Se:nsi 1-.5vi.ty
The high values of Kf (a maximum value of 5.9) determined by the
fatigue tests and the steep rise in the q vs R curves of Fi~ure 7, at
the low values of R, indicate that this 9articular titanium:alloy has a
high fatigue-notch sensitiyity. At notch· radii~ 0.005 in., the room-
temperature fatigue-notch sensitivity is comparable to that of heat-
. treated steel having a .tensile strength of 200,000 psi; 9 th~ 600F notch
sensitivity is comparable to that of heat-treated steel having a tensile
strength of 125,000 psi. 9 There is, however, same relaxation in the
high-notch sensitivity at the larger radii, as the values of q are less
at R = 0.015 in. than at R = ~.005 in. For most mat~rials,the q vs R
curve does not achieve a maximum, but continues to rise with increasing
R,.as,ymptotically approaching a q value of 1.0.·
Another method used fo~ expressing the fatigue-notch sensitivity of
metals is the e-concept •. The 8-concept is based on the premise that the
stress at some finite distance, 8 1 below the notch root is the stress
which limits fatigue life.. The concept considers that because metals are
granular structures·and because fatigue cracks initiate on slip planes in
the grains, it should be necessary to increase the elastic stress amplitude
·so that the fatig~e limit below the notch root is. exceeded to some·finite
. depth, which is de[loted 8.. By considering a finite volume of material
below the notch root, the 8-concept takes into consideration stress
gradient effects.• · That is 1 it predicts that Kf increases as the net-
section thi'ckness increases, and that Kr is a maximum· for some value of ,
notch-root radius >0.
The 8-concept is especially ·useful for estimating the fatigue..,r:lotch
factors for crack-like defects. Experimental values of Kf are generally
not available for· crack-like defects for which R -o and Kt- oo. A paper
by 0' Donnell and Purdy-~ 0· ·demonstrated the use of the 8-concept for

KAPL-3158
 18

estimating fatigue-notch factors for cracks. In that paper, elastic

notch-stress distributions were calculated by the method of Neuber11 for
notched flat bars in bending and tension. The procedure used for
estimating Kf for a crack involves the plotting of K~ (for a given section
~hickness) for a range of Rand the selecting of the maximum value. The
maximum occurs for a finite notch radius, usually in the range from 0.0005
to 0~002 in·. Of course it must be kept in mind the.t the appropriate
value of 8 must be determined experimentally with finite not'ch-radii
specimens.
Extensionof the 8-concept for use with cylindrical bars, including.
the appropriate equations for stress distributions, is described in
Appendix B.· Values of 8 determined from the experimental test results
of this Investigation are 0.0006 in. at roam temperature and 0.0012. in.
at 600F. Agreement between notch factors aetermined using these values
of 8 (K~) and the ~xperimental values (Kf) can be observed in. ColUmns 5
and 6 in Table 4· _The values of .8 quoted have veen selected so that !p
. agre~s with Kf at the smaller notch-root radii and yield conservative
estimates· of Kf for larger notch-root radii.
It should be noted that both of the above concepts for expressing
fatigue-notch sensitivity are based on crack initiation as a failure
criterion, and the data reported here are for fracture. The use of
fract·ure data is . justified on the basis that the major portion of fatigue
life for failure in 106 cycles involves crack initiation. Nevertheless,
the use of crack initiation data could possibly improve correlations of
this type.

SUMMARY AND CONCLUSIONS

The following fatigue design properties have been established at
roam temperature and at 600F (in air) for the titanium base alloy 1 ·
Ti-~Al-6V-2Sn, in the annealed bar stock form!

1. Strain-controlled fatigue curves covering the cyclic-life

region from ~10 2 to 106 cycles.
2. The corrections to the fe.tigue curves for· the "maximum
effect of mean stress;" determined for applied prestrains
of 1.0%.
J. Notch sensitivity for sharply notched cylindrical bars
including factors for estimating Kf of crack-like defects.
The test results support the following-conclusions about the fa~igue
properties of·this material:

KAPL-3158
 19

1. It possesses good~ unnotched fatigue strength in air rela-

tive to other materials, when considered on a strain-range
pasis.
2. It has a high sensitivity to mean stress, particularly at
room temperature.
' .
3. It has a high sensitivity to sharp notches (<0.005. in.
notch-root radius).

REFERENCES
1. Titanium. Engineering Bulleti-n No. 10~ "Properties of Ti-6Al-6V-2Sn,"
Titanium Metals Corporation of America.
2. Peterson~ R. E. Stress Concentration Design Factors. New_York:
John Wiley and ~6ns, lnc. 195J.
3· Manson 1 s; S., and M. H. Hirschberg. "Fatigue Behavior in Strain
· Cycling in ·the Low- and Intermediate-Cycle· Range." Fatigue - -!!!,
Interdisciplinary Approach. Syracuse University Press. 1964.
Pp. 133.-173.
4· Richards, c. w. Engineering Materials Science. Wadsworth Publfshing
Company~ .Inc. Pp. 356~360.
5. !anger, B. F. "Design of Pressure Vessels for ).ow-Cycle Fatigue."
J .. of Basic Engr· ~ September 1962. P. 389.
6. 0' Donnell, W. J. ~ and B. F. Langer. , "Fatigue Design Basis for .
Zircaloy Components." Nuc. Sci. and Eng. ~ 1964. Pp._ 1-12.
7. Mowbray, D. F. "Effects of 1.0% Superimposed Mean· Strai~ o~ the
Bending Fatigue Strength of Zircaloy=4·" Nuclear Applications, l·
February 1, 1965. Pp. 39-48. · · ·
8. Mowbray 1 D. F.~ G. J. Sokol, and R. E. Savidge. "Fatigue Character-
istics·of Ni-Cr-Fe Alloys with Emphasis·on Pressure Vessel-Cladding."
KAPL-3108. July·l, 1965. ·
9. Peterson, R. E. "Analytical Approach to· Stress Concentration Effeat
of Aircraft Materials." WADC-TR ... 59-507. August 1959. Pp. 27.3-299.
10. 0' Donnell, W. J ., and c. M. Purdy. "The Fatigue Strength of Members
Containing Cracks·." ASME Trans.·~· May 1964. P. 205.
11. Neuber, H. "Theory of ~etch Stresses~ Principles for Exact Calcula:..
tion of Strength with Reference to Structural Form and Material~"
AEC-TR.:.4547. 1961.

KAPL=Jl58
 'i

THIS PAGE
WAS INTENTIONALLY
LEFT BLANK
 21

APPENDIX A. THE MAXIMUM EFFECT OF MEAN STRESS BASED

ON THE GERBER PARABOLIC DAMAGE RELATIONSHIP .

The Gerber parabolic law for representing the effect of combined

alternating. and mean stress is 4

=1 (A.l)
••
To determine an expression for the so-called "maximum effect of mean
stress," the point of ~ntersection of the yield line and parabola defined
by·:·Equ.at:l.on A.l must bed:ocated as shown in Part B of Figure ·9, p. 16.
The equation of the yield line is

·
s"a = s.."-1> ~ sm (A.2)

Upon solving for Sm in Equations A.l and A.2 and equating, the following
;relationship ·results:

(A.J)

Solving Equation A•3 for s" results in the following:

a

s' = (s_ - s~ .) + s
a \.,_b 2Sa. ·u

The criterion for evaluating tne me~n-stress correct~on to the S-~

curve using the Gerber parabolic· d~age.relationship· is tqen
I

s~ )
- -
2Sa
+ s u. . · . (s~
-
4S 2
. a
+ 1
·~ - 5t>'
-
· Sa
for Sa < s.. .
-o
:(A. 5)

KAPL-3158
 THIS PAGE
WAS INTENTIONALLY
LEFT BLANK
 23

APPENDIX B. EQUATIONS FOR USE WITH THE 5-CONCEPT

If the elastic stress distribution in the minimum section is ex-

pressed as (see Sketch B.l),

K = K(Y, R, b, T), ( B.l-)

'
the fatigue-notch factor is defined according to the 8-concept as

•'1.•=
TL - K(b I RI lJ I T} = Kf6 .
.
(B.2)

8 is treated as a material property and must be determined.by experiment.

,.-- ~

~
......... IY -R

-
~

T ..._ D jT
I -
. ~
-
.P

Sketch B.l. Notched Cylindrical Bar.

The von Mises criterion of failure was used to take into considera-
tion the multiaxial fatigue stresses in the vicinity of the notches.
Considering the radial stress component as negligible, Kf can be ex-
pressed by the von Mises criterion as (considering fatigue rather than
yielding):

1 (B. 3)
J2

KAPL-3158
 /

24

It is recognized that use of the von Mises criterion with "t!his·material

is not strictly correct because of probable anisotropic behavior. How-
ever, in lieu of fatigue properties in more than one test direction» it
is considered the best that is available.
The method of Neuber.ll. was used to calculate the notch~stress distri-
butions. N~Uber has derived nearly exact· solutions for infinitely wide
members containing shallow and deep notches. For the_iriterm~diate region
·between shallow and deep notches~ he suggests use of an interpolation
formula. It should be noted that the notch-stress distributions are also
based on isotropic theor.y. The appropriate equations for the normal and
tangential stress distri,butions are given below~
1. Normal stress

= (Ks = l}(Ktn - 1) .
~ ""' 1 + =~=,..,....==-=~~=~ (B.4)
n ~(Ks - 1)2 + (Ktn = 1)2

2. Tangential stress

(B. 5.) .

where
I

Ks = 1 +
cosh u 0
=---~-~...;;,.,.-
uo (.e 21l0. ':" 3~ ( 1 + 2-._ cosh u )
[e 2 .Jcosh2u - i
2(cosh2 u + 1)

+ c_osh ·Uo cosh u J (B.6)

~cbsh 2 u - 1

cosh u = (1 + ; ) cosh u0 (B. 7)

cosh 11o -A~ (B.8)

·v~

KAPL-3158
 25

. 1
sinh llo = -;:===- (B.9)
. . .JT/R - 1

K' = v (K8 - 1) (B.lO)

8

. 1
{c+ 21!( -1 + v) }
Ktn = vl- [(1- 21J~ sin vo]'

' + {1 -[(1 - ~) sin Yorr· {- B Cl +

(Boll)

Ktt = f~ [6- F~ sin vo]2 }l/2 { E

--.
(B.l2)

B. = (1 - 2v)(l :f- cos v 0 ) E (B.lJ)

. · C = .B ~ E cos 2 v0 (13.14)

1 [ 1 + cos '170 ]
=- ~
(B.l5)
E 1 +. 2v cos v + cos 2
0
v
0

1
cos v 0 = .. .., (B.l7)
. · ~D/2R -t ·1

· KAPL-3158.