Stress Corrosion Cracking of Welded Joints in High Strength

Variables affecting stress corrosion cracking are studied and conditions recommended for and postweld heat treat to obtain maximum resistance to cracking

BY T. G. GOOCH

ABSTRACT. High strength steels may linear elastic fracture mechanics prin­ detrimental effect on weld metal SCC suffer a form of stress corrosion ciples using precracked specimens. resistance, although segregation may cracking (SCC) due to em­ Testing was carried out in 3% sodium be particularly significant in precipita­ brittlement, the hydrogen being liber­ chloride solution as representative tion hardening systems. SCC failure ated by a cathodic corrosion reaction. of the media causing SCC of high may take place intergranularly, by Most service media will be expected strength steels. Welds were prepared cleavage, or by microvoid coales­ to liberate hydrogen, and the problem in the experimental alloys and the cence, intergranular failure being affords a considerable drawback to pre-existing crack located in various largely associated with the presence the widespread use of high strength regions of the joint, while samples of twinned and high sus­ steels. For a number of reasons, fail­ were also prepared using The Weld­ ceptibility. The results suggest that ure may be particularly likely when ing Institute weld thermal simulator highest SCC resistance will be ob­ welding is used for fabrication. Un­ to reproduce specific heat-affected tained from low , low alloy less the structure is efficiently stress zone (HAZ) microstructures. Suscep­ systems, since these are unlikely to relieved, tensile stresses of yield or tibility was defined in terms of the suffer the development of deep corro­ proof stress magnitude can remain in critical threshold stress intensity to sion pits which may act to initiate the vicinity of the weld, while even cause cracking, the results obtained SCC in service, and will have high with stress relief, a weld will consti­ being related to material composition resistance to SCC by a cleavage tute a region of stress concentration. and microstructure with reference to mechanism. During the welding cycle, metallur­ different heat affected zone and weld Attention has been further paid to gical changes may take place causing metal areas. Fractographic examina­ the practical risk of SCC initiation in local microstructures that are particu­ tion was carried out and susceptibility service, consideration being given to larly sensitive to SCC. The present related to failure mechanism. the effects of environment on SCC investigation was initiated to clarify From the results obtained it has initiation and propagation. A prelim­ the situation, three particular objec­ been possible to derive a general inary correlation has been obtained tives being identified, namely: understanding of the variables affect­ between fracture mechanics data and 1. To define the relative SCC be­ ing weld SCC performance. Suscep­ the SCC behavior of uncracked, un­ havior of a range of high strength tibility is primarily dependent upon dressed welds. steels. microstructure, and, for a given mate­ The information reported thus con­ 2. To assess the effects of welding rial, recommendations may be made stitutes a rational basis for under­ and postweld heat treatment on regarding welding conditions and standing the SCC behavior of welded SCC susceptibility. postweld heat treatment to obtain joints. Although direct testing is still 3. To evaluate the general prac­ maximum SCC resistance. The necessary to establish the behavior of tical implications of the data ob­ presence of twinned martensite in a given joint in a particular environ­ tained. particular should be avoided, since ment, it is possible to maximize SCC A range ot steels was studied in­ this ohase has a highly deleterious resistance in advance by attention to cluding both conventional medium effect on SCC resistance. Provided relevant factors. carbon, low alloy materials and low postweld heat treatment similar to Introduction carbon, precipitation hardening that specified for base metal is ap­ grades. SCC testing was based on plied, HAZs and matching composi­ While it is widely realized that fer­ tion weld metals will generally show ritic steels may be subject to stress T. G. GOOCH is Group Leader — SCC resistance similar to the base corrosion cracking (SCC) in certain Austenitic Group, The Welding institute, metal; without such heat treatment, Abington, Cambridge, Great Britain. Paper specific environments such as caustic increased susceptibility must be an­ conditions (Ref. 1), the environments was presented at the 55th AWS Annual ticipated. The presence of segregation Meeting held in Houston during May 6-10, causing trouble have in the past been and inclusions generally has little 1974. relatively few and the problem has

WELDING RESEARCH S U PP LE M E NT I 287-s not been widespread. However, havior of welds in high strength practical application (Ref. 7). Further, during evaluation of high strength steels. Although a number of evalua­ the crack can be sited in any region of alloys (yield stress > 1000 tion studies have been carried out a weld, and the risk of failure of in­ N/mm2(150 ksi)) for aerospace and relevant to specific alloys welded dividual weld regions assessed large­ other applications, it became appar­ using particular techniques, the rela­ ly independently of surrounding mate­ ent that these materials could suffer a tive importance of the various factors rial. Concurrently with these studies, more general form of SCC in a variety determining weld performance has metallographic and fractographic of media at around room temper­ not been defined. In consequence, it examination was carried out to iden­ ature, cracking arising in environ­ has been difficult to make general tify the factors determining SCC ments that had proved entirely recommendations for the avoidance behavior. innocuous in the past with lower of SCC in service. This lack of knowl­ Second, SCC testing of plain un- strength alloys (Ref. 1). The reason edge constitutes a potential drawback cracked welds was undertaken, and for this susceptibility has been to the more widespread use of the results correlated with the frac­ studied, and two hypotheses pro­ welded medium and high strength ture mechanics data. Compositional pounded, namely that failure is due steels, especially since a very wide and microstructural gradients across either to the presence of a readily range of environments has been an unmachined, uncracked weld may corrodible "active path" within the reported to cause cracking (Ref. 5). modify SCC initiation and propaga­ material, or to a form of hydrogen em­ The present investigation was under­ tion as compared to a precracked brittlement, with the hydrogen being taken to clarify the situation, three specimen where the development of liberated by a cathodic corrosion reac­ particular objectives being identified: SCC is largely controlled by the pre­ tion (Ref. 2). While the former mech­ existing crack and associated plastic anism may be operative in some in­ 1. Evaluation of the relative SCC zone. Further, there is little informa­ tion on the extent to which geomet­ stances, research has indicated that behavior of a range of high rical stress and strain concentrates at in general failure takes place as a strength steels. weld toes affect the corrosion proc­ result of hydrogen embrittlement (Ref. 2. Definition of the effects of esses causing SCC initiation. Thus, it 3). From investigations into the prob­ welding and postweld heat treat­ was considered necessary to estab­ lem of hydrogen embrittlement of ment on SCC susceptibility. lish that relative data obtained from steels, the risk of SCC can be expect­ 3. Assessment of the general practical implications of the data precracked samples were directly ed to increase with increasing mate­ applicable to practice. rial hardness, and the phenomenon obtained. assumes particular significance in Experimental Approach Precracked Sample Studies the case of high strength alloys. Sample Preparation Stress corrosion failure at welded Five high strength steels were joints in high strength steels may be selected as a basis for study, includ­ The base metals were heat treated especially likely for a number of ing both medium carbon, low alloy as in Table 2. Single edge notched reasons (Ref. 4). Unless the fabrica­ material and low carbon, precipita­ and cracked (SEN) samples tion is efficiently stress relieved after tion hardening grades. The alloys (Ref. 7) were machined as in Fig. 1, welding, residual welding stresses chosen were FV520S, NCMV, with adequate dimensions in all will remain. In the weld area, these RS140, 18% nickel maraging, and a cases to give plane strain condi­ stresses will be tensile, and as they copper- Table 1, first five tions during SCC testing, although will normally be of yield stress magni­ items), with the materials supplied as this was not always the case with tude, they can greatly increase the 25 mm (1 in.) and 15 mm (0.6 in.) toughness tests in air (Ref. 8). risk of cracking. Even with stress thick plate (Ref. 6). Other, lower Welds were made in each steel to relief treatment, local stress concen­ strength alloys were used in the cover a range of processes of indus­ trations at welds will usually remain. course of the investigation to clarify trial application as indicated in Table Depending on detail design, welded certain aspects of SCC behavior 3. Matching composition filler metals joints may be associated with (Table 1, last two items). were used in most cases, although crevices where concentration of The program was carried out in two consistent with normal practices, an aggressive chemical species can stages. First, welds were prepared in FV520B was adopted for occur, further increasing the likeli­ the experimental steels, and the SCC the FV520S GTA welds. In addition, a hood of failure. Metallurgical changes resistance of the various regions of lower strength Ni-Mo-V filler metal taking place during the welding ther­ these welds defined, comparison was used for a GMA weld in NCMV mal cycle may result in local micro- being made with base metal be­ steel. The weld metals studied are structures that are particularly sus­ havior. A linear elastic fracture me­ shown in Table 4. ceptible to SCC. Hence, SCC can chanics approach to SCC testing was A K-preparation (double-bevel- occur at welded joints even though adopted, since the use of precracked groove joint) was used for the arc the base metal is relatively immune. samples reproduces the worst prac­ welds to obtain an approximately Despite a general appreciation of tical case of cracking initiating from a planar fusion boundary and HAZ, so these factors, there is little informa­ pre-existing welding defect, while the that the pre-existing crack could be tion available regarding the SCC be­ approach gives quantitative data of located in any particular weld area,

Table 1 — Analyses of Steels Used'a) Material C Si Mn Ni Cr Mo Cu Ti Nb Ca Zr Al Co Cu-Si 0.37 1.7 0.9 — — 0.12 1.75 0.23 — RS140 0.36 0.25 0.58 0.26 3.2 0.96 — 0.21 — FV520S 0.07 0.45 1.26 5.57 15.3 1.73 1.7 — 0.15 Maraging 0.009 0.005 0.09 18.0 — 3.19 — — 0.31 0.01 0.003 0.006 0.07 9.25 NCMV 0.45 0.80 0.47 1.74 1.32 0.94 — 0.25 — BS4360 0.19 0.01 1.4 0.01 0.01 0.02 0.04 0.01 0.015 HY80 0.19 0.19 0.28 2.85 1.46 0.44

(a) Analyses supplied by RARDE, except for BS4360 and HY80, which were analyzed at The Welding Institute.

288-slJULY 1974 with fatigue and stress corrosion Simulated HAZ specimens were the NCMV and HY80 steels were heat crack propagation taking place prepared to assess the effect of peak treated using a thermal simulator to through approximately constant ma­ temperature reached in the HAZ, and reproduce the HAZ cycle created by terial microstructure. The electron to study the worst case likely to arise SMA welding at 1 kJ/mm (25kJ/in.) beam welds were close square- in practice of a single pass weld with with a 250 C preheat, maintained for groove butt joints. no post-weld heat treatment. For 24 h after cycling. Peak temperatures SEN samples were machined as in comparative purposes BS968 (now of 1300 C and 1000 C were repro­ Fig. 2, with dimensions based on identified as BS430, grade 50B or duced. To develop maximum simu­ results obtained from base metal, and 50C) and HY80 materials were lated HAZ hardness, similar size sam­ intended to give plane strain condi­ studied as representing low alloy ma­ ples of Cu-Si, RS140, FV520S, tions during SCC testing. The tip of terials which would not develop such BS968 and HY80 steels were austeni- the pre-existing crack was aimed at high HAZ hardnesses on welding as tized at 950 C, quenched in glyc­ the supercritical (transformed) or sub- the medium carbon steels. Samples erine, and stored in liquid nitrogen critical HAZ, or at the weld metal. 10x 1 Ox 50mm (0.4x 0.4 x 2.0 in.) of prior to testing. These samples were then edge notched and fatigue cracked. Table 2 — Heat Treatment of Experimental Steels Postweld heat treatment was carried out as in Table 5. Some welds Mechanical properties were tested in the as-welded condi­ Hardness(b) Yield stress |CJ (a) 2 tion to examine the effect on SCC be­ Steel Heat treatment HV 20 N/mm ksi havior of during multipass Cu-Si 980 C, 1h OQ: T 350 C, deposition. 560 1490 220 1h AC Cu-Si 980 C, 1h OQ:T 600 C, 548 1370 200 1h AC RS140 900 C, 1h OQ: T 350 C, 590 1310 190 1h AC RS140 900 C, 1h OQ:T 600 C, 442 1150 170 1h AC FV520S 1050 C, 5 min AC: 410

Table 3 — Welding Conditions

Plate Surfacing F asses Final Weld Preheat & Base Welding Thickness, Voltage Current Speed No. of Voltag e Current Speec No. of Interpass Metal Process (£ ' mm V A mm/mir passes V A mm/min passes Temp. C Cu-Si SMA (b| 12.5 22 100 NR 1 22 100 NR 7 250 RS140 SMA(b) 25 20 110 NR 4 20 110 NR 20 250 RSI 40 EB , 20 NA NA NA NA 140*103 35x10"3 250 1 Room FV520S GTA(C) 25 11 110 150 10 12 180 100 14 Room FV520S EB 25 NA NA NA NA 140x103 35x10"3 250 1 Room Maraging SMA(d) 25 22 140 NR 4 22 140 NR 18 Room/100 Maraging GMA(e) 25 20 200 460 3 20 200 460 11 Room Maraging EB 25 NA NA NA NA 140x103 35x10"3 250 1 Room NCMV SMAlb> 12.5 20 110 NR 1 20 110 NR 7 250 NCMV GMA(f| 25 27 230 275 3 24 230 275 11 250/350

(a) SMA = Shielded metal-arc b) 3.2 mm diam GTA = Gas tungsten-arc c) 2.8 mm diam fill 3r wire GMA = Gas metal-arc d) 4 mm diam electrodes EB = Electron Beam e) 1.2 mm diam wire, Ar/2%0 2 shielding gas NA = Not applicable f) 1.6 mm wire, Ar/5%02 s lielding gas NR = Not recorded

Table 4 -- Weld Metal Compositions

Base Welding Element (wt %) metal process C S P Si Mn Ni Cr Mo V Cu Nb Ti Al B Pb Sn Co

(a) Cu-Si SMA 0.42 0.014 0.030 1.57 1.05 0.05 0.06 0.21 0.28 1.47 0.01 0.03 0.016 0.001 0.01 0.02 0.02 (a| RS140 SMA 0.29 0.010 0.024 0.08 0.25 0.38 3.16 2.16 0.16 0.08 0.005 0.01 0.010 0.001 0.01 0.01 0.01 FV520S GTA(b.O 006 0.01 0.02 0.25 0.8 5.9 14.3 1.7 — 1.6 0.25 — (d Maraging GMA > 0.03 0.006 0.004 0.10 0.05 17.7 4.04 — — 0.26 0.08 0.003 — — 8.0 d — — Maraging SMA < ' 0.03 0.013 0.002 0.15 0.08 17.9 — 3.33 — — — 0.14 0.08 — — — 8.4 NCMV SMA (a) 0.59 0.012 0.025 1.04 0.36 2.55 1.91 1.39 0.29 0.07 0.005 0.03 0.011 0.001 0.01 0.01 0.02 (C) NCMV GMA 0.07 0.01 0.01 0.5 1.35 1.3 — 0.45 0.15 — — — — — — — — (a) Weld metal analysis (c) Typical filler metal used (b) FV520B filler metal used (d) Weld metal analysis carried out by International Nickel Co., apart from carbon

WELDING RESEARCH S U P PLE M E NT I 289-s Mechanical Testing complete failure of the sample took Results of SCC Testing place between increments, and the Hardness surveys were made on all threshold stress intensity was not ex­ General Comments. Table 2 gives base metals and welded joints. Yield ceeded by any great margin. Testing base metal yield stress and hardness stress values were obtained for base was carried out until a reasonable data. The toughness and SCC results metals and weld metals. SEN sam­ indication of threshold behavior was obtained for the precracked samples ples were tested in 3 point bend in air obtained; with highly susceptible ma­ are given in Tables 6 and 7 together and the relevant fracture toughness terials, this normally entailed tests of with the average hardness of the determined. up to about 200 h, although testing region sampled in each case by the on SCC resistant materials was car­ fatigue crack. Results established as ried out for up to 2000 h. being under plane strain conditions SCC Testing according to ASTM criteria (Ref. 8) The precracked specimens were are indicated. Most toughness and Metallographic Examination loaded in 3 point bend and exposed to SCC tests were probably under plane a 3% NaCI solution, selected as repre­ Optical metallography was carried strain conditions, as indicated by a sentative of the media causing SCC of out on all welds. Carbon extraction general absence of shear lips on spec­ high strength steels (Ref. 9). The ini­ replicas and thin foils were prepared imen fracture faces, but without tially applied stress intensity was from the various regions tested, and measured yield stress data for the varied, and the time to failure examined in the electron microscope. HAZ of a weld, which is difficult to recorded to determine the critical Fractographic examination was car­ obtain, this cannot be regarded as threshold stress intensity below ried out using the scanning electron established. The stress intensity which SCC would not take place. microscope. The retained results are therefore expressed in Studies were also undertaken using contents of the experimental weld terms of a KQ parameter. KQ is the an incremental load technique, the metals were determined by x-ray dif­ critical stress intensity to cause fail­ rate of load increase being such that fraction. ure in air when it is not known whether the test is under plane strain conditions: KQSCCis the corresponding SCC parameter. While critical stress intensity criteria can be used to compare mate­ rials, the practical application of such data should make allowance for the stresses likely to arise in service. Service loading will normally be related to the base metal properties, and the KQ and KQSccvalues obtained in Table 6 were therefore converted into defect tolerance parameters (DTPs), using the general relation­ ships

DTP = [KQ/ayPor, (1a)

DTP = [KQSCC/o-v Y (1b)

whereay isthe base metal yield stress Fig. 2 — /(-preparation (double-bevel-groove) welding procedure: (a) joint preparation, (b) (Ref. 10). one side surfaced with filler metal, (c) joint completed and (d) joint machined and notched Base Metal Behavior. With the as required alloys studied and heat treatment conditions used, highest SCC resis­ tance was shown by the FV520S and Table 5 - Post Weld Heat Treatment Condit ons and Weld Metal Yield Stress 18% Ni alloys. The RS140 and NCMV steels tempered at 600 C were com­ Base metal Weld Metal parable to these materials, but heat treat Post weld Yield stress showed much greater susceptibility Steel 2 condition Weld heat treatment N/mm (ksi) following tempering at 350 C. The Cu-Si T350C SMA T 350 C, 1 h AC 1500(220) Cu-Si alloy showed high susceptibil­ Cu-Si T600C SMA T 600 C, 1 h AC ND'a' ity following heat treatment at 350 RS140 T350C SMA T350 C, 1h AC ND and 600 C, with relatively high hard­ RS140 T600C SMA None ND ness (550 HV) obtained even after the SMA T 600 C, 1 h AC 920(135) latter heat treatment. RS140 EB None ND HAZ Behavior. It is apparent from EB T 500 C, 1 h AC ND RS140 Table 6 that considerable loss in SCC FV520S S 750 C GTA 750 C, 2h AC: 0C, 2h: ND resistance can occur in the as-welded PH 450 C, 2h AC transformed HAZ of a weld. This was FV520S EB -70 C, 1h: especially the case with the medium PH 450 C, 2h AC ND carbon steels examined, when high Maraging PH 480 C SMA PH 480 C, 3h AC 2150(310) hardnesses were developed as a Maraging PH 480 C GMA PH 480 C, 3h AC 1290(190) result of the welding thermal cycle. Maraging EB PH 480 C, 3h AC ND Susceptibility to SCC was less NCMV T350C SMA T 350 C, 1h AC ND marked with the lower carbon, low NCMV GMA T 350 C, 1 h AC ND alloy and FV520S simulated HAZ NCMV T600C SMA T 600 C. 1 h AC ND samples examined. Two types of behavior within the transformed HAZ (a) ND = Not determined. were observed, as shown in Table 7.

290-slJULY 1974 Table 6 — Summary of Fracture Mechanics Results from Welds

Average D efect tolerance parameter, DTP Material Region Hardness KQ ,_ K QSCC From KQ From K QSCC Condition Weld tested (a) HV Nmm-3'2(ksiVin.) Nmm -3'2 (ksi Vin ) mm (in.) mm (in.) Base Metal Cu-Si T350C NA BM 560 1040* (30) 244* ( 7) 0.49 (0.02) 0.03 (0.001) T350C SMA Tr HAZ 485 1710 (49) 1360 (39) 1.32 (0.05) 0.83 (0.03) T350C SMA Sub HAZ 350 ND 1740 (50) ND 1.36 (0.05) T350C SMA WM 490 ND 870* (25) ND 0.34 (0.01) T350C NA Sim HAZ 709 975* (28) 244* ( 7) 0.42 (0.02) 0.03 (0.001) T600C NA BM 548 1640* (47) 418* (12) 1.44 (0.05) 0.09 (0.004) T600C SMA Tr HAZ 480 ND 1980 (59) ND 2.09 (0.08) T600C SMA WM 470 ND 930 (27) ND 0.45 (0.02) Base Metal RS 140 T350C NA BM 590 1705* (49) 418* (12) 1.69 (0.07) 0.10 (0.004) T350C SMA TR HAZ 490 3270 (107) 1170 (34) 8.07 (0.32) 0.79 (0.03) T350C SMA Sub HAZ 425 ND 2750 (79) ND 4.41 (0.17) T350C SMA WM 510 1950 (56) 1370 (39) 2.22 (0.09) 1.09 (0.04) T600C NA BM 442 4360 (133) 1740* (50) 14.4 (0.59) 2.28 (0.09) T600C SMA Tr HAZ 450 ND 1810 (52) ND 2.47 (0.10) T600C SMA Sub HAZ 405 ND 2610 (75) ND 5.15 (0.20) T600C SMA WM 540 2170 (71) 1290* (37) 4.6 (0.18) 1.25 (0.05) <(bhll T600C SMA WM 544 2430 (69) 870 (25) 4.46 (0.18) 0.57 (0.02) |b T600C EB Tr HAZ > 679 1870 (54) 418 (12) 2.64 (0.10) 0.13 (0.005) b T600C EB WM< > 641 1880 (54) 630 (18) 2.66 (0.10) 0.32 (0.01) T600C EB Tr HAZ10' 548 ND 800 (23) ND 0.48 (0.02) T600C NA Sim HAZ 692 1220 (35) 244* ( 7) 1.12 (0.04) 0.04 (0.002) Base Metal FV520S PH450 C NA BM 410 3410 (98) 2090* (60) 8.94 (0.35) 3.36 (0.13) PH450 C SMA Tr HAZ,d| 390 3480 (100) 2720 (78) 9.31 (0.37) 5.70 (0.22) PH450 C SMA Sub HAZld) 410 3130 (90) 2610 (75) 7.55 (0.30) 5.24 (0.21) PH450 C SMA WM 380 ND 2400 (69) ND 4.43 (0.17) PH450 C EB Tr HAZ(d) 405 4350 (130) 2780 (80) 15.8 (0.62) 5.95 (0.23) PH450 C EB WM 255 4480 (129) 2610 (75) 15.4 (0.61) 5.24 (0.2V PH450 C NA Sim HAZ 374 2960 (85) 1740 (50) 6.75 (0.27) 2.34 (0.09) PH480C NA BM 408 3510 (101) 2260* (65) 8.13 (0.32) 3.38 (0.13) GMA PH480 C Tr HAZ 405 3560 (105) 2780 (80) 8.81 (0.35) 5.11 (0.20) PH480 C GMA Sub HAZ 410 3480 (100) 2610 (75) 8.01 (0.32) 5.00 (0.18) PH480 C GMA WM 393 ND 2230* (64) ND 3.28 (0.13) PH480 C SMA WM 508 1740 (50) 1530* (44) 2.00 (0.08) 1.55 (0.06) PH480 C EB Tr HAZ 410 3480 (100) 2960 (88) 8.01 (0.32) 5.80 (0.23) Ph480 C EB WM 460 2090 (60) 1390 (39) 2.89 (0.11) 1.28 (0.05) Base Metal NCMV T350C NA BM 460 2020* (58) 313* ( 9) 1.54 (0.06) 0.04 (0.001) T350C GMA Tr HAZ 475 2960 (85) 696 (20) 3.30 (0.13) 0.18 (0.007) T350C GMA WM 350 2710 (78) 2440 (70) 2.76 (0.11) 2.24 (0.09) T350C SMA WM 520 1530 (44) 139 ( 4) 0.88 (0.03) 0.01 (0.001) T600C NA BM 430 3790* (109) 1960 (56) 10.1 (0.40) 2.72 (0.11) T600C SMA Tr HAZ 500 ND 1150 (33) ND 0.93 (0.04) T600C SMA WM 570 ND 770 (22) ND 0.42 (0.02) Additional Peak Hardness Studies BS968 NA Sim HAZ 450 1570 (45) 626-M18) 2.03 (0.08) 0.28 (0.01) HY80 NA Sim HAZ 418 2260 (65) 1640* (47) 4.27 (0.17) 2.24 (0.09)

(a) BM = Base Metal (b) As-welded Tr HAZ = transformed HAZ (c) Tested after T500 C 1 h AC Sub HAZ = subcritical HAZ (d) Crack extended laterally and propagated in weld metal. WM = weld metal Sim HAZ = simulated HAZ * - plane strain value

With the highly susceptible NCMV steel, the peak temperature reached in the HAZ had little effect on the Table 7 — Simulated HAZ Samples: Effect of Peak Temperature KQSCC, despite a considerable increase in prior austenite grain size at higher Peak temp. Sample Prior austenite temperature. With the less suscep­ in cycling Hardness grain size K QSCC tible HY80alloy, increasing peak tem­ Steel C HV p.m Nmm-3« (ksivin.) perature and associated grain size led to reduced SCC resistance. NCMV 1300 622 100 520 (15) NCMV 100 614 40 520 (15) In multipass welds, the SCC resis­ HY80 1300 369 100 1570 (45) tance of the reheated transformed HY80 100 368 25 2120 (61) HAZ was considerably higher than in

WELDING RESEARCH S U PPLE M E NT I 291-s twinned martensite (Fig. 3). This structure was found in the simulated HAZ samples in these alloys, and in the BS4360, although transformation ! ;.;:., :;..,.V.,./;;.^:^-,»-V". in the latter material was not uni­ «.J.v«i:-« form, and intermediate transforma­ s tion products were observed locally. HE- IKS Twinned martensite was also observed in the matching NCMV weld metal tempered at 350 C. nfP£ After heat treatment at 600 C, the medium carbon steels showed typical tempered martensite structures, with 1111 extensive carbide precipitation. Re­ heated transformed and subcritical HAZs and weld metals also showed : • - • •"••'• *'-:*:,q»*p5V *:' this structure, apart from the match­ ing NCMV weld metal tempered at - 350 C. A fine precipitate was noted in the Cu-Si steel tempered at 600 C, this being identified as copper by elec­ tron diffraction. The base metal and HAZ samples of the 18% Ni, FV520S and HY80 alloys all showed slipped marten- sites, with fine precipitation in the former two materials after postweld heat treatment. In the 18% Ni steel, the martensite appeared acicular, with occasional large martensite plates (Fig. 4). The martensite in the FV520S was in the form of largely parallel, irregular laths, while that in the HY80was acicular (Ref. 6). F/ij/. 3 — Typical microstructures of low Fig. 4 — Microstructure of maraging steel Apart from the presence of segre­ base metals, (a) Cu-Si steel. base metal (a) X500, reduced 31%, (b) gation of alloying elements, inclu­ X500, reduced 31%, (b) NCMV steel. X20,000(thin foil), reduced31% X40.000(thin foil), reduced31% sions and austenite (Table 8), the weld metal structures were largely similar to those of the base metal the as-welded condition and could ex­ the FV520S EB weld, and the Cu-Si with respect to martensite unit geom­ ceed that of the base metals. Post­ SMA weld, as both of these weld etry and precipitation, although from weld heat treatment further improved metals were appreciably softer and visual assessment, the weld metal the transformed HAZ characteristics, more resistant to SCC than the base dislocation density was rather higher. but the full base metal SCC resis­ metal. A particularly low Kosccwas ob­ In the low alloy steels and FV520S tance was obtained only when the tained for the matching NCMV SMA weld metal, segregation had little dis­ heat treatment caused the HAZ hard­ weld metal tempered at 350 C, in cernible effect on the distribution of ness to fall to the base metal level. contrast to the good resistance of the precipitation following heat reat- All sub-critical HAZs tested showed nonmatching GMA weld. ment. The maraging steel weld higher SCC resistance than did the metals showed nonuniform precipita­ parent material. Metallographic and Fractographic tion, particularly in the case of the EB Weld Metal Behavior. In most Observations weld, with the density of precipitation cases the matching composition weld being highest in the areas last to metals studied were harder than the The steels studied showed three solidify during welding. Segregation corresponding base metals, and main types of microstructure. After showed rather higher SCC suscep­ tempering at 350 C, the medium tibility. The exceptions to this were base metal contained

Table 8 - Results of Retained Austenite Determinations

Austenite Austenite Welding content % content % Material process (Range)(al (Average)

Cu-Si SMA (T 350 C) 4.3 - 8.5 5.7 RS140 SMA (T 600 C) ND

292-slJULY 1974 in the maraging steel weld metals HAZ similar to those associated with SCC Testing the high strength alloys, but in rela­ was quite marked, with reverted aus­ The samples from the high tively soft base metal. For this pur­ tenite being apparent (Fig. 5). The strength steel butt welds were FV520S GTA weld studied contained pose, SMA fillet welds were made in stressed in 3 point bend in 3% NaCI. a very mixed microstructure, with aus­ 50 mm thick steel plate to BS 4360 Two outer fiber stress levels were tenite, 8 ferrite and martensite being Grade 55E, using no preheat and aus­ studied, namely 67% and 85% of the present. Considerably more retained tenitic filler metal to base metal yield stress (ay) as austenite was evident in the FV520S give maximum HAZ hardness (Fig. representative of typical design and EB weld, despite the subzero treat­ 4a). An average HAZ hardness of 440 proof test loading respectively. The ment used. The structure of the non- HV 2/2 was obtained. fillet weld samples were stressed in matching GMA weld in NCMV mate­ bending as in Fig. 9b, at 85% and rials was predominantly a low carbon 100% o-y , the latter being chosen to tempered . reproduce the most severe loading From the results in Table 6, it will likely to be encountered in practice. be noted that there is a general trend The samples were left under con­ for susceptibility to increase with in­ stant load, generally until failure creasing material hardness. Further occurred. Test duration was noted, to­ to this, the weld and base metal sam­ gether with the location of failure ples showing highest SCC suscep­ relative to the weld. tibility contained twinned martensite, with only one exception, namely the Cu-Si steel tempered at 600 C. Results Overall, the highest SCC resistance The results obtained are summar­ was shown by the precipitation hard­ ized in Table 9, giving the applied ening systems, although the low alloy stress level, failure time and failure steels tempered at 600 C to similar location for the welds tested. DTPs hardness levels displayed broadly derived previously for the regions comparable behavior. where cracking initiated are given, Three principal fracture mechan­ together with the lowest DTP for any isms were observed in the high strength steels tested, namely inter­ granular failure with respect to prior austenite grain boundaries, and trans­ granular failure by cleavage or micro- void coalescence (Fig. 6) (Ref. 11). Highest SCC susceptibility was asso­ ciated with intergranular failure and lowest susceptibility largely with fail­ ure by microvoid coalescece, al­ though as the stress intensity under which cracking occurred was in­ creased, a transition was followed from intergranular to microvoid co­ alescence failure. Evidence of the solidification struc­ _..** .. *..™ .., • •" '••••••• •-**• iM ture was observed on some weld metal fracture faces (Fig. 7), but there •lb* was no apparent change in fracture mechanism in such areas. In the maraging steel weld metals, it was not possible to identify any particular Fig. 7 — Evidence of the solidification area of the fracture face as being structure on the fracture face of a SMA maraging steel weld metal. X60, reduced related to the reverted austenite 31% present.

Plain Weld Studies

Sample Preparation Welds were selected from the pre­ pared samples previously described to show a range of SCC behavior on the basis of fracture mechanics test­ ing, with high and low susceptibility in various weld regions. Sections were taken from these welds, and one surface dressed flush for loading \ in 3 point bend, the weld reinforce­ Fig. 6 — Representative fracture faces Rear face ment on the other surface being left showing principal failure mechanisms, (a) machined flush intact to provide a stress concentra­ Intergranular failure (Cu-Si steel). X265, Loading points for tion (Fig. 8). reduced 31%, (bj Cleavage failure 3-poini bending (maraging steel). X290, reduced 31%, (c) In addition, welds were prepared in Microvoid coalescence failure (GMA Fig. 8 — Sketch showing sample prepara­ a lower strength steel under condi­ maraging steel weld metal). X500, tion and loading points relative to weld for tions giving hardness values in the reduced31% high strength steel butt weld samples

WELDING RESEARCH S U PPLE M E NT I 293-s Table 9 — Results of SCC Tests on Plain Welds Failure region Applied Most susceptible Critical flaw size, a Postweld stress Test and cr region and DTP from equation 2, mm (in.) Base Weld heat level, duration DTP, Failure region Most susceptible Metal Process treatment %o- mm (in.) y hours mm (in.) region Cu-Si, SMA T350 C, 1 h 67 134 WM,0.34(0.01) BM.0.03 (0.001) 0.011 (0.004) 0.02 « 0.001) T350 C

RS140, SMA None 67 . WM.0.57 (0.02) WM.0.57 (0.02) 0.19 (0.007) 0.19 (0.007) 0-5 ,a T600C SMA T600C, 1h 67 2370NF NF WM.1.25 (0.05) NF 0.42 (0.02) SMA T600C, 1h 87 1483 WM.1.25 (0.05) WM.1.25 (0.05) 0.26 (0.01) 0.26 (0.01) Maraging, GMA PH480 C, 3h 67 3133NF NF WM,3.28 (0.13) NF 1.18 (0.04) PH480 C 85 5180 Tr HAZ.5.11(0.02),b> WM.3.28 (0.13) 1.06 (0.04) 0.68 (0.03) EB PH480 C, 3h 67 0.01 WM, 1.28 (0.05)(c) WM, 1.28 (0.05) 0.43 (0.017) 0.43 (0.03) NCMV, GMA T350 C, 1 h 67 552 TrHAZ.0.18 (0.007) BM,0.04 (0.001) 0.06 (0.002) 0.03 (0.001) T350C SMA T350C, 1h 67 1.3 WM.0.01 (<0.001) WM.0.01 (< 0.001) <0.01 K0.001) 0.01 (<0.001) BS4360 SMA None 85|d) 1000NF NF TrHAZ,6.81 (0.27)|e NF 1.42 (0.06) Grade 55E SMA None 100|d) 3530NF NF TRHAZ.6.81 (0.27) NF 1.02 (0.04)

(a) NF = no failure (b) Failure initiated in Tr HAZ, but propagated in WM. (cj Sample contained WM crack; applied K estimated to be between 2200 and 5200 Nmm 3/= (63-150 ksi Jin.). (d) Duplicate samples tested. (e) K QSCC determined as 104O Nmm V2 (30 ksiVln.).

region of each weld. These DTPs are acr is the critical crack depth to present work, the crack sizes involved calculated as in equation 1 b above. cause SCC, and in initiating SCC are small in relation In addition, the fracture mechanics is the complete elliptic integral, to the material thickness, and Mk has results were used to obtain an esti­ its value depending on the crack been taken for equal to Kt . mate of the size of surface defect re­ front shape. Stress concentration factors re­ quired to cause SCC under the load­ ported for butt welds vary between ing conditions used in testing the The treatment has not been rigorous about 1.2 and 3.0 (Ref. 13). At high plain weld samples. For this purpose, in view of the uncertainty associated stress levels, as used in the present the following relationship for a semi- with appropriate Mk values, and be­ studies some plastic deformation at elliptical surface flaw was used (Ref. cause the objective was primarily to the weld toe can be anticipated, and 12): (2) assess the approximate size of defect the use of stress concentration fac­ M M M M /ci. that must be considered in causing tors derived from elastic analyses K.QSC C kk s t p SCC initiation in practice. should be regarded with some cau­ where Mk is a magnification factor Equation 2 ignores any contribu­ tion, especially the higher values re­ due to the stress concentration at the tion to the operative stress of residual ported by various workers. For the weld toe; welding stresses, since in the rela­ present butt welds, a Kt value of 1.4 tively small samples used, these will was employed, after the results of Ms M, and Mp are magnification have largely been relieved during ma­ Trufyakov et al for welds stressed in factors to correct for the free chining of the specimens. bending (Ref. 14). In cases where base metal showed highest SCC sus­ surface at the crack mouth, the Mk is dependent on the crack depth free surface ahead of the crack, (Ref. 12), and as the crack size de­ ceptibility, a K, value of 1 was used to obtain a critical crack size. A number and the crack tip plastic zone re­ creases, M k approaches the conven­ spectively; tional stress concentration factor, K,, of butt weld samples failed in the a is the remotely applied stress; associated with a free surface. In the weld metal, the cracks initiating at the junction between parallel weld passes. Stress concentration factors for this situation do not appear to have been investigated, and a value of 1.4 was adopted for such spec­ Austenitic imens. A factor of 1.8 was used for welds

Extension 28mm (1.1In) diam. steel welds Strain gauge T High tensile studding 50mm (2in.) mr £lte>-f H •~. > >-o E0T TJ22>- taSF

Fig. 9 — BS 4360 Grade 55E fillet weld samples, (a) Sample geometry, (b) Loading system for SCC testing

294-slJULY 1974 the fillet welds, as found by Hayes and Maddox to be applicable to a 1 Cu-Sl* 16 RS140 31 18%Ni surface crack at the toe of a 30 deg 2 Cu-Si* 17 RS140 32 18%Ni fillet weld (Ref. 15), similar to the 3 Cu-Si 18 RS140 33 18%Ni present sample geometry. 4 Cu-Sl 19 FV520S 34 NCMV* For the purpose of Table 9, it has 5 Cu-Si* 20 FV520S 35 NCMV* been assumed that the initiating 6 Cu-Si 21 FV520S 36 NCMV* defect has a depth/length ratio 7 Cu-Si 22 FV520S 37 NCMV 8 Cu-Si 23 FV520S 38 NCMV (acr/2c) of 0.1, and as noted above, 9 RS140 24 FV520S 39 A652 that a is small. Under these condi­ cr 10 RS140 25 FV520S 40 BS4360* tions, the term M M,M /ci becomes s p WOO 11 RS140 26 FV520S 41 BS4360 1.04 (Ref. 12). 12 RS140* 27 18 Wl 42 En 16* With substitutions for the various 13 RS140* 28 18%Ni 43 En 20B* magnification factors as indicated, 14 RS140* 29 18%Ni 44 5LX60 „, 2s equation 2 was used to calculate o 15 RS140 30 18%Ni 45 5LX60 ' values of acr for the weld areas ~~mo where SCC occurred during test for Open symbols denote other micro- the area defined as most susceptible structures. by tests on precracked samples. **M .40°° 33A 023° "cAo Closed symbols denote twinned (0'1)- These values are given in Table 9. 37 g A^? martensite. A; 10 °4 33A —^ O Base Metal Discussion 7ST n HAZ &I6 A 38 &6 A Weld m eta I. Microstructure and SCC * Intergranular failure (001) Susceptibility Relationship • 35 •19 The dominant role of microstruc­ 0-1 2o •K • 42 ture in determining the mechanism • >3 by which SCC propagation takes • 34 place and the associated material sus­ • ; •« (0001)- ceptibility is well indicated by the 001 l.*i_J* 36 results in Table 6. Overall, as shown 200 300 4-00 500 600 700 800 by various workers, susceptibility can Hardness HY. be expected to increase with in­ creasing material hardness. This is Fig. 10 — Relationship between hardness and DTP for present and other weld samples illustrated by Fig. 10, which is a sum­ mary diagram prepared to include the present results and those of other investigations at The Welding Insti­ tute. Further to the effect of hardness, which cleavage can occur. Failure by pendent on suppression of intergran­ there is a sharp division between the cleavage has been shown to be ular and cleavage failure (Ref. 11). high susceptibility associated with favored by factors such as the With lower strength steels than structures causing intergranular fail­ presence of suitable crack nucleation primarily considered in the present ure, and other structures. It has been sites and a high material yield stress work, bainitic type structures may be suggested previously that this failure (Ref. 17). The latter is consistent with encountered. While these have not mechanism results from restriction of the general finding that SCC suscep­ been investigated exhaustively from intragranular dislocation movement, tibility tends to increase with in­ the SCC viewpoint, general com­ so that the applied strain is effectively creasing hardness. In respect to the ments can be made from studies concentrated at the prior austenite former, the results in Table 6 indicate on hydrogen induced HAZ cracking grain boundaries (Ref. 11). Disloca­ that precipitation hardened systems on welding. Again, it is probable that tions in twinned martensite are will afford higher SCC resistance reduction in hardness as far as pos­ strongly pinned, and this phase in than medium carbon alloys tempered sible is to be preferred, but at a given particular has a highly deleterious to similar strength levels. This is hardness level, a number of different effect on SCC behavior (Refs. 6, 16). likely to be due to cleavage nucleation types of bainitic or other non-marten- For optimum SCC performance, being facilitated in medium carbon sitic structures can arise. The major compositions and heat treatment con­ steels by the large carbide particles point is that structures containing ditions causing twinned martensite formed during tempering. In general, sheaves of parallel bainite plates should be avoided. Intergranular therefore, low carbon slipped should be avoided, since the resis­ failure can also arise in tempered massive and acicular can tance to cleavage may be relatively martensite at high hardness levels, be expected to be more resistant to low, causing a loss of both toughness as shown by the Cu-Si tempered at SCC than tempered medium carbon per se (Ref. 19) and resistance to 600 C, and it would appear that the martensites. hydrogen cracking (Ref. 10). Further, hardness of structures other than it is desirable to avoid transformation In the softer materials studied, SCC twinned martensite should be kept to structures containing intergranular took place by microvoid coalescence. below about 550 HV to obtain good pro-eutectoid ferrite surrounding an Failure by this mechanism has been SCC resistance in neutral chloride acicular intragranular structure. This found to be dependent on the density media. The relationship between ma­ structure results in applied strain and distribution of nonmetallic inclu­ terial hardness and risk of SCC in ser­ being concentrated at grain boun­ sions, resistance to crack propagation vice is considered further in the fol­ daries analogous to the situation with increasing as the inclusion popula­ lowing. twinned martensite, and a microscop­ tion is reduced (Ref. 18). Thus, the ically "ductile" intergranular form of risk of SCC will be reduced by the use At hardness levels below 550 HV, cracking can result, with a marked in­ of cleaner material, but it must be and in the absence of twinned mar­ crease in susceptibility to hydrogen appreciated that practical benefits tensite, susceptibility to SCC will induced failure (Ref. 11). depend largely on the ease with from cleaner steels will largely be de­

W ELDING RESEARCH S U PPLE M E NT | 295-s Effects of Welding ture is less affected by cooling rate, present maraging steel SMA weld heat input should be minimized to metal should have a much lower KQScc reduce the extent to which grain than the other maraging steel weld General Comments. When con­ coarsening occurs, and the width of metals, (Table 9), and this was not the sidering the probable effects of a the coarse grain HAZ region In both case. It is therefore probable that Toy welding operation on the risk of SCC situations, preheat and postweld and Philips' observations are an of a fabrication, attention should be conditions should be controlled as effect of SCC rather than a cause, paid to two aspects of behavior. First, necessary to avoid hydrogen induced and austenite should be regarded as the susceptibility of individual weld cracking during welding. beneficial in improving SCC resis­ areas must be evaluated so that the In multipass welds, the SCC be­ tance. weak link can be identified. The likeli­ havior of the transformed HAZ is im­ It might be expected that local hood of cracking in service can then proved by the tempering effect of sub­ compositional variations arising from be assessed, and suitable remedial sequent welding passes. The prac­ segregation would affect weld metal measures undertaken. From the tical significance of this will depend SCC behavior. To some extent, this present results, the performance of on the heat input and thermal cycle appears to be true in that evidence of welded high strength steels can be involved in relation to the ageing the solidification structure was ob­ expected to depend principally on the characteristics of the steel, and the served on some fracture faces. How­ SCC resistance obtained in the trans­ thermal cycle during welding may not ever the association between fracture formed HAZ, and in the weld metal if always be adequate for HAZ resis­ path and segregation was not this is of matching composition. Thus, tance to equal that of the base metal. marked, and segregation does not particular consideration should be It is however noteworthy that in seem to play a major role in deter­ given to these areas in any material cases where the base metal was sus­ mining SCC resistance. Segregation evaluation study. ceptible to intergranular failure, the may be of more significance in pre­ Second, recognition must be made reheated transformed HAZ failed by cipitation hardening systems than in of the probable failure path through cleavage, with correspondingly high low alloy steels, as indicated by the SCC resistance. The tempering effect the welded assembly. For example, in nonuniform precipitation observed in of welding is even more marked in multipass welds, softening of under­ the maraging steel EB weld metal. the subcritical HAZ regions, as in all lying material will occur and SCC This weld metal gave low K and KQSCC multipass welds studied, the subcrit­ Q propagation will take place into a re­ values relative to the GMA weld, for ical HAZ showed higher SCC resis­ gion of reduced susceptibility, when example. It has been reported pre­ the crack may arrest. A decision will tance than the base metal. viously that EB welds in this type of then be necessary as to whether the Weld Metal Befiavior. Precisely steel have poor fracture toughness, crack can be tolerated, based on its similar comments apply with respect although the reason has not been probable maximum depth, or whether to the effect of microstructure and clear (Ref. 24). However, segregation it will lead to failure of the fabrication hardness on SCC behavior for weld of hardening elements during solid­ by another mechanism such as metals as for HAZs and base metals. ification will affect the local response fatigue. The low alloy weld metals studied of the weld metal to the precipitation HAZ Behavior. The possible dele­ were generally harder than the base hardening heat treatment, and a dete­ terious effect of microstructural metals and HAZs, despite the multi­ rioration in mechanical properties changes during welding is high­ pass deposition and postweld heat would not be unexpected. Such an lighted by the extreme susceptibility treatment. This increased hardness is effect will be particularly marked in of the hard as-welded transformed probably largely responsible for the EB weld metal, since the high solid­ HAZ structures developed in the weld metals showing rather higher ification rates experienced with this medium carbon steels examined. SCC susceptibility than base metals, process will promote segregation of When the transformed HAZ is sensi­ as indicated in Table 6. alloying elements. tive in intergranular failure, the peak Greatest microstructural differ­ The presence of deoxidation prod­ temperature experienced seems to ences between weld metal and base ucts and other inclusions in weld have no significant effect on suscep­ metal arose with the precipitation metals will reduce SCC resistance to tibility, so that failure is equally likely hardening alloys. The FV520S weld some extent, and this may have con­ in both the coarse and fine grain size metals differed most with their high tributed to the low weld metal K QScc regions of the HAZ (Ref. 21). In retained austenite contents and the contrast, susceptibility to cleavage values observed. However, as pre­ presence of 8 ferrite. Austenite is viously pointed out, inclusions are of failure is affected by peak temper­ only slightly sensitive to hydrogen ature, being highest in the coarse most importance in affecting crack embrittlement, and the high SCC propagation by microvoid coales­ grained HAZ close to the fusion boun­ resistance of these weld metals may dary. Thus, even if weld dressing is cence. Most of the present weld well be due in part to the retained aus­ metals failed largely by cleavage. employed to remove the stress tenite, especially in the soft EB weld concentration at the weld toe, SCC is Hence although a reduction in inclu­ metal. Certainly, the austenite would sion content is presumably of benefit most likely to occur in the high tem­ tend to retard crack propagation, and with respect to weld metal SCC be­ perature HAZ, and when carrying out may be of further benefit by acting as havior, inclusions are likely to be of material evaluation studies, particular a "sink" for hydrogen formed during secondary importance compared to attention should be paid to this region corrosion. Toy and Philips showed microstructure. (Ref. 22). When failure occurs by that hydrogen may be picked up by cleavage, changes in welding condi­ reverted austenite in maraging steel tions will have one of two effects weld metals, with an associated (Ref. 22). With less hardenable alloys, strain-induced transformation to mar­ Postweld Heat Treatment an increase in heat input will lead to tensite (Ref. 23), and they suggested transformation on cooling to a rela­ A salient feature of the present re­ that the SCC susceptibility of such sults is that given postweld heat treat­ tively soft microstructure with low weld metals is due to the austenite susceptibility to hydrogen embrittle­ ment similar to that applied to base present. This is unlikely to be entirely metal, there was no loss in SCC resis­ ment. Thus, increasing heat input correct, since base metal containing will increase the SCC resistance in tance in the weld area, provided only negligible amounts of austenite suf­ that the final hardness was near the the transformed HAZ. With more fers SCC, while if austenite were hardenable steels, where HAZ struc­ base metal level. This proviso was not dominant in causing failure, the fulfilled with the low alloy weld

296-slJULY 1974 metals, nor with the transformed data derived from laboratory studies practical purposes. In the first in­ HAZ of the NCMV steel tempered at on precracked specimens may be stance, this will depend on the limits 600 C. Nonetheless, the results used to assess the probable service of detection afforded by nondestruc­ show that postweld heat treatment performance of a welded joint. The tive testing (NDT). However, from the should be regarded as essential to ob­ principal limitation to this is likely to magnitude of the flaw sizes derived, tain maximum SCC resistance from arise with high alloy materials, where all materials studied must be re­ welded joints in high strength steels. marked pitting corrosion in the weld garded as potentially susceptible to area may occur, with an associated SCC with normal NDT methods. risk of failure (Ref. 6). Changes in the values of stress mag­ Correlation Between Fracture nification factor and flaw shape Mechanics Data and Plain Effect of Weld Toe Stress parameter used in equation 2 to allow Weldment Behavior Concentration on SCC Initiation for different practical situations are unlikely to more than double the crit­ The results in Table 9 show the The practical effects of static and ical flaw sizes given in Table 9, even plain welds to have behaved largely dynamic strain on the corrosion of fer­ accepting that precise analyses are as predicted from consideration of ritic steels have not been unambigu­ not presently available for small sur­ the fracture mechanics data. The ously defined. However, even though face flaws. Thus it will not always be welds found to have the longest lives the corrosion rate may be increased possible to guarantee avoidance of were those with the highest DTPs, in some conditions by tensile strain, pre-existing defects of critical size by while failure generally took place in as will be experienced at a weld toe, NDT. Even with sufficiently sensitive the area established as most suscep­ available data suggest that strain will NDT, it is not possible to be specific tible from the precracked specimen have little effect on the long term cor­ regarding an acceptable DTP for a studies. rosion of low alloy steels in neutral given service duty, since this will In the NCMV GMA and Cu-Si SMA chloride media (Ref. 26). This is cer­ depend on the service environment welds, the lowest DTP was that of the tainly indicated by the lack of prefer­ with respect both to the amount of base metal, yet SCC took place in the ential weld toe attack with the hydrogen taken up by the steel, and to weld region. This variation however, present low alloy materials. The effec­ the likelihood of SCC initiation by is not surprising in view of the very tive anode area/cathode area ratio of pitting corrosion. low DTPs, for these steels; minor sur­ the weld toe and base metal may Nonetheless, it should be possible face irregularities will be sufficient to affect the situation to some extent, to define acceptable materials for dif­ cause SCC, and initiation from the but it is considered that the present ferent service conditions by obtaining rough, as deposited, weld metal must samples were sufficiently large for an empirical correlation between lab­ be regarded as at least as probable as their anode area/cathode area ratio oratory test data and direct practical initiation in the base metal. to adequately reproduce the practical experience. It has been suggested Severe localized weld toe pitting case. that consideration should be given to was noted with the maraging steel It should nonetheless be appre­ the concept of a critical hardness GMA weld stressed at 85% oy , SCC ciated that the pitting corrosion in­ below which SCC does not occur in a taking place from an elongated pit of volved in SCC initiation is to some given medium (Ref. 6). The approach about 1.5 mm (0.06 in.) depth. This pit degree autocatalytic, and as a general is analogous to that used in designing depth would be expected to cause fail­ principle, any local increase in corro­ welding procedures for structural ure, being greater than the critical sion rate due to tensile strain at a sur­ steels, in giving a parameter which flaw sizes derived for the weld metal face irregularity will tend to be con­ effectively describes the situation and HAZ of 0.68 and 0.06 mm (0.03 centrated. Unless stifled by corrosion where the risk of cracking is con­ and 0.04 in.) respectively, although product, therefore, the development trolled to a negligibly low level under the stress intensity associated with a of a pit suitable for SCC initiation is the combined effect of operative pit will presumably be somewhat less more likely at a weld toe than at a stress, size of any discontinuity, than with the sharp planar defect plane surface. The practical signif­ hydrogen content and microstructure geometry assumed in equation 2. No icance of this effect will depend on likely to be encountered in practice. comparable localized attack was the corrosion conditions engendered Without more data on the hydrogen found on any alloy steel tested, only by the material/microstructure/en- producing potential of different media general pitting being observed. vironment combination being consid­ than are presently available, very few The RS140 SMA weld which was ered. As noted earlier, such local points can be located on such a crit­ not heat treated after welding beyond attack was observed in a maraging ical hardness scale. However, it is maintenance of the 250 C preheat steel GMA weld, and it is probable well established that sour oil or gas showed failure initiating between the that the effect will be of greater signif­ media represent a severe case, since two final passes of filler metal. These icance with high alloy materials the sulphide scale formed by corro­ had not benefited from tempering which may be sensitive to pitting sion tends to poison the hydrogen during multipass deposition, and attack than with low alloy steels. It is evolution reaction and cause high ma­ were considerably harder (~620 HV) not possible to make any quantitative terial hydrogen contents (Ref. 5). In than the balance of the weld metal prediction at the present time, but it such duties, failure can take place at (~540 HV), and thus presumably can be expected that SCC resistance above about 240 HV, and in view of more susceptible to SCC. This demon­ in high alloy weldments can be im­ the severity of the environment this strates the danger of retaining hard proved by dressing the weld to re­ hardness can, for most practical pur­ untempered material in the weld move the reinforcement and obtain a poses, be regarded as a lower boun­ area, since despite the relatively high plane surface. dary with respect to SCC arising from DTP for the reheated weld metal, fail­ hydrogen embrittlement. ure occurred after only 0.5 h. Practical Risk of SCC Further, Bloom showed that in ma­ Overall, it may therefore be con­ rine conditions, martensitic stainless cluded that the results shown in It will be appreciated from Tables 6 steels are unlikely to fail by SCC at Table 9 confirm the validity and use­ and 9 that susceptibility of high below 400 HV, while in a salt fog or fulness of the fracture mechanics strength steels to SCC is associated industrial atmosphere, a hardness of approach adopted to define relative with relatively low DTPs and critical 450 HV could be tolerated (Ref. 27). material SCC behavior, and the flaw sizes, and the question arises as These materials may be considered a effects of welding. Thus, comparative to what level of DTP is acceptable for severe case insofar as they are likely

WELDING RESEARCH S U P P LE M E NT I 297-s to suffer marked pitting corrosion, ularfailure. sitic Steels", Corrosion, 25 (9), 1969, pp and it is probable that susceptibility at 5. With the particular alloys 353-358. hardnesses below 400 HV will not be studied, highest SCC resistance was 4. Gooch, T. G., McKeown, D. and significant in marine conditions. Cer­ found with low carbon precipitation Willingham, D., "Stress Corrosion of Welded Materials: Evaluation and Con­ tainly the results in Table 9 are in ac­ hardening materials. If tempered at trol", Metal Construction and British Weld­ high temperature to similar hardness cordance with this, although some ing Journal, 1 (10), 1969, pp 469-475. caution is necessary, in that at 400 levels, medium carbon low alloy 5. Phelps, E. H., "A Review of the HV, the steels may be sensitive to grades will be generally comparable Stress Corrosion Behavior of Steels with defects about 1 mm deep (cf Fig. 10 with precipitation hardening High Yield Strength", Proc. Conf. "Funda­ and Table 6): the sustained rate of systems, and the former materials mental Aspects of Stress Corrosion Crack­ pitting of low alloy steels in seawater may be preferable if marked pitting ing". Ohio State University, NACE, 1967, can approach 0.7 mm/yr. (Ref. 28), attack is likely in service. pp 398-410. and the risk of long term SCC initia­ 6. The presence of alloy element 6. Gooch, T. G., "Stress Corrosion of Welded High Strength Steels", Welding tion must not be overlooked. The crit­ segregation and inclusions may re­ Research International, 3 (2), 1 973, pp 17- ical figure of 400 HV is almost certain­ duce the SCC resistance of high 43. ly conservative for typical offshore ap­ strength steel weld metal, but such 7. Brown, B. F., "The Application of plications involving C-Mn and similar effects are of secondary importance Fracture Mechanics to Stress Corrosion low alloy structural steels, where the compared with that of microstructure. Cracking". Metallurgical Review, No. 129, service stress levels will be lower The adverse effect of segregation 1969. than with high strength steels. This is may be most marked in high alloy pre­ 8. ASTM STP "FractureToughness Test­ indicated by the absence of cracking cipitation hardening steels. ing and Applications" No. 381, 1 965, and in the BS 4360 fillet welds tested 7. The stress concentration at a "Plane Strain Crack Toughness Testing of High Strength Metallic Materials". No. (Table 9), and by the fact that the weld toe does not significantly affect 410, 1967. Welding Institute investigators have the corrosion processes causing SCC 9. Gooch, T. G., "Stress Corrosion Test­ not observed any service SCC failure initiation in low alloy steels in neutral ing of Welded Joints", AGARD Confer­ in structural steelwork exposed to ma­ chloride media. Such stress concen­ ence Procedings No. 98. "Specialists rine conditions, even though hard­ trations can, however, localize attack Meeting on Stress Corrosion Testing nesses approaching 450 HV must in high alloy materials and promote Methods", NATO, 1 972, Paper 10. have existed in some situations. It is SCC initiation at the weld toe. Thus, 10. Burdekin, F. M. and Dawes, M. G., therefore considered that a critical while fracture mechanics tests can "Practical Use of Linear Elastic and Gen­ hardness of 450 HV should be regard­ be used to give comparative data on eral Yielding Fracture Mechanics with Par­ ticular Reference to Pressure Vessels", ed as applicable to offshore structural resistance to crack propagation, Proc. Conf. "Application of Fracture steelwork. Service failures have been conventional tests on uncracked samples are essential to define the Mechanics to Pressure Vessel Tech­ observed in rather higher strength nology" I Mech. E., London, 1971, pp 28- risk of SCC initiation taking place by a alloys than C-Mn structural steels at 37. corrosion mechanism in service. above 400 HV when cathodically pro­ 11. Gooch, T. G. and Cane, M. W. F., tected by zinc, and the possible 8. Consideration has been given to "Fractographic Observations of Hydrogen adverse effect of cathodic polariza­ the risk of SCC in different environ­ Induced Cracking In Steel", Proc. Conf. tion and concomitant hydrogen for­ mental conditions. Other work has "The Practical Implications of Fracture mation upon data shown in Fig. 10 indicated that SCC in marine condi­ Mechanisms", Newcastle, Institution of must be borne in mind. tions will not occur in material softer Metallurgists, 1 973, pp 95-102. than 400 HV: the present results on 12. Maddox, S. J. "An Analysis of Fatigue Cracks in Fillet Welded Joints", to high strength steels are in accor­ be published in International Journal of Conclusions dance with this finding. However, it is Fracture. considered that this critical hardness 1. The SCC susceptibility of high 13. Gurney, T. R. Fatigue of Welded is conservative for structural steels strength steels may be greatly in­ Structures, Cambridge University Press, where the service stresses are lower Cambridge 1968. creased by welding, especially if than with high strength alloys: test 14. Trufyakov, V. I., Osaulenko, L. L. and marked hardening occurs in the HAZ results and practical experience indi­ Koryagin, "Stress Concentration in Butt or weld metal as a result of the weld cate that SCC is unlikely in structural Joints", Automatic Welding, 19 (10), thermal cycle. steels at below 450 HV when 1966, pp 18-21. 2. If postweld heat treatment is ap­ exposed to marine conditions. 15. Hayes, D. J. and Maddox, S. J., "The plied so that the hardness of the weld Stress Intensity Factor of a Crack at the Toe of a Fillet Weld", The Weld Inst. Res. area is restored to that of base metal, A cknowledgements there will be no significant loss in Bulletin, 13 (1), 1 972, pp 1 5-1 7. The work carried out was jointly funded 16. Snape, E., "Sulphide Stress Corro­ SCC resistance of the assembly. If by The Welding Institute and RARDE. sion of Some Medium and Low Alloy such heat treatment is not applicable, Acknowledgement is also made to RARDE Steels", Corrosion, 23 (6), 1 967, pp 1 54- increased susceptibility must be an­ for supplying the experimental steels. 172. ticipated in the transformed HAZ, and The writer thanks Mr. G. Regelous for 17. Knott, J. F. "Second-Phase Particles in matching composition weld metal. his organization of the experimental work, and the Toughness of Structural Steels", 3. The susceptibility of welded and Messrs. B. Ginn, D. Boud, R. Gant and Proc. Conf. "Effect of Second Phase Par­ high strength steels to SCC is primar­ M. Catling for their experimental assis­ ticles on the Mechanical Properties of ily dependent on the microstructure tance. Steel, Scarborough, I.S.I., 1 971, pp 44-53. 18. Steele, A. C. "The Effects of Sulphur produced in the different weld areas. References and Phosphorus on the toughness of Mild 1. Evans, U. R., The Corrosion And Susceptibility generally increases Steel Weld Metal", Welding Research Oxidation of Metals, Edward Arnold (Pub­ with increasing hardness, while International, 2 (3), 1971, pp 37-76. lishers) Ltd., London, 1 960. the presence of twinned martensite 19. Dolby, R. E. and Knott, J. F. "Tough­ 2. Bhatt, H. J. and Phelps, E. H., "Effect in particular is deleterious. ness of Martensitic and Martensitic-Bain- of Solution pH on the Mechanism of 4. SCC can take place intergran- itic Microstructures with Particular Refer­ Stress Corrosion Cracking of a Martensitic ence to Heat Affected Zones in Welded ularly with respect to prior austenite Stainless Steel", Corrosion, 17, (9), 1961, Low-Alloy Steels", J. and Steel Inst., grain boundaries, or in a transgran- 430t-434t. 210(11), 1972, pp 857-865. granular manner by cleavage or 3. Barth, C. F. Steigerwald, E. A., and microvoid coalescence. Highest Troiano, A. R., "Hydrogen Permeability susceptibility arises with intergran- and Delayed Failure of Polarized Marten­ (Continued on p. 306-s)

298-slJULY 1974