AA35548 AN EVALUATION OF OVERLOAD RETARDATION BEHAVIOR AND IlOVERLOA(O RE ARDXTION MU U FOREIGN TECHNOLOG DIVWRIGHT PATTERSON AFB OH G U FJGDA ET AL 17 NOV 83

1 AYIII D OT-DR I-577 83 F V i/f

L~iIS____ -68 -

I 11.25LA.

IAI_

- _ - .- . .. .. , ... . ... -- - .JL....iI irrl ll II..... . -.. ~. I- ]i . ... _ I. . .

I ,

FTD-ID(RS)T-1577-83

IFOREIGN TECHNOLOGY DIVISION

All EVALUATION 0? OVRLAD UTARUTION RENAVIOR ANDOVRLOAD RETMADATION MODELS OF Ti-6A1-4V

SHEUT TITANIUM ALLOY

by.• /

G. tiingda, Z. Yougkui and Y. Minggao

• DTIC-:-Af ~E L5 C TE

Approved for public releae;distrbution unlimited.

- .,-, •

• ~~~~~ ~ 1 .0""9III I" ... III IIIII

FU OP O

!, .' _ . .. ;.., r , --; " ;;-- . ". - --. -- ; : -- " ' --T ' :F .. .. ', .7

- •, , . : :7....

FTD -ID(RS)T-1577-83

EDITED TRANSLATION

FTD-ID(RS)T-1577-83 17 November 1983

MICROFICHE NR: FTD-83-C-001394

AN EVALUATION OF OVERLOAD RETARDATION BEHAVIOR ANDOVERLOAD RETARDATION MODELS OF Ti-6Al-4V SHEETTITANIUM ALLOY

By: G. Mingda, Z. Yongkui and Y. Minggao

English pages: 16

Source: Haqgkong Cailiao Yanjiusuo, Vol. 1,Nr. 2, December 1981, pp. 27-34; 13

Country of origin: China*Translated by: LEO KANNER ASSOCIATES

F33657-81-!D-0264.Requester: FTD/TQTAApproved for public release distribution unliaited.

ELECTE

*1 ~DEC i8

-, . THIS TWANSLAfIOP IS A NINVITIO10 Of TNfE OI#I.NA. MI'ON TEXT WITHOUT ANY ANALYTICAL Oft. 'TOrSAa. OmmWIN. STAIIENTS 0 TNINusm1S PagPAnRE SY:

•.. ATWO W W A011 TNWOOf TOW SO I1

AM ( OpT H NO I EI KtC1TM SI. POREMAN &VCIIAO W

"OILII I PAPIL 1I40.

FM -DO.T-1577-3_

- . . , .

i. V1

GRAPHICS DISCLA1HKR

All figures, graphics. tables, equations, etc. mrged into thistranslation were extracted from the best quality copy available.

Accession VerWTIS GRA&]DTIC TAB ELhaennouncod [

just if bat ion. .

Distribution/Availability Codcs

Dist Speeial

';,

A-

4-. ] - - - i - -

An Evaluation of Overload Retardation Behavior and OverloadRetardation Models of Ti-GAl-4V Sheet Titanium Alloy

Gu Mingda, Zhang Yongkui and Yan Minggao

Abstract

This paper focuses research on overload retardation

behavior under different overload ratios and different cracklengths through overload tests of a Ti-6A1-4V titanium alloy.

It also discusses the effects of some major factors on retard-

ation behavior. Retardation behavior is considered to be the

result of cyclic loads. It is suggested that the retardationprocess can be divided into five stages. Prom an analysis of

the modes of crack growth and other factors, the overloading

process of fatigue crack growth in these tests is regarded asmainly in a plane strain ooidtion or a mixed mode in which the

plane strain occurs predominantly. Thus, at a given overload

ratio 00I the nuwber of delay cycles So caused by overload

increases with the 'decroase of overload level K0 1 under plane

strain conditions.

In this paper, the Wheeler, fillenborg, Hateuoka and

Naarse models were selefted in view of applications and overallomarisons and discussions were carried out regardinq the des-

cribing capacity and application ce itica of each model on

retardation behavior. The Natev*&m model, based on the closure

ofeat, was found to be In roltiely gof reomat with the

test rsults. t aw alas nalwored' that the 40lay effeot

soMe asa In the e of Sa mode l oeaeably

iSMUer thea the- Oat"UaL msue.(rm ~10"ee 66147 effe% nowe.

1.12a, **04 VON"* a Poest 46"a of04 ~ ah es ie[Z-3)~ totego~is f ht* g IM M..s~U~

-• - - --

L.. s. K

varying loads, especially the interaction between the retarda-tion behavior of the crack growth rate caused by overloadingand the cyclic loads in tfie load time course.* That £5, themethod of combining and considering the overload retardationmodels on the basis of the successive accumlation method inorder to predict the fatigue crack growth life under varyingamplitudes. Many researchers have studied the factors affect-ing retardation behavior from different angles. A largeamount of test research has shown that retardation behavior notonly depends on the overload conditions themselves but is alsoclosely related to the fatigue failure caused by the loadcourse before overloading. in short,. retardation behaviorshould be the result of the interaction between the cyclicloads. it is a very complex ghnma and to date the mach-anism of retardation is still not clear. Further, the qatita-tive analysis of certain factors as well an their, measuremtent,is still difficult.

However, in order to explain the'retardation behavior ofoverload and load sequence on crack growth, we successively

pooed many retardation mechasisaw and models and attemptedto etabishtheoretical:. or awirioal overload retardation

-mo*I&in rde toprovite an eff'"tivar metbod to calca~tefatigue Crack 2row-t lit uftder verYin amplitudes i n egift-

"A~ts Pepez begim irem oabe bi t ofViNW. Of liad seleets twe trrpee at U~dOSW ft tpe -IS, th* MOiAe =OL

(4 a" *0I1I toe 51 1** tebs.th inm* MOW060

MOM"in OW~e *a"e awt momi*

mt-ek011 ( *I) AW* WW^ 1

~~IT"- 4WO

physical Coasept. It is 90901Y UseA am but is being devsl-oped. For this roiess, we ctseessed the retarcation, phenmenaand factors for the *vexl tests at fi-mA-4? alloy sheets.We also carried out, overall Lomses and evaluations of thedescribing capacity and cplsisonsditions ae. of theabove mentioned moels on the retardation usno ea

11. Materials and Zxperimental V&ocedure

The mater ial used in the aneiments was a Ti-EAI-4V alloyannealed sheet (thickness is 2 *Ova). Its main chemical compos-ition and mechanical properties are listed in Table, 1.,

jI 1 TV j F9 St 1,US US VPf b.0 KG.MPW.9r

Table 1 Chemical composition. (weight 1A) and .Mechanical Proper-ties.

The test Saple, Ls tiliceatral enetatioa. crack type(CC?) of a Xongit4a.epig-Zsimsa. are30 z1Oa' and it*IntialI 09". 10eapk after pefbstedftigueO creohe is 1. 02*

us toet wer cut"e Imt tGe 0~~ 00% * ftA &to*-

Mti se W th *S,04* "oft to&

carried out at room ten erature and the relative humidity wasless than 600.

See reference (111 for the explanations of the basic prin-ciples, calculation formulas and computer programs of relatedmodels.

Ill. Test Results and Discussion

*Constant amplitude tests were carried out according to theASTM Z647-78T. The test data was processed based on reference[121 wad the processing results are listed in Table 2.

(1)~I (3)J 3*

CLMPtN LL**~

table. 2 Data wrcs ing resultsof Ti-SAl-4V'sheet cons tantKe 1 (1) rit- forms.1as'; (2), Naterial 1 a coasteat;

(3) Ymer of 140"of aracks which grow from2@& fto 64mr (0) wilat; (S) Average

measue valep -I& Za tble, A is $he am ofpartia Aiffereace, sques. at the. fittin valuea"S tho .mesu a4 , o16 t4 value'' V IS the

t.b

9- emu am

WS, tottt., il'

w~

21 yA0 2 4 10 12 14

NX 10 4

Fig. I a-V test curves under different Q 1

3.1 Description of Overload Models in Relation toRetardation Phenomena

ina evaluating oerload retardation imodels, we first

analyze whether or not the model can appropriately describe andexplain the retardation behavior as well as whether or not itgrasps the major factors influenaing retardation behavior.Figure 2 gives the, overload retardation properties of titaniumalloy when single tensile owerlosid ratio %1-2 .0: da/dN-4 anida-V test aurvei at the "e time,, it gives the calculationcutve cocresponding-to the model.

fte retardation ptoe" ohmsn an th. toot curve in Figure 2is divided Into five stages.

-A~d-A

Mr -

ire -----

in faiu rcIrwh

1) Crack growth acceleration stage of overload point(A) (131;

2) Hysteresis stage of overload retardation (B);3) Maximum retardation point of overload (C);4) weakened stage of overload retardation'CD);5) Basic loss stage of overload retardation (3).

Note: the boundary line of stages (D) and (1) is the over-

load's monotohe plane stress plastic sone boundary, that is

4~;-4 (L)It can clearly be seen from Figure 2-that the Wheeler and

Willearg models. which use the otack point' residual, stressathe basis imediatlyr reach the maxiim retardation point

JUet after oMW2.e. a r Wdm, t.ee &. me show the hysteresisstab" 1 lofdetoab t -t tZy eater into the* weaheami

stag ofset tto. "A. Ip"h" thet tbw N aumahloto,0MMUe ""eeSM OO poebw r, ande i e i& mor -'se

he~01W Veep$ MaI thewes. ant V0'

-- Opt -

reteratioa point o5 &/iflA 1 M/' mu enes the Galv-j ated life to be- too lage eM then is. U4nAty -o refereevalue. La short, their eapflty -to 4orfibe eadainiquite poor&. Ccmstiwety vs"It*w ac ^dNWSmodels which use the creek coftn effect as the basis can verywell describe and eapl"a the.Attnr o retrdation, proces.Mowg te two, "te Ntee"stA e le. to, maloser to thek test

cve. This eslata that tbey:gnaw thef crack cloure andother a jar teters, cot~oflil xetardation behavior.f

it can alec- be see in tba fiflre that that; autismretar:datio poi"t aptr nd that thes rising areacrack zowth rato S. tod; 099t.. fta, is! the, result of theirsitting may retatt"ato % *n(e later Atnioaio).

Vaalauemasre

VaiV

et.. 7ie tha

Figure 4 gives the a-N test curve under many singletensile overloads ( 0 l.)as well. as the entire calculationcurv, of each model.

Fig. 4 a-N curve of many single tensile overloads (Q 01 -1.8).

Figure 4 also gives the measured value N* of the number ofcycles required for crack growth to grow from 2a-22mm to 7m

as well as the corresponding estimated value Nca of the re-mode abe t reglat reardtionindx x Cl.6)adoti

tardation model. It can be seen that not only is the Wheeler

satisfact ory results with its relative errors being -1.461% but ~INatsrv'1 a's model. also obtained satisfactory results with rela-tive errors of -1.40%. The errors of Hearse's model were

-10.250; the errors of the Willenborg model, based on the cal-culation of plane stress and plane &train, were separately

PfLgUZ I gives the I of single tensile overload. fe can

se* by coaring the calculation: results of Figs.' 3 and 4that the errors of the etiaton of sI*gl point %~ arela""r than these of the estimations of the entire life.

It aM& be. SM, Ewe it~ AD& 4- t% t f@llew6ft hi~Se @1Ly hs mtiesbleizreass.

fhist"C U" 12tL3w eeei a1 )4 but aheee viii a"

be discussed here.

Figure 3(b) does not list the calculation results of the

willenborg model because it loses effectiveness when Qol 2.

These tests also measured the overload ratio without re-

tardation of this titanium sheet under R=0.1 conditions to be

1.3 and the overload ratio without crack growth to be 2.8 (see

Fig. 1). As regards this point, only Matsuoka's model was

able to suppiy estimations and the results of estimations using

this model are separately 1.40 and 2.66. Its results are still

considered satisfactory.

3.3 Relationship of Crack Growth Mode and ND

Test results prove that under given Q01 conditions, ND

gradually decreases with the changes of Kol from small to large.

Following this, it is basically stable or there is a slight

increase (Fig. 3). This phenomenon is identical to the 2024-T3

overload test results given in reference 115].

In order to distinguish the growth mode, we carry out

verification for the maximum Kol-69NPaV point when Ql-2.0,

2=/B t et

€£

Key: ()- B is the thickness of the testsople. (2) And.

According- to reference (161. it ge.ul4 be iplem strain,that is, the crack growth is tho tensile mode. This can bePCGVea from the fracture of theatest sample (Fig. S, see Plate13): asid* from tho Otic teai" A in Inneou breaking

t I

J .L_

caused by the overload itself on the fracture of the test

sample, the entire crack growth surface (including the over-

load delay effect area) has typical plain strain fatigue

fractures or only a very small part is a little sheared. There-

fore, it is considered that this test belongs to the crackgrowth of the plane strain mode.

The influence of the crack growth mode on ND reflects theinfluence of overload level Kol on ND. Th dependent relation-

ship of ND on Ko, can be indirectly analyzed from the following

formula:

.N~u~mfda'o U,C( AK)"

In the formula, UD is the retardation coefficient. Under the

given conditions of and R, the above formula can be roughlyexplained as follows: on the one hand, %* increases with AKand the assumed negative index relationship d*creases sharply;on the other hand, ND* also gradually increases with the in-creases of D* Therefore, the relationship of a* and Kol aresimilar as shown by the solid line in rig. 6. When 101 isrelatively small, it is the stress condition of the plane strain.1t: Can be. seen that under pln stzain-conditions, ID* increaseswith the decrease of K

o.

a. *. 'f - ' .*i' *"' .WY " Y

rig. 5 (Plate 13) rront view (a) and saaros€copic fracturepiture (b) of test sample with single tensile overload

10

di

K.K

-, a(e,).... - N.-f, (AK )

-- K-

Fig. 6 Schematic diagram of the relationship between ND* andKol with given Qol"

Actually, under the conditions of these tests, the changes

of Ko1 were in essence the changes of the crack growth., Thisis to say that the crack growth and crack growth type are inthe same way factors which influence retardation behavior.

3.4 Relationship of Y , to, and Retardation Behavior

It is generally known that the size of overload plasticarea dimension Y is. closely related to retardation behavior.Yet, at present, the calculation formulas used for Yrb are onlydivided into the two extreme situations of plane stress andplane strain which are not sufficiently rational. On thp other

* hand, growing fatigue cracks are still processed according tothe concept of plane strain and plane stress in fracture tough-4ess which is also not very appropriate. After a large amount

of test 6bservations, we not only consider that the growth pro-

ess of fatigue cracking" should primarily use plane strain butfurtherom, after the discussion in the above section, we alsobelieve that the overload process in these tests should in thesame way be plane strain or a mixed mode primarily of planestrain. Por this reason, using the Willenborg model listed inrig. 4, the error measured aoooding to plane strain were farmaller than those oalulat.d acooding to plan. stress wich

* wll proves this point.

J .U

s . t * . . . .-r "I I

For the !4atsuoka. model, it was considered even norerational to introduce the cycle loading characteristics intothe plastic zone of the opposite direction (16, 171.

6-WILLM~ONG "U91LG. * XASUOXA "UKSE-NAARSS UOs.

0.4

Pig. 7. Relationship of delay effect zon e dimensions dand 1D

Key: (1) Test value.

*Figure 7 gives the relationship of crack growth M D and Kolwhich have been iffected by retardation. it can be seen fromFig. 7 that 0D correspondingly increases with the increases of

Ko After comparisons, it can be seen that the calculationvalue V.of each model is far smaller than the measured value(about 1/2-1/15 of the. meatured value). *moreover, this larger

i~i~. te lrge th crck engh),the larger the differ-

ence value with the tests. The production of this differencevalue is on the one hand related to the formula of the modelselected to calculate the plastic zone dimensional: on the otherhand (which is an even more, important reason), it is created by

1- each model lizmiting the delay ef fect. sone in the monotoneplastic zone caused by a. correspoadjngq overload. Naturally,this does not accord with test results.

ftaet value do0is larger tbA& calculated value 16 of eoverload's plastic scme &Laseifsim.. This, also explains the owtwo]. actimonf tiftarm otUkelese eftects iA the' rtzoo_06-

i --

basic loss stage (Z). This can be explained from Fig. 2.

The five crescent moon shaped darkened crack delay growth

zones are "a flight of steps" formed from the direct breaking

damage of the crack tip material caused by the overload itself

and the delay effect zone: after the crack edge and crack str-

face grow a very small distance in the 450 direction after-

overload, they then gradually reverse to the direction parallel

to the original crack surface and continue to grow until a

growth speed of a constant load is restored. It can also be

seen from the figure that the width of this zone increases withthe increase of the crack length and moreover the leading edgeis even more protruding. In Fig. 5(a), the turning point of

the crack growth direction changing and returning for the

four times and five times overload effect zone is very clear.I

Naturally, the protruding and restoring of the leading edge

of the crack after overload, especially the turning and reversalof the crack growth direction, influence retardation behavior.However, four models do not involve this factor.

To sun up, we consider that to study retardation behaviorit is not only necessary to consider the overload conditions

themselves such as overload ratio, overload number etc. but itis also necessary to consider other important factors such as

the closure stress (including the two sections of closure

stress of the crack and crack tip which pass into and throgh

the overload plastic sone), the crack tip's residual stress,

the crack tip's passivation and sharpening, the crack tip's

lal strain hardening, the, change and reversal of the crack

tip' direction, the leading edge of the crack becoming "bowshape" and the transfdoation of the crack growth mode etc.

These factors are all the result of crack point suroundig

plastic deformation caused by overloading. .nM these,.Closure bebsvier bri"S sbet .LO range" effects for each

IS

Yv. - :

stage in the entire retardation process and thus it in pro-sently being given serious attention.

Furthermore, the overload'sa environmental factors andoverload' s cycle stress ratio (inc luding the negative value) aremajor factors affecting retardation behavior. not all 44 theabove mentioned models were cons idered and thus we must awaitfurther work.

WV. Conclusions

1. Tests proved that the overload retardation process canbe divided into five stages and that retardation behavior isthe result of the interaction between cycle loads. The changesof the crack growth direction and protrusion of the leadingedge of the crack can affect retardation behavior.

2. The overloading process of these tests are planestrain or mixed mode which predominantly use, plane strain. Underplane strain conditions,, the retardation cycle numnber D cot-

eospondingly increases with the decrease, of overload level ofor a given 001 value..

3.* The results of estimsting crack growth life of fourretardation models with many single tensile overloads are stillconsidered satisfactory. nowever, because each model limitsthe delay effect none in the overload monotone plastic sone,the delay -effect sons dimensions calculated for each model are

mch smller than the test values.4.* The Wheeler and Willenhoeg maols are relatively

* lacking in their ability to describe, retardation hm ea.Theeswression ef the formesr is saple anad its applicability iss ttoaW yet It mat ameasure, the retardation indexi the latter issimple and convenient let loses effetivenss whenoa1 2.

5. The eoak olesure behavlar plays the meot impotantrole In the. entire, VeeaOtespies. ftuat the ability ofthe Mteuse wan Werse mels with *lams, of foot "te

it

omn

relatively good for deacribing retadaton pbsena. Among

them, the Matsuia model is oven coer to the test conditions.However', the effects of bot h of these am omplex load spctu

still await further research.

A groat deal of assistance was rendered by V4 Dejun,Shang Shijie, Lu Hui ubng, Be Xiaobe and Omnang Jie as well asconcades of the Room 16 Computer Group during the research

process. We would like to thank them here.

References

(1 1 Oatigue Crack Grovth Under Spectrum Loads" (originalillegible) STP 595 (1976).

[2 1 Huang Yusban, Liu Xuehui and Ki Nuiling, Scientific andI.. Technical. Materials of Western ngineering University,

Issue 919, 1979.9.

[31 eQingzhi, Journal of Beijing Aviation College, 1960, Wo.[4 1 Wheeler, O.3., Trans. AfM, Journal of Basic ingineering,

(1972), I, 181-106.

(S I Willenborg, 3., Bngle, R.N. and Wood, L.., (originall ii illegiLble)., TMJ-71-1-t51971).

16 Narse, J., Fracture, C1'4, 2(1977), 1025-1034.

(7 1 Matsuokar, S. and Tnaka, I., Sag. Tract. aech. 9(1977),-.507-523.

[S ] Nt a, S. ad Ta,=naika, a., Ing. Iract. Nech. 10(1976),* $15-525.

(9 1 Matsuoka, S. and tanka, 1., 3rg. lract. Mach. 11(1979),703-71S.

(101 Wanhill, R.J.3. at I., A TR 74043 u(*riginal illegible)(1974).

fill Shang. YoUgki and o Mita, "CaLOWAtIon Method andProgram for C u ewir, seij ing Cnstituteat of

Aeronaut ") Matesrials (10tera matRIDCal), 1961.(121 Sbag Yughui and *A IagS, aoutia eria8(Ipeelal Issu), 1951, 1.

[131 Sbijve, J., AM04M 595 (34) 3-13.

aterials, 1979, Me. 4.ft5 1111411, @. ad SaX.egyL.-trnsu. 3*A= 094,Cr(197). 1 -V95..zU.

i~l "'" '

1161 Schilve, 3.1 us. pweet. limb. 11(1979)1 147..221.

1171 Jasebyr 0.9. 90 ik, it. and Van Lipsive I.T.M.. "m1

Abstract

Ooelmi #4.51*. behavbio der diffev eied ratios a" differentcrack lengths Is a Tl-eAI-eV titoa. alO he. been izvestigted.?he Wffst *ION"s major aters, o retaedasow behavior is dlismod.The rotardatio. behaviome be eogsldeu# "e the- realts of isteuam"tieffsets between *ooe . a""Olse 66. is l aggtat the rotardatlor prow" ayh dividod, into fiveSurfe. tops as analysis of the so"e of fulpo. fractuter. the Ovealoefa

possof tat igse crack growth is thew. lt$t"aabo regarded " a w inl~ La pla. strin **"ot ea Sized &04.I* 'Whiel the Viae Strain eccursoreiOmlmaaty., Tbhu% at & siVe. ovotlood raoso Q'I the numew *f ielay.1.1. N, aesd by an "verloa LaOSres wIth a de"# Ls of K., adef Plane

to ti maver. the Wheeler. Wifle*Abe. Name". a", motivoha mSi"VWr seleee4 InrVW *1o oftSgift~l aeplWtW%. JAc *98tA&I Of the doserb-ins eseoe"t awl If* pedlatew o *m theeo aemi retardatin, baWlos havbees mde. Tb. *Soo"e we" bMd. "o ea reeb to es~w was &Mad.sowoovelf to hbe -ao agsemat wilk the eami tal roadts It Isalso wos~mpo edsh the IWGoeurlee TOO"~ae of ~fott 4o deaffet amem.else o Somalamrdg Venter amo she einlemlae vee foewatod hp the sot

IdI

f, A