Introduction Welding distortion, a result of thenonuniform expansion and contractionof weld metal and adjacent base metalduring the heating and cooling cycle ofthe welding process, is a major concernduring the fabrication of a welded struc-ture. Welding distortion causes complexconsequences, most of which are detri-mental during fabrication and service.

From the point of view of structural in-tegrity, welding distortion acts as an ini-tial imperfection of welded compo-nents. These initial imperfections canhave a significant influence on the be-havior of the structure under variableloading (Ref. 1) and reduce the bucklingstrength of the structure (Ref. 2). Fromthe perspective of manufacturing tech-nology, the consequences of weldingdistortion are out-of-tolerance geome-

try and variable quality of products.Preventive and corrective actions haveto be taken in order to avoid degradeddimensional quality induced by distor-tion. These actions are time consumingand associated with significant costs. Welding distortions can be classi-fied into transverse and longitudinalshrinkage, angular distortion, longitu-dinal bowing, and buckling (Ref. 3).Transverse and longitudinal shrink-ages are the in-plane modes, and buck-ling, longitudinal bowing, and angularchange are out-of-plane modes (Ref.4). In general, in-plane distortion isnegligible in small parts. In large com-ponents, such as shipbuilding, in-plane distortion can be significant sothat manufacturers have to use theo-retical and empirical formulas (Refs.5–8) and shrinkage data models (Refs.9, 10) to estimate shrinkages and com-pensate for them by starting with big-ger parts. Out-of-plane distortionmodes, often observed in a thin-walled structure where fusion weldingis applied, are very common and needspecial techniques (Refs. 4, 11–15) tocontrol. This paper concentrates on in-plane shrinkage and briefly discussesout-of-plane distortions. A large number of factors affect thetype and extent of in-plane shrinkageand out-of-plane distortion due towelding. The magnitude and distribu-tion of shrinkage and distortion in awelded structure are dependent ongeometric parameters, material prop-erties, welding process parameters,and the degree of restraint duringwelding. Geometric parameters such

WELDING RESEARCH

Material Strength Effect on Weld Shrinkageand Distortion

The effects of material strength and heat input on in­plane shrinkage and out­of­plane dis­tortion were studied by welding and measuring 44 small­scale panels in the laboratory

BY Y. P. YANG, R. DULL, H. CASTNER, T. D. HUANG, AND D. FANGUY

ABSTRACT High­strength steels have been increasingly used to reduce thickness and weightin cars, trucks, cranes, bridges, ships, and other structures that are designed to han­dle large amounts of stress or need a good strength­to­weight ratio. However, thinnerstructures are more likely to deform during welding because of a lack of rigidity.Welding shrinkage and distortion become major issues during fabrication of weldedstructures made of high­strength steels. Manufacturers are interested in estimatingshrinkages and compensate for them by starting with bigger parts. However, thereare no accurate empirical formulas to calculate weld shrinkage for thin structuresmade of high­strength steel since the weld shrinkage allowance data and formulasdeveloped by world­wide researchers are primarily for thick plate and low­strengthsteels. The effects of material strength and heat input on in­plane shrinkage and out­of­plane distortion were studied by welding and measuring 44 small­scale panels inthe laboratory. It was found that in­plane weld shrinkage is lower for high­strengthsteels (HSLA­80, HY­80, and HY­100) than for the typical hull structural steel (ABSGrade DH­36) at normal welding heat input. Higher strength HY­80 and HY­100 steelshave less out­of­plane distortion than lower strength DH36 steel while HSLA­80 steelhas similar out­of­plane distortion compared to DH­36. As welding heat inputincreases, shrinkage and distortion are increased for both lower and higher strengthmaterials. For the same heat input, as thickness increases, both shrinkage and distor­tion are reduced.

KEYWORDS • High­Strength Steels • Residual Stress • Distortion • Flux Cored Arc Welding • Weld Shrinkage • Heat Input

Y. P. YANG (yyang@ewi.org), R. DULL, and H. CASTNER are with Edison Welding Institute (EWI), Columbus, Ohio. T. D. HUANG is with Ingalls Shipbuilding,Pascagoula, Miss. D. FANGUY is with Bollinger Shipyards, Lockport, La.

NOVEMBER 2014 / WELDING JOURNAL 421-s

Yang Supplement November 2014[1]_Layout 1 10/14/14 3:59 PM Page 421

as the dimensions of the structure andthe type and size of welded joints, andwelding parameters such as heat in-put, welding sequence, preheating,and postheating will affect weldshrinkage and distortion. Watanabe and Satoh (Ref. 16) useda combination of empirical and ana-lytical methods to study the effects ofwelding conditions on the distortionin welded structures. Gunaraj andMurugan (Ref. 17) studied the effectof process parameters on angular dis-tortion in gas metal arc welding ofstructural steel plates. Sadat et al.(Ref. 18) modeled the effects of pre-heating on angular distortions inone-sided fillet weld joints. Weldingfixtures are widely used to controldistortion by providing restraintagainst thermal expansion and con-traction (Ref. 19). Material propertiessuch as temperature-dependent ther-mophysical and mechanical propertiesalso contribute to the magnitude ofshrinkage and distortion, althoughthere is little published information inthis regard (Ref. 20).

In recent years, high-strength steels havebeen increasingly usedto reduce thickness andweight in cars, trucks,cranes, bridges, ships, and other struc-tures that are designed to handle largeamounts of stress or need a goodstrength-to-weight ratio. However,thinner structures are more likely todeform because of lack of rigidity. Theshrinkage allowance data and formu-las developed by world-wide re-searchers are primarily for thick plateand low-strength steels (Refs. 6–8). Toinvestigate the effect of materialstrength on shrinkage and distortion,44 small-scale specimens made fromboth high-strength steels (HSLA-80,HY-80, and HY-100) and the typicalhull structural steel (DH-36) were fab-ricated under laboratory conditions.This paper describes controlled weld-ing and shrinkage measurement trialson these small-scale specimens. Thedata gathered were used to determinethe effect of material type and heat in-put on weld shrinkage and distortion

for high-strength steels.

Technical Approach Three high-strength steels and onetypical grade of hull structural steelwere used to compare weld shrinkageand distortion among the differentmaterials. A mockup panel was de-signed with a double-sided butt jointand welded by flux cored arc welding(FCAW). Weld consumables andgrooves were selected based on mate-rial and thickness, respectively. Thedeformation induced by welding wasmeasured, and the data wereprocessed to obtain weld shrinkageand distortion for each panel.

Materials and Mockup PanelGeometry

The steel plate materials evaluated

WELDING RESEARCH

WELDING JOURNAL / NOVEMBER 2014, VOL. 93422-s

Fig. 1 — Double V­groove joint design.

Fig. 2 — Table, clamps, and measurement points.

Table 1 — Experimental Matrix

Material Nominal Thickness (mm) Total Heat Input (kJ/mm) Groove Root Opening (mm) Number of Panel Panel

4.76 1.14 Square 0 3 43,44,45 1.57 0 3 40,41,42 2.09 0 3 46,47,48

HSLA­80 6.35 1.14 Square 0 3 25,26,27 1.69 0 5 8,9,10,17,18 2.13 0 3 28,29,30

9.53 1.65 Double V 0 3 34,35,36 2.20 0 3 31,32,33 2.56 0 3 37,38,39

DH­36 6.35 1.14 Square 0 3 19,20,21 1.65 0 3 2,3,4 2.09 0 3 22,23,24

HY­80 6.35 1.65 Square 0 3 5,6,7

HY­100 6.35 1.65 Square 0 3 11,12,13

Yang Supplement November 2014[1]_Layout 1 10/16/14 9:36 AM Page 422

in this task were DH-36, HSLA-80, HY-80, and HY-100. The DH-36 plate wasused to compare shrinkage and distor-tion with the high-strength materials(HSLA-80, HY-80, and HY-100) andalso called low-strength steel in the fol-lowing discussion. The plate size was24 48 in. (610 1219 mm), providinga 48-in. (1219-mm) square mockuppanel when welded with a single buttjoint. Plate thickness was nominally ¼in. (6.35 mm) for panels fabricated tocompare the effects of material type.Additional HSLA-80 panels were weld-ed using nominally 3⁄16 in. (4.76 mm)and 3⁄8 in. (9.53 mm) thick plates.

Joint Design

Most of the butt-joint welds on the

3⁄16 in. (4.76 mm) and ¼ in.

(6.35 mm) thick plateswere double-sided weldson square groove jointswith no root opening. Theedges of the plates to bewelded were machined toprovide a tight fit, i.e.,minimum root opening

along the length of the weld. The weldjoint design for the 3⁄8-in. (9.53-mm)plate was a double-sided V-groovewith no root opening as shown in Fig.1. It should be pointed out that the9.53 mm is a nominal thickness. Thereal plate thickness used in the experi-ment was about 10 mm. Three replicaswere conducted for each experimentalcondition to ensure the reliability ofmeasured data.

Fixturing and Clamping

A fixture was designed to restrainthe plates adjacent to the weld joint andalong the edges as shown in Fig. 2. Atable was used to support the plate witha 4-in. (101.6 mm) wide slot centeredalong the weld joint. A bridge was fabri-cated from 2-in. (50.8-mm) square steeltubing and welded to the table. Clamp-ing was provided by means of four 1⁄2-in.(12.7-mm) threaded rods on each sideof the joint spaced approximately 15 in.(381 mm) apart. The first plate was located length-wise on the table by means of a locat-ing bar that was tack welded to thetable. The joint length was alignedwith two spacer blocks located at themachined plate edge to the center of

WELDING RESEARCH

NOVEMBER 2014 / WELDING JOURNAL 423-s

Fig. 3 — Measurement point locations. Fig. 4 — Coordinate measuring machine measurement repeatability.

Fig. 6 — Panel deformed shape.

Fig. 5 — Data pairs used to calculate in­plane shrinkage.

Table 2 — Material Strength Properties

Material Property DH­36 HSLA­80 HY­80 HY­100

0.2% Yield Strength MPa 423.9 607.4 629.2 751.0Ultimate Strength MPa 553.1 646.7 713.6 828.3Elongation % 33.0 25.2 22.8 18.9

Yang Supplement November 2014[1]_Layout 1 10/16/14 9:36 AM Page 423

the 4-in. (101.6 mm) wide slot. Thisprovided consistent alignment withthe weld axis for each plate. After thefirst plate was located, it was clampedby means of four C-clamps along theouter edge and four threaded rods onthe bridge. The spacer blocks were re-moved and the second plate wasbutted against the first and clampeddown in the same manner as the first.The clamped plates were tack weldedtogether in four locations using auto-

genous gas tungsten arc welding(GTAW).

Welding Equipment,Consumables, and Parameters

Equipment was set up and weldingparameters were adjusted to providethe appropriate welding conditions de-scribed in the following sections. Thewelding process used for fabricating allof the panels was FCAW. The welding

electrodes used were• AWS E81T1-Ni1MJDH4 for DH-36,

HSLA 80, and HY80• AWS E111T1-K3M for HY100. The shielding gas was 75% Ar and25% CO2. The welding equipment con-sisted of a welding power source, elec-trode wire feeder, welding gun, and aside-beam carriage. The welding parameters were nomi-nally as follows. After welding, allplates were allowed to cool down toless than 100°F (38°C) while still beingclamped. Three travel speeds wereused to achieve high, middle, and lowheat input.• Travel angle: 15-deg dragging• Work angle: 90 deg• Contact tip-to-work distance: ¾ in.

(19 mm)• Electrode wire feed speed (WFS): 233

mm/s• Voltage setting: 33.3 V• Welding current: 290 A• Travel speed: 9.0, 11.6, 16.2 mm/s• Gas flow rate: 40 ft3/h.

Experimental Matrix

Table 1 shows the experimental ma-trix for the mockup panels that werefabricated. The mockup panels includ-ed four different materials, threethicknesses, three heat inputs (low,medium, and high), and two groovetypes. Three replicas were made foreach condition. The total heat inputwas the sum of heat input for weldingboth sides of the joint.

WELDING RESEARCH

WELDING JOURNAL / NOVEMBER 2014, VOL. 93424-s

Fig. 7 — Comparison of weld cross sections of different materials.

Table 3 — Across­Weld and Along­Weld Shrinkage Data (DH­36, HY­80, and HY­100)

Plate Material Thickness Total Heat Input Average Across­Weld Shrinkage (mm) Average Along­Weld Shrinkage (mm/m)(mm) (kJ/mm)

A­H B­G C­F Av Avm I­Q J­P K­O Av Avm

19 DH­36 6.30 6.31 1.16 1.2 0.37 0.34 0.29 0.34 0.25 0.33 0.24 0.45 0.34 0.2420 DH­36 6.28 1.14 0.19 0.16 0.12 0.16 0.23 0.21 0.09 0.1821 DH­36 6.33 1.16 0.37 0.20 0.20 0.25 0.22 0.19 0.16 0.19

2 DH­36 6.30 6.31 1.62 1.6 0.45 0.50 0.40 0.45 0.46 0.63 0.61 0.38 0.54 0.493 DH­36 6.31 1.63 0.39 0.33 0.30 0.34 0.33 0.24 0.21 0.264 DH­36 6.33 1.67 0.83 0.49 0.45 0.59 0.67 0.60 0.70 0.66

22 DH­36 6.31 6.40 2.10 2.1 0.98 0.58 0.58 0.71 0.67 0.61 0.54 0.50 0.55 0.5123 DH­36 6.59 2.13 1.02 0.65 0.56 0.74 0.53 0.51 0.41 0.4824 DH­36 6.30 2.08 0.73 0.47 0.49 0.56 0.52 0.49 0.50 0.50

5 HY­80 6.71 6.70 1.67 1.7 0.25 0.26 0.24 0.23 0.26 0.32 0.35 0.24 0.30 0.346 HY­80 6.67 1.65 0.25 0.23 0.24 0.24 0.35 0.31 0.25 0.307 HY­80 6.72 1.65 0.31 0.32 0.27 0.30 0.48 0.38 0.39 0.42

11 HY­100 6.77 6.77 1.61 1.6 0.30 0.21 0.20 0.25 0.26 0.38 0.35 0.25 0.33 0.2912 HY­100 6.77 1.63 0.34 0.31 0.30 0.30 0.31 0.23 0.16 0.2313 HY­100 6.78 1.64 0.15 0.23 0.26 0.22 0.36 0.24 0.32 0.31

Yang Supplement November 2014[1]_Layout 1 10/16/14 9:36 AM Page 424

The 6.35-mm materials were usedto study the effect of material strengthon shrinkage and distortion. The ef-fect of heat input on shrinkage anddistortion was studied with materialsHSLA-80 and DH-36. The combina-tion effect of thickness and heat inputwas studied with material HSLA-80,which has three thicknesses and threeheat inputs were used.

Measurements

Each mockup panel was measuredwith a coordinate measuring machine(CMM) prior to and after welding thebutt joints. Data were recorded foreach set of measurements. Flat platespecimens were prepared and markedwith the locations of data pointswhere shrinkage measurements wouldbe taken, as shown in Fig. 3. Platespecimens were fit up and tack weldedin preparation for welding of the sin-gle butt joint with the plates re-strained at the edges. Butt-joint weldswere produced and measurements

were taken after each side was welded. The detailed panel fabrication andmeasurement processes used were asfollows:• Clamp down and tack weld• Release clamps and clamp at three

points• Measurement 1• Clamp down• Weld first side and keep clamped• Release fixture and clamp at three

points

WELDING RESEARCH

NOVEMBER 2014 / WELDING JOURNAL 425-s

Fig. 8 — Effect of material on out­of­plane distortion.

Table 4 — Across­Weld and Along­Weld Shrinkage Data (HSLA­80)

Plate Material Thickness Total Heat Input Average Across­Weld Shrinkage (mm) Average Along­Weld Shrinkage (mm/m) (mm) (kJ/mm) A­H B­G C­F Av Avm I­Q J­P K­O Av Avm

43 HSLA­80 5.10 5.08 1.14 1.1 0.60 0.33 0.26 0.34 0.29 0.27 0.32 0.26 0.2844 HSLA­80 5.12 1.13 0.46 0.26 0.30 0.30 0.35 0.30 0.21 0.29 0.3045 HSLA­80 5.01 1.13 0.25 0.28 0.25 0.23 0.38 0.27 0.32 0.32†40 HSLA­80 5.07 5.06 1.60 1.6 0.64 0.51 0.52 0.53 0.60 0.56 0.36 0.25 0.3941 HSLA­80 5.07 1.53 0.87 0.56 0.50 0.60 0.75 0.57 0.36 0.56 0.4442 HSLA­80 5.04 1.55 1.07 0.64 0.53 0.67 0.45 0.38 0.30 0.3846 HSLA­80 5.03 5.05 2.11 2.1 1.23 0.81 0.70 0.86 0.77 0.64 0.56 0.30 0.5047 HSLA­80 5.09 2.08 0.88 0.73 0.62 0.70 0.73 0.57 0.37 0.56 0.5548 HSLA­80 5.03 2.12 0.93 0.71 0.75 0.77 0.82 0.60 0.38 0.60

25 HSLA­80 6.17 6.19 1.14 1.1 0.10 0.09 0.05 0.07 0.10 0.20 0.15 0.13 0.1626 HSLA­80 6.22 1.15 0.15 0.15 0.14 0.15 0.19 0.19 0.20 0.19 0.1827 HSLA­80 6.17 1.15 0.13 0.13 0.08 0.09 0.24 0.17 0.16 0.19

8 HSLA­80 6.41 6.48 1.68 1.7 0.55 0.37 0.31 0.38 0.31 0.47 0.56 0.57 0.539 HSLA­80 6.52 1.68 0.07 0.23 0.32 0.23 0.34 0.34 0.24 0.3110 HSLA­80 6.54 1.72 0.62 0.24 0.32 0.32 0.57 0.50 0.30 0.46 0.3617 HSLA­80 6.46 1.63 0.68 0.24 0.22 0.34 0.21 0.23 0.15 0.2018 HSLA­80 6.50 1.66 0.32 0.28 0.24 0.27 0.33 0.23 0.29 0.28

28 HSLA­80 6.24 6.19 2.11 2.1 1.10 0.71 0.57 0.71 0.66 0.38 0.43 0.34 0.38

29 HSLA­80 6.18 2.13 1.04 0.60 0.49 0.64 0.45 0.42 0.37 0.41 0.3930 HSLA­80 6.15 2.14 0.87 0.63 0.54 0.62 0.39 0.33 0.39 0.37

34 HSLA­80 9.86 9.90 1.62 1.6 0.30 0.19 0.19 0.22 0.17 0.13 0.15 0.15 0.1435 HSLA­80 9.96 1.67 0.13 0.09 0.09 0.09 0.13 0.18 0.10 0.14 0.1436 HSLA­80 9.87 1.66 0.22 0.21 0.19 0.19 0.16 0.15 0.15 0.15

31 HSLA­80 9.98 9.96 2.21 2.2 0.58 0.35 0.33 0.41 0.42 0.25 0.28 0.31 0.2832 HSLA­80 10.07 2.22 0.67 0.38 0.44 0.47 0.16 0.31 0.41 0.29 0.2933 HSLA­80 9.83 2.14 0.51 0.37 0.32 0.37 0.28 0.29 0.31 0.29

37 HSLA­80 9.88 9.89 2.63 2.6 0.70 0.43 0.44 0.50 0.45 0.32 0.32 0.30 0.3138 HSLA­80 9.87 2.54 0.68 0.45 0.36 0.44 0.43 0.47 0.46 0.45 0.3439 HSLA­80 9.92 2.53 0.59 0.36 0.30 0.40 0.25 0.27 0.27 0.26

Yang Supplement November 2014[1]_Layout 1 10/16/14 9:36 AM Page 425

• Measurement 2• Turn panel over, clamp down, weld

second side• Turn panel back, align the panel close

to the original coordinate system (notexact), and clamp at three points

• Measurement 3 (final). A series of measurements was takenusing the portable coordinate measur-ing machine to determine measurementrepeatability. The results of this studyare shown in Fig. 4. The mean measure-ment error is 0.026 mm.

Shrinkage and DistortionCalculation

Corresponding to the welding direc-tion, weld shrinkages include along-weld shrinkage and across-weldshrinkage. Along-weld shrinkage isalso called longitudinal shrinkage andacross-weld shrinkage is also called

transverseshrinkage.

Across-weldand along-weldshrinkage werecalculated bysubtracting thedistance betweenpairs of points af-ter welding from

the distance before welding as shown inFig. 5. The data pairs used for calculat-ing across-weld shrinkage were the dis-tances between rows A-H, B-G, and C-F,respectively. The data pairs used for cal-culating along-weld shrinkage were thedistances between columns I-Q, J-P, andK-O, respectively. The L-N data pairswere not used for calculating shrinkagesbecause the L-N measurements are rela-tively less accurate than the I-Q, J-P,and K-O. Generally, the points in a datapair are far away from each other. Themeasurement has high accuracy. While weld shrinkage and distortionwere calculated after welding side 1 andafter welding side 2, only the totalshrinkage and distortion (after weldingside 2) are discussed in this paper. Out-of-plane distortion was calcu-lated by taking the Z-coordinate dif-ference before welding and after weld-ing. The distortion maps for the four

materials will be contour plotted withthe data points shown in Fig. 5.

Results and Discussion The results of this work are pre-sented in the following sections withrespect to welded panel shape, weldmacrographs, measured shrinkage anddistortion data, material type, andheat input.

Welded Panels andMacrographs

All panels listed in Table 1 were weld-ed with the fixture and procedures dis-cussed in the previous section. Figure 6shows a typical panel deformed shapefor all materials. The panel edges atweld start and weld stop were pulled upwhile the middle of panel edges alongthe welding direction were pusheddown. The deformation was a typicalsaddle buckling distortion, which is of-ten observed in bead-on-plate welding(Refs. 11, 21). The change in buckling distortionmode was observed after welding ¼-in.-(6.4 mm) thick HSLA-80 with high heatinput (Panels 28–30). After the first sidewas welded and the clamps were loos-ened, the panel distorted such that themiddle of the panel rose up while bothends of the weld curved down. Measure-ments were taken in this configuration.When the panel was moved to turn itover, it was noticed that the bucklingmode was changed to the oppositeshape. For a thin sheet, the mode couldbe readily changed and there was nopreference for one mode over the other.

WELDING RESEARCH

WELDING JOURNAL / NOVEMBER 2014, VOL. 93426-s

Fig. 9 — Out­of­plane distortion distributions: A — Low heatinput (1.14 kJ/mm); B — middle heat input (1.65 kJ/mm); C— high heat input (2.09 kJ/mm).

A

C

B

Table 5 — Effect of Materials on Across­Weld and Along­Weld Shrinkage

Panel Material Thickness Total Heat Across­Weld Along­Weld (mm) Input (kJ/mm) Shrinkage (mm) Shrinkage (mm)

2, 3, 4 DH­36 6.31 1.64 0.46 0.495, 6, 7 HY­80 6.70 1.66 0.26 0.34

8, 9, 10, 17, 18 HSLA­80 6.48 1.68 0.31 0.3611, 12, 13 HY­100 6.77 1.62 0.26 0.29

Yang Supplement November 2014[1]_Layout 1 10/16/14 10:21 AM Page 426

A small force or pressure change on theplate can change the buckling modeshape. After turning the plate over, theplate weight could result in the changeof buckling mode. After welding the sec-ond side, the buckling was locked inwith the middle of the plate low, andthe ends of the weld up. Weld cross-section macrographswere prepared to investigate the weldbead profiles for materials of DH-36,HSLA-80, HY-80, and HY-100 at thesame thickness and heat input. Threemacrographs taken in the middle ofthree welded panels were prepared foreach material. Representative crosssections of double-sided square groovewelds on the different materials areshown in Fig. 7. The similar weld-beadshapes were found among the fourmaterials. HSLA-80 had slightly morepenetration than DH-36 and HY-80.HY-100 had the least penetrationamong the four materials.

Tensile tests were conducted tocheck the material strength of DH-36,HSLA-80, HY-80, and HY-100. Table 2shows the average values of threereplicas. It was found that HSLA-80has lower strength than HY-80. Allmaterials met the minimum requiredstrength for the material class.

Measured PanelShrinkages andDistortion Data

Across-weld andalong-weld shrink-age data are shownin Table 3 for theDH-36, HY-80, andHY-100 panels, andin Table 4 for the

HSLA-80 panels. In the tables, Avm isthe averages values for a material.Each table includes the measured platethickness and total heat input. Totalheat input is the sum of side-1 andside-2 heat input. In Table 3, the values in column Avare the average values of column A-H,B-G, and C-F, which are the across-weld shrinkages for panels. The val-ues in the column Avm are the aver-age of values in the column Av, whichare the average across-weld shrink-ages for materials. In Table 4, the values in column Avis the average values of column I-Q,J-P, and K-O, which are the along-weld shrinkages for panels. The val-ues in the column AVM are the aver-age of values in the column Av, whichare the average along-weld shrinkagesfor materials. By processing and understandingthe measured shrinkage data, the re-lationships between materialstrength and shrinkage and betweenheat input and shrinkage were estab-lished. This relationship will be dis-cussed in the following sections. The out-of-plate distortion datafor each panel are discussed in thefollowing section using contour plots

to develop the relationships betweenmaterial strength and distortion andbetween heat input and distortion.

Material Strength Effect onShrinkage and Distortion

The effect of material strength onshrinkage and distortion can be deter-mined from the data from Tables 3and 4 for the same nominal thickness(6.35 mm) for the four materials.

Shrinkage

Table 5 lists the weld shrinkagedata for DH-36, HY-80, HSLA-80, andHY-100 with a nominal ¼ in. (6.35mm) thickness and welded with medi-um heat input. The shrinkage datawere extracted from Tables 3 and 4.The average thickness was 6.31 mmfor DH-36, 6.70 mm for HY-80, 6.48mm for HSLA-80, and 6.77 mm forHY-100. The average total heat inputwas 1.64 kJ/mm for DH-36, 1.66kJ/mm for HY-80, 1.68 kJ/mm forHSLA-80, and 1.62 kJ/mm for HY-100. Both the thicknesses and heat in-puts between the four materials havesome differences. These differenceswill be considered in the comparisonof weld shrinkage between the fourmaterials. It should be noted that data varia-tions were observed on the HSLA-80steel. Therefore, two additional panelswere welded to ensure the reliability ofthe measured shrinkage data. By comparing the across-weldshrinkage and along-weld shrinkagesamong the four materials in Table 5, itwas found that the higher strength

WELDING RESEARCH

NOVEMBER 2014 / WELDING JOURNAL 427-s

Fig. 10 — Effect of heat input on in­plane shrinkage and out­of­plane distortion for DH­36: A — Across­weld shrinkage; B —along­weld shrinkage; C — maximum out­of­plane distortion.

A

C

B

Yang Supplement November 2014[1]_Layout 1 10/16/14 9:36 AM Page 427

materials (HY-80, HSLA-80, and HY-100) shrink less than the lowerstrength material (DH-36). However,the higher strength materials areslightly thicker than the lowerstrength materials when welded usingthe same conditions, including weld-ing heat input. Thus, normalizedshrinkage was calculated with Equa-tion 1 to take into account the minorvariations of thickness and heat inputin order to compare shrinkage be-tween the higher strength materialsand the lower strength material.

where is shrinkage, n is the normal-ized shrinkage, t is thickness, tn is thenormalized thickness 0.25 in. (6.35mm), q is total heat input, and qn is thenormalized total heat input 1.68kJ/mm. The measured and normalizedacross-weld shrinkage data are shownin Table 6. The shrinkage ratio relativeto DH-36 was calculated, which wasused as a multiplier in the shrinkagemodels developed for a uniform panel(Ref. 9) and a complex panel (Ref. 10).HY-80, HSLA-80, and HY-100 have 60,67, and 60% of across-weld shrinkageof DH-36, respectively. HSLA-80 hasmore shrinkage than HY-80. This maybe the result of two factors: HSLA-80

has lowerstrength thanHY-80, as shownin Table 2, andHSLA-80 hasslightly moreweld penetrationthan HY-80, as

shown in Fig. 7.The measured and normalized

along-weld shrinkage data are shownin Table 7. The shrinkage ratio relativeto DH-36 was calculated, which wasused as a multiplier in the shrinkagemodels developed for a uniform panel(Ref. 9) and a complex panel (Ref. 10).HY-80, HSLA-80, and HY-100 have 76,79, and 63% of along-weld shrinkageof DH-36, respectively.

Out­of­Plane Distortion

Out-of-plane distortions were calcu-lated based on the method discussed in

the section titled Shrinkage and Distor-tion Calculation and were contour plot-ted as shown in Fig. 8. Figure 8 showsthe typical out-of-plane distortion dis-tributions for each material. All fourmaterials have the same buckling mode.The numbers in Fig. 8 are the highestand lowest distortion magnitude for thematerials. The maximum distortionscan be calculated by taking the differ-ence between the highest and lowestdistortion. It can be found that themaximum distortion magnitude is 19.1mm for DH-36, 16.7 mm for HY-80,20.4 mm for HSLA-80, and 14 mm forHY-100, respectively. HY-80 and HY-100 material had lower magnitudes ofout-of-plane distortion than DH-36 ma-terial, while HSLA-80 has slightly high-er distortion magnitude than DH-36.

Discussion

The data show that higher strengthtt

qqn

n nδ = δ (1)

WELDING RESEARCH

WELDING JOURNAL / NOVEMBER 2014, VOL. 93428-s

Fig. 11 — Effect of heat input on in­plane shrinkage and out­of­plane distortion for HSLA­80: A — Across­weld shrinkage; B —along­weld shrinkage; C — maximum out­of­plane distortion.

A

C

B

Table 7 — Effect of Materials on Along­Weld Shrinkage

Along­Weld Shrinkage (mm)Material Shrinkage After normalized both thickness and heat input Ratio

DH­36 0.49 0.47 1.00HY­80 0.34 0.36 0.76

HSLA­80 0.36 0.37 0.79HY­100 0.29 0.30 0.63

Table 6 — Effect of Materials on Across­Weld Shrinkage

Across­Weld Shrinkage (mm)Material Shrinkage After normalized Ratio

both thickness and heat input

DH­36 0.46 0.47 1.00HY­80 0.26 0.28 0.60

HSLA­80 0.31 0.32 0.67HY­100 0.26 0.28 0.60

Yang Supplement November 2014[1]_Layout 1 10/16/14 10:18 AM Page 428

steels (HY-80, HSLA-80, and HY-100)shrink less than the lower strengthsteel (DH-36). This fact could be ex-plained by material strength differ-ences between these materials. Higher strength materials will haveless along-weld shrinkage than lowerstrength materials if applying thesame force because more plastic strainwill be induced in the lower strengthmaterial. Since shrinkage is the out-come of plastic strains, lower strengthmaterials will have more shrinkage.

Heat­Input Effect on Shrinkageand Distortion

A comparison of the in-planeshrinkages and out-of-plane distor-tions as a function of heat input forDH-36 and HSLA-80 material was in-vestigated. One thickness (6.35 mm)was conducted for DH-36 and threethicknesses (4.76, 6.35, and 9.53 mm)were conducted for HSLA-80.

DH­36 Steel

The effect of heat input on shrinkageand distortion for 6.35-mm-thick DH-36 was studied by processing the data ofnine panels (panels 19–21, 2–4, and22–24). As shown in Table 3, panels19–21 were welded with a low heat in-put of 1.2 kJ/mm, panels 2–4 with a

middle heat inputof 1.6 kJ/mm, andpanels 22–24 withhigh heat input of2.1 kJ/mm. Themeasured data forthe nine panelswere processed to

calculate the across-weld shrinkage andalong-weld shrinkage with the samemethod illustrated in Table 3. The typical out-of-plane distortionsfor low, middle, and high heat inputwere contour plotted, as shown in Fig.9. As heat input increases, out-of-planedistortion increases. Note that the dis-tortions were magnified five times tobetter visualize the distortion shape.The maximum distortion was also cal-culated as shown in each figure. Panel19 with low heat input had 12.9-mmmaximum distortion, panel 2 with mid-dle heat input had 19.1-mm maximumdistortion, and panel 22 with high heatinput had 21.4-mm maximum distor-tion. The maximum distortions for allnine panels were plotted in Fig. 10C. Only the average across-weldshrinkage and average along-weldshrinkage for each panel were plotted,as shown in Fig. 10. Figure 10 showsthe relationships between across-weldshrinkage, along-weld shrinkage, andout-of-plane distortion and heat in-puts, respectively, for DH-36. This re-lationship will be implemented in theshrinkage model developed for uni-form panel (Ref. 9) and complex panel(Ref. 10). Across-weld shrinkage (Fig.10A) increases linearly as heat inputincreases. Along-weld shrinkage andout-of-plane distortion do not increaselinearly, as shown in Fig. 10B, C. Fig-

ure 10C plots the Delta values (maxi-mum out-of-plane distortion) in Fig. 9.After heat input increases beyond 1.65kJ/mm, along-weld shrinkage and out-of-plane distortion increase slowly.

HSLA­80 Steel

The effect of heat input on shrink-age and distortion for 4.76-, 6.35-, and9.53-mm-thick HSLA-80 was studiedby welding 29 panels as shown inTable 1. The measured data for the 29panels, as shown in Table 4, wereprocessed to calculate the across-weldshrinkage and along-weld shrinkagewith the same method illustrated inTable 3. Only the average across-weldshrinkage and average along-weldshrinkage were plotted in Fig. 11A, B. As shown in Fig. 11A, across-weldshrinkages for 4.76-, 6.4-, and 9.53-mm-thick HSLA-80 increases as heatinput increases. When heat input in-creases beyond 2.2 kJ/mm, the across-weld shrinkage increases slowly. Forthe same heat input, as the plate thick-ness increases, across-weld shrinkagereduces. This is because material melt-ing area will be reduced when materialthickness increases. Figure 11B shows that along-weldshrinkage increases almost linearly asheat input increases. For the same heatinput, as the plate thickness increases,along-weld shrinkage reduces. This isbecause shrinkage resistance increasesas material thickness increases. Figure 11C plots the maximum out-of-plane distortion that was calculatedwith the same method used in Figs. 9and 10C. The contour plots similar toFig. 9 were omitted in this paper and

WELDING RESEARCH

NOVEMBER 2014 / WELDING JOURNAL 429-s

Fig. 12 — Comparison of shrinkage between material DH­36and HSLA­80: A — Across­weld shrinkage ( t = 6.35 mm); B—along­weld shrinkage (t = 6.35 mm); C — maximum out­of­plane distortion.

A

C

B

Yang Supplement November 2014[1]_Layout 1 10/16/14 9:36 AM Page 429

only the maximum out-of-plane distor-tion in panels were plotted, as shown inFig. 11C. Figure 11C shows that maxi-mum out-of-plane distortion increaseslinearly as heat input increases. Highheat input induces high out-of-planedistortion. For the same heat input, asthe plate thickness increases, out-of-plane distortion reduces. This is becausestructural rigidity increases as materialthickness increases. Overall, in all cases, the extent ofshrinkage and distortion increase as afunction of heat input. This relationshipwill be implemented in the shrinkagemodel developed for the uniform (Ref.9) and complex panels (Ref. 10).

Comparison of DH­36 and HSLA­80Steels

The effect of heat input on shrinkageand distortion was conducted on bothDH-36 and HSLA-80 materials with anominal thickness of 6.35 mm. By com-paring the shrinkage and distortion forthe two materials, it was found thatHSLA-80 had less shrinkage than DH-36 at low and middle heat inputs. For high heat input, HSLA-80 andDH-36 have similar across-weld shrink-age, as shown in Fig. 12A. Therefore,the ratio of HSLA-80 and DH-36 inTable 6 depends on the heat input. Itwas found that HSLA-80 has less along-weld shrinkage than DH-36 for all heatinputs, as shown in Fig. 12B. Figure 12Cshows that HSLA-80 has similar maxi-mum out-of-distortion as DH-36 at lowand middle heat inputs. At high heat in-put, HSLA-80 has higher maximumout-of-plane distortion than DH-36.

Conclusions

The following conclusions could bedrawn based on the weld shrinkageand distortion data gathered in thispaper: • High-strength steels (HSLA-80,HY-80, and HY-100) have less across-weld shrinkage than low-strengthsteel (DH-36) for normal heat input.For high heat input (larger than 2kJ/mm), high-strength steel (HSLA-80) has similar across-weld shrinkageas low-strength steel (DH-36). • Higher strength steels have lessalong-weld shrinkage than low-strength steels for all heat inputs stud-ied when the thickness is the same.

• HY-80 and HY-100 steels havelower out-of-plane distortion thanlower strength EH36 steel, whileHSLA-80 steel has similar out-of-planedistortion compared to DH-36. Forhigh heat input (larger than 2kJ/mm), HSLA-80 steel has slightlyhigher out-of-plane distortion thanlower strength EH36 steel. • Shrinkage and out-of-plane distor-tion increase as heat input increases. • For the same heat input, as thick-ness increases, both shrinkage andout-of-plane distortion decreases.

The results of this paper were fromthe project Weld Shrinkage and Dis-tortion Allowance Data Model for NeatConstruction Ship Design Engineer-ing, funded by the National Shipbuild-ing Research Program Advanced Ship-building Enterprise (NSRP-ASE). Theauthors thank NSRP-ASE for support-ing this study.

1. Gurney, T. R. 1979. Fatigue of WeldedStructures, 2nd Ed. Cambridge, UK: Cam-bridge University Press. pp. 226–243. 2. Yin, D., and Qiang, S. 1991. Influenceof residual stresses and deformation onbuckling behaviour of plates. MechanicalEffects of Welding (Eds. L. Karlsson, L. E.Lindgren, and M. Jonsson). IUTAM Sym-posium, Sweden. pp. 269–276. 3. Masubuchi, K. 1980. Analysis of Weld-ed Structures. Oxford, UK: Pergamon Press. 4. Michaleris, P. 2011. Minimization ofWelding Distortion and Buckling: Modelingand Implementation. Cambridge, UK:Woodhead Publishing Limited. 5. Feng, Z. 2005. Processes and Mecha-nisms of Welding Residual Stress and Distor-tion. Woodhead Publishing Limited andCRC Press LLC: 13–24. 6. Verhaeghe, G. 1998. Predictive for-mula for weld distortion — A critical re-view. TWI Research Document, PRAD NO.7346.02/97/961.3. 7. Spraragen, W., and Ettinger, W. G.1950. Shrinkage distortion in welding.Welding Journal 29(6): 292-s to 294-s. 8. White, J. D., Leggatt, R. H., andDwight, J. B. 1980. Weld shrinkage predic-tion. Welding and Metal Fabrication 11:587–596. 9. Yang, Y. P., Castner, H., Dull, R.,Dydo, J., and Fanguy, D. 2013. Uniform-panel weld shrinkage data model for neat

construction ship design engineering.Journal of Ship Production and Design 29(1):1–16. 10. Yang, Y. P., Castner, H., Dull R., Dydo,J., Huang, T. D., Fanguy, D., Dlugokecki, V.,and Hepinstall, L. 2014. Complex-panel weldshrinkage data model for neat constructionship design engineering. Journal of Ship Pro-duction and Design 30(1): 1–24. 11. Yang, Y. P., and Dong, P. 2012. Buck-ling distortions and mitigation techniquesfor thin-section structures. Journal of Ma-terials Engineering and Performance 21(2):153–160. 12. Conrardy, C., Huang, T. D., Harwig,D., Dong, P., Kvidahl, L., Evans, N., andTreaster, A. 2006. Practical welding tech-niques to minimize distortion in light-weight ship structures. Journal of Ship Pro-duction 22(4): 239–247. 13. Yang, Y. P., Dull, R., Conrardy, C.,Porter, N., Dong, P., and Huang, T. D. 2008.Transient thermal tensioning and numericalmodeling of thin steel ship panel structures.Journal of Ship Production 24(1): 37–49. 14. Huang, T. D., Harwig, D. D., Dong,P., and DeCan, L. A. 2005. Engineering andship production technology for lightweightstructures. Technology Review Journal13(1): 1–22. 15. Huang, T. D., Conrardy, C., Dong, P.,Keene, P., Kvidahl, L., and DeCan, L. 2007.Engineering and production technologyfor lightweight ship structures, Part II: Dis-tortion mitigation technique and imple-mentation. J. Ship Production, The Societyof Naval Architects and Marine Engineers23(2): 82–93. 16. Watanabe, M., and Satoh, K. 1961.Effect of welding conditions on the shrink-age and distortion in welded structures.Welding Journal 40(8): 377-s to 384-s. 17. Gunaraj, V., and Murugan, V. V. 2005.Effect of process parameters on angular dis-tortion of GMAW structural steel plates.Welding Journal 84(11): 165-s to 171-s. 18. Sadat, A. M., Rao, S., and Rao, N.2012. Modelling the effects of preheatingon angular distortions in one sided filletwelds. Journal of Achievements in Materialsand Manufacturing Engineering 55(2):578–582. 19. Vural, M., Muzafferoglu, H. F., andTapici, U. C. 2007. The effect of weldingfixtures on welding distortions. Journal ofAchievements in Materials and Manufactur-ing Engineering 20(1-2): 511–514. 20. Zhu, X. K., and Chao, Y. J. 2002. Ef-fects of temperature-dependent materialproperties on welding simulation. Comput-ers and Structures 80: 967–976. 21. Wang, J., Yin, X., and Murakawa, H.2013. Experimental and computationalanalysis of residual buckling distortion ofbead-on-plate welded joint. Journal of Ma-terials Processing Technology 213(8): 1447–1458.

WELDING RESEARCH

WELDING JOURNAL / NOVEMBER 2014, VOL. 93430-s

Acknowledgments

References

Yang Supplement November 2014[1]_Layout 1 10/16/14 9:36 AM Page 430