Induced Marten Site Fatigue

8/3/2019 Induced Marten Site Fatigue

1/8

Low-cycle Fatigue Behavior in Metastable Austenitic SteelAccompanying Deformation-induced Martensitic Transformation*Kaneaki TSUZAKI,** Eixaburo NAKANISHI,*** Tadashi MAKI **and Imao TAMURA**Synopsis

The effect of deformation-inducedmartensitic transformation on thestrain-controlled low-cycle atigue behavior in type 304 metastable aus-tenitic stainless steel at room temperature has beenstudied. The fatiguebehavior n type 310 stable austenitic stainless steel has beenalso examined

for comparison. The fatigue life (N f) of type 304 steel was markedlyaffected by the ca'-martensite ormation. When the total strain range(AEt) was higher than 0.8 %, where the cr'-martensite formation startedbefore the crack initiation, N f of type 304 steel was about 1/5 times shorterthan that of type 310 steel, when comparedat the same 4e (mean plasticstrain range rom first cycle o final cycle). This shorter N f of type 304steel was attributed to the increase n d t (mean cyclic tensile stress) due tothe a'-martensite formation and to the martensite acting as the preferentialsite of crack initiation. The degreeof decrease n N f was larger when theca'-martensite ormation started in the strain hardening stage than in thesaturated stress stage of fatigue process, in connectionwith the larger in-crease n at. At low det such as 0.6 where the a'-martensite forma-tion started after the crack initiation, N f of type 304 steel was about 2times longer than that of type 310 steel. This longer J f, o type 304 steelwas attributed to the suppression of crack propagation due to the a'-mar-tensites omed around the crack tip. The experimental data of N f fell inthe range of factor of 2 of the predicted N f which were obtained on thebasis of Tomkins' model by taking into consideration the increase in 6tdue to the a'-martensite formation.I. Introduction

ecently, the consumption of LNG (Liquid Natu-ral Gas) as new heat energy is increasing, and sheetsof type 304 austenitic stainless steel are used as thefirst wall (membrane) of LNG storage tanks.1,2~ Sincethe cyclic stress and strain are caused in the first wallby the change in gas and liquid pressure due to thevariation of the liquid surface level, the low-cyclefatigue is quite an important problem for the designof the first wall materials.1 LNG is usually held at111 K (-162 C).1) At such a low temperature themartensitic transformation is induced by the fatiguedeformation in type 304 steel. It is thus necessary tomake clear the influence of martensite formationon the low-cycle fatigue life in order to optimizeperformance.There have been several papers'-9) about theinfluence of deformation-induced martensitic transfor-mation on the fatigue behavior of metastable aus-tenitic steels. The strain-controlled low-cycle fatiguetests? 9) showed that the martensitic transformationshortens the fatigue life. It has been reported thatsuch a decrease in fatigue life is closely related withthe increase in the stress amplitude due to the mar-tensite formation7'8 and with the martensite acting as

the preferential site of the crack initiation.9~ Theseprevious investigations indicate that the martensiteformation plays an important role on the fatigue lifein the metastable austenitic steel. However, sys-tematic investigations about the relation between thelow-cycle fatigue behavior and the deformation-in-duced martensitic transformation are few, and so theextensive studies concerning such a problem aredesired.The present authors10~ studied the effect of totalstrain range on the martensite formation behavior dur-ing strain-controlled fatigue deformation in type 304austenitic stainless steel. They10> reported that thecritical number of cycles for the onset of martensiteformation decreased with an increase in the totalstrain range, and that the martensitic transformationstarted in the strain hardening stage or in the saturat-ed stress stage of fatigue process, depending on thetotal strain range. However, the difference in thelow-cycle fatigue behavior between the two cases inwhich the martensitic transformation starts in thestrain hardening stage and/or in the saturated stressstage has not yet been reported. The present studywas thus performed to obtain a clear understanding ofthe effect of martensite formation in such two cases onthe fatigue behavior in type 304 metastable austeniticstainless steel.II. Experimental Procedure

n the present study, type 304 and type 310 aus-tenitic stainless steels were used. These steels wereselected because type 304 steel is metastable austeniticand type 310 steel is stable austenitic at room tem-perature. The chemical compositions and MS tem-peratures of these steels are listed in Table 1. Speci-mens for fatigue and monotonic tensile tests shown inFig. 1 were machined from hot swaged bars 20 mm indiameter. These specimens were solution-treated invacuum at 1 473 K (1200 C) for 3.6 ks and thenquenched into oil. Average austenite grain sizes ofspecimens in type 304 and type 310 steels were about200 ~mand 250 fim, respectively. The specimen sur-face was carefully polished with 80, 180 and 400 gritemery papers in order to minimize surface effect onthe fatigue properties.he fatigue tests were carried out in air with anMTS type machine (Shimadzu Servopulser) underthe tension-compression condition in the strain con-

* Presented to the 104th ISIJ Meeting , September 1982, 51314, at Hokkaido University in Sapporo. Manuscript received March 7,983. 1983 ISIJ* * Department of Metal Science and Technology , Kyoto University, Yoshida-honmachi, Sakyo-ku, Kyoto 606.* * * Graduate School , Kyoto University, Sakyo-ku, Kyoto 606.

( 834 ) Research Article


2/8

Transactions ISIJ, Vol. 23, 1983 (835)

trol mode. Axial strain in a 10 mm gauge length wasmeasured with an LVDT extensometer fitted on theuniform diameter section of the specimen. The cyclicstrain was of a triangular form with time. The totalstrain range was altered in the range from 0.5 to2.5 %. Tests were performed at room temperature(299 K~301 K) and at a strain rate of 3.3 x 10_3 -1.At various fatigue cycles, load-elongation hysteresisloops were recorded. The number of cycles to failure(i.e., fatigue life) was defined as the number of cyclesat which the stress amplitude in tension side decreasesto 3/4 of the maximum stress amplitude. The mono-tonic tensile tests were also carried out with the samemachine, and at the same test temperature and thesame strain rate as those for the fatigue tests.

he fatigue and the monotonic tensile tests werestopped at several periods, and the volume fraction ofa'-martensites was measured with a commercial " fer-rite scope " which measures the magnetic permeabili-ty of specimens. This instrument had been calibrat-ed for the a'-martensite phase in type 304 steel.III. Results1. Martensite Formation during Monotonic Tensile De-

ormationhe true stress-true strain curves of type 304 andtype 310 steels at room temperature are shown inFig. 2. The a'-martensite formation during themonotonic tensile deformation was observed only intype 304 steel and not in type 310 steel. The volumefraction (VM) of the a'-martensite formed in type 304steel is also shown in Fig. 2. As is seen, VM ncreaseswith increasing stress and/or strain. The critical ap-

plied stress and strain for the onset of the a'-marten-site formation in type 304 steel are recognized to beabout 290 MPa and 6 %, respectively. It is clearfrom these results that the room temperature (-300 K)is in the temperature range between Ms and Md fortype 304 steel and is above Md for type 310 steel,where Md is the temperature above which martensiteis not produced by plastic deformation and MS is thetemperature below which martensite forms beforeyielding of austenite, and above which martensiteforms after yielding of austenite.2. Martensite Formation and Change in Cyclic Tensiletress with the Number of Cyclesduring the Fatigue De-

ormationhe change in cyclic tensile stress (i.e., stress am-plitude in tension side, c) with the number of cyclesduring the strain-controlled fatigue deformation intype 310 stable austenitic steel is shown in Fig. 3. Thetotal strain range (art) was altered in the range from0.5 to 2.5 %. In all the tests the cyclic tensile stress(6t) increases with the number of cycles in the initialstage of fatigue process and then reaches the saturatedstress levels. This initial stage of fatigue process is

Table 1. Chemica l compositions (wt%) and MS temperat ures of speci mens.

Fig. 1. Shapes an d dimensions of specimens.

Fig. 2. True stress-true strain curves of the monotonic ten-sile deformations in type 304 and t ype 310 steels.

Fig. 3. Change in cyclic tensile stress with the number ofcycles during fatigue deformation in type 310 stableaustenitic stainless steel.


3/8

(836) Transactions ISIJ, Vol. 23, 1983

hereafter referred to as " strain hardening stage ".After the strain hardening stage, r scarcely changes(or slightly decreases) with the number of cycles upto failure. This later stage of fatigue process is re-ferred to as " saturated stress stage ". The saturatedstress level decreases with a decrease in 4st, and atlow 4~t such as 0.5 % the increase in o't in the strainhardening stage is very small. Such a variation of 6twith the number of cycles is similar to that in an-nealed stable metals and alloys.11,12)igures 4(a) and (b) show the martensite forma-tion and the change in cyclic tensile stress with thenumber of cycles in type 304 metastable austeniticsteel, respectively. 4st was altered in the range from0.6 to 2.5 %. Arrows (~ ) shown in Fig. 4(b) in-dicate the critical number of cycles for the onset ofthe a'-martensite formation. As is seen in Fig. 4(a),the a'-martensite formation is observed in all 4st ex-amined, and the volume fraction (VM) of a'-marten-site increases with the number of cycles. It is notedthat VM at a given number of cycles increases withan increase in 4~t, indicating that the a'-martensiteformation is enhanced as &t is increased.t should be noted in Fig. 4(b) that unlike type 310steel, two types of 6t-N curves (cyclic tensile stress-number of cycles curves) are observed in type 304 steel.At low 4st (i.e., 0.6 %, 0.8 % and 1.0 %) the saturat-ed stress stage is observed after the initial strain hard-ening stage. After some cycles of the saturated stress

stage, the value of 6t again increases with the numberof cycles. It is noted that the start of such a secon-dary hardening corresponds to the onset of the a'-martensite formation, and the tendency of 6t to in-crease with the number of cycles is consistent withthat in VM(Fig. 4(a)). The start of secondary hard-ening (i.e., the onset of the a'-martensite formation) isaccelerated and the degree of secondary hardening isincreased as 4et is increased. This range of 4st from0.6 to 1.0 % is hereafter referred to as " low 4s range."On the other hand, at high 4et (i.e., 1.5 %, 2.0 % and2.5 %) the a'-martensite formation starts in the strainhardening stage of fatigue process. The value of atcontinuously increases with the number of cycles dueto the a'-martensite formation, and thus the saturatedstress stage can not be observed. This range of 4stfrom 1.5 to 2.5 % is hereafter referred to as " high4Ct range." The high and the low 4et ranges corre-sponded to the test condition that the saturated stresslevel of austentte was higher or lower than the criticalstress for the onset of the a'-martensite formation dur-ing monotonic tensile deformation (i.e., 290 MPa inthis particular case), respectively.t can be concluded from the results shown in Figs.3 and 4 that the shape of at-N curve depends not onlyon whether or not the a'-martensite formation takesplace during the fatigue deformation, but also onwhether the a'-martensite formation starts in thestrain hardening stage or in the saturated stress stageof fatigue process.3. Fatigue Lifen the low-cycle fatigue tests, the Coffin-Mansonlow-cycle fatigue law has been used to account forthe fatigue damage. The Coffin-Manson law is giv-en as13~

4apNf= G ...........................(1)where, 4ep : the plastic strain rangef : the number of cycles to failure (i.e.,atigue life), C: constants.n case of the fatigue test under the total strainrange control mode, the plastic strain range (4e,) de-creases with an increase in cyclic tensile stress. Thenin the present study, the mean value of plastic strainrange (4;p) from first cycle to final cycle was measuredfrom the hysteresis loops. The values of 4~ werealmost equal to those of 4ep at mid-life (i.e., 0.5 Nf)of the specimens. The log ()-log (N f) plots fortype 304 and type 310 steels are shown in Fig. 5.The parameter a in Eq. (1) for both steels are listedin Table 2. Respective marks 0 and in type 304steel in Fig. 5 indicate the data for the high and thelow Act anges, which were defined earlier.

s is seen in Fig. 5, in the case of type 310 stableaustenitic steel all of the experimental data lie on astraight line with the slope of -0.54. However, inthe case of type 304 metastable austenitic steel twostages with different slopes are observed. In the rangeof 4~p higher than about 0.7 % the experimental datalie on a straight line with the slope of -0.65, which

Fig. 4. Martensite formation and the change in cyclic ten-sile stress with the number of cycles during fatiguedeformation in type 304 metastable austenitic stain-less steel. Arrows (~ ) in Fig. 4(b) indicate thecritical number of cycles for the onset of a'-marten-site formation.

Research Article


4/8


is somewhat higher than that in type 310 steel. Thefatigue life (N f) is about 1/5 times shorter than thatin type 310 steel, when compared at the sameOn the other hand, in the range of dip lower thanabout 0.7 %, the experimental data lie on anotherstraight line with the lower slope of -0.25. In thisrange of dip the difference in Nf between the twosteels is decreased with a decrease in dip, and at last"If for dip=0.36 % (dot=0.6 %) of type 304 steel be-comes about 2 times longer than JV of type 310 steel.Similar results (i.e., two stages in log (dip)-log (Nf)plot) have been reported in metastable austeniticsteels with the martensite formation by Smith et a1.l4)and by Baudry and Pineau.')he a'-martensite formation took place during thefatigue deformation in type 304 steel, and did not intype 310 steel. Then it is clear that the difference inN f between the two steels in Fig. 5 is attributed to thea'-martensite formation in type 304 steel. Morevoer,it must be emphasized in Fig. 5 that the two stageswith different slopes observed in type 304 steel cor-respond to the high and the low 4Et ranges (i.e., eachrange of det indicated by marks 0 and ), respective-ly. Alternatively, the two stages in log ()-log (N f)curve are classified by whether the a'-martensite for-mation starts in the strain hardening stage or in thesaturated stage of fatigue process. From these re-sults, the fatigue lives in such two cases of type 304steel can be summarized as follows :

1) The case in which the a'-martensite forma-tion starts in the strain hardening stage (i.e., at dst=1.5 %, 2.0 % and 2.5 %) ; N f of type 304 metastableaustenitic steel is much shorter than that of type 310stable austenitic steel.2) The case in which the a'-martensite formationstarts in the saturated stress stage (i.e., at 4a=0.6 %,0.8 % and 1.0 %); the difference in JVf between the

two steels becomes smaller, as 4st is decreased (i.e., asthe start of the a'-martensite formation is delayed).As the start of the cr'-martensite formation is moredelayed at z1=0.6t %, N f of type 304 metastable aus-tenitic steel becomes longer than that of type 310stable austenitic steel.onsequently, it is concluded that the fatigue life

of type 304 metastable austenitic steel is closely relat-ed with when the a'-martensite formation starts dur-ing the fatigue deformation.4. CyclicStress-Strain Curve

t is well known that the saturated stress level(6S)of the fatigue deformation is given by a power lawasl5)

os=kd~p ...........................(2)where, 4sp the plastic strain range in the saturatedtress stage

l: the cyclic hardening exponent: a constant.n case of type 304 steel, in the high dat range thespecimen does not show the saturated stress stage, andin the low 4st range the secondary hardening takes

place due to the a'-martensite formation, as was shownin Fig. 4. In the present study, the mean values ofcyclic tensile stress (Qt) and plastic strain range (Js)from first cycle to final cycle were measured from thehysteresis loops. The cyclic stress-strain curves (i.e.,log (at)-log (dip) curves) of type 304 and type 310steels are shown in Fig. 6. The values of parameter10 n Eq. (2) for both steels are listed in Table 2.

s is seen in Fig. 6, in the case of type 310 stableaustenitic steel all of the experimental data lie on astraight line with the slope of 0.39. It is worth notingfor type 304 metastable austenitic steel that two stagescorresponding to the high and low 4Et ranges are alsoobserved, as were observed in the log (48p)-log (A)curve (Fig. 5). In the high 4~t range (indicated by

Fig. 5. Relation between the mean plastic strain range andhe fatigue life in type 304 and type 310 steels.

arks and Q in type 304 steel indicate the lownd high 4r t ranges, respectively.

Fig. 6. Log (6t) as, log (drp) plot in type 304 and type 310teels. Marks and Q of type 304 steel indicatehe data for the low and high dst range, respective-

y.

Table 2. The values of parameters a in thedspN;=C and 9 in the equationfor type 304 and type 310 steels.

equation6S=kdsp


5/8


mark 0) the experimental data lie on a straight linewith the slope of 0.10. At a given Aep, the meancyclic tensile stress (at) of type 304 steel is from 1.4 to1.7 times higher than that of type 310 steel. The for-mation of a'-martensite took place only in type 304steel. It is thus clear that such a difference in at isattributed to the a'-martensite formation. In the low4et range (indicated by mark ) the experimental datalie on another straight line with the higher slope of0.99. The difference in at between the two steels isdecreased, as 4ep is decreased. At last, 6t for 4~p=0.36 % (4et=0.6 %) of type 304 steel becomes almostthe same as that of type 310 steel.t is well known that the propagation rate of fatiguecrack increases with an increase in stress amplitude(i.e., it). Therefore, it is possible that the two stagesof type 304 steel observed in log (Asp)-log (JV) curve(Fig. 5) is closely associated with the two stages inlog (at)-log (Aep)curve. This problem will be discuss-ed in the following section.Iv. Discussion

here have been several papers 1-9) about the in-fluence of deformation-induced martensitic transfor-mation on the fatigue behaviors of metastable aus-tenitic steels. Baudry and Pineau7) reported that inthe strain-controlled fatigue test both the initiationand the propagation of fatigue crack are acceleratedby the martensitic transformation. In the case ofstrain-controlled tests, the stress amplitude is increasedwith an increase in the amount of martensite. Sincethe crack propagation rate is increased with anincrease in stress amplitude, they') proposed that theincrease in crack propagation rate is related with theincrease in stress amplitude due to the martensiteformation. Moreover, Chanani and Antolovich9) sug-gested that martensites formed before the crack initia-tion act as the preferential site of crack initiation,resulting in a decrease in fatigue life. On the otherhand, Chanani et a1.4) eported in the stress intensityrange (4K) controlled test of TRIP steel that the crackpropagation rate is decreased in the condition wherethe martensitic transformation takes place near thecrack tip. It has been proposed that martensitesforms around the crack tip after the crack initiationabsorb the strain energy at the crack tip16) and sup-press the crack propagation by producing a residualcompressive stress.3,s)he fatigue life (A) is the sum of the number ofcycles spent in the crack initiation (.N2) and in crackpropagation up to failure (J'f ). Then the contribu-tions of martensite formation to Ni and Np should beconsidered in order to make clear the relation betweenthe martensite formation and the fatigue life. Inview of several investigations surveyed above3'4,7-s) hecontributions of martensite formation to the fatiguelife (Nf=N2+Np) can be summarized as follows:(Contribution 1) the increase in stress amplitude dueo martensite formation shortens Np(Contribution 2) the martensite formation before therack initiation shortens A'(Contribution 3) the martensite formation around theResearch Article

crack tip after the crack initiation prolongs NpHere the difference in .N'f between type 304 and type310 steels observed in Fig. 5 will be discussed based onthe above three contributions of martensite forma-tion.First, Contribution 1 is discussed. In the case ofstrain-controlled fatigue test, the crack propagationrate (dl/dJ"f) was proposed by Tomkins17) asldN 22(l)22 24ep( 6t) l ...............(3)8 2 T'

where, 4s p: the plastic strain ranget: the cyclic tensile stress (i.e., the stressmplitude in tension side): the crack length: the mean tensile stress equivalent to theean shear stress acting along flow bandsadiating 45 deg from the crack tip.quation (3) indicates that the crack propagationrate increases with an increase in 6t at a given 4ep,resulting in a decrease in the fatigue life. As wasshown in Fig. 6, at the same Asp the value of at oftype 304 steel was higher than that of type 310 steeldue to the a'-martensite formation. Therefore, inorder to discuss the difference in the fatigue life be-tween the two steels, the term of of should be con-sidered in addition to Asp.he fatigue life equation with the term of 6t canbe derived using Eq. (3) as follows. Integrating Eq.

(3), the number of cycles (JV'p) spent in the crackpropagation from initial crack length (lo) to final cracklength (l f) is expressed as

T2 if 2tp _ - ~c2 h4eQN In )..................The mean plastic strain range (4sp) and the meancyclic tensile stress (Qt) can be used instead of 4ep and6t of Eq. (4). The value of ultimate tensile stress(c) is generally taken as the value of T.17) Usingthese terms and assuming Jfp~Nf,17) Eq. (4) can berewritten as

2Aespat Nf= C* and G* = g~2In fh......(5)Uu

The terms of right hand (G*) in Eq. (5) are constant.The value of c can be obtained by the monotonictensile test. In the present study, l f was taken as aspecimen diameter (5 mm) and to was assumed to bethe initial crack length obseved by Grosskreutz18)(10 nm). Thus C* becomes 5.04. The ultimatetensile stress (an) for type 310 steel was taken as 943MPa which was observed in the tensile test at 300 K.In type 304 steel, the a'-martensitic transformationtakes place during the fatigue deformation and thefatigue crack propagates in the martensite-austenitetwo phase structure. Tomota et a1.19) eported thatin the case of two phase steel the fatigue crack propa-gates preferentially in the softer phase. Since o (i.e.,T) is the value for the front region of crack tip,17) thevalue of Qu or the austenite phase should be used intype 304 steel. The value of c (920 MPa) obtained


6/8

Transactions Is", Vol. 23, 1983 (839)

by extrapolating the data above Md to 300 K wasused for type 304 steel.ow it can be expected from Eq. (5) that the ex-perimental data of type 304 and type 310 steels lie onthe same straight line with the slope of -1 in thelog {QEp(at/Uu)2}-logNf) plot, if the difference in j f,between the two steels observed in the log (dip)-log(Nf) plot (Fig. 5) is caused only by the increase in atdue to the a'-martensite formation (i.e., Contribution1). The log {48p(at/au)2}-log Nf) plots for type 304and type 310 steels are shown in Fig. 7. In this figurethe predicted N f line using C* = 5.04 is also shown.s shown in Fig. 7, the experimental data of type304 and type 310 steels lie on respective straight lineswith the slope of -1, except for the datum for 4st=0.6 % (dip=0.36 %) of type 304 steel indicated by No.6. It should be noted that the difference in J%ff e-tween the two steels is still observed even in thelog {d~p(at/ru}2)-log (Nf) plot with the term of at. Thefatigue life at 4et>_0.8 % of type 304 steel is about0.6 times shorter than that of type 310 steel. On thecontrary, the fatigue life at 4et=0.6 % of type 304steel is about 2 times longer than that of type 310steel. It is clear from these results that the differencein N f between the two steels observed in the log (A)-log (N f) plot (Fig. 5) can not be explained only bythe increase in at due to the a'-martensite formation.However, it is noted in Fig. 7 that the experimentaldata of both steels fall in the range of factor of 2 ofthe predicted Nf line. This result indicates that theTomkins' model gives a good estimation of fatigue lifeand is applicable in the present steels. Thus it can beconcluded that the fatigue life of type 304 steel is ingood agreement with the predicted Xf by consideringthe increase in at due to the a'-martensite formation(i.e., Contribution 1), although Contribution 1 cannot explain all of the differences in Nf between thetwo steels.he decrease in N f of type 304 steel due only toContribution 1 can be estimated as follows. As it wasalready shown in Fig. 5, in the case of log (dip)-log(N f) plot without the term of 6t, the fatigue life oftype 304 steel in the high 4st range (i.e., dot>_ .5 %)

was shorter than that of type 310 steel by about 80 %.However, in the case of log {&,(at/cu)2}-log (Nf) plot

(Fig. 7), the fatigue life of type 304 steel at 4et>_0.8 %is shorter than that of type 310 steel by only about40 %. Therefore, it can be considered that at leasthalf of the decrease in Nf, of type 304 steel observedin Fig. 5 is attributed to the increase in 6t due to thea'-martensite formation (Contribution 1).urthermore, the two stages with different slopes(i.e., different values of a in Eq. (1)) of type 304 steelobserved in Fig. 6 can be now explained using Eq . (2)and Eq. (5) as follows.ombining Eq. (2) with Eq. (5), Tomkinsl7~ derivedthe Coffin-Manson law as

where,

lisp . 2p+1L1a = 2~+1

[()2= Bau In

Q 2 if 1i2~+1= 8~ 1niif 1/28+1to

(6)

It is clear from Eq. (6) that the value of a is dependenton R (cyclic hardening exponent) and decreases withincrease in 3. The values of 11(2/3+1) for type 304and type 310 steels are listed in Table 2. As is seen,the value of 1/(2j3+ 1) is in fairly good agreement withthe experimental value of a. Therefore, it can beconsidered that the difference between the parametera in the low dot and the high dot ranges of type 304steel is attributed to the larger 3 in the low 4Et rangethan that in the high 4s range.s was already described in Fig. 7, even in thelog {dsp(at/Uu)2}-log .N,) plot the difference in JV be-tween type 304 and type 310 steels was still observed.This difference in JV will be tried to explain by con-sidering other contributions of martensite formationto the crack propagation or the crack initiation (i.e.,Contributions 2 and 3 described earlier).s described earlier, martensites formed before thecrack initiation act as the preferential site of crackinitiation,9~ resulting in the decrease of X1 (Contribu-tion 2). Martensites formed around the crack tipafter the crack initiation absorb the strain energy atthe crack tip16) and suppress the crack propagation byproducing a residual compressive stress,3'8~ esultingin the increase of Np. (Contribution 3). It is easilyconsidered that such Contributions 2 and 3 are closelyrelated with whether or not the martensite is formedbefore the crack initiation.

he ratio (Nc/Nf) of the critical number of cyclesfor the onset of a'-martensite formation (J'i) to thenumber of cycles to failure (N f) in type 304 steel isshown in Fig. 8 as a function of dot. Unlike type 310steel, the value of .1V of type 304 steel is influenced bythe a'-martensite formation. Thus in order to knowthe ratio Nc/J%ff ithout the effect of a'-martensite for-mation, the values of Nf of type 310 steel at the same4st as type 304 steel were used in Fig. 8. In thisfigure the ratio Nc/)'T obtained by JV of type 304 steelis also shown for the comparison (indicated by mark0). As is seen, both data of /Nf (i.e., the data in-dicated by marks L and 0) are almost the same, ex-cept for the data at &=0.6 %. It is noted in the

Fig. 7. Log {4ep(Qtlcu)2} vs. log (N f) plot in type 304 andype 310 steels. Marks and 0 in type 304 steelndicate the data for the low and high dst ranges,espectively.

Research Article


7/8


data indicated by mark L that the value of JVc/Nf isalmost constant (about 0.2 %) at dat> 1.5 %. As 4rtis decreased, the value of J'T JV'f s increased and itbecomes about 20 % at 4s1=0.6 %.hompson20~ reported that the crack embryo ispresent in the range from 1 to 10 % of the fatigue life.Nakamura et a1.2~ bserved in type 304 steel that thecrack initiation takes place at about 20 % of the fa-tigue life. As was shown in Fig. 8, at dat>_0.8 % thea'-martensite formation strated at the number ofcycles below 2 % of the fatigue life. Therefore, it canbe considered that the a'-martensite formation occursbefore the crack initiation at dat>_0.8 %. Such mar-tensites formed in the early stage of fatigue life mightact as the preferential site of crack initiation, as sug-gested by Chanani and Antolovich9~ (Contribution 2).It is thus considered that the short Nf at 4a0.8 %of type 304 steel observed in Fig. 7 is closely associatedwith the decrease in Nz due to Contribution 2.On the other hand, at d st =0.6 % the a'-martensiteformation started at about 20 % of the fatigue life(Fig. 8). In this case the contribution of martensiteto the crack initiation is negligibly small because thecritical number of cycles for the onset of a'-martensiteformation is considered to be larger (or almost equal)in comparison with that for the crack initiation.Moreover, it can be expected that in the case of 4e=0.6 % most of the a'-martensites are induced at thecrack tip by the stress concentration after the crackinitiation. Martensites fromed around the crack tipsuppress the crack propagation3,8,16~(Contribution 3).Therefore, the long Nf at 4~t=0.6 % of type 304 steelis considered to be associated with the above reason.onsequently, the effect of a'-martensite forma-tion on the fatigue life (Nf) can be summarized as fol-lows :

1) The case in which the a'-martensite formation

starts before the crack initiation, Nf is decreased.This decrease in N f is caused by Contributions 1 and2. When the a'-martensite formation starts in thestrain hardening stage, the degree of decrease in N,is larger than in the saturated stress stage, in connec-tion with the larger increase in 6t.2) The case in which the a'-martensite forma-tion starts after the crack initiation, Nf is prolonged.This increase in Nf is caused by Contribution 3.

V. Conclusionshe effect of deformation-induced martensitic trans-formation on the strain-controlled low-cycle fatiguebehavior in type 304 metastable austenitic stainlesssteel at room temperature was studied. The fatiguebehavior in type 310 stable austenitic stainless steelwas also examined for comparison. The main resultsobtained are as follows :

1) In the case of type 304 steel accompanying thedeformation-induced martensitic transformation, twotypes of at-N curves (cyclic tensile stress-number ofcycles curves) were observed. In the high 4s t range(i.e., dst=1.5 %, 2.0 % and 2.5 %) the a'-martensiteformation started in the strain hardening stage offatigue process, and the value of of continuously in-creased up to failure. In the low 4s range (i.e.,&t =0.6 %, 0.8 % and 1.0 %) the saturated stressstage was observed. After some cycles of saturatedstress stage the value of 6t was again increased withthe number of cycles due to the a'-martensite forma-tion.

(2) The fatigue life of type 304 steel was marked-ly affected by the a'-martensite formation during thefatigue deformation.3) At dat>_0.8 % the fatigue life (Nf) of type 304steel was about 1/5 times shorter than that of type 310steel, when compared at the same 47p. This range ofdSt was considered to correspond to the case in whichthe a'-martensite formation started before the crackinitiation. The short Nf of type 304 steel was attribut-ed to the increase in Qt due to the a'-martensite for-mation (Contribution 1) and to the martensite actingas the preferential site of crack initiation (Contribu-

tion 2). The degree of decrease in J f of type 304steel was larger in the high 4s range than in the low4Et range, in connection with the larger increase inJt(4) At Oat=0.6 %, Jff of type 304 steel was about2 times longer than that of type 310 steel. The con-dition of Jet=0.6 % was considered to correspond tothe case that the a'-martensite formation started at thecrack tip after the crack initiation. The long Nf oftype 304 steel was attributed to the suppression ofcrack propagation due to the a'-martensite formationaround the crack tip (Contribution 3).5) The experimental data of JV fell in the rangeof factor of 2 of the predicted N f which were obtain-ed on the basis of Tomkins' model by taking intoconsideration the increase in at due to a'-martensiteformation.

(6) In the case of type 304 steel, two stages withthe different slopes (i.e., the different values of a in

Fig. 8. Effect of total strain range on the ratio (Nc/N f) ofhe critical number for the onset of a'-martensiteormation (JV~) n type 304 steel to the number ofycles to failure (N f). L. and Q indicate the databtained by J%/f of type 310 and type 304 steels,espectively.


8/8


Eq. (1)) corresponding to the high 4~t range and thelow dot range were observed in the log (app)-log (.Nf)plot. Such a difference in parameter a in Eq. (1)between the high and the low 4~t ranges could be ex-plained by the result that the cyclic hardening ex-ponent 9 was larger in the low zk range than in thehigh dst range.

REFERENCES) T. Udoguchi,K. Kasahara, M. Kitagawa, T. Tsunenari,. Ichihara and N. Sakamoto Pressure ng. Japan, 19

1981),169.) T. Nakamura, M. Tominaga,H. Murase and Y. Nishi-ama Tetsu-to-Hagane,8 (1982), 71.) A. G. Pineau and R. K. Pelloux: Met. Trans.,5 (1974),103.) G. R. Chanani, S. D. Antolovichand W. W. Gerberichet. Trans., (1972), 661.) R. G. Luther and T.R.G. Williams: MetalSci.11 (1977),19.) G. B. Olson,R. Chait, M. Azrin and R. A. Gage: Met.rans.,11A 1980),1069.) G. Baudryand A. G. Pineau: Mat. Sci. Eng.,28 (1977),29.

8)9)

10)

11)12)13)14)15)

16)17)18)19)20)

D. Hennessy, G. Steckel and C. Altstetter: Met. Trans.,7A (1976), 415.G. Rl Chanani and S. D. Antolovich: Met. Trans., 5(1974), 217.K. Tsuzaki, T. Maki and I. Tamura: Proc, of Int'1 Conf.on Martensitic Transformation (ICOMAT'82), edd by L.Delaey et al., Leuven, (1982), 423.C. E. Feltner and C. Laird: Acta Met., 19 (1970), 673.V. D. Munz, W. Wagner and E. Macherauch: Z Metallk.,60 (1969), 761.L. F. Coffin, Jr.: ASTM Special Tech. Publ. 520, 5 (1973).R. W. Smith, M. H. Hirshberg and S. S. Manson : NASATech. Note D-1575, (1963).E. A. Starke, Jr and G. Lutjering: Fatigue and Micro-structure, 1978 ASM Materials Science Seminar, ASM,Metals Park, Ohio, (1979), 205.S. D. Antolovich and B. Singh : Met. Trans., 2 (1971),2135.B. Tomkins : Phil. Mag.,18 (1968), 1041.J. C. Grosskreutz : J. Appl. Phys., 33 (1962), 1787.Y. Tomota, N. Tachibana, K. Tanabe and K. Kuroki :Tetsu-to-Hagane, 63 (1977), 962.N. Thompson: Fracture, Technology Press of M.I.T.,Mass., (1959), 354.

Documents

Induced Marten Site Fatigue