Fatigue crack propagation in tensile residual stress fields of welded joints under fully compressive cycling

IntJ FatiguelO No 4 (1988) pp 237-242

Fatigue crack propagation in tensile residual stress fields of welded joints under fully c o m p r e s s i v e cycling A. Ohta, M. K o s u g e , T. Mawari and S. Nishijima

Fatigue crack propagation rates in centre-crack-typed transverse butt-welded joints were measured at a constant stress intensity factor range obtained by decreasing the applied and mean loads on test specimens. The propagation rate was maintained constant except at extremely compressed stress ratios. Fatigue crack propagation properties under compressive loading were found to be similar to those under tensile loading. Only under highly compressive cycling did crack propagation rates decrease. Key words: fatigue crack propagation; transverse butt-welded joints; compressive loading; residual stress

Propagation of fatigue cracks initiated by weld defects,/'2 such as lack of penetration, can cause failure of welded structures. Investigations into fatigue crack propagation in welded joints with defects H5 reveal that these properties are strongly affected by residual stresses through the stress ratio effect. 4 In high tensile residual stress fields, fatigue crack propagation properties are generally similar for different applied stress ratio conditions under tensile load cycling. 6 This similarity is due to extremely high tensile stress ratio conditions at the crack tip which is always open during any type of fatigue loading. 4'6qz'ls It is acknowledged that the fatigue crack propagation rates in base metals decrease as the applied stress ratio decreases)6 18 How fatigue cracks behave under highly compressive cyclic loading when they are in high tensile residual stress fields is investigated in this study. Some structural members, such as the top flanges of bridge girders, receive fully compressive load cycling. 19 If the propagation properties of a crack in tensile residual stress fields improve under compressive cycling compared with those under tensile cycling}S design stresses could be increased. Fatigue crack propagation tests in tensile residual stress fields under fully compressive load cycling conditions showed that propagation rates did not reduce except at extremely high compressive stress ratios. The crack opening stress intensity factor, Kop*, was deduced from the decreased ratio of fatigue crack propagation rate. This was difficult to predict using the calculated stress intensity factor, Kr, obtained from the initial residual stress distribution.

Experimental details Material and specimens Using 20 mm thick JIS SB42 steel (Tables 1, 2) welded joint plates were prepared to the specifications given in Table 3. The weld line was perpendicular to the rolling direction

of the material and centre-cracked-type transverse buttwelded specimens were machined from the plates. A centrenotch was located in the middle of the weld metal (Fig.

1). The residual stress distribution in the specimen without a centre notch was measured using stacked rosette-type strain gauges. The residual stresses parallel to the loading direction (Cby) were calculated by: eLY =

E 1 - v 2 (~;y + vs~)

(1)

where E is the elastic modulus, v is Poisson's ratio and ey and ex are the strains measured after mechanical cutting. T h e maximum value of the tensile residual stress is about 60 MPa in the middle part of the specimen. In Fig. 2 the dot-dash curve is the average of the front and back surface data, which is corrected to be symmetrical.

Crack propagation tests Fatigue crack propagation tests were conducted at room temperature in laboratory air on a 400 kN electrohydraulic fatigue testing machine at 1--60 Hz. Fatigue crack length, a, was measured with two sets of travelling microscopes. The stress intensity factor, K, was calculated by:2° K = Px/~[1.77 + 0.227(2a/W) - 0.51(2a/W)2 + 2.6(2a/W)3]/BIF/

(2)

where P is the load, W is the width of specimen, and B is the thickness. To maintain a constant range of the stress intensity factor AK, the load range, AP, and the mean load, Pro, were decreased as the crack extended. Fig. 3 shows the results of decreasing the minimum and maximum stress intensity factors (Kmin and Km~) to maintain a constant stress intensity

0142-1123/88/040237-06 $3.00 © 1988 Butterworth & Co (Publishers) Ltd Int J Fatigue October 1988

237

Table 1. Chemical composition (weight %) Si

Mn

P

S

Ni

Cr

Mo

V

0.19

0.81

0.015

0.007

0.01

0.02

0.001

0.001

C

0.17

Table 2. Mechanical properties

100 I' Edge o f I specimen

Tensile properties Upper yield stress (N/mm =)

Tensile strength (N/mm =)

284

Charpy Elongation absorbed % energy (J)

441

35

39( - 20°C)

~--

~

Edge o f specimen

~

0

A I

E E z

-50

v

-100

.~-150 _ I-

=0o

-200

---

10

/Cracklotch/--}-

- e - - - - Measured on back surface

~,

Used for calculation o f Kr

-250 -300

I 50

100

I 0

I 50

100

Distance from centre of specimen (mm) Fig. 1 Details of specimen (dimensions in mm)

Fig. 2 Residual stress distribution

Table 3. Welding condition

Welding process

Submerged arc welding Weld length (mm)

30"

Plate size (mm) Edge preparation by gas cutting (dimensions in ram)

Welding consumable

Pass sequence

I range, AK = 5.7 MN/m 3/2. Succeeding tests were performed by holding P~a,. Other specimens were used to determine fatigue crack propagation properties for wide-ranges of AK values at stress ratios of - 1, 0 and 0.5. Crack opening load, Pop, was measured with an extensometer straddling the notch at the centre of the specimen, and by using the unloading elastic compliance

238

5OO 500x 500 Fused flux, mesh size: 20 x 200 4.8 mm diameter wire

Welding position

Flat

Welding current (A)

670

Welding speed (m/min)

0.32

Arc voltage (V)

36

Heat input (kJ/cm)

45

Preheating (*C)

125

Re-drying of welding consumable Interpass temperature (*C)

250=C/1 h 125

method 21'z~ which analyses the anomaly in the relationship between load and notch centre deformation.

Results and discussion Fatigue crack propagation properties of the specimens are shown in Fig. 4. The stress ratios were maintained constant for the open symbols shown. Fatigue crack propagation

Int J Fatigue October 1988

4

0

-5

-10 I

• o

E

o

z

-15 o A K = 5.7"1~ Pmin = - 8 9

~

"

i• O

°eeoo e •

T.P, No I ._c

-20

==

& K = 5.7

I Pmin = --115

¢/)

T.P. NO 2

-25

A K = 5.7

T.P. No 1'

Kmax oooooooo

K• p

eeooeeeo

Kop •

[ Pmi. = - 1 0 2

&Kin MN rn -3/2 ) and Pmin in kN

-30

Kmin

- -

-35 24

Kr

I

i

I

I

26

28

30

32

I

I

fj

l

34 36 40 Crack length, a (mm) Fig. :3 Variation of stress intensity factor with crack extension for AKconstant and P,,j, hold test

properties are similar despite the change in the stress ratio, R, when R remains positive. For plots with solid diamonds, the stress ratio was reduced gradually as the stress intensity factor decreased; for example R = - o 0 for AK = 5.4 M N / m 3/2 and R = 1.15 for Stress ratios with an overbar, such as 1.15, are compressive for both the maximum and minimum stresses. The fatigue crack propagation properties for these fully compressive cycling conditions are coincident with those at positive stress ratios. Fig. 5 shows the effects of maintaining a constant stress intensity factor while reducing the stress ratio. The range of the stress intensity factor was kept at 5.7 M N / m 3/2 and the mean load was gradually decreased as crack length increased. The results of the tests are plotted in Fig. 4 with solid circles. The fatigue crack propagation rate could be reduced in certain compressive stress ratio conditions. Half solid symbols in Fig. 4 indicate the result of Pm= hold tests performed after the R decreasing-AK holding tests had reduced the crack propagation rate to 5 x 10-11 m/cycle. Results suggest that the propagation rates of cracks in tensile residual stress field could be reduced under heavily compressive load cycling compared to those under tensile cycling. However, tests under fully compressive cycling gave variable data for different cases. Fig. 6 shows the relationship between fatigue crack

AKth.

Int J Fatigue October 1 9 8 8

I 42

1 44

I 46

48

propagation rate and stress ratio in AK constant and R decreasing tests. It is clear that da/cln does not vary with decreasing R except for R values approaching 1. The R value at which da/dn began to decrease varied with the crack length and the specimen. This R value showed a tendency to approach 1- for longer crack lengths, is Relationship

between

da/dn a n d

AK,.

The stress ratio effect has been explained by the fatigue crack closure effectfl3 or the effective range of stress intensity factor, AK~r, corresponding to the opening of the crack during cycling. Fig. 7 shows an example of crack closure for an extremely low level of da/dn. The measured crack opening load, Pop, was determined from results as a function of load and modified displacement and a relationship between da/dn and AK, u was obtained. Fig. 8 plots the results with solid symbols with a 99% confidence band 12 for data at positive stress ratios. As results at different positive stress ratios coincide, crack closure does not take place for a whole range of load cycling and AK coincides with the true range of the effective stress intensity factor. However, most of the data for compressive stress cycling are located to the left of the 99% confidence band for cycling with positive stress ratios. The discrepancy suggests that the measured Pop under compressive cycling

239

10--6 <

10 7

(.) E

SB42 Stress ratio: R .... 1.15 nR=-I OR=0 AR=0.5 • A K = 5.7 [] P=,~ = - 8 9 , a = 26.6 ~ 27.2 Pmi~ = - 1 0 2 , a = 45.1 ~ 47.3 z~ Pm,~ = - 1 1 5 , a = 34.8 ~ 35.6

.EE 10_9

o

[] -- ~

[]

O13

g

~ 10 ~0

SB42 A K = 5.7 MN m -3/2 [] a = 24.9 ~ 26.6 mm <> a = 39.3 ~ 45.1 mm z~ a = 30.2 ~ 34.8 mm

e~

_v n

10 8

0

r~ O

_

lO-~--<;t,OI

1 ~ 9 9 % confidence interval

10 9

[]rl

(:3

DD 13

I

"I- i ~ f ~

I Illlli

11~_.~l

lilll[

2[ ~ ~ T O - ~ - o ~ - 2 0 - 1 0

i

I

-6-4

I

-2

-1

Stress ratio, R (J Fig. 6 Relationship between fatigue crack propagation rate and stress ratio for fixed range of stress intensity factor (AK = 5.7 M N / m ~ ) .

i==

e&& @

LL 10 ~0

$ []

10 N L 1

2

tiff

4

t6

8 10

20

40

Range of stress intensity factor, A K (MN m -3/2 )

Fig. 4 Relationship between fatigue crack propagation rate and range of stress intensity factor

~ 9

0,5

HtlIIV, R = o I.... ~ 1 ~ n = -1 o/~,,llAII R =-= vvvf,,iW~r, R =

M o d i f i e d displacement, 5' = ~ - o~P

Fig. 7 Example of test record showing fatigue crack closure

', R = 1 . 5 "~ 10 -7

g

•SB42 Pmin = - 8 9 , a = 26.6 ~ 27.2 • Prn,, = - 1 0 2 , a = 45.1 ~ 47.3

E

, R = i.22

APr~" = - 1 1 5 ' a = 3 4 " 8 ~ 3 5 ' 6

~

Q,. <

e f t " e,~S~"

e

e

~

10 .8

"o

\

c-

•~O 10 .9 •

0

Time, t Fig. 5 Schematic illustration of waveform for various stress ratios

cannot give a correct value o f the effective stress intensity range, for the plots should fall within the 99% confidence band w h e n a true effective value is referred to. Residual stress intensity

factor

The stress intensity factor for a crack in a welding residual stress field, Kr has been analysed. 24-26 Trials have also been carried out to predict the fatigue crack propagation rate in welding residual stress fields by c o m p u t i n g the sum o f K , and applied K. 3'7"11'13'14

240

•

~ ~

•

•

10 -1°

,~ 10 -11 0.1

I 0.2

,

I i I,I~I 0.4 0.6 1.0

~

.99"°/°c°,nf!(~ ence interval for AKef f* = AK

~ I ~

2

I lllll

4

6 8 10

I 20

40

A K e f f (MN mm -3/2)

Fig. 8 Relationship between fatigue crack propagation rate and effective range of stress intensity factor

The value o f Kr for this study was calculated by using the following equation z6 and referring to the residual stress distribution in a dot-dash curve in Fig. 2.

Int J Fatigue O c t o b e r 1 9 8 8

1

O

(3) where =

a

2.

Miki, C., Nishino, F. and Hirabayashi, Y. 'Fatigue crack growth analysis of corner weld of truss chord" Proc Col/oq on Fatigue of Steel and Concrete Structures (Int. Association for Bridge and Structural Engineering, Lausanne, 1982) pp 345-352

3.

Glinka, G. 'Effect of residual stresses on fatigue crack growth in steel weldments under constant and variable amplitude loads" in Fracture Mechanics, ASTM ST/=677 (1979) pp 198214 Ohta, A., Sasaki, E., Kamakura, M., Nihei, M., Kosuge, M., Kanao, M . and Inagaki, M . 'Effect of residual tensile stresses on threshold level for fatigue crack propagation in welded joints of SM50B steel" Trans Japan Weld Soc 12 (1981 ) pp 31-38.

= x/a and fJ = a/IV, with u = na~/2 and v

rip~2, f(a, J3) --- [1 + 0.297(1 - a2) ½(1 - cos v)] g(a, ~)

and g(a, ~) =

2(tan v)½

4.

[1 - (eos v/cos u)21* Evaluation

of true

Kop

The measured AK=a did not coincide with the true effective range of stress intensity factor. The true crack opening stress intensity factor, Kop*, was evaluated using the difference in AK values between data plots and the central curve of the 99% confidence band in Fig. 4. The solid circles in Fig. 3 show the evaluated Kop*. Experimentally applied Kmi, and Km,~, measured Kop and calculated Kr are also shown. There is no systematic change in the relationship between Kop* and K, among the three cases examined. This suggests that the discrepancy in Fig. 4 could not be explained only by Kr. Further work is necessary to give a more reasonable explanation of crack propagation mechanisms in tensile residual stress fields under fully compressive load cycling.

Conclusions The tensile residual stresses induced around the crack tip adversely affected the fatigue crack propagation properties of the specimens. Only under extremely high compressive cycling conditions, where the minimum stress was about - 1 0 0 MPa, cculd the fatigue crack propagation rate be reduced. For most cycling conditions, the relationships between da/dn and AK coincided well, despite differences in the stress ratios, even under fully compressive cycling, ie., fatigue crack closure did not occur. Therefore, the true range of the effective stress intensity factor was equal to AK. The effective range of stress intensity factor, AKe~f, obtained from the measured Pop cannot give the correct value of the effective stress intensity factor range. Expecting reduced propagation rates under compressive cycling compared with rates under tensile cycling is not realistic in an engineering sense.

Acknowledgements The authors wish to thank for support the Subgroup on Fatigue Properties of Welded Joints, Technical Advisory Committee for the Fatigue Data Sheets in National Research Institute for Metals, chaired by Dr H. Nakamura, Director of the Welding Division, NRIM. They are also grateful to Dr H. Terada, Head of the Strength Laboratory in National Aerospace Laboratory, for his information on the calculation program for the stress intensity factor induced by welding residual stresses.

References 1.

Maddox, S. J. and Webber, D. 'Fatigue crack propagation in aluminium-zinc-magnesium alloy fillet welded joints' Fatigue Testing of Weldments, ASTM STP 648 (1978) pp 159-184

Int J Fatigue O c t o b e r 1 9 8 8

5.

Fukuda, S., Watari, S. and Horikawa, H. 'Effect of welding residual stresseson fatigue crack propagation' TransJapan Soc Mech Engrs47A 416 (1981) pp 384-390 (in Japanese)

6.

Ohta, A., Sasaki, E., Nihei, M., Kosuge, M., Kanao, M. and Inagaki, i . 'Fatigue crack propagation rates and threshold stress intensity factor for welded joints of HT80 steel at several stress ratios" Int J Fatigue 4 4 (1982) pp 223-237

7.

Nihei, K., One, H. and Tsunenari, T. "Study on prediction of fatigue crack propagation life considering welding residual stress" J Soc NavalArchitects Japan 152 (1983) lop 390-396 (in Japanese)

8.

Kitsunai, Y. 'Fatigue crack growth behavior in mild steel weldments at low temperatures" in Fatigue at Low Temperatures, ASTM STP 857 (1985) pp 274-292

9.

Mori, T. and Horikawa, K. "The effect of welding residual stresses on fatigue crack propagation rate" Quart J Japan Weld Soc I 3 (1983) pp 436-443 (in Japanese)

10.

Kobayashi, A. and Todoroki, A "Effect of weld-induced residual stress on fatigue crack growth in compact specimen' J Soc Material Science Japan 35 391 (1986) pp 401-406 (in Japanese)

11.

Terada, H. 'An analysis of a crack in the residual stress field of welding' Role of Fracture Mechanics in Modern Technology (Elsevier, Amsterdam, 1987) pp 899-910

12.

Ohta, A., Soya, I., Nishijima, S. and Kosuge, i . 'Statistical evaluation of fatigue crack propagation properties including threshold stress intensity factor' Engng Fract Mech 24 6 (1986) pp 789-802

13.

Kitsunai, Y., Kobayashi, H., Narumoto, A., Ishizuka, I. and lida, K. 'Effect of welding residual stresses on fatigue crack growth behaviour in butt-welded joints of STB42 and SMSOA steels'/IWDocXIII-1203--86 (1986) pp 1-24

14.

i u k a i , Y., Nishimura, A. and Kim, E.-J. 'Effect of welding residual stress on fatigue crack opening behaviour" Quart J Japan WeldSoc5 1 (1987) pp 113-118 (in Japanese)

15.

Ohta, A., Kosuge, M. and Nishijima, S. 'Fatigue crack growth in welded joints under compressive applied stresses' Int J Fract (1987) pp R17-R20

16.

Forman, R. G., Kearney, V. and Engle, R. M. "Numerical analysis of crack propagation in cyclic-loaded structures' Trans ASME Set D 89 (1967) pp 459-464

17.

Klesnil, M. and Luke,, M. 'Effect of stress cycle asymmetry on fatigue crack growth' Mater Sci Engng 9 (1972) pp 23124O

18.

Ohta, A. and Sasaki, E. 'Influence of stress ratio on the threshold level for fatigue crack propagation in high strength steel' Engng Fract Mech 9 (1977) pp 307-315 Fisher, J. Fatigue and Fracture in Steel Bridges (John Wiley & Sons, New York, 1984) Brown, W. F. Jr and Srawley, J. E. 'Plane strain crack toughness testing of high strength metallic materials' ASTM STP410(1967) pp 1-66 Kikukawa, M., Jono, M. and Tanaka, K. I. 'Fatigue crack closure behaviour at low stress intensity level' Proc 2nd Int Conf on Mechanical Behaviour of Materials, Boston, August 1976 (ASM, Metals Park, OH, 1978) pp 254-277 Ohta, A., Kosuge, M. and Sasaki, E. "Fatigue crack closure

19. 20.

21.

22.

241

over the range of stress ratios from - 1 to 0.8 down to stress intensity threshold level in HT80 steel and SUS304 stainless steel" IntJ Fract14 3 (1978) pp 251-264 23.

Elber, W. "The significance of fatigue crack closure' ASTM STP486 (1971) pp 230-242

24.

Kanazawa, T., Oba, H. and S u x i , J. "The effect of welding residual stress upon brittle fracture propagation' TransJapan Soc Naval Architects Japan 110 (1961) pp 359-368 (in Japanese)

25.

242

Terada, H. 'An analysis of the stress intensity factor of a

crack perpendicular to the welding bead' Engng Fract Mech 8 (1976) pp 441-444 26.

Tada, H. at al The Stress Analysis of Cracks Handbook (Delaware Research Corp, 1974) 2.33

Authors The authors are with the Fatigue Testing Division, National Research Institute for Metals, 2-3-12, Nakameguro, Meguroku, Tokyo 153, Japan.

Int J F a t i g u e O c t o b e r 1 9 8 8