• Like
Fatigue behavior of welded steel
Upcoming SlideShare
Loading in...5

Thanks for flagging this SlideShare!

Oops! An error has occurred.

Fatigue behavior of welded steel



  • Full Name Full Name Comment goes here.
    Are you sure you want to
    Your message goes here
    Be the first to comment
    Be the first to like this
No Downloads


Total Views
On SlideShare
From Embeds
Number of Embeds



Embeds 0

No embeds

Report content

Flagged as inappropriate Flag as inappropriate
Flag as inappropriate

Select your reason for flagging this presentation as inappropriate.

    No notes for slide


  • 1. Fatigue Behavior of Welded SteelJoints in Air and SeawaterG.H.G. Vaessen, Metaalinstituut-TNOJ. de Back, Delft tJ. of TechnologyJ.L. van Leeuwen, Delft U. of TechnologySummaryConstant-amplitude fatigue tests were carried out in air expensive and time consuming; therefore, it is possibleand artificial seawater on T -shaped welded steel joints. to perform only a very limited number of such tests. TheThe main objective of the test program was to compare vast majority of corrosion fatigue tests has to be per-the behavior of welded steel joints in air and seawater. In formed with laboratory-size specimens under conditionsaddition, the effects of weld-profile/weld-finishing, simulating the conditions for offshore structures in thestress relieving, and stress ratio were investigated. The North Sea.influence of cathodic protection and overprotection on This paper describes the results of fatigue tests withcorrosion fatigue behavior also was studied. welded joints in air and artificial seawater. The tests were carried out on nonload-carrying welded joints. TheIntroduction parameters varied in this investigation are environment The recent use of offshore structures in deeper and (air and artificial seawater), stress ratio R (R=O.l and rougher areas (e.g., the northern part of the North Sea), - 1), stress relief treatment [postweld heat treatment where there are lower air and sea temperatures, high (PWHT)], weld profile, weld finishing [as-welded, winds, and severe marine conditions, has increased the grinding, and tungsten inert gas (TIG) and plasma dress- probability of fatigue failure. For safe design of offshore ings}, and cathodic protection and overprotection. The structures in such areas, knowledge of the corrosion work forms part of a large research program on the cor- fatigue behavior of steels and welded joints under rosion fatigue behavior of welded steel joints. Some representative conditions is of vital importance. The vast preliminary tests results are given. majority of fatigue data on which current fatigue design rules are based have been derived from tests with Material laboratory-size specimens in air. -5 The material used for the fabrication of the test The corrosion effect normally is catered for, in the specimens was in accordance with Euronorm Fe 510 (BS case of offshore structures, by extrapolation of the 4360 Grade D steel). The plate steel thicknesses were 40 design S-N curves beyond the fatigue limit. 6 Otherwise, and 70 mm. The chemical composition and mechanical the same basic concepts as for "in-air" structures are ap- properties of the plates are described in Tables 1 and 2. plied-"stress range philosophy" and Miners rule. The The microstructure of the steel contained about 20 to justification for this approach is a limited number of 25% pearlite (grain ~ize of 10.5 to 11 /tm according to laboratory tests in simulated seawater, very limited ex- ASTM .specification). The impurity content of the steel perience from laboratory test results that corrosion pro- was found to be low. The analysis and properties of the tection is effective in delaying corrosion fatigue failures, steel are within specification values (Euronorm Fe and service experience (which at present does not in- 510).7.elude the North Sea). To obtain fatigue data appropriate to steel offshore Experimental Work structures, it is desirable to obtain more data about the Test Specimens (Design and Fabrication) corrosion fatigue behavior of tubular joints under Fig. 1 illustrates the specimen configurations and weld simulated North Sea conditions. However, these tests are proftles. Each specimen was fabricated such mat the0149-2136l8210002-8621 $00.25 longitudinal direction of the specimen was alignedCopyright 1979 Offshore Technology Conference parallel to the rolli!lg direction of the plate. Manual440 JOURNAL OF PETROLEUM TECHNOLOGY
  • 2. TABLE 1-CHEMICAL ANALYSIS OF PARENT STEEL C Mn P S Si AI Cu Sn Cr Ni Nb 0.17 1.44 0.018 0.004 0.35 40 0.019 0.006 0.039 0.017 0.046 TABLE 2-MECHANICAL PROPERTIES OF PARENT STEEL Elongation R • (d p 5)· Zt Charpy V Energies· (N/m~2)*. (%)** (%) at -3O D C (J) •• 40-mm steel plate 551 31 26 to 46 160 70-mm steel plate 530 30 28 to 62 155 -Mean value . • "Transverse to rolling direction. tThickness direction; range. TABLE 3-SPECIFICATION OF THE TlG AND PLASMA DRESSING METHODS TIG Dressing Plasma Dressing (vertical position, (vertical position, downhill direction) downhill direction) thoriated tungsten nozzle (¢ = 3.2 mm) electrode (¢=3.2 mm) plasma gas (95% argon and 5% hydrogen) cup diameter = 11 mm flow (plasma gas)=0.75 dm 3 /min shielding gas argon shielding gas argon = 10 dm 3/min heat input = 13.1 kJ/cm heat input = 21.5 and (first and second run) 18.1 kJ/cm current = 21 0 A current = 130A voltage = 12.5 V voltage = 28 to 30 V 320 t I t 920 tJ i- ! I I" I II 11 218 520 t t 1. 1520 IJ I II Fig. 2-lIIustration of weld finishing by means of toe burr grinding.~~ lSwldtd illProwed profile Fig. 1-Details of fatigue test specimens and schematic il- lustration of fatigue tests.FEBRUARY 1982 441
  • 3. metal arc welding (vertical, uphill), using covered basic quency of 2 to 5 Hz; the tests in artificial aerated electrodes (4)=3t,4 mm for the root runs and 4>=4 mm seawater were performed at a loading frequency of 0.2 for the following runs) according to AWS Code E 7016, Hz. was employed to fabricate the specimens and a For defining the applied stress, strain gauges were preheat/interpass temperature of 100 to 150°C was used. bonded to the test specimens at a distance of about 25 Three specimens were welded simultaneously without mm from each weld toe. In addition, for obtaining infor- restraint. The effect of a steep and a smooth weld profile mation about crack initiation, strain gauges were bonded on fatigue behavior has been investigated for the 40-mm to some test specimens in the immediate vicinity (about 2 specimens. mm) of the weld toe at both sides of the gusset. Failure Stress relieving of the test specimens was performed was defined to have occurred in a test whenever the by heating the specimens to 580±20°C at a mean rate of crack reached half the specimen thickness. 2oo°C/hr (above 300°C), holding 2.5 hrs, and furnace At this stage, the fatigue crack growth rate was so fast cooling at a rate of l00°C/hr to 300°C. followed by air that the· number of cycles remaining to complete separa- cooling. tion would have been very small compared with the number of cycles already applied. Test Procedure After test completion, some specimens were examined The specimens were tested in four-point bending under for a detailed failure analysis. The corrosion fatigue ex- sinusoidal constant amplitude loading. Seven loading periments were carried out by pumping artificial machines under servovalve control in an MTS ™ or seawater (according to ASTM Sr-cification D 1141-52) Schenck ™ (250-kN machine) closed-loop testing from a reservoir (about 500 dm ) through a transparent system were used. The layouts of these machines were box (about 50 dm 3 ) surrounding the central part of the similar. The force exerted by the actuator was transferred specimens. The seawater flow rate was of the order of 1 to the specimen so that it was loaded in four-point bend- dm 3 /min. ing (Fig. 1). A fresh mix of seawater was substituted periodically The 40-mm test specimens were tested at a stress ratio (about every 3 months). The temperature (20± 10c) and R;;=O.I; the 70-mm test specimens were tested at R=O.l the pH (8 to 8.3) of the seawater were controlled and and - 1. The tests in air were performed at a loading fre- monitored continuously. The chemical composition .nt 30 2 ,,n &" . . . . . . ,. . I., . OM.lair ~ • N.I SNWIbY t 3 on 2 5n 8 f,,11 penetrlUoa wlel. tdtl tit atr 0 .... 1 .. o R-1 MI.. 200 ( 70.) 0.--1 atr "E 2 00 ( .J 0 .1 NIT 01-.. 1 NIT E • Roo .. 1 • ...,.url. l...c: 150 .. .. c: 150 !:!: 1,00 ~ g : 100 -=;c:!• .. ~ Ii c: 50 I 5n 7 10- 10· 10 Number of cycle. -t Number 01 cycl •• -t Fig. 3-lnfluence of stress ratio (as welded). Fig. 5-lnfluence of stress relieving (air).t 3 00 2,!II> & fun. penatrltton ...1•• NfT 0 W ..1.. • .... l ""w t 3 on 25n & . . . ,..-.. . ful1 ....ttltiOll ..ld. • • M.IJ111 a.-1 .,.. 200 ( .J 0.-1 .,. .....1 HWttll" N 200 V J1) . . . . . INIT E E • .. -1 Nfll l.•c::! 150 .. • " :! 150 "-: 1OIl : 100!• . ~ ........... ........ ....... iE ! ~ 50 50 . 10· Number of cycle. -t 10 Number 01 cycle. -t - Fig. 4-lnfluence of stress ratio (stress relieved). Fig. 6-lnfluence of stress relieving (seawater). 442 JOURNAL OF PETROLEUM TECHNOLOGY
  • 4. (salinity, bicarbonate ion concentration, chlorinity, etc.) Figs. 3 and 4 show, for the 70-mm T-shaped.was controlled periodically. The seawater tests were per- specimens as welded (Fig. 3) and stress relieved (Fig . fonned at free corrosion potential. 4), the effect of seawater on fatigue lifetime. The test specimens have been tested at two different stress ratios Fatigue Strength Improvement Techniques (viz., R=O.1 and -1). Fatigue improvement techniques (grinding and TIG and The influence of stress relieving on fatigue lifetime is plasma dressings) were applied to some T -shaped shown in Figs. 5 and 6. The effect of stress relieving has 40-mm test specimens. been investigated in air and seawater and at two different Toe burr grinding is achieved by grinding the weld toe stress ratios (R=O.l and -1). with the burr tool illustrated in Fig. 2. This technique ef- The effect of weld profile and weld finishing (grinding fectively removes the undercut and defects that occur at and TIG and plasma dressings) on fatigue endurance is the weld toe region. The average depth of grinding at the shown in Figs. 7 and 8. weld toe was approximately 0.6 mm. The manual TIG- In Fig. 9, the test results of cathodic overprotection are dressing/plasma-dressing procedure involved remelting given. of the weld toe by one· single weld run, followed The fatigue data obtained in air and seawater have [because of an unacceptable hardness in the heat-affected .been compared with relevant fatigue design curves for zone (HAZ) after the first weld run] by a second weld the joint geometry under consideration (Class F). JO The run with nearly the same heat input. 8,9 The second weld mean design curve and the most commonly used design run was 2.5 to 3 mm from the toe of the first run in the curve (the mean-minus-two-standard-deviations curve) direction of the original groove weld. Table 3 gives some have been used (Fig. 10). In comparing the data ob- details of the applied dressing techniques. tained, remember that the design curves are intended to be applied primarily to axially loaded joints and that the Results present results have been obtained under four-point Because the stress range generally is considered the bending. Results obtained under bending are expected to primary variable detennining the fatigue life of welded correspond to longer lives than those obtained under ax- joints, the results of this investigation have been ex- ial loading. pressed in tenns of the applied stress range against the During the fatigue experiments, the cracks always number of cycles to failure (on logarithmic axes). The began near the toe of the weld, generally at both sides of test results are shown in Figs. 3 through 9. the gusset. Usually, after some time, crack development t 3Inn 2SO ( &"_. ".. . I 40 u I wldtd~ 10 .tress .,..tic W.l balN sertes atr o:r,.ncHag H_ter-. t 300 2~ & . . . . . . _. -......, t~ • . fun """traUon we14 SUU$ N:1ined. R-D~1 .. 200 .. 200 70 yUthOlth: protec:Uon el .. II 150 $I .......r_ : , . . ......1 .....t.er-- .- "E E z 150 • - .- 9OO.v .-1100 .v c: " c , i! .. ":: 100 ~ ", ..... ~ !! ., . 100 --~., ~ .. .::iii c ~ iii c "-i i0:!l z 50 50 i i 10· 10· Numb.r Of eJel. . -+ NUmber 01 cycle. -+ Fig. 7-lnfluence of weld finishing (grinding and plasma Fig. 9-lnfluence of cathodic overprotection. dressing). 300 300 250 ~l1 ....t""""1d •••1.4. ttNSI. ratio buesert.,.air - N.l 250 ..-n..,....,.. . . . . . . . . ." fun fltl8tlath",_4 .--,. -- *40.ai"t 0 ( 40. H • •te,. n~HNSsinjl - t 200 ~ *40_s.-tel.. 20 o , .. • 010_ .11 .- _70 _ selllr,t.l1" e e e .1W..... ·-...fll. e...7lc 150 " "- .... -- ." 7l 150 c: . " .... , If! i! ~: 100 ;:0.;.. :: 100 ""I.: IS 153-- _nj ......~ , ....... "to.. ....~ . ~ IS In. .. -20 , 15: ~ Iiii c: "1 ~ I , ....... f.io i IIz :!l 5:n 5 .ft 11 10· 10· Num".r of eJel. . -+ Numbe, 01 cycle. -+ Fig. 8-lnfluence of weld finishing (TIG dressing and im- Fig. 10-Comparison of results with fatigue design curves. proved profile).FEBRUARY 1982 443
  • 5. 0,7 A 40 - 1 - 5 - L - Ro iE5TSPEC1MEN QqO-l<~-l-RO N(jH1N~l STIIESSAANGE "-144 tl/M"!;? ~o. ~ II: t; _0, ..~h:.:~~. . E W/BASE MAT. I I I I 0, I I I I I I I I I I I 1.0 STR~INS AT CR~CKEO SIDE mm _ Fig. 11-Strain gauge records during fatigue lifetime. Fig. 12-Striation spacing as a function of crack depth as measured by transmission electron microscopy (TEM).became predominant at one side. lifetime between the welded steel joints tested in air and The strain gauge measurements showed that crack in- in seawater of SoC under freely corroding conditions.itiation usually takes place in the early stage of a fatigue This leads to the tentative conclusion that the deleteriousexperiment (about 10 to 2S% of the total fatigue life). A effect of seawater on the fatigue behavior of welded steelcharacteristic strain gauge record is shown in Fig. 11. joints is strongly dependent on seawater temperature. A detailed fractographic analysis of one of the This conclusion also needs further verification.specimens tested in air showed that the cracks originated The fatigue life obtained for the welded steel jointsfrom an undercut in the weld metal and that crack prop- tested in air corresponds well with the Class F (the rele-agation took place afterward in a coarse-grained area of vant joint classification) mean S-N curve of the designthe HAZ. A diagram showing the measured striation curve. The fatigue life of the joints tested in seawaterspacing as a function of the actual crack depth is shown corresponds well with the Class F mean-minus-two-in Fig. 12. Striations smaller than about 0.04 to 0.06 J.tm standard-deviations S-N curve of the design curve (Fig.are usually nearly invisible. For that reason, the mean 10).striation spacing in the vicinity of the crack origin maybe overestimated. Influence of Steel Plate Thickness The S-N curves obtained for the 40- and 70-mm testDiscussion specimens in both air and seawater reveal no significantEffect of Seawater effect of plate steel thickness on fatigue strength,The S-N curves (based on stress range) obtained in air although such an effect has been predicted for similarand in seawater of 20°C under freely corroding condi- joints on the basis of theoretical fracture mechanicstions for the welded steel joints are different. The fatigue analyses. Booth 3,4 found a plate thickness effect in 2S-lifetime has been found to be at least a factor of two to and 38-mm transverse joints and explained this effect onthree times shorter in seawater (Figs. 3 and 4) than in air the basis of the interdependence among stress intensitywhen the same stress range is applied. The seawater ef- factors, stress concentration factor, and plate steelfect on the fatigue behavior of the welded steel joints thickness.seems to be more pronounced for the stress-relieved The steel plate thickness effect also has been observedspecimens than for the as-welded specimens. Stress· in other investigations-however, usually with relativelyrelieving has a favorable effect on air fatigue endurance thin plates. The preliminary conclusion of the present(Figs. 3 and 4). This effect is larger at R= -1 than at work that increasing the thickness above 40 mm has noR=O.I, which can be explained on the basis of crack significant effect on fatigue behavior in air and inclosure effects. In seawater, however, the favorable ef- seawater needs further justification.feet of stress relieving is much smaller (Fig. 6); even atR =0.1, the effect is negligible because of the larger ef- Influence of Stress Ratiofect of the seawater on the fatigue endurance of the The results of the fatigue tests with the 70-mm transversestress-relieved specimens. joints, tested in bending at stress ratios R=O.1 and -1, Booth,3,4 with similar test specimens and test pro- indicate some effect of the applied stress ratio on thecedure, has found no significant difference in fatigue fatigue strength.444 JOURNAL OF PETROLEUM TECHNOLOGY
  • 6. It can be observed that, under bending loading, an in- seawater on fatigue endurance. In seawater, fatigue en-crease in stress ratio from R= -1 to R=O.l tends to durance was found to be at least two or three timesresult in a small decrease in the fatigue strength. It ap- shorter than in air. .pears that the influence of the stress ratio is about the 2. The plate steel thickness does not influencesame for both air and seawater. The results clearly in- significantly the fatigue strength of 40- and 70-mm weld-dicate that the influence of stress ratio is more pro- ed steel joints in air and seawater.nounced for the stress-relieved welded joints than for the 3. The stress ratio has a small influence on the fatigueas-welded joints (even more so in seawater than in air). strength (based on stress range) of welded steel jointsThis may be explained on basis of the fact that, in the lat- loaded in bending in air and seawater. An increase inter joints, high tensile residual stresses exist. Under the stress ratio from R= -1 to 0.1 results in some decreaseapplied loading (R=O.1 and -1), the stresses near the in fatigue strength. The effect was found not to dependweld in the as-welded specimens remain largely tensile on the environment (air or seawater) and to be more pro-even under compressive loading (R= -1). Under these nounced for the stress-relieved specimens than for the as-circumstances, the stress range is assumed to bethe ma- welded specimens.jor variable determining fatigue and no large effect of 4. Grinding and plasma dressing of the weld toe in-stress ratio is expected when the results are expressed in crease the fatigue life in air as well as in seawater. TIGterms of the applied stress range. dressing of the weld toe gives only a slight improvement, especially at long lives. A less steep toe angle (45 vs. 70°) has only a slight beneficial effect. Influence of Fatigue Improvement Techniques 5. The test results in air are in good general agreement Fig. 7 shows that grinding and plasma dressing of the with existing fatigue data and seem to be described safe- weld toe significantly increase the fatigue strength of the ly by the fatigue design rules.,welded steel joints in both air and seawater. The effect 6. Cathodic protection was found to be most effective seems to be more pronounced in air than in seawater. at lower stress ranges and gives an improvement of a fac- Obviously, as with the stress-relieved vs. as-welded tor of four in fatigue endurance. Cathodic overprotection specimens, the detrimental effect of the seawater on has an unfavorable effect on fatigue life compared with fatigue endurance becomes more predominant after cathodic protection. However, compared with the basic fatigue improvement techniques are applied. corrosion fatigue results, no disastrous effect need be In air, TIG dressing seems to be favorable only in feared. short lives (Fig. 8). It is assumed that TIG dressing in thick plates (40 mm) causes high welding stresses, which influences the fatigue life unfavorably, especially at a high number of cycles. The improvement of TIG dress- Acknowledgments ing of welded specimens tested in seawater seems to be This work forms part of a European Offshore Steels small. Changing the weld angle from 70 to 45° seems to Research project sponsored by the European Coal and increase the fatigue life in air as well as in seawater only Steel Community. We are indebted to SMOZ (Founda- at a high number of cycles. The improvement is not very tion Materials Research in the Sea) for coordinating the significant. Dutch part of the ECSC work. We also are indebted to our colleagues J.J.W. Nibbering at Delft U., J.L.Influence of Cathodic Protection Overbeeke at Eindhoven U., and W. Dortland and H.The results of the tests on cathodically protected Wildschut at TNO for their active participation in this( -900-mV) and overprotected (- 11 OO-mV) 70-mm investigation.specimens (stress relieved) are shown in Fig. 9. Theresults indicate that cathodic protection is effective onlyat low(er) stress range values. This is in agreement withBooths findings. 3,4 Cathodic overprotection reveals an Referencesunfavorable effect on fatigue life compared with the 1. "Regulations for the Structural Design of Fixed Structures on theresults of cathodically protected specimens. However, Norwegian Continental Shelf," Norwegian Oil Directorate (April 1977). . ,the end endurance of the overprotected specimens is still 2. Oumey, T.R. and Maddox, S.J.: "A Re-analysis of Fatigue Datasomewhat higher than that of the basic test specimens in fOT Welded Joints in Steel," Report El44172, Welding Inst., Cam-seawater. bridge, England (1972). PWHT could have influenced the effect of cathodic 3. Booth, O.S.: "Constant Amplitude Fatigue Tests on Welded Steel Joints Perfonned in Air," European Offshore Steds Research,protection beneficially; however, test results of Cambridge, England (Nov. 1978).cathodically protected as-welded specimens, reported by 4. Booth, O.S.: "Constant Amplitude Fatigue Tests on Welded SteelBooth,3,4 show a beneficial effect on fatigue life (at Joints Perfonned in Seawater," European Offshore Steelslower stress ranges). Research, Cambridge, England (Nov. 1978). 5. Solli, 0.: "Corrosion Fatigue of Welded Joints in Structural Steel and the Effect of Cathodic Protection," European Offshore Steels Research, Cambridge. England (Nov. 1978).Conclusions 6. Berge, S.: "Constant Amplitude Fatigue Strength of Welds inConstant amplitude fatigue tests were conducted in air Seawater Drip," European Offshore Steels Research, Cambridge,and seawater on transverse nonload-carrying welded England (Nov. 1978). 7. Wildschut, H., Dortland, W., de Bach, J., and van Leeuwen,steel joints. The following conclusions are drawn. J.L.: "Fatigue Behavior of Welded Joints in Air and Seawater," 1. TheS-N curves (based on stress range) obtained for European Offshore Steels Research, Cambridge, England (Nov.air and seawater clearly reveal a deleterious effect of the 1978).FEBRUARY 1982 445
  • 7. 8. Haagensen, P.J.: "Effect of TIG Dressing on Fatigue Perfor- SI Metric Conversion Factors mance and Hardness of Steel Weldments," paper STP 648, ASTM, Philadelphia (1978). Btu x 1.055 056 E+OO kJ 9. Millington, D.: "TIG Dressing for the Improvement of Fatigue OF (OF-32)/L8 °C Properties in Welded High Strength Steels," Contract Report in. x 1.54* E+01 mm CZlS/2ZI7I, Welding Inst., Cambridge, England (July 1971).to. Gurney, T.S.; "Fatigue Design Rules for Welded Steel Joints," L x 1.0* E+OO = dIn 3 Research Bull., Welding Ins~. (May 1976) 17. . lbf x 4.448 222 E+OO N micron X 1.0* E+OO I+mOriginal manuscript received in Society of Petroleum Engineers office Feb. 15, 1979. psi X 6.894 757 E-03 = N/mm 2 (MPa)Paper accepted for publication Jan. 17,1980. Revised manuscript received Nov. 10,1981. Paper (SPE 8621, OTC 3421) first presented altha 11th OffshOre TechnologyConference held in Houston April30-May 3, 1979. *Conversion factor is exact. JPT446 JOURNAL OF PETROLEUM TECHNOLOGY