ensayo root opening

80-s | MARCH 2001

A B S T R AC T. In steel bridge manufactur-ing, dimensional differences caused byweld deformation often occur becausemultipass welding is used to join thickplates. It frequently develops that rootopenings are out of tolerance at buttjoints. For example, to be within toler-ance, root openings must be controlled tob e l ow 6 mm for butt-joint welds in platesunder 20 mm thick, but a root opening of30 mm can develop in the field. In thatcase, a part 24 mm out of tolerance is gen-e rally built up. How e ve r, there has beenno accumulated data and standards de-veloped regarding these built-up weldedparts. In the present case, a study was per-formed to accumulate data on the behav-ior of built-up parts and to verify the ef-fects of root opening on the mech a n i c a lproperties of the welded parts. For thatpurpose, tensile, bend, impact and hard-ness tests were carried out on weld spec-imens having 0-, 6- and 30-mm rootopenings. Additionally, the finite elementcommon code (MARC) was used to studythe effects of 0-, 6- and 30-mm root open-ings on residual stress and weld deforma-tion in multipass welding. The results ofthe experiments and of the analysis werecompared. In the analysis, a 100% ra m pheat input model was used to avoid thenumerical convergence problem causedby an instantaneous increase in tempera-ture near the fusion zone. The effect of am oving arc in a two-dimensional planewas also included. Additionally, to ensurethe accura cy of the analysis and to savetime, small time increments were appliedin a period with instantaneous tempera-ture fluctuations, while large time incre-ments were used in another period. Th eexperimental and analytical results showgood correlation. Weld-included residualstresses and deformation distribution of

the specimen with the 30-mm root open-ing appeared to be asymmetric and themagnitude was larger than those of rootopening specimens in tolerance.

Introduction

When steel structures are welded, alocalized fusion zone is generated in theweld joint because of the high heat inputfrom the arc, and then nonuniform tem-p e rature distribution through heat con-duction is induced. Therefore, nonuni-form heat deformation and thermal stressare included in the as-welded parts. As aresult, plastic deformation is retainedwithin the weldment and nonlinear plas-tic deformations and residual stressesexist after cooling of the welded joint(Refs. 1, 2). Many problems occur in thefield because of dimensional differencesthat occur as a result of these weld de-formations during manufacturing of largesteel structures (Ref. 3). For thick-sectionwelded steel bridges, many welders fre-quently neglect the occurrence of di-mensional differences caused by theweld deformation that results from multi-pass welding. When an out-of-toleranceroot opening arises from the dimensionaldifference through weld deformation,parts out of tolerance are built up in thefield. While this phenomenon is com-

mon, building up the weld is usually per-formed without systematic inspection.

Steel plate 15 mm thick is limited inpractice to a root opening of less than 6mm. How e ve r, dimensional differencescaused by the weld deformation fromheating and cooling of a localized fusionzone result in a root opening of greaterthan 6 mm to a maximum of 30 mm. Inthat case, after grinding, welding is per-formed to build up the weld by 24 mm inorder to make a 6-mm root opening.There are difficulties, however, acquiringsatisfactory data about problems thatarise from this built-up weld and, fur-thermore, no standards exist regardingthe built-up weld.

Although the built-up weld compen-sates for the dimensional difference inthe root opening caused by weld defor-mation, it may impede work progress andcreate safety problems by lowering struc-tural strength.

Many engineers have had difficulty insystematically quantifying weld residualstresses and deformations in largewelded structures that use thick, weldedsteel plates (Ref. 4). Recently, finite ele-ment analysis has been used to quantifydata about weld residual stresses and de-formations (Refs. 5–7). How e ve r, it hasemphasized single-pass welding ra t h e rthan multipass welding. The effects ofroot opening size on weld residual stressand deformation have been analyzedusing MARC, a software for finite ele-ment analysis (Ref. 8). The comparisonbetween finite element analysis and ex-perimental measurements through a sys-tematic approach can be utilized as fun-damental data for quantifying weldresidual stresses and deformations for thedesign of welded steel structures (Refs.9–12). In comparison with the existingheat-input model, this paper suggests anew model as an efficient analysismethod for improving the accuracy ofthe solution and reducing analysis times.

The Effects of Root Opening on MechanicalProperties, Deformation and Residual Stress

of Weldments

BY G. B. JANG, H. K. KIM AND S. S. KANG

Mechanical properties for multipass welds are investigated and weld deformationand residual stress are simulated using finite element analysis

KEY WORDS

Multipass WeldingRoot OpeningResidual StressDeformationSequential Thermal-

Mechanical AnalysisG. B. JANG, H. K. KIM and S. S. KANG arewith the Micro-Plasticity Laboratory in theDept. of Mechanical Engineering, Pusan Na-tional University, Pusan, Korea.

Experimental Procedureand Welding Conditions

The steel, SWS 490A, used in thisstudy is a hot-rolled steel utilized in themanufacture of bridges, vessels, automo-biles, fuel storage tanks and other struc-tures. It has excellent weldability. Th echemical composition and mech a n i c a lproperties of the base metal are shown inTables 1 and 2. Welding was performedwith the flux cored arc welding (FCAW)process. The welding wire used was SF-71. The shielding gas was an Ar-CO2 mix-ture with a flow rate of 20–30 L/min. Thetest specimens had a 60-deg-angle V-groove. Their dimensions were 300 x 500x 15 mm (2W x L x h). A cross section ofthe test specimen is shown in Fig. 1. Thewelding conditions for each specimenare shown in Tables 3–5. The surround-ing atmospheric temperature was ap-proximately 20°C, and postweld heattreatment was not applied.

In the experiments, butt-joint, multi-pass weldments with 0-, 6- and 30-mmroot openings were manufactured withFCAW. Tensile, bend, impact and hard-ness tests were performed to eva l u a t emechanical performance. Moreover, mi-crostructures were observed with an op-tical microscope. Charpy impact andRockwell hardness tests were measuredat distances 2.5, 7.5 and 12.5 mm fromthe upper surface of the specimens. Toprovide data for verification of the ther-mal analysis, thermocouples wereplaced on the specimens at node 84 (at adistance of 5 mm from the weld inter-face). After welding, a three-element,strain-gauge rosette was applied at a dis-tance from the weld interface (Fig. 4), andthe residual stresses were determined byusing the hole-drilling method accordingto ASTM Standard E837. Lastly, weld de-formation of the thickness direction wasmeasured with a dial gauge. Dimensionand shape of the experimental test spec-imen are shown in Fig. 2.

Finite Element Analysis

Finite Element Modelingand Mesh Generation

The coupling between the thermal andm e chanical analyses takes place through

WELDING RESEARCH SUPPLEMENT | 81-s

Fig. 1 — Schematic cross sections of the test specimens. A — 0-mm root opening; B — 6-mmroot opening; C — 30-mm root opening built up with 24 mm of weld material to create a spec-imen with a 6-mm root opening (the numbers in each area of the weld zones represent the weld-ing sequence of the multipass weld).

Table 1 — Chemical Composition of Base Metal

Material Chemical Composition (wt-%)

SWS C Si Mn P S490A 0.15 0.43 1.45 0.03 0.03

Table 2 — Mechanical Properties of Base Metal

Material Mechanical Properties

SWS Yield Strength (N/mm2) Tensile Strength (N/mm2) Elongation (%)490A 310 490 ~ 610 17

Table 3 — Welding Conditions for 0-mm Root Opening

TimePass No. Electrode Diameter of Ampere Volt Interpass Weld Length Speed Heat Input

Polarity Electrode Temperature (mm) min s (cm/min) (kJ/mm)

1 DCEP 1.4 220 26 Ambient 580 4 7 14.1 2.42 DCEP 1.4 230 27 94 610 3 30 17.4 2.13 DCEP 1.4 210 27 66 610 4 6 14.9 2.3

Backgouging4 DCEP 1.4 190 26 45 560 3 15 17.2 1.7

A

B

C

the tempera t u r e -dependent materialproperties in the me-chanical stress prob-lem, wh i ch serves asinput for heat tra n s f e ranalysis. Addition-a l l y, a change in thet e m p e rature distribu-tion contributes todeformation of theb o dy through ther-mal strains and influ-ences the materialproperties. Th r o u g hthis process, sequen-tial thermal-mech a n-ical analysis, fromwh i ch the tempera-ture distribution,stress and displace-ments induced by thethermal strain wereobtained, was per-formed (Refs. 13, 14).Of course, all me-chanical and thermalmaterial propertiesm ay be tempera t u r edependent and theg overning matrixequations may be ex-pressed as the fol-

l owing (Ref. 8):

M + D(T) + K(T)u = f (1)

C(T) + K(T)T = Q + Q1 (2)

where M is the mass matrix, C(T) is thespecific heat matrices, K(T) is the thermalc o n d u c t ivity matrices, D(T) is the tem-perature-dependant displacement matri-ces, T is the nodal temperature vector, uis the displacement vector, f is the plasticwork and Q1 is the amount of heat gen-erated because of the plastic work.

The ch a racteristics of the weld analy-sis require a large amount of time ands t o rage memory to model the weld fusionand heat-affected zones. In the case of asingle-pass weld, 3-D analysis that con-siders moving heat sources can be used,while for a multipass weld it is importantto know many engineers tried to simplifythe problem because of the difficulty of 3-D analysis. In this study, the assumptionthe cross section of each specimen hasthe same temperature distribution wa sused to predict the weld residual stressesand deformations during multipass weld-ing of thick plates. A vertical cross sectionof the weld interface was used for 2-Danalysis. In other words, the assumptionof plain strain is utilized to consider thestress of the thickness direction to avo i d

82-s | MARCH 2001

A

B

C

Fig. 2 — Dimensions and shape of the test specimens. A — Tensilespecimen; B — impact specimen; C — bend specimen. (Test speci-mens were acquired in a perpendicular direction to the weld inter-face.)




1 DCEP 1.4 210 26 Ambient 620 5 51 10.6 3.12 DCEP 1.4 230 27 114 640 3 35 17.9 2.13 DCEP 1.4 230 29 111 660 4 41 14.1 2.84 DCEP 1.4 210 30 42 570 5 43 10.0 3.8




1 DCEP 1.4 180 26 Ambient 620 4 6 15.1 1.92 DCEP 1.4 180 27 123 620 3 54 15.9 1.83 DCEP 1.4 200 27 72 620 3 7 19.9 1.64 DCEP 1.4 200 27 89 620 3 48 16.3 2.05 DCEP 1.4 200 27 120 620 3 6 20.0 1.66 DCEP 1.4 200 27 128 620 3 54 15.9 2.07 DCEP 1.4 200 26 133 620 2 50 21.9 1.48 DCEP 1.4 200 26 87 620 3 17 18.9 1.7

Welding after grinding9 DCEP 1.4 180 26 62 620 7 12 8.6 3.310 DCEP 1.4 210 28 162 620 1 45 35.4 1.011 DCEP 1.4 220 28 135 620 2 0 31.0 1.212 DCEP 1.4 220 28 75 620 3 29 17.8 2.113 DCEP 1.4 210 28 63 620 4 30 13.8 2.614 DCEP 1.4 210 28 110 620 3 20 19/8.6 1.915 DCEP 1.4 210 28 120 620 3 38 17.1 2.116 DCEP 1.4 210 28 60 620 3 59 15.6 2.317 DCEP 1.4 210 28 103 620 4 9 14.9 2.4

errors by generating excessive stress in thet h i ckness direction if the multipass weldis analyzed as plain stress to predict resid-ual stresses. The analysis model of the sizeof the weld specimen and vertical crosssection passing unit length of the weld in-terface is presented in Fig. 3.

Localized heat input creates a finemesh around the weld zone, which hasan abrupt temperature gradient, and thesize of the elements at a distance far fromthe weld fusion and heat-affected zonesis increased to reduce analysis time. Thenode numbers for the analysis results andfinite element model are presented in Fig.4. The mesh for analysis is comprised ofa four-node quadrilateral element anddeactivated element in the weld joint.

D e a c t ivate options deactivate ele-ments except for the first of the multiplel ayers. When an element of the weldzone is deactivated, it does not con-tribute to stiffness and internal forc e .When an element is reactivated afterwelding of the first layer, it will have thestress level from when it was deactivated.The iterated process is performed contin-ually until formation of the weld bead iscomplete. When the 24-mm built-uparea is welded, a connected region as anonbuilt-up mesh is deactivated to in-crease the accuracy of the deformationanalysis. For this study, after the 24-mmbuildup was finished, analysis methodswere used that reactivated a connectednonbuilt-up mesh of 6 mm.

Boundary Conditions

The initial temperature of all nodeswas 20°C, wh i ch was equivalent to at-mospheric temperature. The nonlinearity

in the analysis includes tempera t u r e -dependent properties, latent heat (phasechange) effects, heat convection in thef l ow direction and nonlinear boundaryconditions. In turn, temperature distribu-tions can be used to generate thermalloads in a stress analysis. The thermal ex-pansion coefficient, elastic modulus andyield strength as mechanical propertiesof material and heat convection coeffi-cient, heat conduction coefficient andspecific heat were input as a function oft e m p e rature in order to consider va r i a-tions in material properties. Weld heat islost to the surrounding atmospherethrough convection and radiation, andradiation losses are greatest close to theweld pool. At some distance from theweld pool, convection is the primarym e chanism for heat loss to the atmos-phere. The effects on latent heat throughphase change from solid to solid and

solid to liquid were considered duringwelding. The material is isotropic and itsyield behavior is applied by Von Mises’yield criteria.

The nodes in the center of the bottomsurface were perfect as the boundaryconditions required for mech a n i c a lanalysis of the 0- and 6-mm root openingweldments. Since the specimen wa swelded under a condition where it liftedup at the steel or ceramic plates, thenodes of the region contacted at theplates were limited to the thickness di-rection (+y-direction). The center nodesof the upper surface were constrained be-cause the 0-mm root opening samplewas finished after the third pass and thenb a ckgouged. Fi n a l l y, the 24-mm speci-men was restricted at the center region ofthe built-up and 6-mm weld area. Th eweld zone lifted up at the steel or ceramicplates was restricted in a similar way.


Fig. 3 — Schematic of the weld specimen and analysis zoneshowing the dimensions.

Fig. 4 — Mesh generation and node numbers. A — Model of a 0-mm rootopening specimen that uses 272 elements and 304 nodes; B — model of a 6-mm root opening specimen that uses 326 elements and 364 nodes; C —model showing a 24-mm buildup and 6-mm root opening. It uses 489 ele-ments and 534 nodes. The node index made at the beginning of a model is ameasurement position for thermal cycle, residual stress and displacement.(Distance from weld interface of nodes 46, 84, 87, 119 and 147 is 2, 5, 8, 24and 140 mm, respective l y. Distance from the weld interface of nodes 477,524, 275 and 293 is 2, 8, 20 and 118 mm, respective l y. The distance from thesurface is 4 mm.)

A

B

C

Heat Input Model

It is important that heat input from thearc is exactly duplicated in modeling be-cause the heat input model has an im-portant effect on the accuracy of analysisfrom the temperature distribution, cool-ing rate, size of fusion zone and heat-affected zone to the strength of the weld-ment. The heat input model for analysisof weld residual stresses and deforma-tions was distinguished as a ramp heatinput model and a lumped pass model. Aramp heat input model was developed toavoid numerical convergence problemsbecause of an instantaneous increase int e m p e rature near the fusion zone, andenabled the model to include the effectof a moving arc in a 2-D plane. It takesinto account the variation of plane en-ergy flow in a 2-D model as the arc ap-proaches, travels across and departs fromeach plane under investigation.

The lumped pass model for thick mul-tipass welds developed to reduce analy-sis times and costs is useful in predictingresidual stress, but it is troublesome inpredicting weld deformation. In a single-V groove, the center of the total shrink-age force does not coincide with the neu-t ral axis because the bead deposits arenot symmetric about the neutral axis.This results in larger bend stress and pro-duces different final stress or strain rates.The lumped pass amplifies these effects,so the ramp heat input model is recom-mended for welds that have unsymmetri-cal bead depositions such as a single-V-groove weld. The level is particularlys e vere in cases that have many weldpasses such as the 30-mm root openingweld in this study.

As previously described, the follow-ing heat input model is used to reduce

calculation costsand times throughthe ramp heat inputmodel in this study.Small time incre-ments are applied toa period of instanta-neous tempera t u r efluctuation that gen-e rates stiff tempera-ture gradients due tothe localized highheat input and ra p i dcooling near the fu-sion zone; largetime increments areapplied for other pe-riods. Next, 100%ramp ratio is pro-vided to improve thec o nvergence of theanalyses. These methods make a differ-ence between the total time of actualwelding and the analytical one, includ-ing cooling time. Figure 5 shows the heatinput model of the 100% ramp ra t i o .

In other words, the interpass temper-ature (the temperature to which the weldregion cools between passes) exists in theactual weld. The total weld time is thesum of the welding time of each layer andthe cooling time required for each inter-pass temperature. However, if ramp ratiois increased to improve analyses, thetotal time of analysis is greater than theactual welding time. Therefore, this timedifference must be considered in theanalysis. The heat input rates of the 2-Danalysis are given by the surface flux andexpressed as the following:

Q = ηEI/bL (3)

where η represents 0.8 as the arc effi-

c i e n cy, E and I are arc volts and amperes,and b and L represent the width of the weldbead and the length of the weld directionof the heat input region, that is, unit lengthat the 2-D analysis, respective l y.

Results and Discussion

Experimental Results and Considerations

Tensile test results for each specimenare presented in Fig. 6. The figure showsapproximately 534–540 MPa for the 0-mm root opening, 528–537 MPa for the6-mm root opening and 536–547 MPa forthe weld with the 24-mm buildup and 6-mm root opening. In total, the differencesare not large. The tensile strengths foreach specimen were shown to be satis-fied because fractures were generated inthe base metal, not the weld metal. The30-mm root opening weldment had suf-ficient static strength, but dy n a m i c

84-s | MARCH 2001

Fig. 5 — Shape of ramp heat input function. The area under the rampfunction curve was kept constant to maintain the same net heat inputenergy and to study the effect on thermal and stress responses.

Fig. 6 — Plot of the tensile strength of each weld specimen.

Fig. 7 — Plot of the impact energy of each weld specimen.

strength such as impact and fatiguestrength may be insufficient. For that rea-son, large scattered tensile tests were sug-gested because of the size of the weldpool and fusion zone. The bend test fore a ch specimen was performed fourtimes, and the criteria of no cracks wassatisfied for each specimen. Bend test re-sults are presented in Table 6.

Figure 7 shows the impact energy ofeach weld specimen. There is no differ-ence between the impact energy of thewelds with 0- and 6-mm root openings.In the case of the 30-mm root opening, itis assumed the large scatter between themeasured data is because of the effects ofan enlarged heat-affected zone causedby high heat input. Consequently, it is ex-pected the 30-mm root opening weld-ment will have a problem with dynamicstrength such as fatigue strength becauseof the nonuniformity of its tensile strengthand impact values.

The results of the hardness measure-ment are presented in Fig. 8A–C. Hard-ness was measured at 2.5, 7.5 and 12.5mm from the upper surface of the speci-mens. The hardness of the 0- and 6-mmroot opening weldments showed littledifference. In the case of the weldmentwith a 30-mm root opening, which wasout of tolerance, the total hardness isslightly lower than that of the otherwelds.

Microstructures were acquired fore a ch weld specimen. For the 0-mm rootopening weld, the microstructure of theweld zone is nonuniform because of theeffects of the multiple passes. An inherentsegregation structure of the weld and afinely diffused region caused by multi-pass welding were observed. The mi-crostructure of the heat-affected zones h owed high heat input and the elimina-tion of the initial hot-rolled structure. Th el ower and upper portions of the weld zones h owed severe coarseness and segrega-tion of the microstructure. Because theheat of welding in the lower portion of theweld zone transferred to the supportingplate with ease, weld heat caused only asmall effect. The solidification structure isalso retained because of the small effect ofweld heat in the upper weld zone, wh i chis filled with filler metal.

The microstructure of the weld withthe 6-mm root opening was slightly dif-ferent from that with the 0-mm root open-ing. Since the width of the weld cross-section, wh i ch was filled with all-weldmetal in the upper portion, was increasedby 21 mm, the solidification structure ofthe welded layer disappeared because ofthe relatively greater heat input. How-ever, that of the welded upper layer waswidely distributed. This result is equiva-lent to the fact that the hardness of the

upper solidifiedstructure appearshigh, that is, despitethe coarseness of thestructure, the hard-ness is relative l ywidely distributedbecause of the exis-tence of a marten-sitic structure due torapid cooling.

For the 30-mmroot opening speci-men, solidifiedstructures and re-fined regions by heatappear as a nonuni-form structure. Sincethe solidified struc-tures are not dis-s o l ved by the weldheat, but are insteadretained, they mayh ave a problem ifused in steel fabrica-tions. This phenom-enon may lead to the30-mm root openingspecimen being lessreliable because ofd e t e r i o ration of theweld-included parts.

Analysis Results

Analysis of ThermalCycle

The results of theanalysis of the tem-p e rature cycle fordifferent positions(note the node indexin Fig. 3) of the widthdirection (x-direc-tion) from the weldinterface and a 4-mm distance fromthe surface is repre-sented in Fig. 9. The0- and 6-mm rootopening welds arepresented in Fig. 9A,B. The thermal cycleof the buildup areaand the actual weldin the 30-mm rootopening specimenare presented in Fig.


Table 6 — Result of Bend Test

Welding specimen Test times Crack occurrence Criterion

0-mm root opening 3 No satisfaction6-mm root opening 3 No satisfaction30-mm root opening 3 No satisfaction

Fig. 8 — Comparison of the results and measured position of the hard-ness tests. A — 0-mm root opening weld specimen; B — 6-mm rootopening specimen; C — weld specimen with a 24-mm buildup and 6-mm root opening.

A

B

C

9C, D. The higher the peak temperatureof the thermal cycle, the more it mea-sures the position approach in the weldinterface. Figure 10 shows a comparisonbetween measured and analyzed resultsof the thermal cycle of the initial pass atthe position (node number 84 of each testspecimen) at wh i ch thermocouple wa sinstalled for the 0- and 6-mm welds andthe 6-mm root opening weld after 24-mmof buildup. The analyzed data shows themeasured temperature cycle is biased tothe right side, and the peak temperatureof the built-up weld is higher because ofa more elongated weld time through the100% ramp, as shown in Fig. 5. How-ever, it shows some of the thermal cyclesare very similar to others.

The equilibrium temperature (withoutdifferences in temperature in the variouspositions during the cooling process) isabout 47°C for the 0-mm root opening

and 48°C for the 6-mm root opening. Forthese two cases, the equilibrium temper-ature differed little because the numberof passes and total heat input was nearlyequal.

The equilibrium temperature after the24-mm buildup was approximately 62°C— Fig. 9C, D. In this case, there were sit-uations in which heat was not transferredto the right plates. The equilibrium tem-p e rature reached approximately 40°Cafter the 24-mm buildup and 6-mm rootopening weld. The heat was transferredonly to the left plate, but not the rightone, when the buildup weld was per-formed. It shows equilibrium tempera-ture is lower for the within-tolera n c especimen because of the asymmetricheat transfer to the right plate when the6-mm root opening weld is performed.The lower equilibrium tempera t u r emeans cooling velocity is rapid. Conse-

quently, the maximum value of retainedstress is high when cooling velocity isslow. This can be verified by evaluationof residual stresses.

Residual Stress Evaluation

A weldment is locally heated, there-fore, temperature distribution is not uni-form, and structural and metallurgicalchanges take place in the joint duringmultipass welding. The weld metal andheat-affected zone immediately adjacentto the weld are at temperatures substan-tially above that of the unaffected basemetal. As the weld zone solidifies andshrinks, it begins to exert stress on thesurrounding weld metal and heat-affected zone. Because of that process,the retained Von Mises’ equivalent stressafter cooling of the three types of weldsis retained in the weld metal and listed in

86-s | MARCH 2001

Fig. 9 — Comparison of the thermal cycles at various locations of each weld specimen. A — 0-mm root opening; B — 6-mm root opening; C —left side after 24-mm buildup and 6-mm root opening weld; D — right side after 24-mm buildup and 6-mm root opening weld. (Horizontal axis:time [s x 10000], vertical axis: temperature [°C x 1000].)

A B

DC

Fig. 11. This shows the measured valuesof the experiment and analysis coincidewell. The maximum Von Mises’ equiva-lent stress for the 0- and 6-mm root open-ing welds is approximately 307 MPa, andit is concentrated symmetrically at theweld metal and heat-affected zone of theworkpieces. The initial weld passes pre-heat the base metal of multipass welds.The effects are greatest on the secondpass. As the preheating conditions afterthe second pass are stabilized, the resul-tant change in thermal cycle and coolingrate is less significant. How e ve r, finalwelding of the bottom surface after back-gouging followed by the third pass leadsto considerable reduction of the residualstress and deformation contrary to theothers as shown in Fig. 11A. As a result,when plates with a long, thick butt jointare welded, welding of the bottom sur-face after backgouging can be used to re-duce the residual stresses and deforma-tion of the welded plate.

For the 6-mm root opening weld with-out backgouging, the maximum residualstress is distributed in a wide range be-

cause of the large weld pool. In the 24-mm buildup with a 6-mm root openingweld, the residual stress after cooling isretained as an asymmetric type and bi-ased to the buildup weld as shown in Fig.11B), because heat is not uniformly trans-ferred to the plates. Since the final equi-librium temperature is lower than in thespecimens within tolerance, the maxi-mum Von Mises’ equivalent stress ishigher than that of the 0- and 6-mm rootopening welds.

Weld Deformation Evaluation

The deformation of thickness direc-tion (y-direction) after cooling followedby the final pass is shown in Fig. 12.These results show the measured valuesin the experiments and analysis coincidewell. For the 0-mm root opening speci-men, a small drop in angular deformationwas generated when the final pass wasperformed on the bottom surface afterb a ckgouging, but this value was infini-tesimal. For the 6-mm root opening spec-imen, general types of angular deforma-

tion typical in a butt weld showed dis-placement was limited to weld deforma-tions of less than 1 mm by keeping the in-terpass temperature under 200°C. Welddeformation for the 24-mm buildup, 6-mm root opening specimen was oppositeto the others. The buildup weld gener-ated a positive angular deformation, andthe 6-mm weld after buildup generated anegative angular deformation, that is, adrop due to sudden expansion becauseof high heat input.

To verify a history of weld deforma-tion, the displacement of thickness (y-di-rection) and width (x-direction) at theedge of the upper surface (node number147) is shown in Fig. 13. The horizontalaxis is time (s) and the vertical axis is dis-placement of the thickness direction (cm).Displacement in the thickness directionafter the final pass subsequent to back-gouging approached zero in the case ofthe 0-mm root opening because of itera-tion of expansion and contraction — Fi g .13A. The instantaneous lifting that oc-curred with the heat input of the final passf o l l owed by backgouging was at the bot-


Fig. 10 — Comparison between measured and analyzed results of the thermal cycle. A — 0- and 6-mm root opening weld specimen; B — 30-mmroot opening weld specimen.

Fig. 11 — Comparison of the residual stress distribution by analysis and experiment. A — 0- and 6-mm root opening weld specimen; B — 30-mmroot opening weld specimen.

A

A B

B

tom surface, and then there was a dropbecause cooling occurred. As a result, an-gular deformation approaches zero.

For the 6-mm root opening (Fig. 13B),there was a large displacement in thet h i ckness direction after the final pass asa result of the high heat input from thefinal pass. Asymmetric deformation be-h avior (node number 293 and 147) iss h own in Fig. 13C. Displacement of thet h i ckness direction at node 239, wh i chwas generated by the buildup weld, wa smaintained without change until thefinal situation. Weld deformation, wh i chhad stayed at zero at node 239 duringbuildup welding, occurred suddenlyafter the 6-mm weld. In particular, con-s i d e rable deformation was observed inthe initial pass of the 6-mm weld. Fi n a l l y,angular deformation dropped because ofasymmetric heat tra n s f e r.

Conclusions

The following results were obtainedthrough comparison of the finite elementanalysis and experimental data.

According to the tensile tests, tensilestrength is approximately 534–540 MPafor the 0-mm root opening, 528–537 MPafor the 6-mm root opening and 536–547

M Pa for the 30-mmroot opening. Th e r eis little difference inthe results, but in thecase of the 30-mmroot opening weld,the scatter in the ten-sile strength data ishigher than for theother welds becauseof the extensive weldfusion zone andweld pool. Since noc ra ck propagationwas observed fore a ch specimen, theresults of the bendtests were satisfac-t o r y. According tothe impact tests,there was little differ-ence in the 0- and 6-mm root openingwelds, but the scatter of impact values forthe 30-mm root opening weld was higherand nonuniform. In terms of microstruc-ture, refined regions resulting from weldheat and solidification that were not dis-solved by the heat but were retained, ap-peared as a nonuniform structure. Withregard to mechanical properties, it ap-pears there are problems in attaining

toughness. The static strength such astensile and bend strength in the 24-mmbuildup weld should have no problemsin manufacturing structures, such as thesteel bridges. How e ve r, there may beproblems in reliability of the weldment asa result of reductions in toughness be-cause of nonuniformity in the deterio-rated regions of the weldment.

The residual stress is distributed sym-m e t r i c a l l y, and maximum value is the

88-s | MARCH 2001

Fig. 12 — Comparison of the displacement of the thickness direction(y-direction) by analysis and experiment. A — 0- and 6-mm root open-ing weld specimen; B — 30-mm root opening weld specimen.

Fig. 13 — Comparison of the relation between displacement and timefor each weld specimen. A — 0-mm root opening weld specimen; B— 6-mm root opening weld specimen; C — 24-mm buildup and 6-mm root opening weld specimen. (Horizontal axis: time [s x 10000];vertical axis: displacement [cm].)

A A

B B

C

same for the 0- and 6-mm root openingwelds. In the case of the 30-mm root open-ing weld, the equilibrium temperature isl ower than the within-tolerance speci-mens and the distribution of the residualstress extends over a wide range asym-metrically because of the buildup weld. Itmight have a problem in dynamic strengthbecause most of the residual stress is bi-ased toward the built-up weld part.

Displacement in the thickness directionafter the final pass followed by back g o u g-ing has little effect in the case of the 0-mmroot opening. In the case of the 6-mm rootopening weld without backgouging, a lit-tle angular distortion occurs, but the leve lof deformation is small. Therefore, there isno problem for the specimens within tol-e rance. How e ve r, in the case of the 30-mmroot opening weld, the angular deforma-tion in the asymmetric region where theside of the built-up weld is lifted up and theother side is dropped is greater than theother specimens.

In this study, the predicted weld resid-ual stress and deformation as calculatedfrom finite element analysis are found tobe in close agreement with the experi-mental data.

References

1. Masubuchi, K. 1980. Analysis of WeldedStructure. Pergamon Press.

2. Masubuchi, K. 1991. Recent researchactivities at MIT on residual stresses and dis-tortion in welded structures. Welding Journal70(12): 41–47.

3. Shibata, N. 1991. Prevention and esti-mation of welding deformation —thick platessteel structure (steel bridge). JWS 60(6): 20–25.

4. Natume, M. 1983. Weld distortion con-trol on steel bridges and steel structures. JWS52(8): 30–40.

5. Mahin, K. W., Wi n t e r, W., and Hoden, T.M. 1991. Prediction and measurement of resid-ual elastic strain distribution in gas tungsten arcwelds. Welding Journal 70(9): 245-s to 260-s.

6. Shim, Y., Feng, Z., Lee, S., Kim, D. ,Jaeger, J., Papritan, J. C., and Tsai, C. L. 1992.Determination of residual stresses in thick-sec-tion weldments. Welding Journal 7 1 ( 9 ) :305–312.

7. Free, J. A., and Poter Goff, R. F. D. 1989.Predicting residual stress in multi-pass weld-ments with the finite element method. Com-puter and Structures 32(2): 365–372.

8. MARC User’s Manual. 1996. AnalysisResearch Corp.

9. Shim, Y., Feng, Z., Lee, S., Kim, D., Ja e g e r,J., and Tsai, C. L. 1991. Modeling of weldingresidual stresses. Proceedings of the Winter An-nual Meeting of the ASME, pp. 29–41.

10. Leung, C. K., and Pick, R. J. 1990. Fi-nite element analysis of multipass welds. WRCBulletin 356, pp. 11–33.

11. Ueda, Y., Takahashi, E., Fukuda, K.,and Nakacho, K. 1974. Transient and residualstresses in multi-pass welds. Transactions ofJWRI 3(1): 59–67.

12. The Ohio State Unive r s i t y. 1992. Ex-perimental verification of finite element mod-eling procedures for thick plates. Te chnical re-port submitted to U.S. Army Corps of EngineersDepartment of Welding Engineering.

13. Hibbit, H. D., and Marcal, P. V. 1973.A numerical thermo-mechanical model for thewelding. Computer and Structure s 3 ( 1 1 ) :1145–1147.

14. Friedman, E. 1975. Thermo-mechani-cal analysis of the welding process using finiteelement method. Journal Press. Vessel Te c h .ASME, Series J, pp. 206–243.

15. Hong, J. K., Dong, P., and Tsai, C. L.1994. Finite element simulation of residualstresses in multipass welds. International Con-ference Proceedings on Modeling and Controlof Joining Processes, ed. T. Zacharia. AmericanWelding Society, Miami, Fla., pp. 470–476.

16. Dong, Y., Hong, J. K., Tsai, C. L., andDong, P. 1997. Finite element modeling ofresidual stresses in austenitic stainless steelpipe girth welds. Welding Journal 76(10): 442-s to 449-s.


All authors should address themselves to the followingquestions when writing papers for submission to the WeldingResearch Supplement:

◆ Why was the work done?◆ What was done?◆ What was found?◆ What is the significance of your results?◆ What are your most important conclusions?With those questions in mind, most authors can logically

o r g a n i ze their material along the fo l l owing lines, usingsuitable headings and subheadings to divide the paper.

1) Abstract. A concise summary of the major elementsof the presentation, not exceeding 200 words, to help thereader decide if the information is for him or her.

2) I n t roduction. A short statement giving releva n tbackground, purpose and scope to help orient the reader. Donot duplicate the abstract.

3) Experimental Procedure, Materials, Equipment.4) Results, Discussion. The facts or data obtained and

their evaluation.5) Conclusion. An evaluation and interpretation of your

results. Most often, this is what the readers remember.6) Acknowledgment, References and Appendix.Keep in mind that proper use of terms, abbreviations

and symbols are important considerations in processing a

m a nu s c ript for publication. For welding term i n o l o g y, theWelding Journal adheres to ANSI/AWS A3.0-94, StandardWelding Terms and Definitions.

Papers submitted for consideration in the We l d i n gResearch Supplement are required to undergo Peer Rev i ewb e fore acceptance for publication. Submit an original and onec o py (double-spaced, with 1-in. margins on 81⁄2 x 11-in. or A4paper) of the manu s c ript. Submit only the abstract on acomputer disk. The preferred fo rmat is from any Macintosh®word processor on a 3.5-in. double- or high-density disk.Other acceptable fo rmats include ASCII text, Windows™ orD O S. A manu s c ript submission fo rm should accompany them a nu s c ri p t .

Ta bles and figures should be separate from them a nu s c ript copy and only high-quality figures will bep u blished. Figures should be original line art or glossyp h o t o s. Special instructions are required if figures aresubmitted by electronic means. To receive completei n s t ructions and the manu s c ript submission fo rm, pleasecontact the Peer Rev i ew Coordinator, Doreen Kubish, at(305) 443-9353, ext. 275; FAX 305-443-7404; or write to theAmerican Welding Society, 550 NW LeJeune Rd., Miami, FL33126.

Preparation of Manuscripts for Submissionto the Welding Journal Research Supplement

ensayo root opening

Documents