Quantification of Residual Stresses in External Attachment … · 2018-04-24 · WELDING RESEARCH MAY 2018 / WELDING JOURNAL 145-s such as shot peening, high frequency impact treatment,

WELDING RESEARCH

WELDING JOURNAL / MAY 2018, VOL. 97144-s

https://doi.org/10.29391/2018.97.013

Introduction Fusion welding conditions, such aspreheating and interpass temperature,clamping, and weld bead sequencing,have been investigated and identifiedto affect the residual stress distribu-tion in welded structures (Refs. 1–10).For carbon-manganese and low-alloysteels, preheating has been well identi-fied to reduce the susceptibility to coldhydrogen-assisted cracking (HAC) as

well as reducing, to a limited extent,the weld metal and heat-affected zone(HAZ) thermal strains during weldinginherent thermal cycles. The reductionof thermal strains thereby reduces thelocal stresses in the weld metal andHAZ. Due to constraining the distor-tion displacements, clamping has beenidentified to increase the tensile resid-ual stress value in the weld metal andHAZ within welded structures. Thetypical fusion weld metal and HAZ

residual stresses are tensile and usual-ly significant reaching the local yieldstrength. These stresses, therefore,have been identified to increase thetendency of brittle fracture, fatiguefailures, and environmental stress-cor-rosion cracking (SCC). Along with many researchers, Tenget al. investigated the effect of trans-verse clamping and preheating on theresidual stress distribution in steelplate single-pass autogenous buttjoints using ANSYS finite-elementanalysis (FEA) modeling (Ref. 6). Asshown in Fig. 1A, the effect of thetransverse mechanical clamping atplate edges showed an overall increaseof the transverse residual stress profileby shifting the profile up, through fol-lowing the longitudinal axis of theweldment. The longitudinal axis repre-sents the welding path line accordingto the sample geometry followed byTeng et al. A beneficial effect of pre-heating by reducing the weld metalthermal strains and stresses can be ob-served in Fig. 1B. For instance, at200°C preheating, the transverse ten-sile residual stress peak was reducedby 30% and the transverse compres-sive residual stress values were re-duced, compared to the unpreheatedcondition. Adedayo and Adeyemi alsoinvestigated the effect of preheatingon the residual stress distribution in a6-mm-thick arc welded mild steel plate(Ref. 7). A reduction range of 50 to75% in the weld metal peak longitudi-nal residual stresses compared to theun-preheated condition was observedafter measuring the weld metal resid-ual stresses. Residual stress mitigation methods

Quantification of Residual Stresses in ExternalAttachment Welding Applications — Part II:

Effect of Preheating and Clamping ConditionsThe study investigates the effects of clamping and preheating on residual stress extension behaviors in external attachment welds

BY R. ALHAJRI AND S. LIU

ABSTRACT There has been uncertainty whether postweld heat treatment (PWHT) shouldbe required for external attachment welding applications where stress corrosioncracking (SCC) is a possibility. An industrial criterion established by NACE SP0472,paragraph 3.6.1, indicates PWHT is not required if tensile residual stresses do notextend through the entire wall thickness. To investigate this problem, a finite-ele-ment analysis (FEA) software, SysweldTM, was utilized to predict the extent andlevel of residual stresses of such welds through the thickness of the pressurevessel shell to study the influence of clamping and preheating. Experimental vali-dations following AWS A9.5 standard were performed. The study used ASTM-516grade 70 steel plates of 6.3, 12.7, and 19 mm thicknesses. The effects of rigidclamping and preheating from 95 to 135Con the residual stress distributions fol-lowing the thickness direction were studied.The clamping condition resultsshowed residual stresses were increased consistently in the weld metal andheat-affected zone (HAZ) at significant levels observed at 6.3 mm thickness andminimum levels at 19 mm. Such effect was attributed to lower distortion displace-ment at higher pressure vessel thicknesses compared to the higher distortions atlower thicknesses. Due to thermal strain reduction and insignificant changes ofmetallurgical phase amounts, the residual stresses in the thickness directionwere slightly decreased due to preheating. When welded with preheating orclamping, maximum tensile residual stresses exceeding 80% of yield strength ex-isted in the bottom surface of the 6.3-mm-thick plates but only 19 to 30% yieldstrength in the 19-mm-thick plates. The conclusions indicate that 6.3-mm-thick-ness applications demand PWHT whereas 19 mm or greater thicknesses arepotentially safe depending on the loading conditions.

KEYWORDS • Finite-Element Analysis • Arc Welding • Residual Stress • Preheating • Hole-Drilling Strain Gauge • Clamping

Alhajri 2017118corr.qxp_Layout 1 4/9/18 5:19 PM Page 144

WELDING RESEARCH

MAY 2018 / WELDING JOURNAL 145-s

such as shot peening, high frequencyimpact treatment, and postweld heattreatment (PWHT) have been investi-gated in the literature and were identi-fied to relieve appreciable amounts ofresidual stresses (Refs. 11–15). PWHTin the oil, gas, and petrochemical in-dustries has mostly been utilized to re-lieve the residual stresses of weld-ments used in SCC-inducing environ-ments. According to equipment fabri-cation codes, such as ASME Boiler andPressure Vessel Code (BPVC), SectionVIII; ASME B31.3, Process Piping Guide;and API 650, Welded Tanks for Oil Stor-age, such weldments shall receivePWHT because the weld metal andHAZ, which typically experience hightensile residual stresses, are directlyexposed to the SCC environments. However, such criterion has not beenclearly recognized for other types ofweldments in which a weld metal andHAZ are not directly exposed to the SCCenvironment. Figure 2 schematically il-lustrates an example of an external at-tachment nonpressure retaining weldjoining a reinforcing pad or sleeve to apressure vessel shell, nozzle, or piping.

In such application, the high residualstresses in the weld metal and HAZ mayextend to the bottom surface of a pres-sure vessel or pipe wall where environ-mental SCC is in direct exposure. Thissituation has led to uncertainty onwhether PWHT shall be applied in suchspecial types of applications. A previouswork is being published to demonstratethe residual stress extension behaviorfrom the weld metal to the bottom sur-face of such an application and quantifythe residual stress levels extended tothe bottom surface (Ref. 17). This paperaims to extend the study by investigat-ing the effects of clamping and preheat-ing on residual stress extension behav-iors in external attachment welds.Studying such effects will similarly helpdetermine the PWHT requirement anddraw conclusions on the clamped andpreheated external attachment weldingapplication.

Finite-Element AnalysisModeling Procedures

To study the effects of clamping

and preheating on residual stress dis-tributions, finite-element analysis(FEA) modeling, utilizing theSysweldTM software, was performed.The same single-pass bead-on-plateweldments studied in Part I of this re-search were created to enable a de-tailed relevant comparison with theunpreheated and unclamped weldingcondition (Ref. 17). The meshing de-tails of the bead-on-plate weldmentsare shown in Fig. 3. The base metaland HAZ regions were mostly com-posed of quadrilateral 3D elementswhile both triangular and quadrilateral3D elements were created in the weldmetal region, with no interior anglesless than 20°C. Underneath the weldmetal, fine-scale through-thickness 3D

Fig. 1 — Transverse residual stress distributions along the y-axis (welding path): A — Clamped and unclamped conditions; B — un-clamped and preheated at different temperatures (Ref. 6).

A B

Fig. 2 — Schematic example illustrating an external attachment nonpressure retain-ing weld applied to pressure vessels and piping.

Fig. 3 — FEA model meshing details il-lustrated at the front view plane andisometric view.


WELDING RESEARCH


elements were created to accuratelyand incrementally study the localresidual stress extension behavior inthe through-thickness direction fromthe weld metal bottom weld interfaceto the bottom surface. The FEA usedGoldak’s double-ellipsoid heat sourcemodel, which is applicable to arc weld-ing processes. The heat source param-eters were iteratively calibrated untilthe experimentally measured fusionzone morphology was reached. To study the effect of preheating,2D surface elements, illustrated in Fig.4A, were created at all model exteriorsurfaces where both preheating andfree air cooling at ambient tempera-ture of 20°C were imposed. The pre-heating temperatures studied are de-tailed in Table 1 and were selected byfollowing the minimum necessary pre-heating temperature criteria devel-oped by Yurioka et al. (Ref. 18). Thesecriteria were established to avoid coldHAC in carbon and low-alloy steels. Inthe case of no preheating, 20°C was setprior to welding at the 2D surface ele-ments. To simulate the clamping weld-ing condition, the 2D elements high-lighted in Fig. 4B were set with re-stricted displacements in the x, y, andz axes to impose a 3D rigid clamping. The used FEA modeling computestemperature, metallurgical phases,residual stresses, and distortion tran-sient results through the use of a material-property database that is var-ied as a function of temperature andtransiently present metallurgicalphase. In addition, the effect of phasetransformation is considered in theFEA modeling through the use of con-

tinuous cooling and austenitizationtransformation diagrams. Figures 5and 6 illustrate some of the materialproperties included in the FEA model-ing database of A516 Grade 70 steel(base plate) and ER-70S6 (weld metal)selected in this study, respectively.

Experimental ValidationProcedures

To meet AWS A9.5, Guide for Verifi-cation and Validation in ComputationWeld Mechanics (Ref. 30), experimentalvalidations were performed at the un-preheated and unclamped conditionsand their detailed results were illus-trated in Part I of this study (Ref. 17).The validations included microstruc-tural analysis, as well as temperature,residual stress, and distortion meas-urements. Such validations wereaimed to validate the computations ofthe FEA modeling and the accuracy ofthe material-property database uti-lized. In the microstructural analysis,point counting was used to quantifythe observed weld metal metallurgicalphases by following the ASTM E562-11 standard (Ref. 19). The metallurgical phase identifica-tion followed the criteria detailed inthe International Institute of Weld-ing’s (IIW) classification and terminol-ogy of microstructures in low C, low-alloy steel weld metal (Refs. 20, 21).During phase identification, distinc-tion was made between Widmanstät-ten ferrite and ferrite with secondphase, where Widmanstätten ferritewas agglomerated within the weld

metal total weld metal ferrite. For theresidual stress measurements, thehole-drilling strain gauge method wasperformed at the bottom surface witha setup illustrated in Fig. 7A. Theresidual stress measurement followedthe procedures in the ASTM E837standard (Ref. 23). The distortion pro-files were measured at the bottom sur-face by following the transverse axis,as shown in Fig. 7B. In Fig. 7B, the lon-gitudinal axis shown represents a pro-jection of the weld metal centerline atthe bottom surface. In the experimental validation, single-pass bead-on-plate weldmentswere applied on 178 254 mm ASTM-516 Grade 70 C-Mn pressure vesselquality steel plates. This was to simu-late a single bead external attachmentweld, as a starting point, as elaboratedin Part I. The gas metal arc welding(GMAW) process was used with theconventional GMAW C-Mn steel weld-ing wire, ER70S-6 (1.2 mm diameter),and a shielding gas composition of75%Ar-25%CO2. Different heat inputswithin a range of 327 to 509 J/mmwere followed to study the effect ofheat input on the weld metal mi-crostructure and the weldment resid-ual stress distributions. In addition,three base plate thicknesses — 6.3,12.7, and 19 mm — were studied toanalyze the effect of base plate thick-ness on the microstructure and theresidual stress distributions.

Experimental ValidationResults

Figure 8A and B shows the weldmetal optical micrographs obtained ata heat input of 327 and 509 J/mm andbase plate thickness of 6.3 mm, re-spectively. At 327 J/mm heat input,the weld metal was observed to havean appreciable amount of ferrite withsecond phase and lath martensite inconjunction with a lower amount ofgrain boundary ferrite. Performing ahorizontal Vickers microhardness linetracing, the average weld metal micro-hardness was observed to reach 285HVN with a standard deviation of 14HVN. By increasing the heat input to509 J/mm, a significant decrease inthe amount of lath martensite with anincrease in grain boundary ferrite andferrite with second phase was ob-

Fig. 4 — A — 2D surface elements for preheating and cooling boundary condition; B— 2D surface elements (highlighted red) representing a 3D rigid clamping.

Table 1 — Preheating Temperatures Selected as a Function of Base Plate Thickness

Base Plate Thickness (mm) Preheating Temperature (C)

6.3 95 12.7 117 19 135

A B


WELDING RESEARCH


served in Fig. 8B. The average weldmetal micro-hardness decreased to245 HVN with a standard deviation of 9 HVN. Under similar welding conditions,Glover et al. indicated a local lathmartensite hardness of 390 HVN,whereas Onsøien et al. indicated ahardness range of 215–235 HVN forferrite with second phase (Refs. 24,25). This demonstrates an agreementis reached between phase identifica-tion and the microhardness results. At327 J/mm heat input, the averageweld metal microhardness of 285 HVNlays in the middle between martensiteand ferrite with second phase respec-tive microhardnesses indicating adominant presence of both phases. At509 J/mm heat input, however, the245 HVN average microhardness isclose to the reported ferrite with sec-ond phase microhardness range of215–235 HVN, indicating a dominantpresence of ferrite with second phaseand a significant decrease of the lathmartensite phase in the weld metal.Varied as a function of heat input,

both the FEA and metallographicpoint counting phase amount resultsare illustrated in Fig. 9A–C for marten-site, ferrite with second phase, andferrite phases in the weld metal, re-spectively. The error bars representthe 95% point counting confidence in-terval calculated by following ASTME562. Close trends of phase amountswere observed by both FEA modelingand metallographic point counting. Amaximum deviation of 6% was ob-served in ferrite with second phase, ata heat input of 509 J/mm, indicating ageneral good agreement and validationof FEA phase amount computations. At 6.3-mm base plate thickness and327 J/mm heat input, the bottom sur-face FEA and hole-drilling strain gaugeresidual stress results are shown inFig. 10A–D, following the transverseand longitudinal axes demonstrated inFig. 7B. Close profiles were observedby both FEA and hole-drilling straingauge experimental validation. Themaximum deviation was observed at a30-mm transverse distance, as shownin Fig. 10B, which reached 9% of the

base plate yield strength. Such devia-tion was found to be within the ASTME837, and Schajer (Refs. 23, 26) ex-pected error margins that range from5 to 20%. The temperature and distor-tion validation results can be found inPart I of this study, which also showedgood agreement with respective FEAmodeling results (Ref. 17).

Experimental ValidationDiscussion

Through the experimental valida-tion results obtained and analysis per-formed, a good agreement was consis-tently acquired. One key reasoning ofthe good agreement obtained was theuse of a FEA modeling material data-base that had material propertiesquantified as a function of the tran-siently present metallurgical phaseand transient temperature, rangingfrom room temperature to the meltingpoint. In addition, the use of continu-ous cooling transformation andaustenitization diagrams allowed for

Fig. 5 — ASTM-516 Grade 70 SysweldTM material database: A — Yield strength; B — Young’s modulus; C — thermal strain.

Fig. 6 — ER70S-6 SysweldTM material database: A — Yield strength; B — Young’s modulus; C — thermal strain.

A B C

A B C


WELDING RESEARCH


the consideration of phase transfor-mation in the FEA computation. Theexperimental validation results withtheir small deviations indicated theFEA modeling and the material data-base used were satisfactory and can beused to study the effects of preheatingand clamping.

Finite-Element AnalysisModeling Results

Preheating Effect

At 6.3 mm thickness and 327 J/mmheat input, Fig. 11A and B representsthe through-thickness residual stressdistributions from weld top to bottomsurface at both preheated and unpre-heated welding conditions. The

through-thickness distribution pathfollowed the highlighted 3D elementsin Fig. 12A, taken at the welding mid-path cross-sectional plane, illustratedin Fig. 12B. Superimposed lines are in-cluded in Fig. 11A and B to representthe fusion zone and HAZ boundariesat both preheated and unpreheatedconditions. This is to correlate the ex-planation of the residual stress distri-butions to each specific weldment lo-cal region in the through-thickness direction. The HAZ depth was observed to in-crease by inducing a preheating treat-ment to the weldment. At 6.3 mmthickness and 327 J/mm heat input,the HAZ depth distance was 1.4 mm atthe unpreheated condition, which wasincreased to 1.7 mm with preheatingat 95°C. This represents 18% of HAZ

depth increase. This observation canbe explained by the inherent increaseof the peak temperatures with impos-ing a preheating treatment to theweldment, compared to the unpre-heated condition. Looking into the through-thicknessresidual stress distribution at the un-preheated condition, the weld metallongitudinal residual stresses were ob-served to fall in a range from 376 to400 MPa, while the transverse stressesranged from –86 to 30 MPa. In theHAZ region, the lowest residual stress-es were located in the coarse-grainedHAZ (CGHAZ), where the local peaktemperature exceeded 1100°C. Thisbehavior is attributed to the dominantmartensite present in the CGHAZ byfollowing the FEA modeling metallur-gical computation. In the CGHAZ, the

Fig. 7 — A — Hole-drilling strain gauge residual stress measurement setup with the strain gauge rosette located at the bottom sur-face; B — bottom surface schematic illustration with transverse and longitudinal axes superimposed.

Fig. 8 — Optical micrographs of weld metal with heat input values and 6.3-mm-base plate thickness (M: martensite; FS: ferritewith second phase; PF: primary ferrite): A — 327 J/mm; B — 509 J/mm.

A B

A B


WELDING RESEARCH


austenite started to transform tomartensite at 420°C, which locally induces compressive transformation-induced stresses. These led to residualstresses that are lowest at the CGHAZwhen compared to other sub-HAZ re-gions. In the through-thickness direc-tion, the peak residual stresses are

consistently located at the HAZboundary where both intercritical andsubcritical HAZ regions coexist. Thisbehavior was attributed to the appre-ciable local thermal strains and thelow transformation-induced strainsexperienced during the welding ther-mal cycle at such a region. By further

increasing the distance from the weldtop, the residual stresses were foundto gradually decrease until the bottomsurface was reached. The eventual in-crease in residual stresses prior toreaching the bottom surface was ob-served to be induced by mechanicalequilibrium. By following the through-thickness direction, the local peaktemperature decreased with increasingthe distance from the weld metal fu-sion boundary. This, therefore, indi-cates that higher thermal strains existwithin the base metal regions that arelocated at distances close to the HAZboundary and lower thermal strains,and thereby stresses exist at increaseddistances from the HAZ boundary.Therefore, such an increase of residualstresses prior to reaching the bottomsurface was not caused by the localthermal strains but induced by me-chanical equilibrium. At the bottomsurface, the peak residual stress with-out preheating was observed to reach395 MPa, which represents 95% of thebase metal yield strength. Looking into the preheated weldingcondition, the weld metal longitudinalresidual stresses were lowered byrange of 5 to 15 MPa, compared to theunpreheated condition. However, thetransverse residual stresses were in-creased by 83 MPa, at a maximum lev-el. This increase can be attributed tothe reduction of the weld metalmartensite amount from 27 to 6%, af-ter preheating, following the FEA com-putations. In addition, the CGHAZresidual stresses were observed to besignificantly lowered after preheating.After preheating, the longitudinalresidual stresses at the CGHAZ werereduced by a range of 110 to 130 MPa

Fig. 9 — A — Average FEA and metallography weld metal martensite B — ferrite with second phase; C — ferrite phase fractionsvaried as a function of heat input (6.3-mm base plate thickness).

Fig. 10 — Measured (A), FEA longitudinal (C), and transverse (B and D) residual stressdistributions along the transverse and longitudinal axes for 327 J/mm heat input and6.3-mm base plate thickness.

A B

A B

C D

C


WELDING RESEARCH


while the transverse stresses were re-duced by a range of 20 to 35 MPa. Atboth preheated and unpreheated con-ditions, the CGHAZ was dominantlymartensitic, according to the FEAcomputations. This leads to the con-clusion that, upon preheating, thethermal strain reduction in conjunc-tion with martensite transformation-induced compressive strains led to thesignificant local residual stress reduc-tion in the CGHAZ. In the through-thickness direction, the peak residualstress value at the intercritical andsubcritical HAZ was reduced by 55MPa after preheating. This is attrib-uted to the reduction of thermalstrains, induced by preheating, com-pared to the unpreheated conditionwith no significant transformation-induced strains present at both inter-critical and subcritical HAZ regions.The bottom surface longitudinal resid-ual stress was found to be 377 MPa,90% of base metal yield strength, afterpreheating, which is lower than theunpreheated condition by 5% of thebase metal yield strength. Althoughthe distance between the HAZ bound-ary and bottom surface was slightly re-duced after preheating, which should

tend to increase the bottom surfaceresidual stresses), the bottom surfaceresidual stresses were reduced. Thiscan be attributed to the CGHAZ, inter-critical, and subcritical HAZ localresidual stress reduction. Figure 13A and B represents thecorresponding through-thicknessresidual stress distributions at 19-mmbase plate thickness and 327 J/mmheat input. Similarly, the HAZ depthdistance was increased from 1.1 to 1.4mm, after a preheating treatment,which represents a 27% increase of theHAZ depth distance. According to the FEA computa-tions, the weld metal martensiteamount, after preheating, is still sig-nificant, but was reduced by 30%. Thismay explain the slight increase of weldmetal transverse residual stresses, af-ter preheating, due to the martensiteamount reduction. The weld metal

longitudinal residual stresses, howev-er, were significantly reduced afterpreheating, with a range of 100 to 180MPa residual stress reduction. The CG-HAZ is dominantly martensitic at bothunpreheated and preheated condi-tions. Due to the reduction of thermalstrains and insignificant changes inphase amounts, the residual stresseswere quite significantly reduced at theCGHAZ. After preheating, the peakresidual stresses at the intercriticaland subcritical HAZ regions were re-duced by 28 MPa. The bottom surfacelongitudinal residual stress was ob-served to be 28 MPa, 7% of base metalyield strength, after preheating, whichis lower than the unpreheated condi-tion by 1% of yield strength. In addi-tion, the bottom surface transverseresidual stress, after preheating, was88 MPa, 21% of yield strength, whichis lower than the unpreheated condi-

Fig. 13 — Residual stress through-thickness distributions of 327 J/mm and 19 mmthickness showing both unpreheated and preheated conditions: A — Longitudinal; B— transverse.

Fig. 12 — 3D element: A — Through-thickness path; B — mid-path cross-sectional plane.

A

A

B

B

Fig. 11 — Residual stress through-thickness distributions of 327 J/mm and 6.35 mmthickness showing both unpreheated and preheated conditions: A — Longitudinal; B —transverse.

A B


WELDING RESEARCH


tion by 5% of yield strength. Figure 14A and B illustrates the FEAcomputed weld metal martensite phaseamounts varied as a function of weldingheat input and base plate thickness,

with and without preheating, respec-tively. Figure 15A and B illustrates thecorresponding martensite phaseamount results for the CGHAZ, withand without preheating, respectively.

Looking into Fig. 14B, the weld metalmartensite amounts are reduced withincreasing the heat input value, at allbase plate thickness levels. In addition,the greater the base plate thickness, thegreater the martensite phase amount isobserved, which is attributed to the in-crease of the cooling sink volume withincreasing the thickness level of thebase plate. With introducing preheating(Fig. 14A), the martensite phaseamounts are reduced uniformly at allheat input and base plate thickness val-ues. The most significant decrease ofweld metal martensite amount is ob-served at 19 mm thickness, while theleast decrease is observed at 6.3 mm.This behavior is attributed to the high-est preheating temperature, 135°C, in-troduced to the 19-mm thickness levelwhile lowest temperature, 95°C, was im-posed to the 6.3 mm thickness. Looking into Fig. 15B, the CGHAZis found to be dominantly martensiticat all heat input values at the 12.7-and 19-mm thickness levels. At 6.3-mm thickness, the martensite amountwas observed to be high at heat inputvalues less than 440 J/mm and de-creased significantly with increasingheat input, reaching 52%, at 509J/mm heat input. At the preheatedcondition, Fig. 15A, the martensitephase amount at the 6.3-mm thick-ness level were observed to be uni-formly reduced while at 12.7 and 19mm thickness levels martensite phaseamounts were not found to be affect-ed. In the CGHAZ, preheating effectwas pronounced at the 6.3-mm thick-ness level because of the smaller cool-ing sink volume in which the coolingrates can be more significantly reducedwith a preheating treatment. At 12.7-and 19-mm thicknesses, the coolingsink volume is large enough where thecooling rates at the CGHAZ are stillgreater than the martensite transfor-mation critical cooling rate, with pre-heating.

Rigid Clamping Effect

At 6.3-mm thickness and 327J/mm heat input, Fig. 16A and B rep-resents the corresponding through-thickness residual stress distributionsat both clamped and unclamped weld-ing conditions. Similarly, the distribu-tion path followed the highlighted 3Delements in Fig. 11A.

Fig. 14 — FEA weld metal martensite phase amounts varied as a function of thewelding heat input and base plate thickness: A — Preheated; B — unpreheatedwelding condition.

Fig. 15 — FEA CGHAZ martensite phase amounts varied as a function of welding heatinput and base plate thickness: A — Preheated; B — unpreheated welding conditions.

Fig. 16 — Residual stress through-thickness distributions of 327 J/mm and 6.3 mmthickness illustrating both unclamped and clamped welding conditions: A — Longi-tudinal; B — transverse.

A B

A B

A B


Looking into the clamped weldingcondition, clamping slightly increasedthe longitudinal residual stresses atthe HAZ and base metal regions. Thetransverse residual stresses at the weldmetal, HAZ, and base metal were,however, significantly increased. In ad-dition, the peak transverse residualstress values at the intercritical andsubcritical HAZ regions were also in-creased significantly. The significanttransverse residual stress increase isattributed to the clamping configura-tion (Fig. 4B) in which the distortiondisplacement restriction was clearlypronounced in the transverse axismore than the longitudinal axis. Look-ing into the bottom surface, the resid-ual stress rise prior to reaching thebottom surface was absent at theclamped condition. This can be attrib-uted to the overall residual stress in-crease in the weld metal, HAZ, andbase metal regions close to the HAZboundary. This absence of a residualstress rise, which was induced by me-chanical equilibrium, led to a decreasein bottom surface local residual stress-es. The bottom surface longitudinalresidual stress reached 359 MPa, afterclamping, representing 85% of basemetal yield strength, whereas thetransverse stress reached –52 MPa, –12% of yield strength. This corre-sponds to a reduction of 10% of yieldstrength in the bottom surface longi-tudinal residual stress compared tounclamped welding condition. At 6.3-mm thickness, such behaviorof reducing the bottom surface resid-ual stress, after clamping, was notmanifested at higher heat input valueswhere the residual stress rise prior toreaching the bottom was originally ab-sent, at the unclamped condition. Atsuch condition, clamping caused anoverall increase of residual stresses inthe through-thickness direction, in-cluding the local bottom surface re-gion. Figure 17A and B represents anexample at 441 J/mm heat input and6.3-mm base plate thickness. At theclamped condition, the bottom surfacelongitudinal residual stress reached avalue of 488 MPa, 116% of yieldstrength, while the transverse stressreached 85 MPa, 20% of yieldstrength. This corresponds to an in-crease of 20 and 36% of yield strengthin the bottom surface longitudinal andtransverse residual stresses, respec-

tively, compared to the unclampedcondition. The corresponding longitudinal and

transverse residual stress distributionsat 12.7-mm base plate thickness and327 J/mm are illustrated in Fig. 18A

WELDING RESEARCH



Fig. 19 — Residual stress through-thickness distributions of 327 J/mm and 19 mmthickness illustrating both unclamped and clamped welding conditions: A — Longi-tudinal; B — transverse.

A B

A B


A B


and B, respectively. Clamping caused aslight increase of both the weld metaland HAZ longitudinal and transverse

residual stresses. However, comparedto 6.3-mm base metal thickness, theincrease of residual stresses at 12.7

mm was significantly less pronounced.Such lower level of residual stress increase was attributed to the signifi-cantly lower distortion level that wasoriginally present at 12.7-mm baseplate thickness, at the unclamped,compared to 6.3 mm thickness. There-fore, the distortion reduction byclamping at 12.7-mm was significantlylower than 6.3 mm thickness. Due toincreased base plate thickness andthereby longer distance between theHAZ boundary and the bottom sur-face, the equilibrium induced residualstresses, close to the bottom surface,were slightly altered. Compared to the6.3-mm thickness level, the changes inbottom surface residual stresses werenot as significant. After clamping, thebottom surface longitudinal and trans-verse residual stresses reached 115and 220 MPa, 27 and 52% of basemetal yield strength, respectively.Compared to the unclamped condi-tion, this results in a reduction of 8%of yield strength for the longitudinalresidual stress, whereas transverseresidual stress increased by 5% of yieldstrength. Figure 19A and B illustrates thecorresponding longitudinal and trans-verse residual stress distribution re-sults at the 19-mm thickness level and327 J/mm heat input, respectively.Similar to the 12.7 mm thickness,clamping slightly increased both weldmetal and HAZ longitudinal and trans-verse residual stresses. The increase ofresidual stresses was less pronouncedthan the 6.3 mm thickness. This is,similarly, attributed to the less distor-

WELDING RESEARCH


Fig. 20 — Distortion profiles following the transverse axis in Fig. 7B. Base plate thickness (unclamped: solid curve, clamped:dashed curve): A — 6.3; B — 12.7; C — 19 mm.

A B C

Fig. 21 — Effect of preheating on bottom surface longitudinal (A) and transverse (B)residual stresses as a function of heat input and base plate thickness (un-pre-heated: solid curve, preheated: dashed curve).

Fig. 22 — Effect of clamping on bottom surface longitudinal (A) and transverse (B)residual stresses as a function of heat input and base plate thickness (unclamped:solid curve, clamped: dashed curve).

A B

A B


tion level and reduction at 19 mmthickness compared to 6.3 mm. Due tothe changes of residual stress values atthe weld metal and HAZ, the mechani-cal equilibrium induced residualstresses at distances greater than 8and 6 mm for longitudinal and trans-verse stresses, respectively, wereslightly altered. At the clamped weld-ing condition, the bottom surface lon-gitudinal residual stress reached 7MPa, 2% of yield strength, while thetransverse residual stress reached 79MPa, 19% of yield strength. This led toa reduction of 7% of yield strength inboth longitudinal and transverseresidual stresses, compared to the un-clamped welding condition. Figure 20A–C illustrates the distor-tion profiles of both clamped and un-clamped conditions by following thetransverse axis illustrated in Fig. 7Band at 6.3-, 12.7-, and 19-mm baseplate thickness, respectively. Distor-tion displacements significantly in-creased with decreasing base platethickness. In addition, the reductionof distortion displacements afterclamping decreased with increasingbase plate thickness. This explains theamount of weld metal and HAZ resid-ual stress increase, which was lowestat 19 mm thickness and highest at 6.3mm thickness.

Finite-Element AnalysisModeling Discussion

The effect of preheating on bottomsurface longitudinal and transverseresidual stresses is illustrated in Fig.21A and B, respectively, as a functionof both heat input and base platethickness. The residual stress valuesare taken at the transverse and longi-tudinal axes cross section, in Fig. 7B,which represents the weldment ther-mal and residual stress steady state re-gion. The bottom surface residualstresses consistently decreased withinducing a preheating treatment at allheat input values and base plate thick-nesses. The reduction of residualstresses at the bottom surface is at-tributed to the reduction of the ther-mal strains and, thereby, thermalresidual stresses in the weld metal,HAZ, and base metal regions that ex-perienced high peak temperatures.However, the reduction of bottom sur-

face residual stresses was insignificantand ranged from 2 to 8% of base metalyield strength only. Corresponding to Fig. 21, Fig. 22Aand B represents the effect of clamp-ing on the bottom surface longitudinaland transverse residual stresses, re-spectively, as a function of both heatinput and base plate thickness. At 6.3-mm base plate thickness and heat in-put equal and lower than 367 J/mm,the bottom surface longitudinal resid-ual stresses were reduced after clamp-ing. This reduction is attributed to theabsence of the mechanical equilibriuminduced residual stress rise prior toreaching the bottom surface, afterclamping, as shown in Fig. 16A. At aheat input greater than 367 J/mm, theresidual stresses were significantly in-creased due to clamping. At such heatinput range, no mechanical equilibri-um induced residual stress rise was ob-served originally at the unclampedcondition. At the 12.7-mm thicknesslevel, the bottom surface longitudinalstresses were reduced, while trans-verse residual stresses were increased,after clamping. At 19-mm base platethickness, however, both longitudinaland transverse bottom surface resid-ual stresses were slightly reduced. Thebottom surface residual stresschanges, at the 12.7- and 19-mmthickness levels, were slightly alteredto encounter the slight increase inweld metal and HAZ residual stressesafter clamping. With preheating and clamping con-ditions and following the through-thickness direction, the peak residualstresses were consistently observed atthe intercritical and subcritical HAZregion. This behavior was similarly ob-served at the unclamped and unpre-heated condition and is attributed tolocal high thermal strains and lowtransformation induced strains pres-ent at such local regions. From both Figs. 21 and 22, it can beobserved that increasing the heat in-put increases the peak residual stress-es reached at the bottom surface, irre-spective of clamping and preheatingimposition. In addition, Figs. 21 and22 indicate that the base plate thick-ness is observed to constitute themost significant factor in the peakresidual stresses reached at the bot-tom surface. This can be observed bycomparing the bottom surface residual

stresses at 6.3 mm thickness with 19mm thickness level. At 6.3 mm baseplate thickness, the bottom surfacepeak residual stresses exceeded 80% ofthe yield strength, with and withoutpreheating and clamping. At 19 mmthickness, however, the residualstresses were lower than 35% of theyield strength, with and withoutclamping and preheating. Clamping and preheating did notresult in significant changes to thepeak residual stresses reached at thebottom surface, at all heat input andbase plate thickness, except at 6.3-mmbase plate thickness and heat inputvalues greater than 400 J/mm. At 6.3-mm thickness level and heat inputgreater than 400 J/mm, clamping in-creased the bottom surface peak resid-ual stress by a range from 22 to 35% ofbase plate yield strength, which consti-tutes the highest increase of bottomsurface residual stresses. Both clamping and preheating didnot generally affect the conclusions ofPart I of this study. According to NACESP0472, paragraph 3.6.1, it is requiredto perform PWHT when tensile resid-ual stresses exist in the entirethrough-thickness direction (Ref. 31).According to this criterion, externalsingle-pass attachment welds at 6.3-and 12.7-mm thickness levels werefound to require PWHT because of thethrough-thickness propagation of ten-sile residual stresses, by following thewelding conditions followed in thisstudy. At 19-mm thickness level, how-ever, both compressive and tensileresidual stresses coexisted in thethrough-thickness direction. Thisplaces an ambiguity whether the 19-mm thickness level would require aPWHT. In addition to the residualstress extension behavior, it would bebetter if the NACE SP0472 criterionrelied on the amount of residualstresses evolved at the bottom surface. In the 6.3-mm thickness applica-tion and following the welding condi-tions of the study, the peak residualstresses at the bottom surface werefound to exceed 80% of the base metalyield strength, with and withoutclamping and preheating. Such highresidual stress exceeds the usual SCCthreshold stress levels identified byJones et al. (Ref. 27), which usuallyreaches 80% of yield strength. Hence,at 6.3-mm base plate thickness, PWHT

WELDING RESEARCH



is deemed necessary to avoid SCC at aheat input value within or higher thanthe range studied here. At the 12.7-mm base plate thickness, a peak resid-ual stress range of 36 to 69% of yieldstrength existed at the bottom sur-face, with and without preheating andclamping. At such a base plate thickness level,the PWHT demand can be determinedas a function of the specific SCC envi-ronment threshold stress as well asthe design loading conditions. At the19-mm thickness level, however, thepeak residual stress reached a rangefrom 19 to 35% only at the bottomsurface, with and without clampingand preheating. At such residual stresslevels, the risk of SCC is significantlylower. At the 19-mm or greater thick-ness level, the PWHT requirement canbe considered optional or be waived,provided residual and design loadingstresses are found not to exceed theSCC threshold stress. According to API579-1/ASME FFS-1 Annex 9D, Fitnessfor Service (Ref. 28), a weld metal andHAZ residual stress of 20 to 30% ofyield strength can be estimated to re-main after a PWHT. This range ofresidual stress is quite close to that ob-tained at the 19-mm thickness level,which indicates that the PWHT de-mand at 19 mm thickness is signifi-cantly low.

Conclusions

Generally, preheating was foundbeneficial in reducing the residualstresses in the through-thickness direction. A residual stress reductionrange from 2 to 8% of base metal yieldstrength existed at the bottom surfaceat the preheated condition. Clamping was found to increase theresidual stresses at the weld metal andHAZ at all heat input values and baseplate thicknesses studied. Its effect onthe bottom surface peak residualstress was insignificant, however, ex-cept at 6.3 mm thickness and a heatinput value greater than 400 J/mmwhere the peak stresses increased by arange from 22 to 35% of yieldstrength. Among other welding conditions,the base plate thickness level consti-tuted the most significant factor onthe peak residual stress amountreached at the bottom surface, irre-

spective of clamping and preheating.In addition, in the through-thicknessdirection, the peak residual stresseswere invariably observed at the inter-critical and subcritical HAZ regions. With and without clamping andpreheating, tensile longitudinal resid-ual stresses existed in the through-thickness direction at 6.3-mm and12.7-mm base plate thickness. Bothtensile and compressive stresses exist-ed at 19 mm thickness. According tothe NACE SP0472 criterion, PWHT isrequired for 6.3 and 12.7 mm thick-nesses. The PWHT requirement doesnot explicitly apply for 19-mm-thickplates, however. It is suggested thatsuch criterion should consider theamount of residual stresses reached atthe bottom surface in addition toresidual stress extension behavior inthe through-thickness direction. At the 6.3-mm thickness level, thebottom surface peak residual stressesreached and exceeded 80% of the basemetal yield strength, with and withoutclamping and preheating. At suchthickness level, PWHT is necessary toavoid SCC. At the 12.7-mm thicknesslevel, the peak residual stresses at thebottom surface were found to rangefrom 36 and 69% of the base metalyield strength, with and withoutclamping and preheating. At suchthickness level, the PWHT should berequired as a function of the SCCthreshold stress and the design load-ing conditions. At the 19-mm thick-ness level, however, the peak stressesreached a range of 19 to 35% of basemetal yield strength only. At this lowstress level, PWHT can be potentiallywaived or optional depending on load-ing conditions.

1. Javadi, Y., Hasani, M., and Sadeghi,S. 2015. Investigation of clamping on thewelding sub-surface residual stress and de-formation by using the ultrasonic stressmeasurement and finite element method.Journal of Nondestructive Evaluation 34(1):3. DOI: 10.1007/s10921-015-0277-9. 2. Casavola, C., Cazzato, A., Moramar-co, V., and Pappalettere, C. 2015. Influenceof the clamps configuration on residualstresses field in friction stir weldingprocess. The Journal of Strain Analysis forEngineering Design 50(4): 232–242. DOI:10.1177/0309324715573361.

3. Aalami-Aleagha, M. E., and Eslam-panah, A. H. 2012. Mechanical constraint ef-fect on residual stress and distortion in T-fil-let welds by three-dimensional finite ele-ment analysis. Proceedings of the Institutionof Mechanical Engineers, Part B: Journal of En-gineering Manufacture 227(2): 315–323.DOI: 10.1177/0954405412466 781 4. Reemsnyder, H. S., and Throop, J. F.1982. Residual Stress Effects in Fatigue.American Society for Testing and Materi-als: Philadelphia, Pa. DOI: 10.1520/STP776-EB. 5. Aalami-Aleagha, M. E., Foroutan, M.M., Feli, S. and Nikabadi, S. 2014. Analysispreheat effect on thermal cycle and resid-ual stress in a welded connection by FEsimulation. International Journal of PressureVessels and Piping, 114–115: 69–75. DOI:10.1016/j.ijpvp.2013.12.003. 6. Teng, T.-L., and Lin, C.-C. 1998. Ef-fect of welding conditions on residualstresses due to butt welds. InternationalJournal of Pressure Vessel and Piping 75:857–864. DOI: 10.1016/S0308-0161(98)00084-2. 7. Adedayo, S. M., and Adeyemi, M. B.2000. Effect of preheat on residual stressdistributions in arc-welded mild steelplates. Journal of Materials Engineering andPerformance 9: 7–11. DOI: 10.1361/105994900770346213. 8. Totten, G., Howes, M., and Inoue, T.2002. Handbook of Residual Stress and De-formation of Steel. ASM International: Ma-terials Park, Ohio: pp. 391–397. 9. Leggatt, R. H. 2008. Residual stress-es in welded structures. International Jour-nal of Pressure Vessel and Piping 85:144–151. DOI: 10.1016/j.ijpvp.2007.10.004. 10. Radaj, D. 1992. Heat Effects of Weld-ing. Springer-Verlag Berlin Heidelberg: pp.5–17. 11. Cheng, X., Fisher, J. W., and Prask,H. J. 2003. Residual stress modification bypost-weld treatment and its beneficial ef-fect on fatigue strength of welded struc-tures. International Journal of Fatigue25(9–11): 1259–1269. DOI: 10.1016/j.ijfatigue.2003.08.020. 12. Dong, P., Song, S., and Zhang, J.2014. Analysis of residual stress reliefmechanisms in post-weld heat treatment.International Journal of Pressure Vessel andPiping 122: 6–14. 13. Suominen, L., Khurshid, M., andParantainen, J. 2013. Residual stesses inwelded components following post-weldtreatment methods. Procedia Engineering66: 181–191. DOI: 10.1016/j.proeng.2013.12.073. 14. Adeyemi, M. B., Stark, R. A., andModlen, G. F. 1980. Proceeding of heattreatment ‘79. Metals Society: 122–125. 15. Molzen, M. S., and Hornbach, D.2001. Shot peening and heat treatment re-

WELDING RESEARCH


References


duce stress. Welding Journal 80(1): 38–42. 16. Alhajri, R., and Liu, S. 2016. Predic-tion of Residual Stresses Distribution inExternal Attachment Welding Applicationsto Evaluate the Need of PWHT. Proceedingon the 10th International Conference onTrends in Welding Research, Tokyo, Japan. 17. Alhajri, R., Liu, S., Yu, Z., and An-dreassen, M. J. 2017. Quantification ofresidual stresses in external attachmentwelding applications, Part I. Welding Jour-nal 96(12): 451-s to 466-s. 18. Yurioka, N., and Kasuya, T. 1995. Achart method to determine necessary pre-heat temperature in steel welding. Quarter-ly Journal of Japan Welding Society 13(3):347–357. DOI: 10.2207/qjjws.13.347. 19. ASTM International. 2011. ASTME562-11, Standard Test Method for Deter-mining Volume Fraction by Systematic Manu-al Point Count. West Conshohocken, Pa. 20. International Institute of Welding.1988. Guide to the light microscope exami-nation of ferritic steel weld metals. IIWDoc. IX-1533-88. 21. International Institute of Welding.

1983. Classification of microstructures inlow C-low alloy steel weld metal and termi-nology. IIW Doc. IX-1282-83. 22. Thewlis, G. 2004. Classification andquantification of microstructures in steels:Material perspective. Materials Science andTechnology 20: 143–160. DOI: 10.1179/026708304225010325. 23. ASTM International. 2013. ASTME837-13a, Standard Test Method for Deter-mining Residual Stress by the Hole-DrillingStrain-Gauge Method. West Conshohocken,Pa. DOI: 10.1520/E0837. 24. Glover, A. G., McGrath, J. T., Tin-kler, M. J., and Weatherly, G. C. 1977. Theinfluence of cooling rate and compositionon weld metal microstructures in a C/Mnand a HSLA steel. Welding Journal 56(9):267-s to 273-s. 25. Onsøien, M. I., Liu. S., and Olson, D.L. 1996. Shielding gas oxygen equivalent inweld metal microstructure optimizationWelding Journal 75(7): 216-s to 224-s. 26. Schajer, G., and Ruud, C. 2013. Prac-tical Residual Stress Measurement Methods.Wiley Publication: New Jersey: pp. 1–27.

DOI: 10.1002/9781118402832 27. Jones, R., and Ricker, R. 1992.Stress-Corrosion Cracking: Material Perform-ance and Evaluation. ASM International:Metal Park, Ohio: pp. 1–38. 28. American Petroleum Institute —American Society of Mechanical Engineers.2016. API 579-1/ASME FFS-1: Fitness-For-Service. 29. Chang, P., and Teng, T. 2004. Nu-merical and experimental investigation onthe residual stresses of the butt-weldedjoints. Computational Materials Science29(4): 511–522. DOI: 10.1016/j.commatsci.2003.12.005. 30. AWS A9.5.2013, Guide for Verifica-tion and Validation in Computation Weld Me-chanics. 2013. Miami, Fla.: American Weld-ing Society. 31. NACE International. 2015. SP0472,Methods and Controls to Prevent In-ServiceEnvironmental Cracking of Carbon SteelWeldments in Corrosive Petroleum RefiningEnvironments. Tex.

WELDING RESEARCH


RASHED ALHAJRI ([email protected]) and STEPHEN LIU ([email protected]) are with the Colorado School of Mines, Center forWelding, Joining and Coatings Research, Golden, Colorado.

Authors: Submit Research Papers Online

Peer review of research papers is now managed through an online system using EditorialManager software. Papers can be submitted into the system directly from the Welding Journalpage on the AWS website (aws.org) by clicking on “submit papers.” You can also access thenew site directly at editorialmanager.com/wj/. Follow the instructions to register or log in.This online system streamlines the review process, and makes it easier to submit papers andtrack their progress. By publishing in the Welding Journal, more than 70,000 members willreceive the results of your research.

Additionally, your full paper is posted on the American Welding Society website for FREEaccess around the globe. There are no page charges, and articles are published in full color. Byfar, the most people, at the least cost, will recognize your research when you publish in theworld-respected Welding Journal.


Documents

Quantification of Residual Stresses in External Attachment … · 2018-04-24 · WELDING RESEARCH MAY 2018 / WELDING JOURNAL 145-s such as shot peening, high frequency impact treatment,