Astm stp 176 1955

By defining transition temperature as the highest temperature at which an initiating crack can propagate under action of the elastic-stress energy alone, that is, sudden and complete fra

Trang 1

Trang 2

SYMPOSIUM ON IMPACT TESTING

Presented at theFIFTY-EIGHTH ANNUAL MEETINGAMERICAN SOCIETY FOR TESTING MATERIALS

Atlantic City, N J., June 27 1955

ASTM Special Technical Publication No 176

Published by the AMERICAN SOCIETY FOR TESTING MATERIALS

1916 Race St., Philadelphia 3, Pa.

Trang 3

This page intentionally left blank

Downloaded/printed by

Trang 4

This publication is based on a Symposium on Impact Testing that washeld at the Fifty-eighth Annual Meeting of the American Society for TestingMaterials in Atlantic City, N J., June 27, 1955 comprising the First andSecond sessions The symposium was sponsored by Committee E-l onMethods of Testing with Mr F G Tatnall, Baldwin-Lima-Hamilton Corp.,Philadelphia, Pa., serving as Symposium chairman

Mr Tatnall also presided at the Second session with H L Fry, BethlehemSteel Co., Inc., Bethlehem, Pa., as secretary and Mr W W Werring, Bell

Telephone Laboratories, Inc., New York, N Y., presided at the.First session with W H Mayo, U S Steel Corp., Pittsburgh, Pa., as secretary.

In addition to the papers presented as a part of the Symposium, five other papers, being appropriate to the general theme of the Symposium, have been

included on "Effects of Manganese and Aluminum Contents on TransitionTemperature of Normalized Nickel Steel," by T N Armstrong, and O 0.Miller, International Nickel Co.; "Low-Temperature Transition of Normal-ized Carbon—Manganese Steel," by T N Armstrong International Nickel

Co and W L Warner, Watertown Arsenal; "Effect of Specimen Width

on the Notched Bar Impact Properties of Quenched-and-Tempered andNormalized Steels," by R S Zeno, General Electric Co.; "Stress-Strain Rela-tionships in Yarns Subjected to Rapid Impact Loading," by Herbert F.Schiefer, Jack C Smith, Frank McCrackin, and W K Stone, NationalBureau of Standards; and "Shock Testing with the Rocket-Powered Pendu-lum," by R W Hager, Sandia Corp

Trang 5

NOTE.—The Society is not responsible, as a body, for the statements

and opinions advanced in this publication.

Trang 6

Stress-Strain Relationships in Yarns Subjected to Rapid Impact Loading—Herbert

F Schiefer, Jack C Smith, Frank L McCrackin and W K Stone

Shock Tester for Shipping Containers—W H Cross and Max McWhirter

Shock Testing with the Rocket-Powered Pendulum—R W Hager.

Properties of Concrete at High Rates of Loading—D Watstein.

Discussion.

PAGE

1

3 7

10 23 25 40 59 70 75 76 84 93 94 110 111 125 126 141 149 156 170

Trang 7

Trang 8

Over the years there have been many

discussions of the technique and

signifi-cance of the impact test, including several

formal symposia

The last Impact Symposium was held

in 1938 Much technical development

since that time has been attributed to

information contained in those

sympo-sium papers

Now seventeen years later another Im

pact Symposium steps into a field that

is already brimming with interest and

activity Some phases of this field carry

the designation "Environmental

Test-ing."

Along this line, it was suggested sometime ago by members of the ImpactCommittee, of Committee E-l, that pa-pers on shock tests be included in thissymposium which would encompass im-pact in parts, components, and completestructures, and not confine_the sympo-sium to notched bar testing This broad-ened scope has been undertaken withwhat appears to be very beneficial en-hancement of the parctical application

of the impact teststraining rates The end result is an ex-cellent balance between theory and ex-perimental results

SYMPOSIUM ON IMPACT TESTING

Trang 9

Trang 10

NOTCHED-BAR TESTING—THEORY AND PRACTICE

BY S L HoYT1HISTORICAL BACKGROUND

This paper covers a few of the points

thought to be especially significant in

handling the practical phases of brittle

fracture

When metallurgists and engineers were

first confronted with the brittle fracture

of steel, it was thought that impact

must be the cause But leading

experi-menters soon found this concept to be

inadequate and, at the turn of the

century, the notch was recognized in its

true light and was introduced into formal

tests of steel quality At the same time,

the energy absorbed in breaking the test

bar was universally adopted as the

meas-ure of quality; this led to the simple

device of breaking the test bar by means

of a swinging pendulum Doubtless due

to the earlier notions, the test thus

be-came commonly known as the

"im-pact test" and so perpetuated the idea

that brittle fracture was the result of

impact This was in spite of

demonstra-tions to the contrary, such as the work of

Considere,2 who found that increasing

the strain rate simply raised the

tem-perature at which brittle fracture

oc-curred

Two other misconceptions impeded

understanding of notch brittleness and

delayed the general acceptance of

notched-bar testing First was the failure

to fully recognize the rigidity or thestiffening effect, at the notch section,which is caused by the third dimensionalstress This is the stress, amplified by thestress concentration, which must exceedthe cohesive or brittle strength for brittlefracture to result Attention becamecentered on stress raisers, and the theoryseems to have been that, since steel isductile, a slight deformation at the rootwould relieve the stress, and that nothingmore was important An example whichbrings out the effect of rigidity is thefailed forging, which while ductile withthe single or standard-width test bar wasbrittle, with the same notch, when testedwith the double-width bar Another isthe Navy Tear Test,3 which gives a muchhigher transition temperature than theCharpy keyhole test bar does with thesame notch acuity Another misconcep-tion has been that steels having the sametensile properties would all performsimilarly in the presence of a notch Thiswas negated at an early date by a formaland authoritative series of tests con-ducted by the German Society for Test-ing Materials, the results of which werepublished in 1907.4 In more recent times,

1 Metallurgical Consultant, Columbus, Ohio.

2 Considere, "Contribution a PltJtude de la

Fragility dans les Fers et les Aciers," (1904).

3 R H Frazier, J R Spretnak, and F W Boulger, "Reproducibility of Keyhole Charpy and Tear-Test Data on Laboratory Heats on Semikilled Steel," Symposium on Metallic Ma- terials at Low Temperatures, Am Soc Testing Mats., p 286 (1953) (Issued as separate publi-

cation ASTM STP No 158.)

4 See for example, S L Hoyt, "Principles of Metallography, McGraw-Hill Book Co., Inc., New York, N Y., p 226 (1920).

Trang 11

SYMPOSIUM ON IMPACT TESTINGthe fallacy of this assumption has been

demonstrated many times

Thus, while stress concentrations were

recognized, the important effect of notch

rigidity was largely ignored and the easy

assumption was made that ductility

would provide steel with what was

needed to handle notch conditions

Im-pacts were held to be the prime service

hazard, which, of course, led to a further

discounting of the effects of notches

This line of thought, except in a few

cases, set up a block to the correct

under-standing of "brittle fracture."

A major object in noting these details

of the history of brittle fracture is to

emphasize that the service problem has

been the brittle behavior and low

energy-absorbing capacity of ductile steel

Hence, the proper handling of this

prob-lem requires an understanding of the

conditions that produce the brittle

fail-ure The problem of ductile failure under

notch conditions is much more complex,

less well understood, and furthermore,

seldom encountered in practice

If steel retained its ductile behavior

under all conditions, as many

non-fer-rous metals do, the problem would be

one of mechanics of materials and not of

metallurgy This is true of steel when it

is tested above its transition

tempera-ture, since such tests determine the

be-havior during, and as a result of, plastic

deformation of some severity In reality,

the significant properties of the steel

must be those of the initial and

unde-formed condition, from which it follows

that to be fruitful, laboratory studies of

brittle fracture should develop the same

brittle behavior as is developed in

serv-ice

It is advanced here that these concepts

are vital to the selection of steel when

the service hazard is brittle fracture and

also to the interpretation of laboratory

test results

STEEL SELECTION

A somewhat unusual example ofnotched-bar testing to select a steel for aspecific application may be illustrated

by means of a hypothetical case It is cussed to illustrate how these principlescan be employed

dis-In this example, a weldment made ofcommon steel was exhibiting brittlefailures at — 30 F When tested with theV-notch Charpy bar, the 10 ft-lb transi-tion temperatures of a group of thesesteels ran from 30 to 90 F, all tempera-tures being idealized for purposes of dis-cussion At the service temperature of

— 30 F, all of the steels tested includingsome of presumably "better" quality,were completely brittle While one couldsay that one steel was better thananother, there was no way of interpretingthe results to decide which steels, if any,were "good enough" for the application.Obviously, these transition temperatures

do not give a sound basis for selecting theproper steel to use At this stage of theexample, there was no opportunity tosecure samples from a known satisfactoryweldment for test purposes

According to one theory, one couldassume that the steel should have atransition temperature, with the stand-ard bar, which came at a safe marginbelow —30 F While that might be a safesteel to apply, there would still be noassurance that it would be "goodenough" or, what is more important inindustrial practice, that it would be

"just good enough." Furthermore, aglance at steel costs would show that asteel with that low a transition tempera-ture would price the weldment off themarket So it became necessary to seekthe solution by other means

A search through past records gested that some steels were performingbetter than others under much the sameconditions This revealed that the weld-4

Trang 12

HOYT ON NOTCHED-BAR TESTINGment was differentiating between steels

at — 30 F, whereas the Charpy test was

not This indicated a possibility of

de-veloping a test bar, with a less drastic

notch effect than that of the V-notch

bar, which would differentiate between

the steels at — 30 F as the weldment did

That test bar could be used as a model

or a guide and, in additional testing,

make it possible to select the cheapest

steel that was "just good enough." In

this case, steel costs dictated that brittle

failure must be prevented by not letting

a fracture start and run, though that

placed the burden on design, fabrication,

and inspection

With reference to

transition-tempera-ture testing for selecting the steel, it

maybe noted, first, that there was noway

of knowing whether to use steel with a

transition temperature of 0 F, — 30 F, or

— 60 F, etc., and, second, that the costs

of these steels vary so much that a wrong

choice could be fatal

The element of cost is so important in

steel selection that a special study is

frequently called for In one case, it was

found that a better steel, although it

cost more per ton, was also stronger than

the steel it was to replace and, therefore,

would give a weldment of the same

strength that would be lighter, less

ex-pensive, and less vulnerable to brittle

fracture than the cheaper and inferior

steel

The case of steel for merchant ships

may be taken as another example of the

use of a notched-bar test of steel for

acceptability in a specific application

It has been established that the plates in

which fracture started has a Charpy

V-notch value of about 10 f t-lb at the air

temperature at which failure occurred

The corresponding value for the plates in

which fracture stopped was about 20

ft-lb Consequently, one can say that

acceptable steel should have values of

either 10 ft-lb or 20 ft-lb at the low

ex-pected service temperature, depending

on which approach is used Steel for ceptance would then have to be tested

ac-at only the one temperac-ature The fication of this procedure is the findingthat the V-notch bar is a reasonablysatisfactory replica of the ship structure,which means that both break in essen-tially the same fashion at the same tem-perature The V-notch for ship steelcorresponds to the special notch fortesting steel for the weldment at —30 F.These are examples of the use of con-stant- or single-temperature testing forengineering applications The method isthat of determining and then using thenotched bar that approximates the com-ponent or structure to be tested, whetherthe test bar is a so-called standard bar

justi-or one developed specifically fjusti-or the plication However, it is to be recognizedthat in the latter case, the design may bebased on the principle of insuring againststarting a fracture and that, therefore, thesteel thus selected may not have theability to stop a fracture from running

ap-If one does not rely on avoiding starting

a fracture, then a steel should be used inwhich a crack does not run under seriousconditions The latter will usually bemore expensive

RATING STEELS FORNOTCH TOUGHNESS

In the great majority of cases, differentnotched bars rate a series of steels inabout the same order of merit Con-sequently, studies of the effects of com-position, heat treatment, and steel-making practice generally employ someone test bar for rating purposes It iswell that this is so, and it points up thefirst requirement of work on brittlefracture, namely, that an effective notch

of some kind be used for testing ever, in some exceptions two differenttest bars may give different ratings forthe same steel, and this raises the ques-tion of which to believe or use

How-An example of the latter situation is

5

Trang 13

SYMPOSIUM ON IMPACT TESTINGgiven by data of Christensen and Aug-

land, which were recently cited by

Houdremont.5 These are reproduced in

Table I

The V-notch rates steels Nos 3 and 6

as equivalents, since they have the same

transition temperature of 66 F, whereas

the same steels behave quite differently

when tested with the keyhole notch

This cannot be too surprising, even

though exceptional, since the test bars*

sample the behavior with two different

stress systems and it is fundamental

that the notch properties of the

unde-formed metal (cohesive strength and

shear strength or critical shear stress) are

different at the different transition

tem-peratures This amounts to testing two

different materials of sufficiently

differ-ent properties to give this difference in

behavior

SUMMARY

In those cases where design and

free-dom from dangerous notches are relied

upon to avoid starting a brittle fracture

and the steel is selected accordingly,

the brittle transition temperature will,

in nearly all cases, select a suitably

tough steel provided it has been

ade-quately calibrated against service

per-formance However, should there be a

significant difference between the

transi-tion temperature and the low service

temperature, the selection could not be

on a sound basis Furthermore, thisapproach can lead to the selection of toocostly a steel In case there is no directcorrelation with service (design, strainrate, and temperature), the procedureoutlined above is available

The case of selecting a steel which willstop a running crack is different WhileTABLE I.—TRANSITION TEMPERA-

TURES, DEG FAHR.

5 E Houdremont, Consultant, Essen,

Ger-many, Metal Progress, January, 1955, p 116.

one is still concerned with the presence

of notches and accidental or unavoidableinitiators of fracture, the problem issimpler because the notch (the runningfracture) has a specific and constantgeometry, as long as the thickness re-mains constant For this case, laboratorywork has developed the V-notch as aguide The indications are that a betterquality steel is needed here because therunning fracture is a worse hazard thandesign and fabrication notches, whenflaws and cracks are eliminated Eco-nomics may step in and play an importantrole in steel selection because steel andother costs may be significantly differentfor each case

6

Steel

No 3 No 6Steel

ence Be- tween

Differ-V-notch Keyhole Difference between notches

66 14 52

66 59 7

0 45

Trang 14

MR S J ROSENBERG.1—I should like

to ask Mr Hoyt a question—would he be

brave enough to venture to give some

definition of transition temperature that

may be satisfactory to engineers and

metallurgists? Every time there is a

dis-cussion of the impact properties of steel,

the phrase "transition temperature" is

bandied around rather freely, yet we have

such a conglomeration of meanings for

this phrase that very few of us know

what the other means when he uses the

term

There is an erroneous impression that

may be gained from Mr Hoyt's

phrase-ology If I understood him correctly, I

gathered that if a steel were to be used

at a certain low temperature, its use

would be suitable provided its transition

temperature were lower than the service

temperature Conversely, if its transition

temperature were higher than the service

temperature, its use would not be

suit-able This would imply that, whatever

the criterion used to determine transition

temperature, this value should be as

low or lower than the service

tempera-ture I do not believe Mr Hoyt

in-tended to convey this impression

Mr Hoyt's proposal, namely, that

steels should be tested at the minimum

temperature of service by some means so

that differentiation can be made between

the relative toughness of the steels at

that temperature, has real merit This

would involve changes in the geometry

of the test specimen, with concomitantchanges in severity in the stress system,until a brittle fracture is obtained at thetemperature in question If the severalsteels considered for use are so evaluated

at the specific temperature involved, itappears logical to assume that the steelneeding the greatest severity in the stresssystem to induce brittleness at thattemperature is the most suitable for use

MR A B WILDER2 (presented in

written form).—Mr Hoyt, for many

years, has given much thought and sideration to the toughness of steel; and

con-in so docon-ing, he has been devotcon-ing histime to a fundamental property, aproperty which is just as importanttoday as it was when he started on thiscourse of dealing with the toughness ofsteel

The steel producers in general canprovide a steel that is tough It will betough enough to suit most requirements,but such a steel may be expensive andrequire special treatment The problem

is not necessarily to provide a tough steelbut to provide a steel so that its use will

be justifiable on an economic basis Inother words, the problem is to developtougher steels at a lower cost There aretough steels available, but economicsvery often do not justify their commer-cial use Furthermore, some of thesesteels would not be available in the large

1 Metallurgist, U S Dept of Commerce,

National Bureau of Standards, Washington,

D C.

2 Chief Metallurgist, National Tube sion, U S Steel, William Perm Place, Pitts- burgh, Pa.

Trang 15

SYMPOSIUM ON IMPACT TESTINGtonnages that certain applications would

require For example, a steel utilizing

nickel to provide toughness may be

available in only limited quantities

be-cause of Governmental restrictions on

the use of nickel

With regard to the general subject of

the toughness of steels, I believe that

most steel producers recognize the

im-portance of this problem in their

steel-making processes There is much work

being conducted in various steelmaking

plants throughout the nation on the

subject This work has been in progress

for many years In other words, steel

mill metallurgists are concerned with the

toughness of steel and are doing many

things to try to improve that property;

and if we are going to consider a new

steelmaking process we must evaluate

the property Mr Hoyt discussed, that is,

the toughness of steel

Another important consideration in

the development of new steelmaking

processes is the factor of utilizing

exist-ing processes and special treatments

available in the various plants For

example, there are two new steelmaking

plants recently constructed, using the

oxygen-lance method—a plant in Canada

and in the United States In Europe there

are several plants using this method,

which is often called the L-D process In

these new processes the toughness

char-acteristics of steel have been given

con-sideration, and in this respect, I would

say the steel industry was traveling in

the right direction with respect to the

factors which are related to the toughness

of steel

In the matter of crack initiation and

crack propagation, there are two

funda-mental characteristics involved, and

when you deal with crack initiation the

requirement is less with regard to

tough-ness than with crack propagation In

other words, tougher steels are required

to stop the propagation of a crack after

it has started than to prevent one from

starting There has been, and will

con-tinue to be, considerable emphasis onthis matter of the crack initiation Effortswill continue in an attempt to develop asteel in which the crack will not initiate.That is less difficult for the steel makerthan to develop a steel which will stopthe crack, although both of these objec-tives can be achieved at a cost In addi-tion to the toughness characteristics ofsteel, it is frequently more important toimprove design and require carefulhandling during installation in order toachieve the best results Very often greatemphasis is placed on the toughness ofsteel when other factors are more impor-tant

MR T B REYNOLDS.3—I should like

to ask Mr Hoyt which type of Charpyspecimen is most used, the V-notch orthe keyhole, and is there a trend towardone or the other?

I should also like to ask if there areapplications in which one is superior tothe other

MR S L HOYT (author}.—Mr

Rosen-berg asked for a definition of transitiontemperature To me the correct transi-tion temperature for any particularnotched test bar is what I have chosen

to call the match point, because that ties

up definitely with the notch properties

of cohesive strength and critical shearstress It is the upper temperature of thebrittle temperature range and is theonly point on the whole transition tem-perature curve that has precise signifi-cance The match point of a structure inservice is presumably the temperature

at which brittle failure occurs, thoughthis can also happen at lower tempera-tures

Mr Rosenberg asked a question aboutthe statement that I made of using asteel whose transition temperature camebelow the service temperature That issafe provided the proper test bar is used

3 Inspector, Ingersoll-Rand Co., Phillipsburg,

N J.

8

Trang 16

DISCUSSION ON NOTCHED-BAR TESTSfor determining the transition tempera-

ture I believe the paper shows what is

meant by the "proper" test bar and also

that the transition temperature must be

the temperature of brittle fracture, that

is, the match point

He also mentioned that testing at the

service temperature means changing the

test specimen I think that is unavoidable

As I read various articles that are

pub-lished from time to time, it seems to me

that that concept is being recognized, at

least for guarding against design notches

To guard against cracks and running

fractures, a more nearly standard, though

more sharply notched, test bar should

be adequate

Mr Reynolds asked which notch is

most used I am not sure what the

correct answer is, but it is my impression

that it is the V-notch at present The

keyhole was the most popular one some

years ago but when we began getting

these results on ship plate with a fairly

good correlation between the V-notch

and the behavior of ships, I think that

the V-notch acquired a background, or a

pedigree, that has never been produced

for the keyhole notch

However, that has to do with fracture

propagation If one were dealing with

initiation of a fracture and if that had

been the point in the study, it might

have been the keyhole notch

He also asked which one is superior

I believe it is the one which best matches

the application, and it might be the

V-notch, it might be the standard keyhole,

or it might be some other

In the case that I cite in the paper,

neither one was satisfactory, or neither

one was superior If one were studying

deoxidation practice, for example, it is

not so critical which notch you use The

main thing there is to use a notch bar

Mr Wilder opens his discussion on a

cardinal point—how to produce, or

procure, steel which has adequate

tough-ness for a given application and at a costthat will justify its use on an economicbasis The problem of steel plate formerchant ships is now a classic exampleand it was to meet the situation that

Mr Wilder mentions that** I proposed,

in the Committe on Ship Steel, the study

of the effects of manufacturing variables

on the toughness of commercial platesteel The results of that project haveissued from time to time as publications

of Battelle Memorial Institute

Steel melting is one of the controllingvariables and, in this connection, it isnecessary to determine the toughness ofthe steel which is produced by a newprocess, if the product is to be ade-quately evaluated The oxygen lance or

L/D process is an example It appears, as

Mr Wilder states, that the steel industry

is traveling in the right direction, yetthere is evidence which indicates that astill greater improvement may be at-tained and with no prejudicial costburden

It was especially gratifying to have

Mr Wilder's comments on crack tion and crack propagation There hasbeen a large amount of investigationalwork done on these points, but little hasbeen published on the practical aspects.That is why, along with their impor-tance, the latter were dwelt on specifi-cally in the paper If one limits himself,

initia-in the selection of steel, to one whosetoughness is sufficient to stop a crackfrom running, he may well find thateconomic considerations preclude theuse of that steel As Mr Wilder com-ments, "it is frequently more important

to improve design and require carefulhandling during installation in order toachieve the best results." An example of

a case of this kind was discussed in thepaper The problem of applying tonnagesteels, where toughness is a factor, ismore acute and difficult than the problem

of applying specially processed steels

9

Trang 17

TRANSITION BEHAVIOR IN V-NOTCH CHARPY SLOW-BEND TESTS

BY CARL E HARTBOWER1

SYNOPSISV-notch Charpy slow-bend tests (see also Armstrong (I) 2 ) have been conducted over a range of temperatures, using three commercial structural steels representing the full range of commercial deoxidation practice, and a number of carbon-manganese steels containing 0.04 to 0.45 per cent carbon and 0.43 to 1.05 per cent manganese The criteria used in evaluating slow- bend performance were (1) lateral expansion, (2) fracture appearance, and (3) temperature In particular, attention was focused on the deformation, fibrosity, and temperature attending catastrophic crack propagation (sudden and complete fracture of the test specimen at maximum load).

It was found that considerable plastic deformation occurs in the V-notch Charpy slow-bend bar even at temperatures resulting in 90 per cent crystallinity and catastrophic crack-propagation By defining transition temperature

as the highest temperature at which an initiating crack can propagate under action of the elastic-stress energy alone, that is, sudden and complete fracture

of the test specimen at maximum load, and by noting the deformation ring in specimens which fractured at maximum load, it was possible to differentiate between the deformation attending the crack initiation and crack propagation stages of fracture in slow bend.

occur-The objective of this investigation was

to evaluate certain performance criteria

and a particular definition of transition

temperature for use in connection with

V-notch Charpy slow-bend tests The

performance criteria were evaluated on

the basis of their response to selected

met-allurgical variables

SELECTION or PERFORMANCE CRITERIA

For the past ten years or more, the

relative merits of various laboratory

tests for notch-toughness have been

1 Chief, Metals Fabrication Branch,

Water-town Arsenal Laboratories, WaterWater-town, Mass.

2 The boldface numbers in parentheses refer

to the list of references appended to this paper,

see p 22.

under investigation in the United Statesand Europe All of the tests are more orless consistent in rating a given series

of steels in the same order However,the choice of performance criteria andthe attendant definitions of transitiontemperature have been an ever-presentsource of confusion Attempts to corre-late data from different test specimens,

or from a single type of specimen butbased upon different criteria, or upondifferent definitions of transition tem-perature with a single criterion, have led

to anomalies and probably erroneousconclusions Stout and McGeady (2)were among the first to bring the transi-tion-temperature dilemma into focus.They contended that the choice of test10

STP176-EB/Jan 1956

Petrofac (Petrofac) pursuant to License Agreement No further reproductions authorized.

Trang 18

HARTBOWER ON TRANSITION BEHAVIOR 11

specimen is far less critical for the

cor-relation of test results than is the choice

of criterion used in establishing the

transition temperature

The selection of the criteria used in

this investigation for evaluating

slow-bend performance was based upon the

following observations regarding

crack-ing in the V-notch Charpy specimen and

regarding the difference between ductile

and brittle fracture:

1 Crack initiation occurs as a ductile

gated by continuous release of the stress field surrounding the crack,whereas a ductile-fracture crack requiresplastic-deformation work to extend thecrack (4)

elastic-In conventional notch-bar impact

testing, results are usually expressed in

terms of (1) the energy absorbed in ducing fracture, or (2) the amount offibrosity (or crystallinity) in the fracture

pro-surface In slow-bending, on the other

hand, certain additional information can

crack in the root of the notch at the

midpoint of the specimen well before

maximum load is reached; the crack

then grows laterally as the load is

in-creased until it reaches the sides of the

specimen, at which tune a maximum in

the load-deflection curve occurs; and

finally, beyond maximum load, the

crack deepens while extending down the

sides of the specimen (Fig 1) (3).

2 Brittle fracture may be

differenti-ated from ductile fracture in the manner

in which the initiating crack is

gated; a brittle-fracture crack is

propa-be obtained from the load-deflectiondiagram, such as the maximum loadsustained, the deflection to maximumload or fracture, and from the area underthe load-deflection diagram, the workdone in producing fracture Raring'sfindings (3) suggest that the energy ab-sorbed after-maximum-load is worthy ofspecial consideration in that it provides

a measure of the work done in ing a crack through the test specimen.Thus, if there is zero energy after maxi-mum load, the crack is self-propagating;that is, the elastic-stress energy storedFIG 1.—Typical V-Notch Charpy Slow-Bend Load Deflection Curves.

Trang 19

12 SYMPOSIUM ON IMPACT TESTING

in the system is sufficient to drive the

crack through the specimen

The differentiation between ductile

and brittle fractures suggested by

Orowan appears particularly useful for

the purpose of denning transition

tern-crack is propagated Thus, when theload-deflection diagram indicates energyafter maximum load (plastic-deforma-tion work), the elastic-stress energy inthe bar was insufficient to propagatethe crack completely through the speci-

TABLE I.—COMMERCIAL STEELS SUBJECTED TO SUBCRITICAL HEAT

TREATMENT.

CHEMICAL ANALYSIS

TENSILE PROPERTIES (After Normalizing at 1650 F for 1^ hr)

LABORATORY CARBON-MANGANESE STEELS

perature Both types of fracture are

initiated by the occurrence of a crack,

and the presence of an elastic-stress

field in the surrounding regions is

neces-sary for the occurrence of the crack The

difference, however, between a ductile

and a brittle fracture, as stated above,

lies in the manner in which the initiating

men At lower temperatures, however,brittle fracture occurs at maximum load3

3 For sake of brevity throughout this report, when a brittle-fracture crack propagates sud- denly and completely through the specimen at maximum load (zero slope of the load-deflection

curve), the crack is referred to as ing, and the highest temperature permitting a self-propagating crack is referred to as the onset

self-propagat-of brittle fracture.

Material per centCarbon Manganese,per cent per centSilicon, Aluminum,per cent

Nitrogen, per cent Soluble

(1:1HC1) Insoluble

Fully-killed 1^ in plate

Semikilled 1 in plate .

Rimmed % in plate

0.20 0.23 0.19

0.58 0.54 0.35

0.21 0.12 0.01

0.04 0.01 0.01

0.006 0.005 0.007

ml nil 0.003

Heat Number Carbon, per cent Manganese, per cent ASTM Grain Size Pearlite, per cent

0.43 0.43 0.49 0.47 0.43 0.74 0.78 0.69 0.98 1.03 0.92 1.05

6 6-7 7-8 8 1-3 7 8 8 7 8 7-8 8

4.0 10.0 18.0 25.0 60.0 4.0 15.0 25.0 4.0 12.0 25.0 40.0

67.2 63.8 58.0

39.5 38.0 36.5

Trang 20

or on the ascending part of the

load-deflection curve (see Fig 1) In other

words, at the lower temperatures the

material is embrittled to such an extent

that the elastic-stress energy is sufficient

to propagate the crack On the basis of

these considerations, the deformation,

fibrosity and temperature attending

nese content (1) The latter,4 containingcarbon ranging from 0.04 to 0.45 percent and manganese from 0.43 to 1.05per cent, were prepared as laboratoryheats, deoxidized with silicon and alumi-num to simulate commercial fine-grainpractice, forged to f-in square bar stock,and normalized from 1700 F Metallurgi-

fracture at maximum load are deemed

less arbitrary as performance criteria

than, for example, a specified level of

fibrosity (50 per cent shear) or a specified

level of energy (15 ft-lb)

Materials:

The steels used in this investigation

were three commercial structural steels

representing the full range of commercial

deoxidation practice (rimmed,

semi-killed and fully semi-killed), and a series of

12 heats of controlled carbon and

manga-cal and physimanga-cal property data are listed

in Table I

Test Method:

The standard V-notch Charpy men was used exclusively Charpy im-pact data on the 12 steels of variedcarbon and manganese content were

speci-4 Supplied through the courtesy of Messrs.

O O Miller and T N Armstrong of the search and Development Division, International Nickel Co., Inc., New York, N Y.

Re-NOTE.—The intersection of the energy-after maximum load data band (•) with the abscissa indicates the amount of deformation occurring at the highest temperature producing a self-propagating crack (at lower temperatures fracture would have occurred on the ascending portion of the load-deflection diagram and the data would have appeared on the abscissa approaching 0.00-in deflection).

FIG 2.—Relation Between Lateral Deformation and Energy to Fracture.

Trang 21

obtained from the earlier work of T N

Armstrong and W L Warner (1)

For purposes of slow-bend testing, the

bars were supported horizontally by a

fixture on the platen of a hydraulic

testing machine and were loaded by a

ram fastened to the fixed head of the

machine All dimensions of the

support-fixed head and the moving platen of thetesting machine The energy absorbed to

a specific deflection or to fracture wascalculated from planimeter readings ofthe area under the load-deflection curve.The slow-bend testing, in contradistinc-tion to the case of impact where adiabaticdeformation may occur, was essentially

ing fixture and loading ram were made

to conform with those specified for the

anvil and striking hammer in the Charpy

impact test The supporting fixture was

contained in a tank so that the

speci-mens could be tested while totally

im-mersed in a liquid The slow-bend

speci-mens were loaded at a deflection rate of

0.01 in per min The load-deflection

dia-gram was automatically recorded by a

microformer-type deflectometer actuated

by the change hi distance between the

isothermal hi that the rate of loadingwas slow and the specimen was com-pletely immersed hi liquid

Trang 22

was tested in slow bend over a range of that a simple relationship exists betweentemperatures encompassing the transi- energy absorbed and lateral deformationtion from ductile to brittle behavior, throughout the range of energy in whichThe data as plotted in Fig 2 indicate the specimen was completely fractured

NOTE,—The groups containing 0.04 and 0.20 per cent carbon consist of three heats each with manganese ranging from 0.43 to 0.92 per cent.

FIG 4.—Effect of Carbon and Manganese on the Energy-Deformation Relationship.

Trang 23

and also reveal that a substantial amount

of deformation occurred in the test

specimen, even at temperatures

produc-ing a self-propagatproduc-ing crack (sudden

and complete fracture at maximum load)

The latter observation is based upon the

fact that the energy-after-maximum-load

band intersects the abscissa at mately 0.04 in (10 per cent) lateralexpansion Furthermore, it is deducedthat the deformation measured in frac-tured test specimens which suddenlyand completely separated at maximum

approxi-load was prefracture deformation; that

FIG 6.—Effect of Carbon and Manganese on the Fibrosity-Energy Relationship NOTE.—Data are for 12 laboratory heats tested in slow bend over a range of temperature encompassing the ductile-to-brittle transition.

FIG 5.—Effect of Carbon and Manganese on the Fibrosity-Energy Relationship NOTE.—The groups containing 0.04, 0.10 and 0.20 per cent carbon consist of three heats each with manganese ranging from 0.43 to 1.03 per cent.

Trang 24

is, it was the deformation attendant

upon the bending and crack-initiation

stage of the fracture process

Three commercial pearlitic steels of

fully-killed, semikilled and rimmed

de-oxidation practice, subjected to a variety

of subcritical heat treatments, were

tested in slow bending over a range of

temperatures encompassing the

transi-tion from ductile to brittle behavior

The data as plotted hi Fig 3 confirm the

observation that appreciable

deforma-tion occurred in the test specimen, even

at temperatures producing brittle

frac-ture, and show that the amount of

de-formation is affected by metallurgical

variables introduced by deoxidation

practice and subcritical heat treatment

Likewise, in the case of the laboratory

heats of carbon-manganese steel (Fig 4),

an appreciable amount of deformation

occurred in specimens tested at

tempera-tures producing brittle fracture Note

that as the carbon (pearlite) content

increased, the deformation

correspond-ing to the case of a self-propagatcorrespond-ing

crack decreased from approximately 15

per cent (with 0.04 per cent carbon) to

somewhat less than 5 per cent (with 0.45

per cent carbon).5

Amount of Fibrosity Attending Brittle

Fracture:

The laboratory heats containing

vary-ing amounts of carbon and manganese

were examined as to the relationship

between per cent fibrosity and energy

after maximum load (work required for

crack propagation) The steels as plotted

in Fig 5 are grouped according to carbon

content; the groups containing 0.04, 0.10

and 0.20 per cent carbon consist of three

heats each, with magnanese ranging

from 0.43 to 1.03 per cent The data

8 In the case of the 0.04 and 0.20 carbon plots,

each contains data from three steels with

man-ganese content ranging from 0.43 to 0.98 per

cent These data when plotted separately

(ac-cording to manganese content) showed little or

no separation.

indicate that the amount of fibrosityattending the onset of brittle fracture(highest temperature permitting a self-propagating crack) consistently fellwithin the range of 0 to 20 per cent forall carbon contents investigated Theaverage for all steels was 10 per centfibrosity (90 per cent crystallmity).Thus, the amount of fibrosity attendingthe onset of brittle fracture was lowand relatively unaffected by carboncontent

Relationship Between Fibrosity and Total Energy:

For a more complete understanding

of the two performance criteria, therelationship between per cent fibrosityand total energy was plotted for several

of the carbon-manganese heats (Fig 6).Again, the heats were grouped according

to carbon content (seven heats fell intothe group containing 0.04 to 0.15 percent carbon, and although manganese inthis group ranged from 0.43 to 1.03 percent, separate plotting revealed no dis-tinction between heats) The data asplotted in Fig 6 indicate that: (1) thework required to produce fracture de-creased with increasing crystallmity,(2) the rate at which the work decreasedwith increasing crystallmity was afunction of carbon (pearlite) content,and (3) the maximum amount of workattending 90 per cent crystallinityvaried from approximately 10 ft-lb for0.45 per cent carbon to 85 ft-lb for0.04 per cent carbon

Ejfect of Carbon and Manganese on the Shape of Transition Curve:

Total energy as measured by eter from the slow-bend load-deflectiondiagram was plotted against testingtemperature and, for purposes of com-parison, the data obtained from impacttests by Warner and Armstrong (l) werereplotted on the same coordinate system.The five heats containing carbon rangingHARTBOWER ON TRANSITION BEHAVIOR

Trang 25

FIG 7.—Effect of Carbon on the Energy-Temperature Relationship in Slow Bend and Impact NOTE.—The steels shown contain essentially constant manganese (0.43 to 0.49 per cent).

FIG 8.—Effect of Carbon on the Fibrosity-Temperature Relationship in Slow Bend and Impact NOTE.—The steels shown contain essentially constant manganese (0.43 to 0.49 per cent) Copyright by ASTM Int'l (all rights reserved); Tue Apr 12 02:07:06 EDT 2016

Trang 26

HARTBOWER ON TRANSITION BEHAVIOR 19between 0.04 and 0.45 per cent with

essentially constant manganese (0.43 to

0.49 per cent), are compared in Fig 7.6

From these conventional plots of total

energy versus testing temperature, it is

evident that both the difference in

deflection rate between slow bend and

impact and the metallurgical

varia-bles introduced by composition affected

the transition from ductile to brittle

behavior However, the curves do not

differentiate between the various stages

involved in the fracture process; that is,

the total energy (or work) in impact and

slow-bend provides only an integration of

the work required for bending, crack

initiation and crack propagation

More-over, as a result of the changes in shape

of the transition curves produced by

various metallurgical variables, it is

apparent that both the choice of

per-formance criterion and the definition of

transition temperature determine to a

large degree the conclusions which are

drawn from the conventional transition

curves Similar observations can be

made from Fig & which presents the

fibrosity-temperature relationships for

slow bend and impact

Definition of Transition Temperature for

Slow Bend:

The amounts of deformation and

fibrosity attending sudden and complete

TABLE II—TRANSITION TURES IN SLOW BEND AND IMPACT.

TEMPERA-6 Other combinations of the twelve

carbon-manganese heats were evaluated; both elements

were found to affect the transition curve Carbon

was found to both shift and change the shape of

the curve; whereas, manganese only shifted the

curve in impact, but both shifted and somewhat

changed the shape of the curve in slow-bend.

The effect of increasing carbon in changing the

shape of the transition curve consisted of a

lowering of the maximum energy and a widening

of the range of temperature over which the

transition from ductile to brittle behavior

oc-curred With increasing carbon the transition

occurred at increasingly higher temperature;

with manganese, on the other hand, the

transi-tion was generally lowered.

(Based on Onset of Brittle Fracture 0 )

Carbon, per cent

ture, deg Cent, Slow Bend

Tempera-(Based on 90 per cent Crystallinity)

Carbon, per cent

Temperature, deg Cent Slow Bend Impact

0.43 TO 0.49 PER CENT Mn 0.04

0.09 0.15 0.21 0.45

-90 -100 -60 -60 -20

0.04 0.09 0.15 0.21 0.45

-100 -110 -60 -60 -40

-40 -40 -60 -60 -20 0.69 TO 0.78 PER CENT Mn 0.04

0.11 0.20

-120 -120 -100

0.04 0.11 0.20

-120 -120 -100

-80 -80 -70

0 92 TO 1 05 PER CENT Mn 0.10

0.20 0.31

-140 0.10 -100 0.20 -80 0.30

-140 -100 -80

-80 -80 -80

nese, per cent

Manga- ture, deg Cent, Slow Bend

Tempera- ganese, per cent

Man-Temperature, deg Cent Slow Bend Impact 0.04 PER CENT CARBON

0.43 0.74 0.98

-9a0 0.43 -120 0.74 -140 0.98

-100 -120 -140

-40 -80 -80

0.09 TO 0.11 PER CENT C

0.43 0.78 1.03

-100 -120 -140

0.43 0.78 1.03

-110 -120 -140

-40 -80 -80

0.20 TO 0.21 PER CENT C

0.47 0.69 0.92

-60 -100 -100

0.47 0.69 0.92

-60 -100 -100

-60 -70 -80

0 Highest temperature producing a propagating crack (zero energy after maximum load).

self-0.874 0.9+8

Trang 27

fracture of the V-notch Charpy

slow-bend bar at maximum load have been

discussed It has been shown that the

amount of deformation attending the

self-propagating crack is dependent upon

certain matallurgical variables, whereas,

fibrosity is relatively independent of the

metallurgical variables investigated To

fulfill the objective of this investigation,

temperature is the final consideration as

a criterion of performance in slow bend

Transition temperature has been

de-fined herein as the highest temperature

resulting in sudden and complete fracture

at maximum load In other words,

transi-tion temperature is the highest

tempera-ture at which the initiating crack can

propagate under action of the

elastic-stress energy alone For sake of brevity

throughout this report, such a crack is

referred to as self-propagating, and the

highest temperature permitting a

self-propagating crack is referred to as the

onset of brittle fracture.

Table II shows the effect on the onset

of brittle fracture of carbon at three

levels of manganese and of manganese

at three levels of carbon With increasing

carbon content, self-propagating cracks

tended to occur at increasingly higher

temperature, whereas, with increasing

manganese content, the onset of brittle

fracture tended to occur at progressively

lower temperatures Thus, transition

temperature based upon the condition of

a self-propagating crack appears to be a

useful and sensitive performance

crite-rion for slow bend

In order to make a quantitative

com-parison between impact and slow rates

of loading, it is necessary to apply the

same definition of transition temperature

to both tests Unfortunately, the

condi-tion of a self-propagating crack cannot

be readily determined from conventional

V-notch Charpy impact tests without

special instrumentation for determining

the load-deflection curve The percentage

of fibrosity, on the other hand, can beestimated directly from the fracturesurfaces in both the slow bend and im-pact test specimens Figure 5, whichpresented the relationship between fi-brosity and energy after maximum load,suggests a definition of transition temper-ature based upon per cent fibrosity Theonset of brittle fracture (zero energyafter maximum load) is shown to corre-spond to 90 per cent crystallinity for allcarbon-manganese compositions investi-gated Thus, if transition temperature isdefined as that temperature correspond-ing to 90 per cent crystallihity, the

transition temperatures in slow bend

should be the same as those determined

on the basis of a self-propagating crack.The data in Table II confirm this ob-

servation In impact, however, the

temperatures corresponding to 90 percent crystallinity did not show a con-sistent trend as regards the effects ofcarbon Yet, increasing manganese didproduce a trend toward lower transi-tion temperature Until the relationshipbetween per cent crystallinity and onset

of brittle fracture is established for

im-pact, the grounds for selecting 90 per cent

crystallinity as a criterion for impacttesting are open to question

FUTURE WORKUsing a modification of the techniquedeveloped at the Naval Research Labora-tory by Harris, Rinebolt, and Raring(5), it may be possible to differentiatebetween the various stages of fracture inthe V-notch Charpy impact test, and atthe same time to determine the highesttemperature producing a self-propagat-ing crack The technique developed atthe Naval Research Laboratory demon-strated that for a given steel there is acritical energy level above which a crackoccurs in the V-notch Charpy specimenand below which no crack occurs Dif-ferent amounts of kinetic energy wereCopyright by ASTM Int'l (all rights reserved); Tue Apr 12 02:07:06 EDT 2016

Trang 28

HARTBOWER ON TRANSITION BEHAVIOR 21delivered to the specimen by allowing

the anvil of the impact testing machine

to fall from different heights Using this

technique, the NRL investigators

ob-served that brittle fracture by "low

blows" occurred at somewhat lower

temperature than indicated by the

con-ventional transition curve Later work

by Raring (2), indicated that the lower

transition temperature may be a function

of deflection velocity (a 240-ft-lb blow

using a 60-lb hammer is delivered at

approximately 16 ft per sec whereas a

15-ft-lb blow, using the same hammer

falling from a lesser height, is delivered

at approximately 4 ft per sec) Of

par-ticular significance to the

self-propagat-ing-crack concept is the fact that crack

initiation by a "low blow" appears to be

independent of testing temperature over

the range producing ductile fracture,

whereas at a lower temperature a

brittle-fracture crack is formed which

propa-gates readily to produce complete

frac-ture under the "low blow." Thus, at the

crack-initiating energy level it may be

possible to determine the temperature at

which the elastic-stress energy is-

suffi-cient to propagate the crack and thereby

differentiate between the energy

re-quired to produce bending, crack

initia-tion, and crack propagation

Based on the above observations, the

following modification of the Naval

Research Laboratory "low blow"

tech-nique is proposed: The first step will be

to determine the minimum energy level

which will initiate cracking (in the

temperature range producing ductile

fracture), and the second step will be to

determine the temperature at which the

initiated crack (using the minimum

crack-initiating blow) readily propagates

to complete fracture of the specimen

Correlation between this temperature

and the temperature obtained in slow

bending using the brittle fracture

(self-propagating crack) criterion is

antici-pated, at least in those steels wheredeflection velocity is not a controllingfactor

SUMMARY

1 Slow-bend tests provide a measure

of the temperature, fibrosity and formation attending brittle fracture(formation of a self-propagating crack)

de-2 An appreciable amount of tion occurs in the V-notch Charpyslow-bend bar at the onset of brittlefracture; the amount of deformation de-creases with decreasing temperature and

deforma-is directly proportional to the work quired to produce fracture

re-3 The amount of deformation tending the onset of brittle fracture pro-vides a measure of the relative embrittle-ment of a series of steels or of a givensteel under various metallurgical condi-tions

at-4 The amount of fibrosity attendingthe onset of brittle fracture is largelyindependent of metallurgical variables

In the steels investigated, the amount offibrosity occurring at the onset of brittlefracture fell within the range of 0 to 20per cent Consequently, slow-bend cri-teria based upon a self-propagating crackand 90 per cent crystallinity resulted hiapproximately the same values of transi-tion temperature

5 From the changes in shape duced in the transition-temperaturecurve by the metallurgical variables in-vestigated, it is apparent that both thechoice of performance criterion and theattendant definition of transition tem-perature determine to a large degree theconclusions which may be drawn withrespect to the effect of metallurgicalvariables on the transition behavior ofsteels

Trang 29

the formation of cleavage fracture is a

phenomenon common to all laboratory

test specimens, even at low temperature

and impact rates of loading In the case

of catastrophic service failures, on the

other hand, it has been commonly

ob-served that brittle fracture is surrounded

by non-deformed metal These facts are

not inconsistent if it is recognized that

local plastic deformation precedes brittle

fracture in both service structures and

laboratory test specimens

From previous work, it was established

that the amount of plastic deformation

occurring in the Charpy bar is directly

proportional to the amount of energy

expended in fracturing the test specimen

(6) Since the plastic deformation

at-tending brittle fracture preceded the

crack-propagatibn stage, it follows that

the energy absorbed and the

deforma-tion measured in specimens developingbrittle fracture reflect the resistance ofthe material to bending and to formation

of the initiating crack Thus, ance criteria based upon the formation

perform-of a self-propagating crack should beuseful in connection with the brittle-fracture problem by separating the crack-propagation and crack-initiation stages

of fracture

Because of the economy of impacttesting (as opposed to the time-consum-ing slow-bend operation) and because ofthe detrimental effects of higher deflec-tion velocity in some metallurgical conditions, it is recommended that a modifi-cation of conventional V-notch Charpyimpact testing technique be attempted inorder to permit differentiation betweenthe various stages of the fracture process.REFERENCES

(1) T N Armstrong and W L Warner, "Low

Temperature Transition of Normalized

Car-bon-Manganese Steels," this Symposium,

p 40.

(2) R D Stout and L J McGeady, "The

Mean-ing and Measurement of Transition

Tem-perature," The Welding Journal, Vol 27,

No 6, June, 1948, p 299-s.

(3) R Raring, "The Load-Deflection

Relation-ship in Slow-Bend Tests of Charpy V Notch

Specimens," Proceedings, Am Soc Testing

Mats., Vol 52, p 1034 (1952).

(4) E Orowan, "Fracture and Strength of

Solids," Reports on Progress in Physics,

Physical Society of London, Vol 12, pp 185-232 (1949).

(5) W Harris, J Rinebolt, and R Raring,

"Upper and Lower Transitions in Charpy

Tests," The Welding Journal, Vol 30, No.

9, p 417-s (1951).

(6) C E Hartbower, "The Poisson Effect in

the Charpy Test," Proceedings, Am Soc.

Testing Mats., Vol 54, p 929 (1954).

Trang 30

MR S L HoYT.1—This paper throws

a lot of light on this problem of brittle

behavior

There is one point I should like to

dis-cuss, and that is the effect of the low

blows

Steels retain their ductile behavior

over a wider temperature range with

static loading than they do with impact

loading I tried to account for that by

means of a diagram which shows that as

you increase the speed of testing you

raise this match point Increasing the

strain rate raises the yield strength or

the critical shear stress, and the match

point then comes at a higher

tempera-ture, at which the lowering of the yield

point is just equalled by the amount by

which it was raised due to the higher

strain rate

Work done at the Naval Research

Laboratory seems to indicate that the

low blow effect comes at the lower end

of the transition

With impact at a given temperature,

the yield strength is at a high level and the

bar is brittle But with a low-velocity

blow, the metal is strained at a lower

rate, the yield strength is left at its low

level, and therefore the bar is able to

deform

There is one other question Mr

Hartbower has shown slides of test

bars that have a small amount of

plas-tic deformation at the root of the notch

I once saw a series of test specimens

which showed that, as the

tempera-ture dropped, the amount of that

plas-tically deformed metal at the base of

1 Metallurgical Consultant, Columbus, Ohio.

the notch decreased continuously, but—and here is the point—at the low tem-perature which I regard as the matchpoint, the plastically strained metal dis-appeared

In other words, it appears that there

is a temperature at which the crack isnot initiated by that small amount ofplastic deformation That point can besettled experimentally, and it seems to

me that in developing the mechanism ofcrack formation and crack initiation, itwould be well to do so

MR D K FELBECK.2—Additional formation is available which should fur-ther confirm the statements about de-formation

in-Some work3 done at the MassachusettsInstitute of Technology a few years ago,with which Mr Hartbower may beacquainted, involved producing a crack

at liquid nitrogen temperature in shipplate specimens This was done by driv-ing a wedge into a notch The specimenshad been annealed before the crack wasproduced The X-ray back reflectionphotographs, which were taken of thefracture surface, produced at low tem-perature, demonstrated that there was avery small deformation at the surface.More recent work4 done in Switzerland

by Felix and Geiger appears to confirmthis

2 Committee on Ship Steel, National Academy

of Sciences, Washington, D C.

J D K Felbeck and E Orowan, "Experiments

on Brittle Fracture of Steel Plates," Welding Journal, Research Supplement, Vol 34, No 11,

November, 1955, pp 570S-575S.

4 W Felix and Th Geiger, "Brittle Fracture

of Steel," Schweiz Arch.,Vol 21, No 2, February,

1955, pp 33-49.

Trang 31

'24 SYMPOSIUM ON IMPACT TESTING

A visual examination of the liquid

njfK^en crack did not show any

duc-tility of this nature However, when

specimens containing such a sharp crack

were tested at very low head speeds at

room temperature (which was possible

because of the poor quality of the steel),

a shear fracture visible to the unaided

eye Always preceded cleavage

propaga-tion^ the crack

This is contradictory to reports that

have been given in the past of service

fractures which were said to exhibit no

ductile crack propagation

^ 'I think that when service fractures are

examined it should be done with great

care A casual examination is quite

dif-ferent from a careful examination by a

trained observer Fracture occurring

under service conditions—not at liquid

air temperatures—may always exhibit

this initial shear deformation prior to

high-speed fracture, because of the

es-sential contribution of the resultant

triaxiality of stress to the initiation of

cleavage fracture

MR CARL E HARTBOWER (author's

closure).—Mr Hoyt's discussion is

fo-cused hi large part on the description of

futurejvork that has been suggested as a

result of the study of V-notch Charpy

impact versus slow bend.

First, with regard to Mr Hoyt's

com-ments on the difference hi velocity

be-tween low-energy blows: If a full blow of

the impact pendulum is delivered, the

deflection velocity is 16 ft per sec,

whereas, with a 15 ft-lb low blow thevelocity is only 4 ft per sec Thus, witheach low blow there is a different ve-locity However, data obtained by means

of drop-weight tests and by means ofpendulum impact with different weights

of hammer have shown that differences

in velocity hi the range of 4 to 16 ft persec do not have a significant effect on theonset of brittle fracture Further details

on these findings will be contained hi aforthcoming report which is being sub-mitted for presentation at the 1956Annual Meeting of the ASTM.5

Mr Hoyt's second point was that at asufficiently low temperature the plasticdeformation preceding fracture would benegligible This is undoubtedly true, but

in laboratory-size specimens the perature at which ductility disappears istoo low to have engineering significance.For example, in the quench-and-tem-pered AISI 4340 steel which I ampresently investigating, plastic deforma-tion and fibrous cracking occurs at theapex of the Charpy V-notch at tempera-tures down to the lowest testing tem-perature investigated (—196 C) Even in

tem-a notch-sensitive condition (quench-tem-and-tempered to Rockwell hardness C 45),brittle fracture in this steel initiated

(quench-and-from a shear crack at temperatures well

below practical service temperatures.These observations tend to confirm theobservations of Mr Felbeck

6 Carl E Hartbower, "Crack Initiation and Propagation in the V-Notch Charpy Impact

Specimen," to be published in Proceedings, Am.

Soc Testing Mats., Vol 56 (1956).

Trang 32

EFFECTS OF MANGANESE AND ALUMINUM CONTENTS ONTRANSITION TEMPERATURE OF NORMALIZED

NICKEL STEEL

BY T N ARMSTRONG 1 AND O O MlLLER 1

It has been fairly well established that

fine ferrite grain size is conducive to low

temperature of transition from ductile to

brittle behavior in normalized carbon

steels of moderate to low carbon content

(I).2 The generally accepted practice for

producing fine grained steels is to add

aluminum to the melt after preliminary

deoxidation with silicon Although the

amount of the addition will depend upon

the condition of the melt, a residual or

acid-soluble aluminum content of 0.015

per cent usually will ensure fine austenite

grains at the temperatures associated

with normalizing and fine ferrite grains

after air cooling (2, 3) There has been

recurring evidence, however, that the

ef-fect of aluminum on transition

tempera-ture may be more extensive than that of

grain refinement alone (4); consequently,

the minimum aluminum content required

to obtain fine grain size might not

neces-sarily be the optimum quantity to

pro-duce the lowest transition temperature

The transition temperature of mild

carbon steel also may be lowered by

in-creasing the manganese content (5, 6),

and the addition of nickel long has been

recognized as one of the most effective

means of lowering the temperature of

embrittlement (7) What has not been

determined is whether the effects of

man-1 Development and Research Division, and

Research Laboratory, respectively, The

In-ternational Nickel Co., Inc., New York, N Y.

2 The boldface numbers in parentheses refer

to the list of references appended to this paper,

see p 37.

ganese and nickel are additive andwhether the optimum aluminum contentfor lowering the transition temperature

of carbon steels is independent of themanganese content and whether theoptimum aluminum addition is the samefor nickel steel as for carbon steel.The plan evolved to study the effect

of nickel, manganese, and aluminum ontransition temperature called for 24heats, half of which contained no nickel

and the other half 2\ per cent nickel,

with each type at three manganese levels

of 0.4, 0.95, and 1.5 per cent, and each

of these six combinations at the fourlevels of acid-soluble aluminum of nil to0.005, 0.030, 0.055 and 0.090 per cent.Carbon level of all melts was maintained

at 0.15 ± 0.01 per cent This block-typedesign improves validity on the effects

of nickel, manganese, and aluminumbecause it provides data on the effect ofeach one at all combinations of the othersand permits easy comparison when theresults are presented graphically Thisdesign has the additional advantage thatthe data can be analyzed statistically bythe powerful "analysis of variance"technique (8) to indicate the degree ofcertainty of the conclusions, provided, ofcourse, that no large uncontrolled varia-bles are involved Unfortunately, nor-malizing produced a significant quantity

of martensite in four heats, and theaimed-for deoxidation practice was notquite achieved in one heat These rela-tively large uncontrolled variables made

Trang 33

TABLE I.—CHEMICAL COMPOSITION, NORMALIZED GRAIN SIZE,

AND HARDNESS OF EXPERIMENTAL STEELS.

Steel

Chemical Composition, per cent

Carbon Manganese Silicon Nickel Aluminum 0

ASTM Grain Size

Austenitic Ferritic

Brinell Hardness

CARBON-MANGANESE SERIES 0.40 MANGANESE

0.41 0.42 0.40 0.42

0.18 0.24 0.22 0.19

0.07 0.03 0.03 0.09

0.008 0.024 0.062 0.093

3-5 6-8 6-8 6-8

5-8 7-9 7-8 7-9

108 117 112 114

0.94 0.93 0.87 0.89

0.22 0.20 0.15 0.17

0.05 0.06 0.09 0.06

0.004 0.035 0.052 0.088

2-6 7-8 8-9 7-8

5-8

8-10

8-9 8-9

128 127 122 122

1.46 1.53 1.55 1.53

0.25 0.25 0.22 0.24

0.05 0.08 0.14 0.06

0.004 0.030 0.052 0.090

3-7 8 6-8 7-8

6-9

8-10 8-10 8-10

143 144 139 141

2.25 NICKEL SERIES 0.40 MANGANESE

0.45 0.43 0.43 0.42

0.26 0.23 0.24 0.25

2.29 2.25 2.26 2.35

0.007 0.030 0.055 0.100

4-8 8-9 7-8 7-8

8-10 8-10 8-10 8-10

143 142 140 141

0.96 1.00 0.99 1.00

0.19 0.23 0.23 0.22

2.31 2.28 2.25 2.26

0.004 0.025 0.052 0.099

4-7 8 7-8 6-8

7-10 9-10 9-10 8-10

153 158 155 156

1.46 1.51 1.59 1.53

0.23 0.23 0.25 0.23

2.16 2.30 2.26 2.28

b

Aluminum reported is acid-soluble aluminum Sulfur and phosphorus not under 0.04 per cent.

determined-Martensite interferes to give questionable ratings.

Trang 34

ARMSTRONG AND MILLER ON NICKEL STEEL 27

it impossible to apply successfully the

analysis of variance; however, several

simpler statistical tests were applied to

the data, as shown later in the paper

STEEL-MAKING PROCEDURE

The steels were made in a 30-lb,

high-frequency induction furnace The ingot

iron charge was melted, the heat blocked

with a small ferrosilicon addition,

high-carbon ferromanganese was added, and

the remainder of the ferrosilicon was

charged The surface was then skimmed,

the power turned off, aluminum

intro-duced, and the heat poured into

pre-heated 4 by 4 by 7-in cast iron ingot

molds with double hot tops The nickel

steels were made in identically the same

manner as the carbon steels, except that

electrolytic nickel was added just after

the meltdown The bottom half of each

ingot was forged and rolled to 6| ft of

f-in square bar for the testing program

The top half was reduced to 1 j-in rounds

and held in reserve

A total of 43 heats was made before 26

satisfactory steels were provided since,

in some cases, difficulty was experienced

in obtaining the desired acid-soluble

aluminum content Chemical analysis for

carbon, manganese, silicon, and nickel

was made from the top or middle of each

ingot, but each f-in bar was checked for

manganese, nickel, and aluminum for

positive identification after rolling

Chemical analyses are listed in Table I

The aluminum analyses reported are

from the middle of the ingots and are

not significantly different from

incom-plete^ results from top and bottom Each

f-m bar was rated for size and number

of inclusions, and it was found that the

inclusion differences among the heats

were not marked

Heat Treatment:

The effect of temperature on theaustenite grain size was determined foreach steel in order to select suitableaustenitizing temperatures for heat treat-ment After careful study, a normalizingtemperature of 1625 F was selected forsteels Nos 1 to 12 (carbon-manganeseseries) and 1600 F for steels Nos 13 to

24 (2.25 per cent nickel series) Thesetemperatures were safely above the AStemperatures, yet low enough to avoidsignificant austenite grain coarsening.The austenite and ferrite grain sizes foreach steel were measured and are recorded

in Table I

The f-m bars were normalized byheating to 1600 or 1625 F, holding for

45 min, and cooling in still air A total

of 567 Charpy impact specimens wasprepared The specimens were ran-domized for drilling the keyhole notches

to prevent the possibility of drillingvariability being concentrated on anyone of the steels All specimens wereprepared with the notch transverse tothe direction of rolling

CHARPY TESTSThe Charpy specimens of each steelwere tested at a series of temperatureswithin the range 212 to —320 F Theenergy absorbed in breaking each speci-men and the test temperature are given

in the Appendix at the end of the paper.From these data, energy-temperaturetransition curves were plotted for eachsteel The fracture appearance also wasrecorded and, although not included here,the percentage of granular fracture in-creased as the absorbed energy decreased,

as would be expected For convenience,the curves of steels of the same man-ganese content are assembled in groups

in Fig 1 The temperature

Trang 35

FIG 1.—Effect of Aluminum Content on Transition Temperature of Normalized 0.15 per cent Carbon, Carbon-Manganese, and 2.25 per cent Nickel Steels.

Petrofac (Petrofac) pursuant to License Agreement No further reproductions authorized.

Trang 36

ARMSTRONG AND MILLER ON NICKEL STEEL 29ing to a value of 15 ft-lb was arbitrarily

selected as the transition temperature

because it is the value required in ASTM

specifications

DISCUSSION OF RESULTS

Effect of A luminum:

The grouping of curves for steels of

the same manganese content in Fig 1

permits ready comparison of the effect of

aluminum on the impact properties of

obtained with aluminum content of0.030 per cent but it is believed that this

is of no particular significance

Although there is no significant ference in transition temperature attrib-utable to differences of aluminum withinthe range 0.030 to 0.090 per cent, thesteels in this range have transition tem-peratures which on the average are 70 Fbelow that of steels with 0.005 per centaluminum

dif-FIG 2.—Effect of Manganese on Transition Temperature of Normalized 0.15 per cent Carbon Steels.

these steels In both the

carbon-manga-nese steels and the nickel steels, the

heats with less than 0.01 per cent

acid-soluble aluminum (steels Nos 1, 5, 9, 13,

17 and 21) have the highest transition

temperature in each manganese group,

because these steels are the silicon-killed

or coarse-grain type, as suggested by

their grain size in Table I The three

remaining steels of each manganese

group show no major differences

attrib-utable to the variation in acid-soluble

aluminum within the nominal range of

0.030 to 0.090 per cent In the nickel

steels with 0.95 and 1.5 per cent

manga-nese, lowest transition temperatures were

Effect of Manganese:

The effect of manganese is shown inFig 2 These curves are the same asshown in Fig 1 for heats containingnominally 0.03 per cent aluminum butthe curves representing the three differ-ent levels of manganese for both thecarbon-manganese steels and the nickelsteels are grouped to show more readilythe manganese effect

The three carbon-manganese steels,Nos 2, 6, and 10, show progressivelylower transition temperatures as themanganese is increased up to 1.50 percent For the nickel steel, there is a lower-ing of the transition temperature as the

Trang 37

manganese is increased from 0.4 to 0.95

per cent for two aluminum levels, but

the effect is not consistent with results

of the other melts, as shown in the

fol-owing tabulation:

The mean or average drop in tion temperature due to nickel is 104 F,the drop being greater at 0.4 than at0.95 per cent manganese It is quitepossible that some manganese content

0.95 Mn

-75 -130 -125 -140 -125 -255 -235 -170

1.5 Mn

-95 -165 -165 -190

Drop in Transition Temperature

0.4 to 0.95 Mn

80 40 55 50 -55 55 40 -20

0.95 to 1.5 Mn

20 35 40 50

0.4 to 1.5 Mn

100 75 95 100

On the basis of these values, the

aver-age drop in transition temperature of

carbon-manganese steel for each 0.10 per

cent increase in manganese within the

range 0.4 to 0.95 per cent is 10.2 F and

within the range 0.95 to 1.50 per cent

6.6 F

Effect of Nickel:

Since only one level of nickel was used

in these tests, comparison to show the

effect of nickel can be made only between

the steels with no nickel and those with

2.25 per cent nickel The following

tabulation permits comparison on the

basis of temperature required for 15 ft-lb:

Temper-2.25 per cent Nickel -180 -200

-195 -190 -125 -255 -235 -170

Drop in Transition Tempera- ture Caused by 2.25 per cent Nickel, deg Fahr

185 110 125 100 50

120 110 30

between 0.4 and 0.95 per cent would bethe optimum for producing the lowesttransition temperature in 0.15 per centcarbon - 2.25 per cent nickel steel Onthe basis of these results, each 1 per centnickel within the range 0 to 2.25 per centlowers the transition temperature 46 F

Microstructure:

The microstructure of steels Nos 1through 20 comprised equiaxed ferriteand pearlite with some Widmanstattenstructure in the steels of larger austenitegrain size There was some banding, but

it was not particularly pronounced Themicrostructures of steels Nos 5, 7, 17,and 19 in Fig 3 are representative ofsteels Nos 1 through 20 Steels Nos 17through 20, containing 0.95 per centmanganese and 2.25 per cent nickel, havesufficient hardenability to produce asmall quantity of martensite which doesnot appear to be particularly detrimental,although the scatter in results indicatesthat this combination may be borderline.Steels Nos 21 through 24, containing1.5 per cent manganese and 2.25 per centnickel, have hardenability sufficientlyhigh to produce 10 to 15 per cent marten-

Trang 38

FIG 3.—Microstructure of Normalized 0.15 per cent Carbon Steels Steels Nos 5 7 17 and fernte plus fernte, picral-metal etch Steels Nos 21 and 22: ferrite (white) plus pearlite (black)' plus bainite (white with black dots), plus martensite (gray), picral etch Austenite may be retained

19-in the martensite.

Trang 39

site with some bainite This

microstruc-ture raises the transition temperamicrostruc-ture

significantly

There appears to be no correlation

between hardness and transition

tem-perature in these data The presence of

appreciable martensite in the four steels

Nos 20 through 24 is reflected in higher

hardness; however, the high transition

previous discussion that this is an desirable combination for low-tempera-ture toughness due to the presence ofappreciable martensite after normaliz-ing

un-Application of the well-known tistical test for the significance of asingle mean (8) to the effect of adding

sta-FIG 4.—Effect of Manganese, Nickel, and Aluminum on the Ductility Transition (15 ft-lb) of Normalized 0.15 per cent Carbon Steels.

temperatures are believed to be related

to martensite rather than to hardness

per se.

Summary of Effects of Manganese,

Nickel, and A luminum:

The effect of the three variables,

manganese, nickel, and aluminum, on

the transition temperature (15 ft-lb) of

0.15 per cent carbon steels is presented

graphically in the bar chart, Fig 4, and

by the curves in Fig 5 The data for 1.5

per cent manganese - 2.25 per cent nickel

steels are not included in Fig 5, since

it is obvious from Fig 4 and from the

FIG 5.—Effect of Manganese, Nickel, and Aluminum on Transition Temperature of Nor- malized 0.15 per cent Carbon Steels with Micro- structure of Ferrite plus Pearlite.

Trang 40

ARMSTRONG AND MILLER ON NICKEL STEEL 33

aluminum to the 0.4 and 0.95 per cent

manganese steels shows that, if nickel

had no real effect in lowering transition

temperature, the results obtained would

occur by chance alone less than one tune

in 1000 Such odds amount to certainty;

in fact, the statistician most frequently

considers an effect to be real if it would

occur by chance alone one tune in 20

Likewise, the effect of manganese in

lowering transition temperature could

occur by chance alone only one time in

100 The same type of test shows that

the steels containing 0.030, 0.055, and

tempered at 1100 F for 1 hr Impact testswere then made on these steels and thedata used to plot the curves shown inFig 6 There is no significant difference

in curves representing the tempered anduntempered condition for the 0.4 percent manganese - 2.25 per cent nickelsteel, steel No 14 Both of the steelscontaining 1.5 per cent manganese,carbon-manganese steel No 10 andnickel steel No 22, show the effect oftempering by lower transition tempera-tures The effect on the carbon-manga-nese steel is moderate, the transition

FIG 6.—Effect of Tempering on Impact Properties of Three Normalized Steels.

0.090 per cent aluminum are not

sig-nificantly different in transition

tem-perature; however, there is a real

differ-ence between the coarse-grained (0.005

per cent aluminum) and the fine-grained

types (0.030 to 0.090 per cent aluminum)

which could happen by chance alone, if

there were no real effect of aluminum in

lowering transition temperature, less

than one time in 50

Effect of Tempering:

The relatively high transition

tempera-tures of the nickel steels with high

man-ganese content have been attributed to

the presence of appreciable percentages

of bainite and untempered martensite

To study the effect of tempering, steels

Nos 10, 14 and 22 were normalized and

temperature being lowered from —165

to —200 F The effect is more nounced in the nickel steel in whichtransition temperature was lowered from-180 to -270 F by tempering

pro-There is no significant difference tween the microstructures of the nor-malized and the normalized and tem-pered conditions of steels Nos 10 and 14,

be-as will be noted in the upper four micrographs of Fig 7, all of which show

photo-a structure of pephoto-arlite photo-and ferrite.The situation is quite different forsteel No 22 for which the transitiontemperature dropped 90 F on tempering.The reason for this large change is quiteevident on examining the two photo-micrographs at the bottom of Fig 7.Copyright by ASTM Int'l (all rights reserved); Tue Apr 12 02:07:06 EDT 2016

Tiêu đề	Symposium on Impact Testing
Tác giả	T. N. Armstrong, O. 0. Miller, T. N. Armstrong, W. L. Warner, R. S. Zeno, Herbert F. Schiefer, Jack C. Smith, Frank McCrackin, W. K. Stone, R. W. Hager
Người hướng dẫn	Mr. F. G. Tatnall, Symposium Chairman, H. L. Fry, Secretary, Mr. W. W. Werring, Presided at the First Session, W. H. Mayo, Secretary
Trường học	American Society for Testing Materials
Thể loại	special technical publication
Năm xuất bản	1955
Thành phố	Atlantic City

Định dạng
Số trang	182
Dung lượng	6,74 MB