0~,.CWC,$ THE SOCIETY OF NAVALARCHITECTS AND MARINE ENGINEERS

* One World Trade Center, Suite 1369, New York, N,Y, 1D04S$’

410

~.e PaPers to k ~r,,e”tedatExtremeLoadsR,,p.an,.symw,t”m$ : Arlington,VA,October19-20,19817* ●.“o= &

% .SS*$

Some Extreme Effects of Residual Stressesin ShipbuildingAlfred H. Wells, Jr., Newport News Shipbuilding, Newport News, VA

ABSTRACT

Duringthepastfourdecades, a wealth of informationhas been written on the subject of residual stresses andtheir effects cm structural materials. Of particular interesthas been tbe role of residual stresses in contributing to brit-tle failures. Such failures usualiy occur without warning inlow stress conditions, and often are catastrophic. Thesefailures, or fractures, are the more dramatic manifestationsof a number of contributing factors which may imluderesidual stresses. A number of different sources may gener.ate residual stresses, but the one sowce having the greatestsignificance m shipbuilding is welding. Tbe distortioneKects of residual s&scs a;e visibly evi-derd after weld!ng a

;

fabricated structure and it is this distortion that is a came ofcontinual straightening problems.

8U11 WELD

A more pragmatic approach is used in this paper tolook at residual stresses than is usually found in cdhcr pa-pers on the subject. Fotlowing a brief description of thenature of resid ual stresses in welded construction &nd theireffects on structures, three recent case histories of the kindof problems facing shipbuilders are disc” ssed, Thefirstcasereviewsthemultiplefracturesexperiencedon a ham%? pipeweld overlay during fabrication and the subsequent changesin procedures emptoyed to remedy the crwk~ng problem.The second concerns the angular distortion produced bybutt welding thick steel plates and describes a welding testundertaken to mhimize the pmblcm, The third casedescribes the fracture of the trailing edge casting of a“ oiltankerduring construction and reviews the contributingcauses and their effects.

Y

TENSION

Y

DISTRIBUTION OF u.. ALONO YY

—

I7ENS10N

x— —x

01STR10UTt(X4OFmy ALONG xx

RESIDUAL STRESS DISTRIBUTIONSIN A BUTT WELD

FIGIJRE!

I

189

INTRODUCTION

Residual stresses are the internal stresses whichremain within the material ofa plate or shape after the ther-mal cycle of welding. Theyaregeneratedbythehighlyh-calized heating and subsequent cooling OF the weld metalcausing thermal contraction of the surmundi”g base metalin the heat-affected zone. As the molten weld metal solid-ifies, it exerts tensile shrinkage stresses on the adjacentmetal which usually increase to the yield point of the

.material as the ambient temperature is rezched. Welding isnormally performed linearly i“ a progressive manner sothat the completed portion of the weld resists tbe weld beaddeposited beyond it. Thisproduces Iongit”dinal temilestresses in the weld and in the adjacent base metal. Othertensile residual stresses are atso produced in the transversedirection of a butt weld as the cooler underlyin~ weld passesresist the shrinkage of subsequent depositions. Thesestressesexisti“any weldmentwithout[hepresence ofexter”dforcesand consist of a system of opposing stresseswhich balance one another.Figure 1 illustratesthelongitudinaland transverseresidualstressdistributionsi“abuttweldwithout external restraint (1).

Residualstressesarealsokmwn by a variety of othernames such as initial stresses, locked-in stresses, reacticmstresxes, shrinkage or contraction stresses, and inherentstresses. Thesestressesmay alsobe thermal stresses, whichare imernal stresses that exist during the process of heatingand cooling when the temperate change is nonuniforma“d a stress remains. Residual stresses exist in variouskinds of metals and are produced by many sources, Amongthese are the rolling, forging, and casting operations nems.sary to produce shapes and plates. In a.dditirm, the wmki”gand shaping processes applied m metals such as machi”i”g,cutting, grinding, burning, and bending wilt also developresidual stresses m v m ying levels.

There are two major effects of residual stresses inwelded structmes. The first is distortion which is caused bythe shrinkage stresses acting on a welded joint from 10.calized heating and cooling. The weldment deviates from itsprim position as the stresses produce local yielding and apermment redistribution occurs. The resulting dimensiomdchange is usually a reduction in size accompanied by overaOwarpage in several directions. Distortion is always presemin varying degrees when welding materials of any thtcknessand it must he taken into consideration i“ order to remainwithin most mmtructicm mlerames.

The second major effect of residual stresses is thatthey may be the cause of prernatwe failure by brittle frac-ture at low applied stress levels. All wddments containmultiaxial stresses at or near the beat-affected zone of theweldandthesestressesmay heofyieldpointmagnitude.When a small flaw occurs in a regicm of high tensile residualstress, the stress intensity is increased considerably. Thedefect or crack may grow i“ m unstable m.mner until itpasses out of the tensile residu,d stress region. If the appliedstress level is low, the crack may arrest or may continue togrow depending cm tbe overall state of stress at the crack tipand the inherent m“ghness of the material. Brittle fractwemay be the End result of the crack propagationn wh~ch maynot occur without the presence of residual stress (1,2).

CASE HISTORIES OF SOME EXTREME EFFECTSOF RESIDUAL STRESSES

Weldedco”suuctionofnavaland commercialshipstructuresinvolves a wide variety of applications, many ofwhih require state-of-the-ml techniques needed to effe.c.lively join numerous types of materials. The effectsofresidualstressesprovide challenges to the shipbuilder toimprove the reliability of welded structures. Thethreecasestudiesdiscussedbelowaresome ofthosechallenges.

Aircraft Carrier Hawse Pipe Fractures

During the construction of an aircraft carrier atNewport News Shipbuilding, severe fractures occurred d“r-ing the fa.bricaticm of the port hawse pipe prim to shipboardinstdlatiom Residual stresses caused by the weld deprmiticmof the required hard-surface overlay were initially thoughtto be the principal cause of the problem, but subsequentexwninatiom revealed that there were other contribWi”gfactors as well. The exterior portion of the hmvse pipe as itextends through the ship’s shell near the bow is shown inFigwe 2.

The hawse pipe was comtructed by welding twodifferent castings to the ends of a center section 1220 mm(48 in.) in d,ameter and about 2.4 m (8 ft) long. This cemcrsection is comprised of two semi-circular, rolled, mild-steelplates, 44 mm (1 .75 in.) thick above and 70 mm (2,75 i“.)below, welded together longitudinally. The total weight of15.38 Mgl (33 900 lb) is almost equalty distributed amongthe three main pieces. The material used for the upper a“dlower castings is similar to mild steel and meets the require-ments of Military Specification MIL-S-15083B(NAVY) forGrade B steel with a yield strength of 207 MPa (30 000 psi)and a tensile stre”gtb of 414 MPa (60 000 psi).

The ptmmed sequence of assembly was to butt weldthe upper casting m thecentecrolledsecdo”,addtheweldoverlaytotheIowcrinteriorsurfacein the fabrication shop,and then install these two sections into position on boardtbe ship. The tower casting, forming the exterior portion ofthe assembly, was also clad i“ the shop a“d then erectedonboard the ship by welding to the center section and to theship’s shell plating. Figure 3 illustrates the compl.sted porthawse pipe assembly after erection.

No cracking problems were encountered with theuPPercastinguntiltherequiredhard-surfaceoverlaywasdepositedby weldingtotheinteriorsurfaceof thehawsepipe.This510-mm (20-in. ) wide weld deposit extends forthe fdl length of the assembly a“d spreads cwt to 1020 mm(40 in.) where the upper casting flattens out and widem toaccommodate the two different lead angles of the anchorchain. The purpose of this 6.4-mm (1/4-in.) thick welddeposit is to provide an abrasion-resistant surface to pre-vent excessive wearing of the hawse pipe by the anchorchain. After preheating the assembly to a minimum of121”C (250°F), the overlay was deposited“~i”ga Semi..“tmnaticfluxcoredarcweldingprocesswitha qualified,commercial,hard-surfacingelectrode.Themaximwn inter-Passtemperatureallowedwas 204°C (400”F).

1 Megagrams (1 Mg = 1 metric ton). The IntematiomdSystem of Units (S1) used i“ this paper followsthe guidance of the ASTM E380.79 “Standard forMETRIC PRACTICE,,.

190

AIRCRAFT CARRIER HAWSE

FIGURE 2

191

PIPE EXTERIOR

FWDROLLED PLATE

44 mm (1.75 in, ) 7 ~’ UPPER CASTINGnsrv 0, Amr.I~— -—- u._

/.— 7. ~

~-—-T7’- 5.%/’1Q

-----..-,,,.—..+/(

3-”$ROLLED PLATE

WELD DEPOSIT 7D mm (2.75 in.)—.6.4 mm (X in. )

SECTION A-A 3 $

O..* /

pe>%,

,“

/

m,,-” . . . . . . . .

I

WELD DEPOSIT

6.4 mm (U in.)

SHELL PLATINGJ

4

“.sm I- LA, ,I.JQ

/>

71 SECTION C-c/

IDECK PLATING

WELD DEPOSIT6.4 mm (X in.)

SECTION B-BFR 5-U

PORT HAWSE PIPEELEVATION AT CENTERLINE

FIGURE 3

—, --- . .r

Afterthehard-surfaceoverlaywasdepositedalongapartiallength of the hawse pipe, a fracture occurred in theuPper casting during a weekend when no work was beingperformed on it, Thecrackhadformedatthetopsurface ofthe overlay in the transverse direction and had propagatedinto the casting to a depth of 25 mm (1 in.) and hadextended for a length of 610 mm (24 in.). The casting wasweld repaired in accordance with approved procedures withpeening being employed on alternate layers of weld metal.A magnetic particle inspection was performed on the motzf~e.rgouging to sound metal and also on the final surface.The hard-surface overlay was reapplied using a wanderingsequence to avoid local overheating and, upon completionof the welding, the preheat was slowly removed.

During the next 6 months, 10 additionalfmct”resoccurredin the upper casting at different locations as shownin Figure3.Thesecracksrangedfrom25mm (1in.) to 79mm (3-1/8 in.) deep and from 406 mm (16 in.) to 840 mm(33 in.) in length and were spread out over an arc. of 1,1-m2 (12-ft2) in theuppercasting.Thecastingthicknesses inthe affected areas varied from 83 mm (3.1/4 in.) to 95 mm(3-3/4 in.) Plus 6.4 mm (1/4 in.) for the weld overlay. Afterseveral fractures were repaired with the flux coredelectrode, the weld overlay was deposited using a covertdweid!ng electrode conforming to Military SpecificationMlL-E-19141C, type MIL-I-A21, which has a requiredminimum BrineO hardness of 200.400 (as deposited). Thiselectrode is made from an iron-based alloy and the weldwas deposited in a wandering sequence by shielded metalarc welding. Since additional fractures continued to occurafter the electrode was changed, it was concluded that thehard-surfacing deposit was being subjected to tensilestresses after welding and that the removal of the preheatand subsequent movement of the assembly was the likelycause of the cracking. Thelastweldrepairwasaccomplishedafterthehawsepipewas welded into its find positiononboard the SKIP in an effort to restrain the movementfrom preheating. The hard-surfacing was applied after thecasting was preheated to only 16-C (60°F) and skip weld-in’& using short increments, was employed with a 66-C(lSO”F) maximum interpass temperature. The final weldrepair, which was later ground smooth to preclude wearingof the chain, is visible in Figure 4.

The causes of the multiple fractures experienced onthe port hawse pipe described above are interrelated mdtheir combhed efTects contributed to the unfortwmte e“dresdt. Residual stresses due to shrinkage of the weld over-lay are high on the list of likely contributing factors. Someof the other causes that were believed to add to the problemwere:

● the basic metallurgical structu~c of the iron-basedhard-surfacing alloy which is inherently prone tocracking,

● thecharacteristic shape of the surface and subsur-face weld deposit from flux cored arc weldingwhich creates potential discontin”ities as corm

●

●

●

paredtoshieldedmetalarcwelding;

the complex shape of the upper casting since it isa compound curve and varies rapidly fmm a cir-cular section to a “U” shape and then to a flatsection,

the preheating of the underside of the castingcausing thermal expansion to chmge the shape ofthe casting, and

the lack of restraint of the hawse pipe since itrested on blocks in the shop permitting the cast-ing to expand and contract freely.

The initial response to the problem was to excavatethe cracks a“d complete tbe necessary weld repairs. This

apprOach WaS insufficient as the fractures continued tooccur without a specific pattern from which definite conclu-sions could be drawn. The resolution to the cracking prob-lem described above involved eliminating as many of thelikely tames m possible. The various changesthatwereemployed were:

switchhg from flux cored arc to shielded metalarc welding with a corresponding change inelectrodes;

reducing the preheat from 121“C (250”F) to 16-C(60”F) prior to welding;

installing the hawse pipe onboard the ship to pre-vent movement during the repair welding of theoverlay, and

controlling the local heat build-up by usimg a wan.dering welding sequence and a 66°C ~15TF)maximum interpass temperature.

It should be pointed out that these solutions evolvedover a period of time and are not necessarily hard-and-fastrules to be applied elsewhere. In fact, there are many vag-aries in welding that still defy explanation. Forinstance, thestarboard hawse pipe, which is a mirror image of the porthawse pipe, was fabricated in a similar manner with fewercracking problems. The upper casting was overlaid withweld deposited by shielded metal arc welding in the shopafter preheating the casting to 121“C (250”F). Twotransverse cracks developed that were 25 mm (1 in.) deepand 610 mm (24 in.) long which were repaired. No furtheroccurrences were encountered as the overlay weldingprogressed to completion, The starboard hawse pipe assem-bly unit was then success ftdly welded to the ship’s deck andshell structure as originally planned without inducing anyadditional frmtu~es to the casting or to the hard-surfaceoverlay.

193

_ ——–-

PORT HAWSE PIPE-FINAL WELD REPAIR

FIGURE 4

194

,

Angular Distortion in Thick Plates

The welding of materials produces undesirable sideeffects commonly called distortion. Occurring in manyforms, distortion is the visible result of residual stressesproduced by the welding process. Due to the extreme heat.ing of the material during weld metal deposition and thesubsequent cooling to ambient temperature, thermalstrains are imposed on the grain structure throughout thematerial within the heat-affected zone. As the weld metalsolidifies, it shrinks slightly, but it cannot generate thestresses that are responsible for reduced dimensionalchanges or other changes to the weldment’s shape. Aftersolidification occurs, tbe weld metal continues to cool andcontract. This thermal contraction can generate stress in thesolidified metal, but the maximum stress is limited to theyield strength of the weld metal. The internal forces gener.ated result in plastic deformation which causes various dis.placements by bending, buckting, rotating, or shrinking theplate or shape. Various types of weld distortion are showni“ Figure 5.

.- ———.——-I 1

l---dROTATIONAL.,5,0,,,0.

L’JNO,, U.,NAL ,,.0,..

.,,,0,,,0,

BLONG,, U.INAL ,“,,.,..,

e,“CK,,NG .,s,.3 .,,0.

TYPESOf WELDDISTORTION,,. ”,, 5

As thethkknessofthematerialincreases,theprob-lem of distortion becomes more significant because thedegree of d]focuhy of eliminating or controlling theseunwanted displacements also increases. Distortion in awelded structure cannot be eliminated e“tircly became ofthe complex variables associated with the welding process.A prudent approach to resolving the problem of distortionis to minimize or reduce it to an acceptable level thd is ap-propriate for the intended service of the stmcture. It is

generally easier and more cost effective to achieve reduc-tions in distortion through proper planning of the weld jointdesign rather than postweld plate straightening during theconstruction phase.

One of the most common types of dhtortion found inwelding thick pkdes is angular distorticm at butt-weldedjoints. Rotation about the longitudinal axis of tbe weldoccurs when the weld metal at the face of the weld shrinksmore than at the root. Traditional shipbuilding practice fortbe welding of thcck plates in excess of 25 mm (1 in,) hasbeen to use an offset double-beveled joint design. Thebsvel is offset in order to control distortion and theincluded angle of the bevel is held to a minimum which isusually 45”. The larger bevel is partially welded on the firstside and then tbe plate is turned over to backgouge the motto sound metal. The smaller bevel is filled with weld metaland then the plate is turned back over to the first side tocomplete the joint. If excessive angular distortion is experi-enced during the welding operation on a particular side, thewelding is halted prematurely and the plate is tumcd overto weld on the other side to “pull” the plate back into posi-tion, thereby requiring several additional turns to finish thejoint. It is relatively easy to visualize how distortion can dis-rupt productirm scheddes and quickly increase costs. Theassociated problems become more acute as platethicknesses increase since straightening thick plates afterwelding is diftkult and time-consuming. As a last resort,the weld must be cut out to a sufficient depth to permit ahydraulic press to flatten out the plates which are thenrewddedtoachkvetherequiredflatnesstolerance.This

approachcannotbe implementedonboarda ship u“dtrconstruction and, therefore, mechanical straighteningmethods must be employed.

The angular distortion in butt welds was extensivelystudied by the Shlpbuild,ng Research Association of Japanas reported by Mmubuchi (3). Double-beveled butt weldswith several offset groove depths were investigated in orderto minimize the angular change due to welding from bothsides with and without jriint restraint. ThegraphinFigure6illustratestheweld joint design for various thicknesseswhich most successfully minimized the angdar distortionwith a single turn of the plate (3). For example, when weld-ing a butt joint in a 25-mm (l-in.) plate without jointrestraint, the ratio of bevel depths for minimizing angulardistortion, expressed as a percentage of the total plate thick-ness, is 60 to 40. The corresponding ratio of the weight ofweld metal deposited (wl/w2) will be 0.36 to 0.16, or 2.25,Therefore, it takes over twice as much weld metal on thebacking side to eliminate the distortion. A balanced weldjoint design should be used on a 32-mm (1.25 -in.) plate.For thicker plates “p to 64 mm (2.5 in.), 40/60 md 30/70joints are recommended to reduce distortion which posi-tions the larger bevel on the finishing side.

In order to further investigate the potential of weldingthick plates and reducing plate handling without increasingthe angular distortion, a welding test was recently con-ducted at Newport News Shipbuilding. Two 102-mm (4-in.)plates made of HY-80 steel were beveled to a 30170 jointconfiguration. Each plate was approximately 305 mm(12 in.) wide and 1520 mr.? (60 in.) long. Measurementstaken prior to welding i“dicatcd that tbe plates were fair towithin 0.80 mm (1/32 in.). Figure 7 illustrates tbe testplates in their distorted positions (exaggerated) after eachside was welded. The backing passes on the 30 percent sidewere deposited in the flat position hy submerged arc weld-

195

1{2

,?

\

FREE—

—

THICKNESS OF PLATE, inches

1 1-1/2 2 2-1/2

I I

rBACKING PASSES

L_ FINISHING PASSES

r

RESTRAINED

W1/v42a(hl/h2)2

+r--hl/h2

770/30

0 10 20 30 40 50 60 70

THICKNESS OF PLATE, mm

OPTIMUM BUTT WELD GEOMETRYTO MINIMIZE ANGULAR DISTORTION

FIGURE 6

ing in 12passeswithoutrestrainingtheplate.A dkplace-mentof6.4mm (1/4i~.)wasmeasuredaltheedgeoftheweldbyplacingastraightedgebetweentheouteredgesoftheplateand then measuring down totheplate.The plateswereturnedand therootoftheweldwas inspectedfmcracking after removing the ceramic backing tape used toeliminate the backgouging to sound metal. The finishingpasseson the70 percentside were completed in 43 sddi-tiomd passes using a 2.4-mm (3132-in.) wire electrode des-ignated as 1OOS-I. The preheat temperature was 93-C

(200°F) minimum and the interpass temperature was heldto 149-C (301TF) maximmn. The welding of the finishingside reduced the distortion by 3.2 mm (1/8 in.) leaving anthe overall distortion of 3.2 mm (1/8 in.). Since the widthof the plates must be taken into consideration when dis-cussing measurements of angular distortion, it is more ap-propriate to refer to the angle formed by tbe intersectingsurface planes of the plates. This angle, representing theoverall distortion, was approximately 2.0” which wouldresult in significant unfairness when welding wide plates.

6.4 mm (1/4 in.)

[~

. .

E?E,:,

ml~~

--—--

.

1305 mm 305 mm J(12 in.) (12 in.)

BACKING PASSES COMPLETED

2.0”

-- —--

3.2 mm (1/8 in.)

FINISHING PASSES COMPLETED

ANGULAR DISTORTION FROM BUTT

FIGURE 7

A subsequent welding test using a 25175jointcon-figurationwasperformedh ordertofurtherreduce the an-gular distortion. The test parameters were the same asthose used for the previous 30/70 joint test except that a4.8.mm (3/16 .in.) root gap was added 10 the joint a“d a 3,2-mm (1/8 -in.) wire electrode was used for the submergedarc welding, Tbe first side was completed in 17 passes andthesecondsiderequired65passes.A transverseshrinkageof4mm (O.16-in.) was measured on the finishing side. Theangular distortion. for all practical purposes, was eliminatedas the resulting angular change was reduced to 0.62” whichis significantly less than the angle of 2.0” remaining after the30/70 joint test. While the angular distortion wasdiminished with the 25175 joint configuration, the numberof total weld passes increased by almost 50°h. This imreased

WELDING THICK PLATES

weld depositioncanbeattributedtothelargerrootgapusedtofacilitateaccesstotheweldrooton theIl”khingside.

Wbik theselimitedtestsdidnotcompletely$ubsta”-tialethatangulardistortioncouldbe eliminatedin thickplates, they did exhibit a potential fol rtd”cing plate hamdiing. It was evident as the tests were begun that the firstfew passes on the backing side would be the most criticat, asit is these initial passes thatproducedtbe most angular dis-tortion, In the actual practice, wider plates than those testedwill normally be used and the weight of tbe ptate will pro-vide a substantial restraint cm the joint. Tbe tests did i“di-cate thnt the previous joint design, requiring welding of thelarger beveled side first, ccmld be modified to take advan-tage of a single turn of tbe plates,

197

\

I

ULCC STERN VIEW

FIGURE 8

I

I

198 p-

ULCC Trailing Edge Casting Fracture

In 1978, during the latter phase of hull construction,the ultralarge crude carrier U,S.T.ATLANTIC experienceda severefractureinitstrailingedgecastingconnecdngthesterntubeand the rudder born castings. Forming the ter.minus of the longitudinal bulkhead and the port and star-board shell plating at the ship’s stern, tbe trailing edge cast.ing had been welded into its final position onboard the shipand was undergoing stress relief of the last welded buttjoint. Tbermat stresses set up during the postweld heattreatment were the principal cause of tbe casting fractureduring the cooling-down period, although residual stresses

, from instailat ion may have hcen an important contributingfactor. Figure 8 shows an overall view of the area underconsideration between the stern tube casting and the rud-derhorncasting.

SECTION B-6

+=3+SECTION C-C

Tbe trailing edge casting actuaity consists of threeseparate castings located between other massive structuralcastings as illustrated in Figure 9. Tbe lower piece is amake-up section that connects the shoe casting with tbestern tube casting and is 1220 mm (48 in.) long and weighs1.13 Mg (2500 lb). Just above the slern tube casting there isanother make-up section of about the same size and weightwhich is designated as the middle piece of the trailing edgecasting. Above the middle piece is the 7.6-m (25-ft) long,curved upper piece of tbe traiting edge castins which termi-nates at the rudder horn casting and weighs about 8.71 Mg(19 200 lb). The three prongs of the casting are designed tobe welded to tbe centerline longitudinal bulkhead and theport and starboard shell plating, thereby completing the ter-mimtion of the triple plate intersection. The characteristicpitchfork shape of the cross section at the bottom of thelower piece changes as it rises, flattening out substantially at

h

<%

1..

I I,,,,,,,, ,,, ,,, ,,(, (,AP s 10 15 20

TRAILING EDGE CASTING

FIGURE9

p--199

1

the top of theupperpiece.The64.mm (2.5 -in.) thickness isconstant throughout the casting length except at the transi-tion radti, The distance along the outer surface from theport shell butt weld to the starboard shell butt weld variesfrom 1520 mm (6o in.) at the bottom to 1220 mm (48 in. )at the top. The casting material selected meets the req”ire-nwnts of ASTM A27 for Grade 60-30 steel which has aminimum yield strength of 207 MPa (30 000 psi) and aminimum tensile strength of 414 MPa (60 000 psi).

The threesectionsofthetrailingedgecastinguwrebuttwelded at five locations which required postwddstress-relief heat treatment. Stress relieving is acmmplishcdby heatimg to a suitable temperature, botding the tem-perature constant long enough to reduce the residmlstresses, a“d the” cooling at a slow rate in order to minim-ize the development of new residual stresses. The postweldheat treatment was conducted using resistance type heatingover a 305-mm (12-in.) wide band monitored by ther-mocouples according to the following procedure:

●

●

●

●

●

The rateof temperatureriseabove 26WC(50WF)shallbe lessthan149-C(300”F)perhour.

The hotding time shall be 3 hours for the lowerpiece and 5 hours for the middle and upper piecesat a temperature of 621 ~ 28°c (1 150 ~ 50”F).

The maximum mte of temperate decrease shallbe tess than 93°C (20t3’F) per hour.

The maximum temperature difference betweenany two thermocouples shaO not excmd 93-C(200”F) while below the stress reliewi”g tern.perature and 38°C (1OO”F) while at the stressrelieving temperature.

Stress relieving shatl be considered comptetewhen the weldment has cooled to betow 2600c

(50WF).

Theshipboardinstallationofthethreesectionsoftbetrailingedgecasting began with the lower unit and the workprogressed upward toward the rudder horn casting. Tbetive major butt joints at the ends of the sections were pre-heated to a temperature range of 121-C to 149°c (250”F to300”F) and the” were wetded with low hydrogen E7018electrodes by the manual shielded metal arc welding pro.cess. No problems were encountered during the weldingoperations on any of these joints and the residual stressesdue to transverse shrinkage were relieved according to theprocedure described above.

As the last major weld Joint just forward of tbe rudderhorn casting was being cooled down after stress relieving, abrittle fracture occurred that was accompanied by a loudreport as if from a cannon. The separation of the casting waslocated about 1220 mm (48 in.) forward of the weld join!undergoing stress relief and extended completely tbmughthe 839-cm2 (130-in.2) cross section. A 8ap width of ap-proximately 6,4 mm (1/4 in. ) remained after the castingcooled down to the ambient temperature of 24”c (75”F).

An examination of the circumstances leading “p tothe fracture may be of value to aid i“ “nderstmding thereasons for the occurrence of the faitme. The casting wesnot welded to the centerlim longitudinal bdkhead mm mthe shell plating for a length of 1830 mm (72 i“.) aft of the

fracture.The upper end of the casting was welded to therudder hom casting and the remainder below the fracturehad been joined to tbe ship’s hull structure. Tbe model ofthe structure, in essence, was a bar, ftdty restrained at itsends, which was heated m as much ?.s 64YC (1200”F) onone end. The butt wetd at the aft end of tbe bar wasundergoing stress relief to eliminate the residual mshrinkage stresses fmm we,lding the joint. Since this buttweld was joining two massive sections that were fixed inplace, reaction stresses were also present in the bar andwere added to the local residud stresses at tbe weld ioi” t.Reaction stresses are tensite stresses which exist after weld.ing O“IY because of the restraint at the ends of tbe bar.When combined with residual stresses some permanentdeformation may hmre resulted if the yietd stren~th wasexceeded.

TEMPERATURE,00 400 600

1+-ww

~c800 Ioc

F!!:.I!II

1

400

TEMPERATURE. ‘F

STRESSES FROM HEATING AND COOLINGA RESTRAINED MILD STEEL BAR

PIOIJRE10

The heat-affected zone of base metal adjacent to aweld, which is s“rmunded by tooter metal, is analogous ma bar with restrained ends which is bested and then cooledto ambient tempemt”re. Figure 10 illustrates the stressesdeveloped in a mild steel bar during the heating a“d codingcycle (2). When the bar is heated, compressive ~tre~~e~ areproduced due to the end restraint. At 135-C (275”F) theyield stress is reached and maintained as the temperatureincreases. This is d“e to the decreasing yield point in steelcorresponding to increasing temperatures as shown i“Figure 11 (2). Other changes to the physical and mcchanicdpropemes also occur as the temperature rises; the coef?I-

200

‘1

cient of expansion increases, while the modulus of elasticitydecreases. At 64’YC (120WF) the yield point approacheszero and the compressive stress is low, decreasing furtheras the temperature reaches 927°C (1700”F). Upon coolingunder restraint, tensile shrinkage stresses are created sincecontraction is prevented. At 649°C (1200-F) the steelregains its elastic properties as the yield point rises. As theambient temperature is reached, the tensile stresses rise tothe yield strength where they remain thereafter until therestraint is removed (2).

.TEMPERATURE,‘C

0 200 ,0.2 600 800), 1 (

When the postweld stress retief heat treatment wasapplied to the welded joint, the residual and reactionstresses at the joint were reduced to a low level, Throu&hthe first 204°C (4000F) the shrinkage stresses decrease dueto a phenomenon called recovery which k a reduction ininternal stress associated with a rise in temperate. At649°C (1200”F) the shrinkage stresses are dissipated almostcompletely as relaxation occurs in the material and ptnna-nent yielding results. Unevenorrapidcooling from stress.relieving tcmpemtures may reimpose internal stresses a“dtherefore excessive thermal gradients must be avoided.

Since the fracture occurred about 1220 mm (48 in.)from the weld undergoing stress relief as it was coolingdown to the ambient temperature, it is likely that the ther.mal stresses generated during contraction comhi”ed withthe tensile reaction stresses from welding to produce thefailure. If the emire length of the casting from the butt weldto the fracture had been heated to the stress-relieving tem-perature, the fracture may not have happemed as it wouldhave been uniformly rdieved over its length, The weld

joint was etTectively stress relieved, but the entire bar was

not.

There were otherfactomwhkh may have beenresponsible for the failure. One is that the cmss-sectiomdshape as shown in Figure9isnotstructurallyefficientin itsinherent ability to resist a bending nmmem Due to the cur-vature at the upper end of lhe casting, the centroids of thecross-sectional areas were not completely in alignment,therchy inducing a bending moment from any axial forcespresent. The rcstilting bending strtsses would have corn.bined with the thermal stresses during contraction andcould have quickly reached the breaking strength of themild steel material.

A seccmd factor comributing to the failure may havebeen the jacking and stmngbacking needed to bring the

uPPcrend Of the castingintoalignment with the rudderhornprimtowelding.Inordertoachieve the required fair-ness without having a short make-up piece to facilitate tit-ting, it is possible that high bending stresses were lockedinto the material before the f~nal joint was welded. therebysetting the stage for failure.

A third possibility that requires mentioning is that asignificant defect or flaw may have been present at the fmc-ture location. Thisisa common explanation and is oftengiven when a casting is involved with a fracture. Neverthe-less, casting defects do exist all too often and, in spite of thebest quatity controls and inspections, will continue to causeproblems.

Wh,le this failure was not catastrophic in tbe Se”Sethat it did “ot result in the loss of tife or threaten the shipitself, it was of some concern. With only three months untilLmncb, this 396 000-Mg (390 000-dwt) oil tanker, thelargest vessel ever constmcted in the WesternHemisphere,had a fractured stern in need of immediate repair. A 910-mm (36-in.) section of the cracked casting was cut out andreplaced by a matching section ‘<borrowed’, from a followship of the same design. The edges of the new, weld jointswere clzd with 6.4 mm (1/4 in.) of weld and the section wasstress relieved in the fumacc. The edges of the ship castingjoints were also clad welded and stress relieved Iocdly bypostheating. The new section was fitted into place, and theaft weld joint was completed and stress relieved hy heatingas previomly performed, The find joint was welded andpeening was heavily employed to mechanically stressrelieve each layer of weld except the first and last Layers, Amagnetic particle inspection was performed on the last twojoints on the backgouged weld root, cm the final weld sur.face, and atso during the course of the welding. The findcasting-to-casting joint was not subjected to any postheattmatrnem other than the normal preheat for weldlng. Uponthe s“ccessf”l completion of the fimzljoint, the internal shipstructure was fitted imd wetded to the casting which was fol-lowed by the attachment of the port and starboard shellplating.

The solution employed to solve the fracture problemon this casting was to create an opening and then to insert ashort make-up section without postweld heat treatment onthe find weld joint, This ttchniq”ewas s“ccessf”llyemployedduringtheco”strucdonof thefollowship.Aquestionwkes, however,asto why fractureswerenotencounteredon thelowerand middle trailing edge castingstocated above and below the stem tube. The welds of thesemake-up sections were stress rdieved by postweld heattreatment on both ships. One explanation may be that since

201

I

theirlengthand degreeof curvaturewere substantiallyreduced, the induced bendl”g stresses were less severe.Anotherexplanationmay bethatthesesectionswem lessdi!iicult to tit into place and did not require extensive jack.ing, creating locked-in residual stresses in the material. Athird exp!mation may he that these make-up sections weredefect-free, whereas the upper casting may have containeda small defect or dkcontinuity at the location of the fmctwswhich only required the presence of a temile force to causefailure.

The above account is just cme example of howditXculL it is m accurately assess the etYects of residualstresses. In the attempt to rdieve residual stresses fromwelding at one location, more severe stresses were inducedelsewhere. Unfortunately, it is surmised that this is not aumque occurrence in sh!pb”ildr”g.

SUMMARY

The effects of residual stresses on welded stmctureshave been studied extensively in numerous areas related tothe cases discussed above. While it may appear that exter.redly applied tensileloadsshouldproduce stresses thatwould be superimposed on local residual stresses to causefailure, the research conducted does not support thk con-tentiom Since the internal system of tensile and cormpressive stresses is in equilibrium, a redistribution resultsfrom stresses caused by .mtemal loads ad the residualstresses are significantly increased rmly in areas away fromthe weld. When the level of applied stress is increasedbeyond yielding, the effect of residual stress is negligibleand essentially disappears as the stress distributionbecomes uniform thmughtcmt the cross section. After the

applied Stress has been removed, the residual stressesdecrease to a point below their initial stress levels.

The significance of residual stress effects becomes ofimportance under low applied stress wnditions and mayresult in brittle fracture. Extensive experimental studieshave demonstrated that an umtable fracture may developfrom a smalt defect or flaw which would be stable if therewere im residual stresses pres.mt,

In order to realize the goal of improving the reliabilityof welded structures under demanding service conditions,the etTects of residual stresses and distortion mmt beminimized. This may be accomplished by expanding tbe

working knowledge on the subject of residual stress and

aPPIYi:g new techniques and effective procedures to sowrecurring prohlenm

ACKNOWLEDGEMENTS

The authorwouldlike to express his appreciation toall of his associates at Newport News Shipbuilding for theirvaluable contributions made during the preparation of thispaper. Special thanks are extended to Arthur Price for hisfine preparation of the ill”stratiom md the Word process.ing Center for their typing assistance.

REFERENCES

1.

2.

3.

K. Masubuchi, “Residual Stress.ss and Distortion,, inWelding Handbook, seventh edition, volume 1, edited

by C. Weisman, Miami, America” Welding Society,1976

George E. Linnert, Welding Metallurgy, third edition,VOIWIW 2, New Ymk, American Welding Society,1967

Koichi Mambuchi, Analysis of Welded Stmctures, firstedition, London, Pergamo” Press Ltd., 1980

SUPPLEMENTARY READING LIS’f

1.

2.

3,

ResidwdStressesinMetalsand Metal Construction,edited by W. R. Osgood, New York, Reinhold Pub-lishing Corp., 1954

W. J. Hall, H. Kihara, W. %ete, A. A. Wells, BrittleFracture of Welded Plite, Englewood Cliffs, NewJersey, Prentice-Hdl, Inc., 1967

Residual Strewes in Welded Construction and TheirEtT..cts, [mernatiomd Conference, London, Novem-ber 15- 1‘1, 1977, Volume 1 - Papers, Tbe WeldingInstitute, Cambridge, England, 1978

202