BRITTLE FRACTURE PROPAGATION STUDIES

BY

F=. F. ViDEON, F. W. BARTON AND W’. J. HALL

SHIP STRUCTURE COMMITTEE>’

MEMBER AGENCIES: ADDRESS CORRESPONDENCE TO:

BUREAU OF SHIPS, DEPT. OF NAVYSEcRETARY

MILITARY SEA TRANSPORTATION SERVICE. DEPT. OF NAVYSHIP STRUCTURE COMMITTEE

UNITED STATES COAST GUARD. TREASURY DEPT.U. S. COAST GUARD HEADQUARTERS

MARITIME AOM1NISTRATION. DEPT. OF COMMERCEWASHINGTON 25, D. C.

AMERICAN BUREAU OF SHIPPING

20 August 1963

Dear Sir

As part of its research program directed toward the improvement’.of hull structures of shipsi’ the Ship Structure Committee has been sPo,n-,~soring a brittle-fracture ,mechanics study at the University of Illinois.. .. ... ..’ -“ :?:,., ,

Herewith is the First Progress Report on the low-velocity ‘phase:;:.SSC-1 48, Brittle Fracture Propagation Studies, by F. F. Videon, E=..W.;,.,;.,..., ., .,,.,Barton, and W. J. Hall.

This pro ject was conducted under the advisory guidance-’’~f,;;~y;iCommittee on Ship Structural De sign of the National Academy of sci.+c~s-,;~National Research Co unci 1.

.,..., ~..,.,’.“:i.,,-.,,.... ,:,~,, . ,., ,-, .1,

., .;,,.+. “:::.’,.This report is being distributed to’ the individuals and agencies

,..- .2.

associated with the project, and to those interested in the Ship Structure ,j~:Committee program. Questions or comments regarding this report,wotild’;be appreciated and should be sent to the Secretary, Ship Structur::$Orn.- +‘mittee..

.,, , .,, .,... ‘.,’,.’,,:”,,~... :..,. ..., ..”,.,. ,<, . ,,

. Sincere ly yours,

*

,..

k

. .:,! “,.. ,,,

,.

Rear Admiral, U. S. Coast Guard’ .“ ~‘ . ~~Chairman, Ship structure committee ,‘: ‘”

,,.

SSC-148

First Progress Reportof

Project SR-155“Low-Velocity Fracture”

to the

Ship Structure Committee

BRITTLE FRACTURE PROPAGATION STUDIES

by

F. F. VideonF. W. Barton

andW. ~. Hall

University of IllinoisUrbana, Illinois

under

Department of the NavyBureau of Ships Contract NObs-65790

Washington, D. C.U. S .Department of Commerce, Office of Technical Services

20 August 1963

ABSTRACT

This investigation was undertaken to study low-velocitybrittle fracture in wide steel plates. The detailed results of twotests of 6-ft-wide prestressed steel plates are presented alongwith pertinent observations from tests af similar specimens con-ducted as one of the last phases of Project SR-1 37. Also, resultsof nineteen tests of 2-ft-wide centrally notched plates, a majorityof which had a longitudinal butt weld are presented.

In 6-ft-wide prestressed plates, the residual stress field(longitudinal tensile stresses at the edges of the plate balancedby compressive stresses throughout the central portion) made ini-tiation possible with no external applied stress, and had a sig-nificant effect on the fracture propagation. Fracture speeds werehigh (4000-6000 fps) near the edges of the plates and decreasedrapidly to as low as 165 fps as the fracture propagated into theregion of compressive stresses. In the low-speed regions themagnitude and extent of the dynamic strain field as sociated withthe crack tip was considerably less than had been recorded inearlier tests (Project SR–1 37) of high–speed fractures in plainplates.

For the 2-ft-wide centrally notched and welded platesin which the fractures were initiated statically, fracture speedsas high as 5000 fps were recorded in the zone of high residualtensile stresses near the weld; the speed apparently stabilizedat about 1800 f ps after the fractures had propagated dut of thehigh tensile stress field. The dynamic strain field associatedwith the intermediate–speed fractures (1 800 fps) were roughlycommensurate with that which would be expected for this fracture-velocity level. The fracture initiation observations indicatedthat the tensile residual stress alone was not sufficient to insurelow-applied- stress fracture initiation; also, metallurgical effectsassociated with high–heat input during welding are not necessaryin all cases for low–stress initiation. Indications are that straincycling of the material in the notch region arising from weldingmay play an important role in the iti’tiation of brittle fractures.

/ .

.

Norne nc lature

Part A. Low-Velocity-Fracture Studies in Wide Steel Plates:

Introduction . . . . . . . . . . . . . . . . . . . 9 *4....* . . . . . 1

Description of Specimens, Instrumentation and Te stProcedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

Brittle-Fracture Tests and Results . . . . . . . . . . . . . . . . . 4

Analysis and Discussion of Test Results . . . . . . . . . . . . . 13

Summary. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17

Part B. Fracture Propagation in Centrally Notched and WeldedSteel Plates:

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . .● 9*. 18’

Description of Specimens, Instrumentation and TestProcedure . . .* *..... ● ...**.. .. *...*. ..*. 18

Presentation and Discussion of Test Results . . . . . . . . . . 22

Summary. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31

Acknowledgment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34

Reference s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34

/

NATIONAL ACADEMY OF SCIENCES-NATIONAL RESEARCH COUNCIL

Division of Engineering & Industrial Research

SR-155 Project Advisory Committee“Low-Velocity Fracture”

for the

Committee on Ship Structural Design

Chairman:

Dr. Dana YoungSouthwest Research Institute

Members:

Mr. J. S. ClarkeESSO Research & Engineering CO.

Dr. J. M. FranklandNational Bureau of Standards

Dr. J. M. KrafftNaval Research Laboratory

me ct and Background.—

Tne following terms are commonly usedthroughout the text:

Dynamic Strain Gaqe - Strain gage whose sig-nal IS monitored with respect to time on anoscilloscope during the fracture test; this samegage also may be us~d for static monitoring bef-ore and after the te St.

~t~tlc Stra~n Gaqe - Strain gage used only tomonitor the static strain level.

_onent Strain Gage - One of the threeindividual strain gages of the rectangularstrain rosette.

Crack Detector - A s~ngle wme strain gage.——located on the plate surface perpendicular tothe expected fracture path and intended to bebroken by the fracture. A measure of thefracture speed and crack TIP location may beobtained from a knowledge of the distance be-tween detectors and the time interval cor-responding to the breaking of adjacent de-te ctors.

Initiation E- - In Part A, the edge of thespeclrnen at which the brittle fracture lS ini-tiate d. In Part B; fracture was initiated froma central-notch.

Notch Line - In Part A, an imaginary straightline connecting the fracture initiation not cheson opposite edges of the plate specimen. InPart B, an imaginary transverse line passingthrough the Initiation notch.

Base Strain - For any gage, the strain cor-responding to the applied test load plus thelnltial residual strain, with due regard forsign.

~ro Strain Level - In Part A, the referencecondition of zero strain corre spondlng ‘to theas-rolled, slotted, but preweld~d state. InPart B, the reference condition of zero strainlevel corresponding to the as-rolled, notched,b~ve led but prewe~d~d state .

PART A. LOW-VELOCITY-FRACTURE STUDI~~——IN WIDE STEEL PLATES

The series of tests describsd in tlms part ofthe report was conduc~ed to evaluate the pa-rameters associated with low-veloclty brittle-fracture propagation in 6-ft-wide steel plates.The study consisted of running the fracturethrough a field of compressive residual stressand making measurements of surface strainsand fracture speeds during the crack propaga-tion.

The earli~r pilot studies of low-velocityfracture propagation made as a part of ProjectSR-1 37 have been reported in References 1 and2. Other recent work at the University ofIllinois leading up to the present Investigation1s reported in References 3 through 7. Althoughthe preliminary studiesl’2 provided consider-able strain and spe~d data, measurementswhich would permit the determination of prin-cipal strains near the propagating fracturefront were lacking. The re SUIIS from two platetests undertaken to supply the desired infor-mation are presented here~n. In addition, thesalient obs~rvations from the earl~er 6-ft-wideprestre ssed plate tests are incorporated inorder to provide as complete an analysis aspossible. At the time this study was under-taken, to the authors’ knowledge, there hadbeen no wide-pLate investigations devoted tostudy of the effect of a residual stress fieldon fracture propagation. There had been sev-eral studles lnvolvlng fracture touglxmss ofwelded plates (involving residual stress asone of the variables) in Great Britain, Japan,and the United States, and a summary of muchof this work is presented in Reference 8. Otherwork concerning studies of residual stress andfracture have appeared recently .9-I 0 ‘1 ~ Forthe most part the work reported in the latterpap~rs are “go” or “no go” tests but in Refer-ence 10 the authors reported that tlm fracturepath more or loss traversed the plate at rightangles to the du-ection of principal stressesarlslng from the residual stress system priorto ‘l’h@ time the fracture s~arted--an observa-tion which also was made in the early stagesof the present investigation. However, in thereferenced work no measur~ments were madeof the change in the residual stress system orother fracture charac~eristics during the timethe fracture -was propagating.

INTRODUCTION The study was undertaken to determine the

effect of residual tensile stress on irutiationand propagation and at the same time allow astudy of the behavior of tba propagating frac-ture through both tensile and compressive re-sidual stress regions. Earlier work On 2-ft-wide speclmenss indicated that welding Oftapered slots cut in each edge of the speci-men was ths most satisfactory method for ob-tmnlng the desired residual stress Pattern.This procedure, described brieflY in a latersection of this report, was adopted In prepa-ration of all specimens in this ~nvestigation.

Two brittle-fracture tests of 6-ft-wideplates containing a residual stress fie Id wereconducted as a part of this particular study.For identification purposes the specimens arehereafter des~gnatedas Tests 49 and 50 to con-form to the numbering of tests conducted as apart of the recent brittle-fracture mechanicsprogram at the University of Illinois (SSCProjects SR-137 and SR-155). The plates w-ereinstrumen~ed with strain gages and crackspeed detec L’ors to provide information abouttheir bshavior during propagation. The de-tailed results on Tests 49 and 50 are presentedherein; for purposes of completeness, theanalysls and discussion are based not only onthe results of Tests 49 and 50 but also onpertinent observations from the earlier Tests42 through 48.1’2

DESCRIPTION OF SPECIMENS, INSTRUMENTA-TION AND TEST PROCEDURE

Material propsrt~es

The plate specimens used in this investi-gation were prepared from 3/4-in. semlkilledsteel plate. The mechanical, chemical andCharpy V–~otch data for this material wereobtained during the course of an earlier in-vestigation and the average values are pre–sented in Fig. 1 .

Fabrication and Pro ce dure

The plate spscimens used in this investig-ation were 72-in. wide, 3/4-in. thick and54-in. long. The net wid-:h of each specimenalong the notch line was approximately 2 1/4in. less than the gross width as a result ofthe notches cut in each @dge Of the sPecimen.These notches were used for the notcll-wedge-irnpact method of fracture initiation which hasbeen described previously. s

TENSlm mm! Wl!A

Percent PercentYield ~ nlarlgatlom Retictfon

L* w L T L! T L T

33.8 34.0 61.8 6LB 40.7 39.1 66.9 .%0

*L - Io@tudiWT = !Craumcrse

C~K AHALYS13——

c Mn P s Si Cu Ni Al

0.19 0.71! 0.019 0.029 0.055 6.02 TmCe 0.03

lm, 1 I I

BO

40

0-40 0 40

Ttq@m+mre -

Ci#.rPYV-NotchCurVtS

1. MATERIAL PROPERTIES .

The general preparat~on of each test SpQCi-

men was similar and consisted of sawing taper-ed S1OIS to the desired dimensions, welchngthese S1OTS to produce the residual stress pat-tern In the plate, and finally cutting the edgenotches used In crack initiation.

Four tapered S1OIS, two per edge, were cutin each plat”e, The location and dimensions ofthe slots were identical for Tests 49 and 50 as

shown in the sketches of these specimens InFlg.2and3. The depth of all four slots wasth~ same and their locations were symmetricalwith respect to the longitudinal center line andthe transverse notch line.

The welding procedure employed was simi-lar for both specimens and the typ~cal sequencewas as follows. Weldlng irutlally began at apoirit 4-in. from the tip of one slot and pro-ceeded to the tip. This 4-in. length was weld-ed with alternating passes on opposite faces ofthe plate. For the particular dim~nsions of theslot and specimen, a total of six passes wasrequired to completely C1Ose the slot. Thissame procedure was then followed on similarlylocated 4-in. segments of the remaining threeslots. Four-inch lengths were chosen so thatone electrode would last for a complete pass;E7016 electrodes were used throughout. All

3

four slots were then welded again in the samemanner, this time bsginning at a point 8-in.from the tlp af the slot and proceeding to theprevious lY completed weld. The remainder ofthe slot length was welded following the samepro cedure . This pmticu lar welding te chniqu@was planned to keep bending to a minimum bythe symmetrical placing of the weld metal andyet to produce maximum contraction at theedges of the plate.

Measurements of Residual Strains—.———.————

The residual strains in the Plate sPecimens,

re suiting from the pre stres sing procedure, weremeasured by means of Baldwin SR-4 straingages and were checked in some cases by a6-In. Berry mechanical gage . SR-4 gages wereplaced back-to-back at selected points. Longi-tudinal and transverse strain measurements weremade across ‘the notch line as well as at se-lected pmnts abzwe tlw notch line . The in-

strumentation layout for Tests 49 and 50 IS

shown in Fig. 2 and 3.

Initial strain read~ngs from all gages weretaken afte- the taper@d slots had been sawed

-@in.

—“’-l ‘ ‘ ‘

FIG. 2. INSTRUMENTATION UIYOUT AND S LOT CONFTGURATION - TEST 49.

I westpace IdFIG. 3. INSTRIJNTENTATIONLAYOUTAND SLOT CONJ?1GUR4TION- TEST 50.

4

and w]th the specimen in an unresb-alned con-cbtlon. These strain readings established thezero strain level, and all other strains werereferenced to this prewe Ided state. After thespecimen had been prestres sed by welding thetapered slots, the final static strain readingswere taken at room temperature.

Dynamic Instrurne ntati~

Thirty-four channe 1S of cathode-ray oscil-loscope recording instrumentation were usedfor recording dynamic strains; Baldwln SR-4Type A7 strain gages and Type AR7 strain ro-settes (both types l/4-in .gage length) wereused to measure the dynamic strains.

Crack speeds were measured by using asystem of surface crack detectors which brokeas the fracture traversed the plate; thesecrack detectors consisted of Baldwin SR-4Type A9 strain gages (6-in. gage length). Thebreaking of a crack detector opened an e lec-trical circuit and fed a st~p change in voltageto the recording channel. From a knowledgeof the distance m~asured along the. crack b@-tween detectors and the elapsed time betweensuccessive in’terrupt~ons of the circuit, theaverage fracture speed could be computed.The fracture speed also could be obtained bynoting the elapsed time between pea,klng ofsuccessive SRL4 strain gages at a fixed dis-tance from th~ fracture path.

Data R~duction

Reductiomof the strain data recorded on35mm str~p film was facilitat~d by the utiliza-tion of a decimal converter and the Universityof Ilhnois high– speed dlgltal computer, theILHAC . A brief summary of the data reductionprocedure 1s given be low.

The 35mm film strips were enlarged andtiming marks were scaled on th~ enlargements .Each individual trace had tirrung Interruptionsto insure timing synchronization of all traces.With the aid of the decimal converter, valuesof gage strain versus time were simultaneouslyplotted on an X-Y plotter and typed in tabularform. In Tests 49 and 50, in which strain

rosettes were used, the strain–tim~ data alsowere punched on IBM cards; these data weresubsequently transferred from ‘the IBM cardsto punched paper tape and processed throughthe ILIJAC which computed values of theprincipal strains. The ILLIAC results con-

sisted of tabulated principal strain data, aswell as scaled oscilloscope displays of com-pallent gage and principal strain tm.ces whichmmre photographed for later enlargement andprocessing .

The test procedure used for these testscomslsted of cooling the spec~men (withcrushed dry ica) to ‘the desired test tempera-ture, applylng the test load lf an externalstress was to b= employed, and initiating thefracture by means of an impact produced bydriving a wedge into a notch in the edge of theplate. A more detailed description of the cool-ing technique and the notch-wedge-impactmethod qf initiation may be found in earlierpubhcations .3-4

13RIT’TLE-FR4CTUllE TESTS AND RESULTS

General

Brittle-fracture ‘tests were conducted ontwo specmens containing a residual strainfie Id produced by the weld~ng of th~ taperedslots .

Although the magnitude of the measuredresidual strains varied slightly for the speci-mens, the overall patterns were slml lar forboth plates. The lmghe st r~sidual strainswere produced at the edges of the plates andreachsd y~eld magnitude for a distance of a fewinches in from each edge. As a result of thesshigh “tensile strains at the edges of the speci-mens, it has been demonstrated’2 that it Wa. S

POs sible to inltiat~ fractures with no external-ly appli@d load,

A nominal stress of 3000 psi was appliedin Tests 49 and 50 to keep the plates taut inthe machine and LO obtain a fracture that wouldtraverse the entire plate or at least a majorportion of the plate before arresting; this 10ad-ing also reduced bsnding to a limited ext~nt.

Fractures were successfully lniciated inTests 49 and 50. In Test 50 the fracture prop-agat~d across the entire width of the plate and111Test 49, the fracture arrested in the com-

pre ssive strain region after traversing a majorportion of th~ plate width. The re suits ofTests 49 and 50 are presented in detail herein.A tabulated summary of the test conditions andfracture details for these tests and Tests 43-48

are presented in Table 1.

Test 49

Tests 49 and 50 were cleslgned primarily toprovide information regarding the principalstrain behavior In Lhe specimens during frac-ture propaga hen, and to provide addl’tio rmlinformahon about the type of residual strainpattern resultlng from we ld~ng of the taperedslots. The major portion of the instrumenta-tion, includ~ng all gages used for dy~lamicrecording, was located on the west face of thespecimen, and in all subseqLlent discussionregarding the residual strain field and the dy-namic strain records, it should be kept In mindthat these readings were taken from one sideof the plate only.

The average longitudinal residual sL’rainacross the notch line at the test load is pre-

5

sented in F’~g. 4 (a) . AS may be noted from

this plot, the average residual compressivestrain (back–to-back strain gage readings)across the central 24 in. of the specimen wasapproximately -300 microin./ln. The longl–tudinal and transverse residual strain d~stri-butions at selected polrrts on th~ west plateface are shown in l?lg. 4 (b) and 4 (c). It willbe observed from these plots that both ‘thelongitudinal and transverse strains along andaway from the notch llne are cpte uruformthroughou~ the central portion 01 the specimen.Thus, it appears Lhat the residual compre Ss!vestrain fle Id through which the fracture propa-gated was of famly constant magrntude.

This specimen was tested at an appliedstress of 3000 ps~, at a temperature of -10 Fusing ‘the standard notch-wedge-impact inl–tlatlon Lechnique. An external apphed s’rresswas utiliz~d in an at~ernpt to obtain a fracture

TAB LE 1. SUMMARY OF SIX- FOOT-WIDE PRESTRESSED PLATE TEST.

AU specimens wer@ 3/4 in. thiak by 6-ft wide, s~mikilled steel plates..“. ,

TestAverageNet ~qTest

Avezage SWedmApplied Strmm

Ho. (psi) (%)HiP

- I.OwSpacimn Description *E) Rmarka

L,=j o -12

46 3000

w o -l-z!

45 X00 -m

o

-8

0

-10

-no50 500Q

~ i.. plate length; 18 in. tapredslots, 6 in. above end Wlow notchMne; welded sad teetad in verticalWaition on latmatory floor, withanib of plate unrestrained

% in. p~te length; Plmti phti -W Wda; tested in V8rtiCZd

pooition on laboratom floor, withamM of plata unrestrained

&3 in. p@te length; 18 in.taperedslots, 6 in. above aud lmbw notc”h-8; specimen velded to @l plate-in machine before alot6 wme welded

48 in. plate lenghh; 20 in. tqmred

olote$ 6 in. almve and below notch~~= j EpeC&n Wdtid tO m phbU

in WckhIe bUfOrU slotu were wehied

63 b. p~ts le@hj M ti. ti~radwlot~, 6 in. almm and below notch-j $~ctien chmpcd W @l platesin machine L-&ore slots were welded

I&JIn. plate length; 20 in. taperedslots} 6 in. abve and Oekw mkchHne; apecimcn clamped h pull platiain Machine berore do’h were wd&&

5h Ii. plate Leugth; 16 in. %~rsdOlokE> 10 ill.above and below notchMm; specimen Wekum ‘Go@J. phtcain machine ‘mfore ‘Lnt.swere welded

54 in. “plate lml@h; 18 in. -@&srca$Iotn, Ml ic. above ad below Eckchline; s~cimem chmpd to pull plwke

h machine tifore clots were welded

Fracture propagated approxi.mately 56 in. beforearreating

First shot, no fracture;second *hot, fl-actwmpropzgs.w approxlmateti19 in. into plate

Fractwre ~ha cm-pletely acrom sp5ctien

Frarture pro~tid c--pletely ncrom qechen

Fracture propaga$e& app?osi -

mately 25 in. “wfore

arreatiag

6

1s00

— A,fm Tent lad App).i.d—Aft,, 2.0. Tad Appuca

I . ..-.”< . .... .“.-....—— ,....4

FIG.

Dl”m,e mm Id.i.lti.n Edge - i..

[*) kndtuoind Strain Along m’c.h Lime

4a. RESIDUAL STRAIN DISTRIBUTE ON -49.

‘d l?’-4 ;-4 I-4 M FIG. 5a. FRACTURE PATH - TEST 49.

(b)Longlt.dln81 Strain at SelectidFointm on Plate surface

4b. RESIDUAL STRAIN DISTRIBUTION -

that would propagate through the region of thespecimen where the dynamic strain readingswere to be taken. On the basis of previoustests, lt was felt that 3000 psi would be suf-ficient to Insure complete fracture. The frac-ture, did not complet~ ly traverse The plate,but did travel through the instrumented region,arresting approximately 52 in. from the ini-tiation edge. A photograph of the fracture anda close-up of the area of arrest are shown inFig. 5. Note that ‘th@ fracture curved sharplyupward before arre sting, apparently travelingnormal to the direction of max~mum principalstress.

TEST 49.

(c) ‘l!mneverm Strain d selectedPoiw%s on P1.e,tesurf&ce

FIG. 4C . RESIDUAL STRAIN DISTRIBUTION -TEST 49.

FIG. 5b. FRACTURE PATH - TEST 49.(Closeup of Arrest Area)

The instrumentation layout arid crack pathare show-n in the sketch in Fig . 2 and theStrain-time records obtained from ten of therosette gages are shown in Fig . 6. Also in–eluded In the latter plots are the computedprincipal strains and maximum shear strainsfor each gage point during fracture propagation.

7

[@m“.=b 1

FIG. 6a. STRAIN-TIME RECORDS - TEST 49.

FIG. 6b. STRAIN-TIME RECORDS - TEST 49.

FIG. 6c. STRAIN-TIME RECORDS - TEST 49.

(.) II-.* ~

FIG. 6d. STRAIN-TTME RECORDS - TEST 49.

(c) W,.it.6

FIG. 6f. STRAIN-TIME RECORDS - TEST 49.

hx, (c,) cm mm, [r,,,,..,PJ $L+Mam w, sktrm?.,, (7)

f!r ~,

I \,—z_. _.

—... .. .. . .. .—.—.cl

:fi ,... -....\ ,.-. ,

011812 16mm, . “mu,,..

(d R...e.e 1

FIG. 6g . STRAIN-TIME RECORDS - TEST 49.

m .—

s.,.,.mm.,. mmc-w-n t ,-.,”

Am

lWC!___ ~___

,,\.,~..~m—---- . . . . . . —..— ,,

0— -

, ,,. . . .*

.500

.,000. ‘, 1, ,G

m.. Liec,(,) n-i.,,. 8

FIG. 6h. STRAIN-TIME RECORDS - TEST 49.

(i) R..et.e 9

FIG. 6i. STRAIN-TIME RECORDS - TEST 49,

8

2003“H. (r,) W. “in. [c,

..,>.,,, 0,,,,..,, mmc-mm Cmcd

FYhc,pd ,.,.,W“d . . . . maws,,.,. (7)

, m

It@ .—

II

< ji, ~

r $ f

w —- . .—

,

~ J

j.“““ ‘-3--1 ~ ---- :,

y.,.,,, ~ ,fm

v :! !O ,.. ~:;~;

. ~oo . . . ._

mmoh ,.,60L

M.” :.l), ”...19 16

, h. :,U,DCC

(J) R.*.,,. m

FIG. 6j. STRAIN-TIME RECORDS - TEST 49.

This was the first time satisfactory and com-plete records were obtained from a specimenin WIUch the fracture arre sled. The directionof maximum principal strain also was computedfrom the strain information and representativedirections for various crack lengths are de-picted lD Fig. 7. The slqnificance of “theseresults is discussed in “the next section ofthis report.

Fracture speed information as determinedfrom the dynamic crack detector records ispresented in Fig. 8. Even though the frac-ture arres!ed in this test, the pattern of frac-ture speeds shows no ,narked change fromthose obtained In tests where fracture wascmnplete. The maximum average recordedspcvad was approximately 5650 fps and thelowest average speed was approximately 160fps . From Fig. 8, it may be noted that, as in

\ ‘%: H: %% ;:2I L D,r.ct,m., ,Mh mmc,pti,h,, /:,Tn,,.,,. “,,mi”,l“,I

1,.-—tL

.

,’” I,’

,,”/’

J J

9

1 ! c

FIG. S. AVERAGE FRACTURE SPEEDS - TEST 49.

all earlier tests, the highest speeds occurrednear [he initiation edge and the lower spseds

occurred in the central compressive strainregion. Th~ general shape of the detectorlocation versus breaking Urne plot is approx~-mately the same as was observed in earhertests.

Tes~ 50—.

This test was a duphcation of Test 49 withtwo minor exceptions; the location Of irmtru -mentation was changed slightly and the taperedslots were w~ldecl before the specimen inserthad been welded to the pull plates, ra~her thanafter. Because the ends were free during weld-ing of the slots, the resultlng magnitude of theresidual strain field was considerably lessthan that obta~ned in Test 49. The averageresidual compressive strain across the centralportion of th~ plate was only about -100mlcroin./in. as shown in Fig. 9 (a). Again,these readings represent average values, andin this test, the major partlon of instrumenta-tion was located on the s~de of the specimenon which the lower compressive strain valueswere recorded. The d~stributlon of longitudinaland transverse strains at selected points onthe plate surface are plotted in Flg. 9 (b) and9 (c) . It can be not~d from this P1O: that the

cii stribution of strain In the central portion ofthe plate 1s reasonably uniform.

This specimen was tested at an appliedstress of 30~o psi and a temperature of -10 Fusing the same Impact initiation procedure.lJncler these conditions, a br~ttle fracturepropagated completely across the specimen.It is felt that complete fracture in tkus test..- .,—.

FIG. 9a. RESIDUAL STRAIN DISTRIBUTION -

(b) La@tuWICJ Stmin at SRICCt*d

Point. on Plata Surface

FIG. 9b. RESIDUAL STRAIN DISTRIBUTION -TEST 50.

..7

.

Stratis Fran West

Face at Test had

Struim In 7440. i-klo ,-410.Microin./in. ; (

.gm -,- -, ;-w- \,i-450 /-440-?70.

1 ( \

-+7~ -4, ;* (-441- ;m- -18Q]

-i . -Im’l\

+m !-410 :370. -lYJ\18 24 m 42

Dia’kmceFrm InitiationEdge - in.

(c) !kammeme Strainat Selected

Points on Plate Surface

FIG. 9C . RESIDUAL STRAIN DISTRIBUTION -TEST 50.

10

and only a partial fracture In Test 49 can heattributed to the different res~dual stress pat–terns in the plate since all other test condl–tions for these two tests were identical, Aphotograph of the fractur~ path 1s shown inFig. 10. The part~cular shape of the crack

FIG. 10. FRACTURE PATH - TEST 50.

path is similar TO the fracture paths observedIn Tests 45 and 46 in wluch complete fracturesalso occurred. As may ba noted from the pho-tograph, the fracture always remalnecl abovethe notch line during propagation, showingdefinite changes in direct~on at the points ofmaximum strain gradient.

A sketch of the instrumentation Layout andcrack path for this specimen 1s presented inFig. 3. In this test the fracture propagatedthrough the middle of the dynam~cally instru-mented region and excellent stra~n respons~records were obtained for this low-velocityfracture . The strain-time records, including‘the computed principal strain information, arepresent~d in Fig. 11. From the recorded

(.) R.e.tt.2

FIG. 1 lb. STRA-IN-TIME RECORDS - TEST 50.

FIG. llC. STRAIN-TIME RECORDS - TEST 50.

FIG. 11 d. STRAIN-TIME REC ORDS - TEST 50,

(?.) ?.a”.ttf L

FIG. 11. STRAIN-TIME RECORDS - TEST 50.

(.) n..”=k 5

FIG. he. STRAIN-TIME RECORDS - TEST 50.

FIG. 1 If. STRAIN-TIME RECORDS - TEST 50.

FIG. Ilg. STRAIN-TIME RECORIX - TEST 50.

,mm

lm

w

-$3

.Im+

“--Tr/ u- - 4

A

/dl/.w~, ‘V!ll.-‘-i ~ .,y/

/.,,,,,-.I-’.-i;. ,,;:!j

..,, -.,.-,.,,

12 1mu. L.,c

FIG. llh. STRAIN-TIME RECORDS - TEST 50,

TEST 50.

FIG. llj. STK41N-TIME RECORDS - TEST 50.

strain information the directions as well as Themagnitudes of the maximum prlncipa~ strainswere comput~d; this information for selectedcrack lengths is presented in Fig. 12.

I(n) mrom?..s

,,=.,-,,..,.=,../,,. I

.,,!, ,..,,. ) ,“”, 1.,.,, $,,.,, ,,,, ,,>.

[a]% k. ,,.,,, b=-

FIG. 12. RESIDUAL PRINCIPAL STRAIN DIREC-TIONS AND MAGNITUDES - TEST 50.

I ..,,,,,,,,>M,r,. i”./,,,. I

FIG. 12. RESIDUAL PRINCIPAL STRAIN DIREC-

TIONS AND MAGNITUDES - TEST 50.

Although the fracture propagated completelyacross the specimen, incomplete fracture

speed irLformation was obtained because thefracture passed above some of the crack de-tectors; only four of the twelve crack dQtQC-tors were broken by the fracture. The spQQdinformation, as obtained from the crack de”L’ec–tors and also peaking of the strain gages, ispresented in Fig. 13. Although the speed in-formation is not as complete as that fromearlier tests, the data are sufficient to showthat the speeds are similar to previously ob-tained fracture spe@ds both in rnaqrntude andpattern. The highest average recorded speedwas approximately 6400 fps and the lowestspeed, averaged over a distance of 24 in. in

FIG. 13. AVERAGE FRACTURE SPEEDS - TEST 50.

the central region of the plate, was approxi-mately 330 fps. Since this average speed was

measured over a considerable distance, Itseems reason~~le to assume that the actual

lowest speed occurring dur~ng fracture wasquite likely less than 330 fps.

Photographs of ~he fracture surface for Test50 are shown in Fig. 14 along with the arrest

FIG. 14a. FRACTURE SURFACE.(a) Test 50.

FIG. 14b. FRACTURE SURFACE,(b) Arrest Region - Test 48.

13

region of Test 48 wh~ch is discussed later. Asin all previous tests in this ser~es, the roughertexture was observed to occur adjacent to theinitiation edg~ with a much srnoo~her textureoccurring in the central portion of the speci-men. Ths location of each photograph withrespect to the initiation edge may be deter-mined from the folding rule included. In therougher tmxture near the irutiation edge, thechevron markings of the fracture surface areeasily visible, while their presence in thesmoother region is extremely difficult to dls-stern.

ANALYSIS AND DISCUSSION OF TEST RESULTS

The analysis and discussion which follows1s based on the results of Tests 49 and 50 aswell as the pertinent observations from theearlier Tests 43-48. All of the tests were con-ducted on 6- ft-wide plate specimens which hadbeen prestressed by welding tapered s lots cutin each plate edge. Even though manual weld-ing was used, and the locatlon and dimensionsof the tapered slots varied slightly, the pat-tern of residual strains resulting from thisparticular prestres sing techruque was, in 9en-eral, similar for all specimens and consistedof high tension near the edges and moderateto low strain fields which throughout the cen-tral portion of the specimen could be attrib-uted largely to the restraint present at the timethe tapered slots were welded. For the twospecimens (Tests 43 and 50) in which no endfix~ty was provided during welding of the slots,the magnitudes of the residual compressivestrains acress the central section were muchless than for the remaining specimens in whichthe ends were either welded or clamp~d to thepull plates (see Table 1 for details) Prior tOthe welding of the slots.

The presence of high residual tensilestresses in the vicinity of the ~rutiation notchmade lt possible to consistently initiate brittlefractures using the notch-wedge-impact meth-od, even without external apphed stress. BYway of comparison, in the earher 6-ft–wideplain plate specimens, where no residualstra~n fie ld had been induced, an average ap-plied stress In excess of 15, 000 psi was nec-e ssary .t,o insure irutiation under sirrular testcond~t~ons. Apparently, the presence” of theseresidual tensile stresses has the same effecton initiation and to some extent, on propaga-tion, as an externally applied load of the samemagnitude. Although the impact effect of the

wedge was a necessary factor for successfulinitiation under the test conditions employed,the fact that br~tt Ie fractures occurred in everypre stres sed specimen indicates that a largefield of residual tensile stress 1s an importantfactor in the initiation process.

While the residual tens~on served as anaid to initiation and propagation, the presenceof residual compression in the central regionof the specimens had just the oPPosite effecton the fracture characteristics. It not onlycaused a reduction in the magnitude and extentof the dynamic strain field and resulted ingreatly reduced speeds throughout the com-pressive region, but also was capable of pro-ducing arrest.

In the prestre ssed Plates, th@ fractur~speeds in the compressive strain region werecnns~derably lower than those recorded in the6-ft-wide plain plate tests conduc~ed as apart of Project SR-1 37. A plot of detectorlocation versus detector breaking time for atypical prestressed plate test is presented inFig. 1 5; for purposes of comparison, a similar

FIG. 15. FRACTURE SPEEDS ACROSS PLATEWIDTHS - TESTS 37 and 46.

plot is presented for a typical plain plate testfrom Project SR- 137. In th~s plot, the slope ofthe curve is a measure of the fracture speed atany particular time. As may be seen, thefracture spseds for the two tests are almostidentical for approximately the first 20 in. ofcrack trave 1. This h~gh speed continuedthroughout the fracture travel in the plain platespecimen, whereas the fracture speed in theprestressed specimen decreased sharply andcontinued for the remainder of fracture trave 1at a greatly reduced value . The same generaltrend in fracture speeds was observed in allprestressed specimens.

The high initial speeds and the sudden de-crease to a much lower value are probably theresult of two factors, name lY, the residualstress pattern in the specimen and the initia-tion procedure @mploy@d. High fracture speedswould hs expected near the initiation edgewhere residual tension was present and also adecrease in speed would be expected as thefracture propagated into a zone of residualcompression. The effect of the initiation pro-cedure on initiation was i Illustrated by theresults of Test 44, in which, after the notchroot area had been highly strained by one im-pact, the initiation technique was sufficientto drive a fracture approximately “19 in. intothe specimen.

Because about the same impact energy wasused in initiating fractures In all 6–ft-widespecimens, any variation in fracture velocltlesfor the different tests can probably be attrib-uted large Iy to the particular residual strainpattern present in each specimen. In Fig. 16are shown plots of both the average residuallongitudinal straifi across the notch line attest load and the frachuw speed across theplate width for Tests 46 and 49; these plotsare typical of similar plots for all prestressedspecimens. Exact comparisons between the

residual strains across the notch line andfracture speeds at corresponding points ar~ notpossible because the residual strain shown InFig. 16 changed somewhat during fracturepropagation as a result of redistr~bution, par-ticu larly near the far edge of the plate; how-ever, certain relationships can be observedfrom the information available. From Fig. 16it may be noted that the highest fracture speedsoccurred where the highest r~sidual tensilestrains were present; also, the sudden de-crease In speed corresponded to the pointwhere the fracture encountered the maximum

I600

— c ,..1,SW

--–--- Im.&*ud.l

—.o 12 %4 6 49 72°&l

Mctnwe Fm Initl.t%m E*C In.

(.) T“.% 46

FIG. 16a. VARIATION IN SPEED AND RESIDUALSTRAIN ACROSS PLATE .

FIG. 16b. VARIATION IN SPEED AND RESIDUALSTRAIN ACR 0SS PLATE ,

residual compressive strain region. Also it

WaS noted that the fracture speed at ~imes in-creased even though the residual strain wasdecreasing; this was probably a result of thehigh energy level present in the plate at thatp~int.

As a result of the apparent correlation be-tween residual strain and fracture speed nearthe initiation edge, it was fe It that perhaps abetter indicator of the residual strain versusfracture speed re latlonship could be obtainedby considering only the speed and strain values

15

recorded C1Osest to the point of initiationwhere effects of load redistribution are elim-inated. Although the effect of wedging actionwas present, it was identical for all t~sts andthus would not affect a study of other param-eters. In Fig. 17 is presented a plot of theaverage fracture speed occurr~ng 6 in. from ‘theirutiakion edge versus the average longitudinalresidual strain at the same location for six ofthe prestressecl plates. Six in. was chosenbecause this was the closest poin~ to theinitiation edge at which average speeds couldbe determined. From this plot j“t appears that

FIG. 17. SPEED VERSUS RESIDUAL STRAIN6 IN. FROM INITIATION EDGE.

the relationship between these parameters isalmost linear, wh~ch indicates that the frac-ture speed is at least partially a function ofthe residual strains (or better, energy level)present in a specimen. This one set of ob-servations should be interpreted with care,because the relationship is apparently some-what different at other locations; some differ–ences would be expected by virtue of load re-distribution alone.

From the results discussed thus \ar, it canbe seen that fracture speeds were high in ten-sile strain regions and low in compressivestrain regions, generally showing a continueddecrease as long as the fracture remained in

the compr~ssive zone. It is likely that themagnitude and extent of the initial residualcompressive strain fields were sufficient forarre ST in every case after the fracb-we hadpropagated well into the compressive stra~nfield; however, the elapsed time from the be-ginning of fracture was appar~ntly long enoughto allow redistribution of both residual stressesand any applied stresses to prov~de sufficientenergy for propagation to continue . The re -

suits of the various tests in which the speci–men fractured completely show ihat in everycase, the fracture speed was ~xtremely low inthe compressive strain region and, except forthe effects of stress redistribution, all frac-tures probably would have been arrested.

In the prestres sed plate tests the fracturetexture in the compressive strain region wasconsiderably smoother “than that observed inany previous tests of plain plate specimens.Near the initiation edge where high residualtens~le strains were present, and where thehighest fracture speeds occurred, the fracturetexture was fairly rough showing the fam~ liarherring-bone pattern evident in the plain platetests. This difference in texture in regionsof residual tension and compression may beobserved in the photographs of the fracturesurface of Test 50 in Fig. 14. Also shown inFig. 14 is the arrest region of Test 48; it isevident that the fracture texture at the pointof arrest also is extremely smooth.

Initially it was felt that the difference infracture texture noted abov~ was a re suit ofthe difference in fracture speed across theplate, since the rougher texture always waspresent in regions of high fracture speed andthe sm~~ther texture in regions of much lowerspeed. An attempt was made to relate thetexture roughness with corre spending fracturespe~d but no conclusive results could be ob-tained. Since that time, further study has in-dicated that the roughness of the surface tex-ture appears to be re Iated riot to fracture speed,but to the stress or energy leve 1 present in thespecimen. This fact was first observed infracture tests of centrally notched and weldedspecimens which are described in Part B ofthis report. The aforementioned test data plusdata from the wide plain plate tests suggestsstrong ly that fracture ‘texture is indicative ofthe stress cn- energy level, and that it is notrelated dire ct ly to fracture speed.

Examination of the fracture surfaces for all

16

pr~stres sed plate tests show a rough texturein the v~clnlty of the initiation edge. Previoustests, includlng Test 44 in this series, haveindicated this region close to the initiationedge is affected by “the ~nit~ation technique.Thus, the additional energy suppLied to theregion of the plate near “the point of Impact,along with the high r@sidual termle stress,1s felt to be responsible for the rough texturein that region. Elsewhere, where the energyor stress level was relatively low, ~ncludingthe far edge of the plate, the fracture texturesare observed to be much smoother.

The dynamic strain records from the varioustests give an indication of the marked eHectthat the residual compression had on the strainfield associated with the moving crack. Forexample, in these tests the peak dynamicstrains racorded from component gages 1/2to 1 in. from the fracture were orI the order of1000 to 1500 microln./in. and gages locatedmore than 6 in. from the fracture showedpractically no response. On the other hand,in the earlier plain plate tests in which noresidual stresses were present, peak dynamicstrains on the order of 2000 to 4000 microin. /In. were recorded by gages 1/2 to 1 In. awayfrom the frac~ure, and gages 6 in. from thefracture still showed a noticeable response.An examination of the dynamic strain recordsfrom this series of tests also indicates thatin every case the peak strain magnitudesdecreased as the fracture entered the com-pressive strain region. Thus, as the fracturespeed decreased, the strain field associatedwith ‘the fracture also diminished in magnitudeand extent; Fig. 18 illustrates this observa-tion extreme ly well.

Another intere s;lng feature, not recognizedin earlier plain plate tests, was the fact “thatthe strain fie Id quite near the tip of the propa-gating fracture seemed to be approaching aS“ta”tewhere tensile strains were recgrded forboth vertical and hor~zontal component gages.This phenomenon could be observed only indata taken from gages in the immediate vicin-ity of the fracture and hence only limited in–formation 1s ava~lable. This phenomenon maybe seen by comparing strain-time records ofrosette gages 1 and 5 from Test 50 shown inFig. 11 (a) and 11 (e). It will be noted thatthe response of the vertical gage of Rosette 1peaks in tension at the same time the horizon-tal gage peaks in compression. This is theusual behavior for almost all strain time rec-

I

mm

WEE9 = Wm rt/*ee.(6 ft. tide plnti - Test 55)

Sccu ------ - - 1

— m.- ,.

.——

m = Em mkt.(6 ft. .i&piatc. ‘T,!.LB )

—— ... .——

/SpeCd~ EQ fij.ec.

.tmo— /--—-..”....— (6 ft. tide preazrmadPIE* - !&St w

I ISper! . I&Y) rqte,.

\,

(2 ft. r,OtcbcL alla

W*l&a Pla p - TcSt 5.-.. —

—

4

\0 - .——

.,. -

V——.

---- .-— _ ~_

0 2 4 6 a 10

FIG. 18. COMPARISON OF STRAIN-TIME REC -

ORDS FOR LOW-INTERMEDIATE-AND HIGH-SPEE D FRACTURES. STRAINS WERE MEASUREDAPPROXIMATELY 1/2 IN. FROM THE FRACTUREPATH .

orals observed to da~e . The records from Ro-sette 5, however, show both verhcal andhorizontal gages peaking in tension, althoughthe magnitude of ~he peak from the horizontalgage is smaller. An examination of previousstrain-t~me records from this and earlier in-vestigations revealed the same phenomenonbut apparently strain records show tensionpeaks on horizontal component gages only ifthe monitored gages are located less thanabout 1 In. from the fracture . From the limitedinformation available lt also seemed that ahigher fracture speed caused this condition tobe detected further from the fracture than for alower fracture speed. This may be noted byexamlnlng the record of Rosette 1 of Test 49

(Fig. 6 (a)) where the fracture speed was ap-proximately 2000 fPs and the record of Rosette5 of Test 50 (Fig. 11 (e)) where the fracturespeed was less than 1000 fps. Although Ro-sette 1 of Test 49 was located at a greaterdistance from the fracture, a much higherstrain peak was noted for the horizontal gagethan for the horizontal gage of Rosette 5 ofTest 50. Comparison of other records indi-cates the same trend.

17

In addition to the component strain gagerecords obtained, corresponding principalstrains and them directions were also comput~clin the tests where rosette gages were used.As may be noted from Fig. 7 and 12, the wekl-ing of the slots in Tests 49 and 50 resulted ina completely different pattern of direction ofmaximum principal strains. In Test 49 thedirections of maximum principal strains in thmcentral portion of the plate are all horizontalwhile in Test 50, ‘the directions are vertical.Also from l?~g. 7 and 1Z, the effect of the ap-proaching fracture on the maximum principalstrain and its direction may be noted. Althougha zone of residual compression existed through-out the central portion of the test specimens,during propagation maximum principal strains

were always tensile in the vicinity of the cracktip. In these figures, the heavy solid linesrepresent the fracture POsition corre spendingto the depicted strain magnitudes and direc-tions; the extended dashed lines represent thefinal path of the fracture. On the basis of thedata obtained, it seemed that, in general, thepath of the fracture in Tests 49 and 50 wasnormal to “the direction of maxunum principalstrain. This effect of princlpa 1 strain dmec-tion on the path of the fracture also would seemto explain the noticeable changes In the direc-tion of the fracture as it passed through theregions of maximum residual compress~vestrain.

SUMMARY

The purpose of this investigation was tostudy the Iow-ve Iocity propagation of a brittlefracture in 6-ft-wide steel plates in whichtensile and compressive residual stress eswerepre serrt, and to determine from the re suits ofthes~ tests, the effect of residual stress onfracture behavior. AS a part of the Project

SR-I 55 investigation two specimens of semi-skilled steel were prestressed by weldingtapered slots in the edg~s of the plate. Theresidual stress field consisted of high tensilestresses at the edges and compressiv~ stressesthroughout the centra 1 portion of the plate.

The two specimens were subjected to anapplied load of 3000 psi and were cooled toabout -10 1? prior to testing. The fracture wasinitiated by the notch-wedge-impact method.Instrumentation was provided to determine thecrack speed and the strain response.

The results of fracture studies of b-ft-wide

prestressed steel specimens as conducted as

a part of this program and the latter phases ofProject SR-1 37 may be summarized as follows:

1 ) Brittle fractures were initiated in all of thespecimens containing a residual stress field;no external load was applied in three of thetests while four specimens were tested at anapplied stress of only 3000 psi. By compari-son, in the 6-ft-wide plain plate specimenstested under similar conditions (S R- 137) anapplied stress in excess of 15, 000 psi hadbeen found necessary to insure fracture initia-tion. Thus, the high residual tension in the

vicinl~y of the initiation edge appears to havethe same effect on initiation as an ~xternallyapplied stress.

Z) That residual tensile stress is a definiteaid to init~ation was confirmed by the test of aplain p~ate specimen (Test 44) . This specimen,containing no residual stress, was tested onthe laboratory floor and required two impactsto drive a fracture approximate Iy 19 in. NOfracture occurred on the first impact but ap-parently the material at the root of the notchwas damaged sufficient ly to facilitate fractureupon application of the second Impact.

3) The strain field associated with the tip ofthe propagating fracture and The fracture speedare considerably influenced by the residualstrain field present in the specimen. The

fracture speed decreased and the magnitudeand extent of the strain field around the tip ofthe crack was considerably reduced as thefracture traversed the region of residual com-

pressive stress.

4) Within about 1 in. of the crack tensionpeak strains were recorded by both verticaland hor~zontal gages as the crack passed thegage; generally the effect was more pronouncedfor the higher crack speeds. This is in con-trast to the usual strain behavior noted forgages located further from The crack in plainplate tests where the strain response from thehorizontal gage normally showed a compres-sive peak,

5) A brittle fracture apparently tends to fo 11OWa path normal to the direct~on of the maximumprincipal strain ex~sting immediately aheadof the crack tip.

6) Fracture speeds varied from as high as6000 fps near the initiation edge to as low as

18

50 fps in the compressive strain reg~on. Theresidual compression in the central portion ofthe plate caused a large reduction in speed.P. nearly linear relationship between residualstrain and crack speed appeared to exist 6 in.from the initiation edge In these tests.

7) Comparison of test results makes it appearas though the residual compression in thecentral portion of the plates tested was suf-ficient in extent and magnitude to cause ar-rest had there been no redistribution of stress.

8) The texture of the fracture surface was

rough near the initiation edge and quite smoothin the central portion of the plates. It is feltthat the texture of the fracture is dependent uP-on the stress or energy level in the materialand not upon the speed. This observation issubstantiated by results of other tests de-scribed m Part B of this report.

PART B . FRACTURE PRO PAGATION IN CENTFLAL-&Y NOTCHED AND WELDED STEELPLATES

INTRODUCTION

Background

The studies of the effect of residual stresson fracture propagation conducted thus far onProjects SR-I 37 and SR-155 had employed thenotch–wedge-impact method of fracture initia-tion. From time 10 time research workers haveraised questions as to the effects of this arti–ficia 1 cracbstarting process on the recordedspeeds and strains.

During the past several years work in GreatBritain, s Japan, 13 and the United States=2 hasshown that brittle fractures can he initiatedfrom centrally notched and welded wide platespecimens at low average applied stresses.Although the initiation procass is not completel-y understood as yet, the stress concentrationresulting from the notch, the residual tensilestresses re su lling from the welding procedure,and perhaps metallurgical changes resultingfrom the welding process, make it possible forfractures to initiate in these specimens at lowapplied stresses. Specimens generally frac-ture either by a short arre steal fracture at a low–stress leve 1 fo ilowed by complete fracture afterthe specimen has undergone general yielding or,

at the time of fracture Initiation, it is possiblefor the specimen to undergo complete single-stage fracture. Quite obviously, the tempera-ture of the test has considerable Influence onthe fracture mechanism noted. Re suits ofrecent studies conducted at the University ofIllinois as a part of the Welding ResearchCouncil Program utilizing centrally not ched andwelded plates indicated that this ‘type of speci-men might be desirable for use in propagationstudies as a part of this brittle fracture me–chanics study. By employing this type ofspecimen it would be possible to obta~n low-stress brittle fractures which could be stati-cally initiated thereby eliminating any externalinitiation device. Since these fractures wouldbe expected to irutiate and propagate at afairly low- stress level it would be expectedthat the resulting fracture speeds would be inthe low to intermediate range which is the pri-mary area of interest in this phase of the in-ve stigation.

~bject and Scope

The purpose of the tests described in thispart of the report was 10 study the parametersassociated with the initiation and propagationof brittle fractures in centrally notched andwelded steel plates; more specifically, to

evaluate and assess the parameters directlyaffecting fracture speed. Consideration alsonecessarily was given to certain aspects offracture lrntiation in these specimens and thisphase of the fracture process is discussed insome detail in this report.

A total of nineteen specimens was testedas a part of this phase of the program. Thetests were all conducted on either 3/4–in. or5/8-in. steel plate specimens which containeda central precut notch; the majority of thespecimens contained either a complete or in-terrupted longitudinal butt weld to produce theresidual tensile stress necessary for low-stress static fracture initiation. Other fabri-cation procedures were used in a limitednumber of specimens ‘co faci Iitate the study ofthe various parameters under investigation.The specimens are designated as Tests 51through 69 to conform to the designation oftests conducted as a part of the Brittle FractureMechanics Program.

DESCRIPTION OF SPECIMENS, INSTRUMENTA-TION AND TEST PROCEDURE

in cases where the stress level is high enough

NIateria 1 Propert&e&

The specimens employed in this investiga-tion were fabricated from 3/4-in. and 5/8-in.thick s~mikilled steel plates. The chemical,mechanical, and Charpy V-notch data for thismaterial are shown in Fig. 19.

Fabrication Proce -

Of the nineteen specimens tested as a partof this phase of the program, seventeen werefabricated of 3/4-in. plat~ and two of 5/8-in.plate. With the exception of Test 64 which

Yield Pmmeutm Percent. ma wttcm

L*FLhdueti~n

v33.2 3L6

L!c LT63.7 62.6 s ~ 63 63

Tvrdurm . ‘%Cb-py v-Not,h cm.?.

FIG. ~9. ~TER~L pROPERTIEs.

was a 1 -ft-wide plate, the specimens were two

feet in width, and three feet long. All of thQspecimens contained a central notch which wascut with either a straight jeweler’s saw or acircular cutting wheel, both 0.006 in. thick.A double-Vee notch cross section was employ-ed in all specimens except Test 6Z which hada straight notch. The notch geometry for thetwo types of notches is shown in Fig. 20, andsketches Of the various types Of specimens

are shown in Fig. 21.

The general preparation of all specimens,except for Tests 61 and 63, consisted of cut-

\ /

FIG. 20. NOTCH I)IMENSIONS.

mL-.24 in__J

~=.t.51.52,53,5)1,5565,67,69and@

TmEA

-—24 m.._mm a

TYPED

TYTEB

(-.24 —~

Tests59an,!65

mc

FIG. 21. SPECIMEN LAYOUT.

20

ting the specimen along its longl’hidinal centerline, bevellng and notching ‘the adjacent edgesof the resulting two tmlves, and rejoining thetwo halves by various methods. In joining thetwo halves of the notched and beveled speci-mens, three fabrication Wocedures (used forSpecimen Types A, B and C in Fig. 21) wereemployed. Th@ Type A specimens were joinedwith a continuous double-Vee butt weld; theType B specimens were joined with an inter-rupted double–Vee butt weld in which a 3 in.section at the notch was left unwelded; in theType C specimens the two ha Ives of eachspecimen were merely welded to the pull platesof the “testing machine leav~ng the adjacentnotched and beveled edges unwelded. Test 61(Type D) was prepared by first cutting thecentral double-Vee uotch in a plain plate afterwhich two 6-in. -long transverse slots werecut 5 in. above and below the notch and weld-ed to produce a residual stress f~eld. Test 63(T.yPe E) was a Plain Plate specimen in whicha central double-Vee notch was cut. All of thespecimens were fabricated such that the longi-tudinal axis was parallel to the direction ofrolhng of the plate material and parallel to th~dire c~ion of testing.

A total of six welding passes with 5/32 in.and 3/16 in. diameter E7018 electrodes wererequired for all of the specimens which con–tained longitudinal w@lds, with the 5/32 in.diameter electrodes being used for the first andsecond passes. In welding, subsequent

pas ses were made on opposite surfaces of thespecimen and the maximum interpass ternpsra-ture was Limited to 100 F. The welding PrO-cedure fo Hewed was s lightly different for theType A and Type B specimens. In the former,each weld pass proceeded from one end of thespecimen to the other, whereas the Type Bspecimens were joined such that each weldpass was placed in {wo-sta.ges each of whichbegan at the end of the plate and terminatedapproximately 1 1/2 in. from the notch.

Instrumentation and Mea sureme nt Techniques

Instrumentation for this series of tests wasprovided for the measurement of cyclic strainsduring welding, longitudinal residual strains,static strains and elongations during loading,and dynamic strains and speeds during frac-ture propagation. All of the above variableshowever were not measured for every speci-men.

Instrume~tation included both electricaland mechanical strain gages. The electricalstrain gages employed In this investigationwere either Baldwin SR-4 Type A7 (1/4 in.gage length) or Budd Metalfilm Types C6-14.1 B and C6-lXI 32 A (1/4 in. and 1/32 in.gage lengths respectively) strain gages. BuddMetalfilm Type C6-I 21 MR3A strain rosetteswere used for the measurement of dynamicstrains in two tests. Crack speed detectors

were Baldwin SR-4 Type A9 gages (6 in. gagelength). NIecharucal gages employ~d in themeasurement of residual strains were a 2-in.md a 6-in. Berry gage.

Residual Strain Measurements--The resid-ual strains in ‘the specimens resulting fromwelding were measured by both electrical andmechanical strain gages. The gages wereselected in order to Provide a wide range ingage lengths which served as a check on theuniformity of strains. Also, the high tempera-tures induced during welding made it neces-sary to use mechanical gages at locationsnear the weld. In determining the longitudinal

residual strain pattern, strains were measuredonly across the notch line.

Initial strain readings were taken prior towelding with the plate in an unrestrainedposition. The initial strain readings estab-lished the “zero” strain level and all othwstrains were referenced to this prewe ldeclstate. After welding had b~en completed, an-other set of strain readings were ‘taken at roomtemperature from which the re sidua I strainpattern was obtained.

Measurement of Cyclic Strains and Tem–peratures Durinq Welding--Strain and tempera-ture readings within approximate lY 1/8 in. ofthe notch root on the plate surface were re-cord~d during welding on selected specimens.

Continuous records of both temperature andstrain as a function of time were obtained duringwelding by connecting the stfain gage andthermocouple leads through bridge circuits totwo mi llivo It strip chart recorders, one of whichrecorded strain and the other temperature. Theparticular strain gages used were temperaturecompensated to 250 F.

I&lea surement of Static Strain and Elonqa-tion--Static stress- strain records were ob-tained for selected specimens during loadingStrain records from eight gages could be ob-tained by tonne cting the leads from the strain

2:

gages to an eight-point motor driven sarnp hngswitch which m turn was connected through abridge circuit to a millivolt strip chart recorder.

In addition to recording static strains duringloading, several specimens were instrumentedwith a 36 in. extensometer. The extensometerprovided a continuous record of ‘the specimenelongation during loading.

Dynamic Instrumentation -- Nine channe 1s ofcathode ray OSCIIIO scope recording instrument~tion for recording the dynamic strains and frac-ture speeds were utilized in some of the testsdescr~bed herein. Tkus method of recordingdynamic strains and speeds consisted of photo-graphing the strain response and dete CtOr

breaking times as they appeared on the oscil-los copes during fracture propagation and wasidentical to the technique used in previoustests on Projects SR-137 and SR-155. Thesystem employed in triggering the oscilloscopesweeps, however, was somewhat different.Because fractures were statically initiated inthis series of tests and the stl-ain level at thetime of fracture was unknown, it was neces-sary to trigger the sweeps with a signal pro-pcutlonal to strain rate, independent of thestrain magnitude. The triggering was ac-complished by connecting the trigger gagesthrough a bridge to a differentiating circuitwhich produced a signal roughly proportional tothe strain rate at the trigger gages. The signalfrom the differentiating circuit after being am-plified by a high-gain amplifier was fed into asweep generator which triggered the horizontalsweeps of all nine oscilloscope channels .

TWO trigger gages were mountsd on thenotch line of each specimen about 1/4 In. and3/4 in. from th~ notch tip where high strainrates would occur with the onset of rap~d frac-“ture propagation. The trigger gages were

connected in a bridge circuit as shown in Flg.22* By wiring the trigger gages in this way, asweeP would be irutiated by a high Strain ra~fi

at either of the trigger gages.

Te St Fro cedure

TWO testing machines of 600, 000-lb and3, 000, 000-lb capacity were employed In thisstudy. The sPecimens were installed in theappropriate testing machine (depending uponthe anticipated fracture load), cooled, andloaded statically to failure.

The specimens were cooled by a mixture ofsolvent and dry ice placed in cooling tanks; atotal of four tanks were used, two on eitherside of the specimen above and be low thenotch . This cooling technique result~d in auniform temperature distribution along thenotch line.

When the desired test temperature had beenattained, loading began and the recordingequipment was started. Because the fractureload in these specimens could not be predict-ed, the film used to record dynamic strains andspeeds had to run continuously throughout thetest. This recording procedure resulted insatisfactory test records in most cases but oc-casionally records were 10 st because of mal-function of the tr~gger, a short arrested frac-

- TOHorizontal

L-L—

!criggerGage J

Difler-enthxtingcircuit

Strain Rataat Trigger

Efi-Gain

kplifier

L z Axis un-ManU.nS forall C-M

w Deflection Sptaaof 9 RecordinsChmnelc

JSweepGenerator

-’v’

——

FIG. 22. TRIGGER CIRCUIT.

22

ture occurred, or because of excessive speci-men loading time and the film supply was ex-pended.

PRESENTATION AND DISCUSSION OF TEST

RESULTS

General

Britt Ie-fracture tests were conducted onnineteen centrally notched specimens in whichfractures were initiated statically. Nine ofthe se specimens contained a continuous lo Wi-tudinal weld (Type A), six contained an inter-rupted longitudinal weld with a 3 in. gap at thenotch left unwelded (Type B), one containedtwo transverse welds dove and below thenot ch (Type D) and three were unwelded (twoof Type C and one of Type E). In these testsconsideration was given to both initiation andpropagation aspects and in the process anumber of items, including thermal straincycling during welding, residual strains re–suiting from welding, load, strain and deform-ation at fracture, and dvnamic strains andfracture speeds were studied. Table 2 con-tains a summary of the results of these “tests.

Residual Strains

The residual strain field was similar in allof the longitudinally welded specimens (TypeA and B) regardless of whe~her the weld wascorminuous or interrupted. With reference to atransverse section, the central one-third ofthese specimens had a residual longitudinal

tensile strain, balanced by residual compres-sive strains in the outer two-thirds of theplate. The residual strain decreased rapidlyfrom a maximum value of 1000 to 2000 microin./in. near the weld to a fairly uniform compres -

/,sive strain of –300 to -500 microin. ~n. towardthe edges of the plate as shown in the typ~calresidual strain plots presented in Fig. 23. Themechanical and electrical residual strain read-ings were in general agreement with some dif-ferences noted, most lik~ ly arising from dif-ferent gage lengths and inherent difficultiess inmaking ‘the measurements in highly distortedregions.

Unlike the long~tudinally welded speci-mens, the residual strain field in the central

portion of Test 61, which contained two trans-verse welds above and below the notch wasfairly uniform except near the notch where ahigh strain concentration occurred. As shownin Fig. 23 (c) the residual strain was extreme-ly high (approximately 6000 microin. /in. ) nearthe notch tip, decreased rapidly a small dis-tance from the notch tip to a residual tensilestrain of about 800 microin./in. and then de-creased less rapidly until a fairly uniform re-sidual compressive strain of about -500microin. /in. was attained 8-in. from the cen-ter line of the specimen. It should be pointedout that the gage used to measure the strainnear the tip of the notch was located closerto the notch in this specimen than similargages in the other specimens, a fact whichmay account, in part at least, for the extreme-ly high residual strain observed at this point.

I I

!I1

Metalflla GuEs I

M(1/4 in. Gu&

hsth) UOedFor Strain

~2

witsnc. m-m centerline - in.

1 1 II ,.. . I I

I l\lMctdfilm (?ases

I

(1/4in.ma L/5la.cageI.aqthU.,dforstrainm

02 468101zD%ntmceFrm Cen’cerlhe- in.

(a) Swcimen55 (lTp A) (b) SWcimen & (~ B)

FI~. 23. RESIDUAL STRAINS .

23

TABLE 2. SUMMARY OF CENTRALLY NOTCHED AND WELDED PLATE TESTS.

AU sp~cimens were 3/4 x 24 x 36 in. and contained double -Vee notches with the following excep-tions: Tests 51 and 52 were 5/8 x 24 x 36 in, ; Test 64 was 3/4 x 12 x 36 in. ; Test 62 containeda straight notch. AU tests conducted at -40 F.

Fracturestrum ReaMual Average SpeedsTemt on Wet Area Specimen Strain Righ-mNo. (kni) W* (Microti./in.) (fps) Ramarko

51

52

53

54

55

56

57

5$

59

60

61

62

63

64

65

66

67

69

69

40.0

la.o

10.0

9.2

10.2

9.0

5.0 Initial fracture33.6 Final &3C+JlrU

17. b Initial fracture>1.8 FinaI fracture

37.0

35.9

35.7

A

A

A

A

A

B

B

B

c

B

D

B

notmeasured

notmeasured

notmeasured

notmeasured

w1/4 in. fronnotch

20(KI1./8im. frannotch

I.&a1/8 im. fra notch

lao1/8 in. fimu notch

o

lW1/8 ITI. &cm notch

@m0.04 in. frcm notch

17001/8 in. frm notch

39.0

13.9

37.L

9.7

8.5

rL.6

9.0

E

B

c

A

A

A

A

o

-1./I6 in. frcm notch

o

notmeam-wed

notmeeaured

netmeasured

notmeasured

notmeasured

55m - 31m

Wo-lml

notmeasured

—-1800

notmeafjured

notmeamured

notmeasured

notmeasured

notmeasured

notmeamxred

nottseaaured

notmeaeured

notmeamured

notmeamured

5ooo-18w

,notmeasured

notmeaeured

5/8 in. tttlck;miasligned notch;general ~elding occurred; roughtextme.

~/8 im. thick; fairly smooth texture

3sooth texture.

Wd dynamir strain records; moothtextme.

&moth texture.

Good dynealc ●train records; smwthtextore.

Initial fracture was 11.5 in. long;

rnooth texture obaened for Initialfracture end rough for finalfracture.

Misali#.h?d notch; initial fractureutm 1 in. long on one aide of.gpc-; rough texture.

General yleldimg occurred; O.Y5 in.elongation in X in. gage length;rough terhu-c.

ILmger gap in weld than othsr TYPEepecimenm, therefore leus cycling;rough texture.

General yielding occurred; 0.2 in.elongation In N in. gage lengkh;rough texture.

Straight notch instead of Vee-notchnpecimen fractured at yield mtreusbut ulthout general yielding; roughtexture.

klimligned notch; general yieldingoccurred; rough texture.

M In. wide awclmen; smooth texture

General yleldlng occurred; 0.27 in.elongation in 36 in. we length;rough texture.

9sootn teitun?.

GcOd dynamic #train records; smoothtexture.

.%ooth texture.

Oood dynamic strain recordm; mootltexture.

24

Thermal Strain Cyclinq

Among the most interesting observationsmade during this series of ‘tests were measurem-ents of the thermally induced strains duringfabrication of four Type B specimens, and thespecimen containing the transverse welds(Test 61). Iri all of these specimens it waspossible .~o ob!a>n straio ~eg?~ds. during we~din~from electrical strain gages mounted near thenotch orI the surface of the plate since the“temperature of the material in this region rare-ly exceeded 200 F. The strain and tempera-ture records obtained during welding of theType B and D specimens are plotted in Flg.24. In presenting these records, Portions Ofthe stra~n data recorded between weld passeswhere the strain was approximately constanthave been omitted; by way of illustration ofone complete cycle, the record for one weldingpass is shown in Fiq. 25.

1

FUG. 24. STRAIN AND TEMPERATURE DURINGWELDING .

Fun No.(c)’r-t*

me. no.(a)‘k.t62

FIG. 24. STRAINAND T’EMPERATUREDURINGWELDING.

(.)l-t*

FIG. 24. STRAINAND TEMPERATUREDURINGWELDING.

Note that in Fig. 24 (a) through 24 (d) thereare two complete strain cycles for each weld-ing pass; this arises from the particular weld-ing procedure employed in which one weldpass was placed in two stages, namely, fromone end of th~ specimen to within I 1/2 ID. ofthe notch followed by an identical second halfpass from the other end on the same side of theplate. Consequently, the recording strain

gage was approached twice by the weldingelectrode for each welding pass, thus pro-ducing two strain pulses for each pass as maybe seen in F’~g. 24 or 25.

Another lnt&esting observation is that dur-ing welding a total strain range of over 2000microin./ln. was recorded for a single pass

,.and as much as 3600 microin. in. for the com-plete welding sequence. This range is nodoLlbtquite d~pendent upon gage location withrespect to the notch root and therefore lt 1sdifficult to predict the amount of cyclic strainexperience by the volume of material at thenotch root.

Welding of the transverse slo~s in Test 61produced little cycling as shown in Fig. 24 (e).There was a slight variation in strain duringwelding of this specimen but compared to therecords obtained from the TYP@ B specimensit was negligible, especially when consideredin light of the fact that the gage for Test 61was C1Oser to the notch than those on the TyP@B specimens.

Static Stress- Strain Records

Typical plots of average apphed stressversus strain obtained from static gages dur~ngloading of several specimens ar~ shown inFig. 26. The locations of ‘the 9a9es are ln-

dicatecl on each P1OT. Where possible, thastrain values for zero stress correspond to theresidual Strain eXi S’hn9 in the Plate beforeloading. For the specimens in which fracturesinitiated at about 10 ksi average applied stress>the static stress-strain PIO”LSwere nearly linearUp to fracture. For specimens in which frac-tures occurred at or above yield, the strain

records indicated a definite yielding of th~material with ‘the load at which a particulargage yielded dep~nding upon the location ofthe gage. Note particularly Fig. 26 (a) whichis the stahc record obtained during loading ofTest 51. As shown In this plot, four of thegages which were loca~ed three or more inchesaway from the we Id yielded at approximately38 ksi (approximately yield stress at the -40 Ftemperature), whereas one gage located on theweld Indicated yie Iding at an applied stress ofabout 16 ksi.

In Test 57 the static strain gages with theexception of gage 9 were located above thenotch Line as shown in Fig. 26 (c) . FractureM this specimen occurred in two stages; thefirst fracture propagated about 6 in. from thecenter line of the specimen before arresting.In the same flgur~ it can be seen that gages1 and 2 experienced greater compressive strainsafter the initial fracture than before loadingbegan. These measurements would seem to in-dicate that a considerable residual strain fieldcontinued to exist ‘throughout the plate, evenin the semirelaxed material directly above thefracture. As is usually the case with two-stage fracturing,, complete fracture occurredafter general yielding of the remaining portionsof the specimen.

In Fig . 26 (d), the static strain records forTest 61 are presented. This specimen (Type D)was prepared by welding 6 in. long transverseslots 5 in. above and below the notch. Thetransverse section through the welded s lotswas shghtly thinner than the thickness of theplate, being about 5/S in. thick at the weld asopposed to an overall plate thickness of 3/4 in.

26

Dk..$.h C.m,.ru” -“i..h

.EElii!3m—

m

—W.9----- C!W1O—-— Lw.II

.

[/’o A.--,-J

Wrd. W.rd.. /i.:

FIG. 26. STRESS-STRAINDURING LOADING.

I m.-?.w.i.m kwt

(.1 ,,.% 55—

PLOTS OBTIMNED

I Ihfl.-.ti,,..hp..

FIG. 26, STRESS-STRAIN PLOTS OBTAINEDDURING LOADING.

With This fact In mind, lt can be observed thatthe ra~e of straining of gages 2, 3 and 4 be-tween the slots was slightly less than that ofthe other gages in the pre-yield range, probablybecause th~s sechon was not resls’ung as muchof the apphed load as the edge areas of theplate. Although not described here, lt Is in-teresting to note in Flg. 26 (d) the effect of thespecimen geometry and relaxation on the strainbehavior across the cross section at the notchwith the onset of general yieldlng.

In Fig. 27 are presented load-deformationcurves for the “three specimens (Types C and E)which did not have Induced residual stressfields. It w1ll be noted that general y~e ldingand considerable inelastic deformation occurred’prior to fracture at’ -40 F.

Fracture Stresses

The fracture stress data for all specimens

FIG. 27. LOAD DEFORMATION CURVES .

1s summarized in Table 2. Th@ Type A speci-mens with contlnLlous welds contained highresidual tens~ 15 stresses at the notch and thematerial near the notch had been subjected tothermal stra~n cychng and high tempera ~uresduring we ldlng. All seven 3/4-in. specimensfractured consistently at an apph~d tens~lestress of abOUt 10 ksl. Tests 51 and 52 were5/8 in. thick and fractured at stresses of 18and 40 ksl; the latter lmgh stress quite hkelycan be attrlbut~d to a rmsalign~d notch whichwas observed after fracture.

In properly Vee-notched specimens con-ta~rnng interrupted welds (Type B), completeor part~al brittle fractures occurred at con–slstently low apphed stresses (5 – 13 ks~) aswas the case for the previous specimens con–talnlng complete welds. In the three remaln-~ng Type B specimens, one contained a mis-aligned notch and a first stage fracture occur-red at approx~mately 17 ksi; in the remairungtwo specimens, complete fractures occurred at32 and 35.7 ksi. This abnormally high stresslevel required for fracture may be attributed tothe straight notch used in one specimen and toa larger gap, result~ng in reduced strain cy-cling, in the second specimen. Offhand, thestrain cychng record in Flg. 24 (c) does notsubstant~ate th~s conclusion; however, thegage from which this record was obtained wascloser to the notch than similar gages in otherspecimens, and thus m a region of higherstrain Concentration.

While the properly notched and longitudinalbutt-we lded specimens (with the exception ofTest 51) failed at low stresses well below

27

Yield, the unwelded specimens and the notchedspecimens containing transverse welds allfractured after general yie Iding. Test 61, aspreviously noted, contained extremely highresidual tensile stresses near the notch tipbut had not been subjected to thermal cychngor high temperatures. This specimen fracturedat a stress above yield after O. 20 in. of elon-gation in a 36 in. gage length, indicating thatresidual stress in itself was not sufficient to

cause fracture at a low applied stress.

Dynamic Strains and Speeds

Dynamic strain records from Longltudina lly ,oriented companent gages were obtained fromTests 54 and 56, and from strain rosettes inTests 67 and 69. These records are plotted InFig. 28. Because of the various difficultiesInvolved with certain aspects of the dynamic

(b) TeBt 54

FIG. 28. STRAIN-TIME RECORDS.

4 I 4 I

1G90

4.,,.-.-.~,~.-,,<,-,-..,>

./ \, ,

/m /

<

2

~

2g

-5W

\_ 1— —

.1OGO

o 0.1 0.2 0.5 0.4 0.5mm - Millisec.

(

o 0.1 0.2 0.3 0.4 0.5 0 0.1 0.2 0.3 0.II 0.5 0 0.1 0.2 0.3 0.4 0.5Time - 14iU.iaec. !Mme - MU’+.,.. Time - ~i~iBCC .

(d) ‘rest67 (e) T.mt67 (f) !amt67

FiUG. ?.8. STRAIN-TIME RECORDS.

‘“”m

.J-LrrlJo 0.1 0.2 0.5 O.h 0.5

‘rime- MIllimec.

(g) Tent @

FIG. 28. STRAIN-TIME REC ORiX .

I notch line.—” — .— .— .

~+d

T3 in.,

J.

150Q

~1033

% —

-5qo 0.1 0.2 0.5 0.4 0.5

Time - M111!i9tc.

(h) m-t 69

recording, ~t was not possible to obtain thecomplete strain history for each of the dynamicgages from the beginning of loading until frac-ture initiation. In other words, it was possibleto monitor the strain h~story from the time ofweldlng up to the time of testing; however, be-caus~ of the recording techniques employed ltwas not POssible to obtain the strain corre-sponding to the exact test load level. For Lhisreason all strtun records are plotted with thezero strain level corresponding to the strainlevel of the measuring gage at the time offracture imtiation. This limitation does notdect the vahdity of the strain records nordoes ~t restrict in any way the interpretation ofthese strain pulses when compared to similarpulses obtained on other tests. However, inthe case of rosette gages it does proh~bit cal-culation of principal strains in the sense thatprincipal strains calculated from the observedcomponent strains have little mean~ng. Be–cause this sltuatlon was recognized quiteear~y for those specimens instrumentated withstrain rosettes, for convenience, the gageswere placed on the specimens after the speci-mens had been fabricated.

As far as the dynamic s~rain records them-selves are concerned, a study of Fig . 28 willshow that the component gage traces ari singfrom tests in which only longl~uchnally orientedgages were employed, or tests in which ro-sette gages were employed, all show about thesame trend. At first glance examination of the

traces In Fig. 28 suggest a difference betweenthe se dynamic records and those of earliertests, e specially for the strain traces for hori-zontal and diagonal componen! gages. How-ever, lf one estima!es the actual strain levelcorresponding to the time of Initiation, all thetraces appear to fall more in proper perspec-tive; any interpretation of the shaln traces, asfor @xarnple study~ng trends across the platewidth, must include consideration of the prob-able base strain level. In general, the straingage traces are similar in nature to those whichhave been found in earlier tests Involving thenotcll–wedge–impact methd of lrut~ation.

Several observations can be made aboutthe se strain records . For the vertical compo-nent gages It was found that the peak verticalstrains ranged from 300 to 1000 microin. /m.,but apparently the peak strains were not asdependent upon the distance from the fractureas hsd been found in earher wide plate testsas indicated in Fig. 29. However, it will benoted in Fig. 28 (c) that for gages located 1/2in. from the crack the peak strain was about700 microlr,. /in. This is in line with observa-tions made for the 6-ft-wide pre stressedplates. There are some anomalies in the rec-ords; for @xamplQ, in Fig. 28 (c) Gage ~ showsa trace wh~ch lS not consistent with whatwouldbe expected from the large number of otherrecords and probably this is caused by a me-chanical malfunction, such as stretching ofvmres or some other phenomenon during the

29

o 1 2 3 4DiEtagce Frm Fracture - In.

FIG. 29. RELATIONSHIP BETWEEN PEAK LONGI-TUDINAL STRAINS AND DISTANCE OF THE MEA-SURING GAGE FROM THE FRACTURE PATH.

testing operation. Also, Gage 1 which isquite close to the initiation source does notshow an extremely high peak which one m~,ghiexpect In the region of high speeds as ob–served in a large number of previous tests.However, for another specimen for which therecords are shown in Fig. 28 (d) a verticalgage 3 1/2 in. from the source of inl~iatlonshows a strain peak which 1s more in hne withwhat one would expect in tkus region.

With regard to the response of the compo-nent gages of the rosettes, the traces werequite similar to those observed in other testsmade as a part of this program. The horizontalgage of Rosette 2 In Tests 67 and 69 (Fig.

28 (e) and 28 (g)) peaked in tension as hadbeen observed in earlier tests for rosetteslocated close to the fracture. The behavior ofthe other component strains was quite Iyplcalof those recorded in earlier tests on ProjectsSR-i37 and SR-155.

Average fractur~ speeds were determinedfrom th~ peaking t~mes of the longitudinal.lyoriented gages and in Test 53 from the det@c-tor breaking times. The strain gage peak~ngtimes and detector breaking times are plottedas a function of the distar.cc of the particularskain gag~ or detector from the trigger gagelocatlon In Fug. 30. The slopes of thesecurves are an indica~ion of the zverage frac-ture speeds.

x

Test54

/——

ze~t 56

Test53

..-—

—, ......-.—

TI

0.1 0.?2 0.3 0.4PeakingTime - l.lUl.is@c.

FIG. 30. FRICTURE SPEEDS.

The fracture speeds as Indicated by F~g.30 appear to have decreased whenthe fracture

propagated into ~ne region of low reslclualstrain. In general, speeds ranged from 5000to 6000 fps near the lmti~tion source LO about1800 fps after the fracture had propagated a few-inches. This variation In frac~ure speed a~sois rmflec~ed 1P. the dynamic strain records inwhich it can be observed that the length oftire@ requmed for Peaking was lon9er for Strain

gages located near the edges of the specimensthan for those located near the center.

Although the recorded data lS average lr.nature, one extreme ly impor~ant observationarises from the se est~mations of fracturespeeds. It wIII be noted in Fig. 30 that th~fracture speed attains a high value, in therange of 5000 fps within a distance of approxi-mately 1 to 2 in. of the fractur~ source. It maybe r~callecl that in tests involvlng the notch-wedge-impact method of lnlL”iaElon, lt had notb~en possible to measure speeds at distancescloser than about six ~nches to the ~nitialionsource. Thus, we now have evidence thateven in the case of a statically inl~iated frac-ture the speed can become very high w~thln ashort dlstanc~ from the source of mtia”t~on.

Another indication of the re ievant magnitudeof the speeds involved in these tests is givenby the width of the strain pulses. It has beenobserved previously in plain plate tests that inthe high-speed region the tensile pulses forvertically oriented gages can be as short as100 mlcrosec. or !.ess. On the other hand, inthe prestressed 6-ft-wide pla~e tests, wherethe speeds were as low as 50 fps, some of the

30

strain pulses were observed to be aS long as2 to 5 rnillisec. In the present series of teststhe comparable traces in the compress~vestrain region were observed to have base pulsewidths on the order of 100 rnicrosec. or morewhich would tend to indicate that these pulseswere associated with an interrnedj ate speedrange.

Fracture Appearance

IrI genera 1 the fracture texture becamerougher and the fracture path tended to wanderslightly as the fracture stress increased.Figure 31 (a) whic3 is a photograph of the frac-ture surface of Test 53 which failed at 10 ksi,shows the smooth texture typical of all of the

FTG.

(a) Test 53

FIG. 31a. FRACTURE APPEARANCE .

(b) Test 52

31b. FRACTURE APPEi%FLANCE.

(c) Test 51

FIG. 31c. FRACTURE APPEARANCE .

31

specimens which fractured at quite low aver-age applied s’cre sses. The chevron markingsof this fracture are barely visible and only aslight shear lip can be observed near thesurfaces of this plate. Low-s h-ess fractures,in addition to having a smooth texture, dis-played a uruform texture across ‘the entireplate width. The fracture texture of Test 52(Fig . 31 (b)) which failed at 18 ksl, when comp-ared with that of the low-stress fractures justnoted, is shghtly rougher with a wider shearlip and more discernible chevron markings.The typical Imgh-stress fractures (y~eld pointand above) are character~zed by a very roughtexture and clearly visible shear hps andchevron markings as can be seen in F~g. 31 (c).

The fracture paths ~n the various specimensalso apparently wer~ influenced by the frac-ture stress as shown In Fig. 32. The low-stress fractures generally followQd a fairlystraight path whereas those fractures occurring

near the yield po~nt of the material were likelyto wander slightly as may be seen in the fig-ure.

Figure 33 shows the fracture surface ofTest 57 in which a partial length fracture OC-curred at 5.0 ksi. Note the extr~mely smooth

texture of the initial fracture as compared tothat of the final s~ag~ which occurred at a netstress of 33.6 ksi. The zone of arrest notedby the arrow and the thumbnail also is inter-esting to observe because the tip of the ar–rested crack was no~ symmetrical with respec!to the thickness. Reinitiation of this fracture

appears to have or~ginated a~ mid depth of theplate even though the geometry of the cracktip was not symmetrical.

SUMMARY

In this particular study considerable in-formation relating to ‘the propagation n and

FIG.

(b) High Stress Fracture - Test 51

FIG. 32b. FRACTURE PATH.

32

(a) Fracture Appearance

FUG. 33a FRACTURE APPEARANCE FOR TWO STAGE FRACTURE -TEST 57.

,, “,’,,, ! ,,. ,,”,

,,

(b) Arrest Region

FIG. 33b FRACTURE APPEARANCE FOR TWO STAGE FRACTURE - TEST 57.

initiation aspects of brittle fractures was ob-tained. For convenience of presentation, thissummary is subdivided into two sections,namely, propagation and initiation.

Propagation Studies

In the following discussion, it should bekept in mind that all dynamic measurementswere obtained from specimens which fracturedat a low average applied stress (8 to 12 ksi).

The recorded fracture speeds in these testsranged from about 5000 fps in the central por-tions of the specunens, where residual tensilestresses were high, to approximately 1800 fpsthroughout the rgmaining portion of the plate.

This decrease in speed as the fracture propa-gated Into the region of residual compressivestra~ns was similar to the behavior of thefractures discussed in Part A of this reportand serves to verify the point of view ~hatchanges In speed are dependent to some extenton the stress level existing in the plate. An-other interesting observation about the frac-ture speeds is that in these tests high fracturespeeds on the order of 5000 fps were recorded

at distances as close as 2 in. from the jnlti-ation source. This is the first time that anindication of the fracture speeds this near asource of initiation, parhcularly in the case ofa statically initiated fracture, has been ob-tained in the work of Projects SR-I 37 andSR-155.

3 3

From the strain traces recorded from theindividual vertically oriented strain gages andfrom the compone rit gages of strain TOsette,s,it was observed that the traces from the com-ponent gages were similar to those recordedin the earlier plain and prestiessed plate testsinvolving the notch-wedge-impact method ofinitiation. In the case of vertical componentgages, the peak strain mwnitude recorded WaSon the order of 700 to 800 microin. /in. Incomparing these peak values with the resultsfrom other tests for gages at about the samelocation with respect to the fracture path, itis concluded that the peak strain values are ofabut the same order of magnitude for the ap-plied strain field conditions encountered. Forthe horizontal and diagonal component gages,the re suiting strain-time records were some-what different than would have been expectedon the basis of previous tests. It ShOU~d bekept in mind, however, that anY interpretationor comparison of this dynamic strain data mustinclude a consideration of the base strainlevel at the time of fracture.

The effect of residual compression on thestrain fie Id surrounding the tip of the propa-gating fracture, could not be fully identifiedin these tests because the dynamic strainswere recorded after the fracture had propagatedinto the zone of residual compression and con-sequently no strain records, recorded in aregion of high residual tension, were availablefor comparison.

The fracture texture observed from thespecimens appeared to be dependent upan thestress or energy level present in t,he plate“prior to fracture. The texture became rougher

with more easily identifiable chevron markingsas the stress at the time of fracture increased.The fracture path also appeared to depend uponthe stress level; the path was quite straightfor the low stress fractures and had a tendencyto wander slightly for the high-stress frac-tures.

Initiation Studies

The fabrication ‘procedures employed in thisinvestigation re suited in various conditions ofresidual strain in the several specimens, aswell as different strain and temperature his-tories for the material in the vicinity of thenotches.

The following statements apply in general

to the 3/4-in .-thick specimens which com-prised the majority of the tests. The resultsfrom the two 5/8-in. specimens fall in line,as discussed ~n the text, when the peculiar-ities of the particular specimens are con-sidered.

In Type A specimens the fabrication pro-cedure employed (continuous longitudinal weld)resulted in high residual tensile strains in thevicinity of the welds as well as high tempera-tures and high strain cycling at the notch. Thefractures always irutiated at an average ap-plied stress of between 8 and 12 ksi whentested at -40 F. This consistency of fracturestress at initiation is indeed remarkable con-sidering all the variables involved.

In the Type B specimens, which werefabricated with an interrupted 10ngltudinal weldleaving a gap of about 3 in. in the vicinity ofthe notch, low stress, high stress, and sometwo-stage fractures occurred. In all of thesespecimens the gap in the weld resulted intemperatures not exceeding about 200 F in thevicinity of the notch; it was observed that theresidual tensile strain field ~n the vicinity ofthe notch was almost identical to that inducedin the Type A specimens. Low stress and two-stage fractures were observed in the properlynotched specimens which had been subjectedto high strain cycling in the vicinity of thenotch. Of the two Type B specimens exhibit-ing high- stress fractures, the strain cychngduring welding quite likely was less in onebecause the we Id gap was longer, and theother contained a straight notch of differentgeometry.

Type D specimens, which cormsted of aVee-notch in the center of the plate, had aresidual strain field induced by the welding oftransverse s lots 5 in. above and below thenot ch. This we Iding of the slots induced ex-tremely high residual tensile stresses at thetips of the notch. The temperature rise at the

notch associated with the welding of theseslots was negligible. Also, it was observed

that the strain cycling at the tips of the notchwas much less than any of the strain cyclingobserved in the other types of welded sP@ci-mens. The one specimen made by this methodfailed at a high stress.

The Type C specimens consisted of twoplate halves containing the central Vee-notchbut without any welding; the ends of the plateswere welded to the pu 11 plates prior ‘co testing.

34

The Type E specimen was a Plain Plate with aVee-notch in the center. In these plates, ob-viously, there was nO effect arising from we~d-mg to create residual stresses, temperatureeffects, or strain cyc~in9 effects. In all casesthese plates failed at high stresses after under-going considerable plastic deformation.

A study of the results of these hmitedinitiation studies leads one to conclude thatthe residual tensile strain field alone is notsufficient to initiate low average stress brittlefractures. From the studies made as a part of

this investigation and other related studies itappears that the notch geometry, and straincycling in the vicinity of the notch play a sig-nificant role in the initiation process.

ACKNOWLEDGMENT

This brittle-fracture-me chanics study is apart of the structural research program of theDepartment of Civil Engineering at the Uni-versity of Illinois. The investigation was

sponsored by the Ship Structure Committee,through the Bureau of Ships, U. S. Navy,Contract NObs. 65790. The members of thePro ject Advisory Committee, Dr. Dana youn9 YChairman, under the cognizance of the Com-mittee on Ship Structural Design of the Nation-al Academy of Sciences-National ResearchCouncil, have acted in an advisory caPacitYfor this program.

REFERENCES

1.

2.

3.

Barton, F. W., and Hall, W. J., “Brittle-Fracture Tests of Six-Foot Wide Frestre ssedSteel Plates, “ The Welding JournaL 39:9,Research Supplement, PP. 379-s- 384-s(September 196CI).

Barton, F. W., and Hall, W. J., Studiesof Brittle Fracture Propagation in Six-FootWide Steel Plates With a Residual StrainFie Id (Ship Structure Committee ReportSerial No. SSC-1 30), Washington: Na-tional Academy of Sciences-National RQ-search Council, April 3, 1961.

Hall, W. J., Mosborg, R. J., and Mc-Donald, V. J., “Brittle Fracture Propagationin Wide Steel Plates, “ The Weldinq Journal,36:1, Research Supplement, PP. l-s-8-s(January 1957).

4.

5.

6.

7.

8.

9.

10.

11.

Hall, W. J., Godden, W. G., and Fettah-lioglu, O. A., Preliminary Studies of Bri~-tle Fracture Propagation in Structural Steel(Ship Structure Committee Report Serial NO.SSC-1 11), Washington: National Academyof Sciences-National Research Council,May 15, 1958.

Rolfe, S. T., Hall, W. J., and Newmark,N. M.j Brittle-Fracture Tests of Stee 1Plates Containi~e sidual Compressive.—Strain (Ship Structure Committee ReportSerial No. SSC-117), Washington: Na-tional Academy of Sciences-National Re-search Council, July Z, 1959 (Reprintedfrom The Weldinq Journal, 38:4, ResearchSupplement, pp. 169-s- 175-s (April 1959)),

Lazar, R., and Hall, W. J., Studies ofBritt Ie Fracture _Propaqation in Six-FootWide Structural Stee 1 Plais (Ship Struc-ture Committee Report Serial No. SSC-11 Z),Washington: National Academy of Sciences-National Research Council, September 17,1959.

Rolfe, S. T., Lynam, T. M. , and Hall,W. J., S-iudles of the Strain Distribution igWide Plate s’ Durinq Brltt le Fracture Prop@-qation (Ship Structure Committee ReportSerial No. SSC-ll S), Washington: Na-tional Academy of Sciences-National Re-search Council, December 30, 1959.

Wells, A. A.,. “Influence of ResidualStress and Metallurgical Changes on Law-Stress Brittle Fracture in Welded SteelPlates, “ The Weldinq Journal, 40:4, Re-search Supplement, pp. 182-s -192-s(April 1961).

Kihara, H., Yoshida, T., and Oba, H.,“Initiation and Propagation of Brittle Frac-ture in Welded Stee 1 Plates, “ HW DocumentNo. X-217-59.

Iida, K., Kusudo, T., Matsuoka, T.,Maeda, T., and Kihara, H., “Effect ofResidual Stresses on the Behavior of Brit-tle Fracture Propagation, “ (January 196 1),IIW Document No. X-263-61 .

Yamauchi, S., and Nakai, T., “Propaga-tion Characteristics of Cracks in the Weld-ed Zone of Quenched and Tempered HighTensile Steel, “ (January 1961), INV DOCU-

35

ment No. X-264-61.

12. Hall, W. J., Nordell, W. J., and Murise,w. ET.,“studies of Welding Procedures, “The Welding Tournal, 41:11, Research Suppl.pp. 505-s -518-s (November 1962).

13. Kihara, H., “Recent Studies in Japan onBrittle Fracture of Welded Stee 1 StructureUnder Law Applied Stress Level, “ JapanInstitute of Welding, (September 1961),HIM Document No. X-291-62.

14. Mylorlas, c., and Fbckey, K. C., “Exh-austion of Ductility by Hot Straining -An Explanation of Fracture Initiation Closeto Welds, “ TI’e_We ldinq Journal, 40:7,Research Supplement, p. 306-s (July 1961).

+ U.S. GOVERNMENT PRINTING OFFICE, 1963 0 704-994

Chairman:

‘A. R. Lytle R..W. Rumke: ””’Director ,Executive Set

~>‘,, ,..!/;.,.