Fifth
PROGRESS REPORT (Project SR-i 10)
on
REPRODUCIBILITY OF KEYHOLE CHARPY AND TEAR TEST DATA ON
LABORATORY HEATS OF SEMIKILLED STEEL
by
R. H. FRAZIER, F. W. BOULGER and J. W. SPRETNAK Battelle Memorial Institute
Transmitted through
NATIONAL RESEARCH COUNCIL'S COMMITTEE ON SHIP STEEL
Advisory to
SHIP STRUCTURE COMMITTEE
Division of Engineering and Industrial Research National Academy of Sciences - National Research council
Washington, D. c.
LABORATORIUM VOOR
SCHEEPSCONSTRUCTIES
FEBRUARY 7, 1955 SERIAL NO, SSC-830107
SHIP STRUCTURE COMMITTEE
February
7, 1955
Dear Sir:
As part of its research program related to the im-provement of hull structures of ships, the Ship Structure Committee is sponsoring an investigation of the influence of chemical composition and deoxidation on the notched bar properties of ship plate steels at Battelle Memorial
In-stitute. A paper covering portions of this work was presented at the Siposium on Effect of Temperature on the Brittle Be-havior of Metals with Particular Reference to Low Temperatures
held. at the annual meeting of the American Society for Testing Materials, Atlantic City, June 28--30,
1953.
This paper,entitled "Reproducibility of Keyhole Charpy and Tear-Test Data
on Laboratory Heats of Semikilled Steel" by R. H. Frazier, J. . pretnak and F. W. Boulger, constitutes Part I of the
attached report, SSC-83, the progress report on the project. Part II of the report describes the results of later calcula-tions.
The project is being conducted with the advisory assistance of the Committee on Ship Steel of the National
Acedemy of Sciences-National Research Council.
Comments concerning this report ar solicited and
should be addressed to the Secretary, Ship Structure Committee. This report is being distributed to those individuals
and agencies associated with and interested in the work of
the Ship Structure Committee.
Yours sincerely,
Rear Aarnirai, U. S. Coast Guard Chairman, Ship Structure
Comm i t t e e
MEMBER AGENCIES: ADDRESS CORRESPONDENCE TO:
BUREAU OP SHIPS. DEPT. OP NAVY SECRETARY
MILITARY SEA TRANSPORTATION SERVICE. DEPT. OP NAVY SHIP STRUCTURE COMMITTEE UNITED STATES COAST GUARD. TREASURY DEPT. U. S. COAST GUARD HEADQUARTERS MARITIME ADMINISTRATION. DEPT. or COMMERCE WASHINGTON 25. D. C.
FIFTh
Progress Report
(Project SR-ib)
on
REPRODUCiBILITY OF KEYI-IOLE CHARPY AND TEAR TEST DATA ON LABORATORI HEATS CF SEHIKILLED STEEL
by
R0 H0
Frazier, J W0 Spretnak and F0 W0 BoulgerBattelie Memorial Institute
und e r
Department of the Navy
Bureau of Ships NObs3239
BuShips Project No0
NS-Oli-078
for
TABLE OF CONTENTS
Pa g e
PART I
Reprint, Symposium on Effect of Temperature on the
Brittle Behavior of Ieta1s with Particular Ref-erence to Low Temperatures, American Society for Testing }'Iaterials, June
28-30, 1953.
PART II
Suiriniary i
1l:aterials 2
Testing Procedures 2
Expected Uncertainty in Tear Test Transition
Temperatures Q G O O O O Q 8
Correlation Analysis of Tear Test Data
Correlation of Charpy and Tear Test Transition
Temperatures 17
LABOATORUM VOOR
SCHEEPSCONSTRUCTES
PART II
REPRODUCIBILITY OF KEYHOLE CHARPY A
DATA Ql LABORATORY HEATS 0F SIKILLED
SUMÌ4ARY
Part I of this report
concluded
that the method of definIng the tear test transition temperature should be reconsidered0
Later calculations, based on 130 observations on Type A steel
indicate the advantages of defining the transition temperature
on a probability basis0 If twenty specimens are tested at temperatures covering the transition zone9 the transition tem.
perature, based on equal probabilities of ductile and brittle
fractures, can be determined with a standard deviation of 606°F0
The comparable limit of uncertainty for
transition
temperaturescorresponding to the highest temperature where at least one
brittle specimen is found when testing groups of four, is 11F0
A large number of tear test data were used to establish
the relationships between transition temperatures defined In
different ways0 The study indicates that transition tempera
tures, based on 5O per cent brittle fracture texture, are
equivalent to those based on p = 05 for brittle specimens
Both give lower transition temperature values than the Kahn-Imbeinbo criterion0
Even when tear test transition temperatures are defined
on a probability basis, they do not correlate very precisely
2
transition temperature, based on data obtained in the other
type of test, are likely to be in serious error0
MATERIALS
Data obtained on
95
plates of steel were used to compare different criteria for defining the transition temperature inNavy tear tests0 The 3/+-in0 plates were rolled from 2-3- experi
mental openhearth ingots0 The steels represented two combina
tions of carbon and manganese which gave tensile strengths of
approximately 62,5OO lb0 per sq0 in The nominal compositions
were O2O% carbon, O76% manganese, and O22% carbon and
o1+5%manganese0 The steels varied in aluminum contents and had been
rolled at different finishing temperatures0 The effects of the
variations in composition and processing treatments are de
scribed in a separate
report*o
The compositions of the plates tested during this study
are listed in Table 70
The steel plants supplied 3/+-in0 plates
and i 3/+-in0 slabs0 The latter were rolled to 3/-iin0 plates
in the laboratory, using finishing temperatures of i65'O0 1850°, and 20500F0
TESTING PROCEDTJRES
Tests were made on keyhole Charpy bars taken parallel to
the rolling direction of the plates and notched through the
TABLE 70 COMPOSITIONS AND ROLLING TEMPERATURES OF EXPERIMENTAL OPEN-HEARTH STEELS
*The aluminum values for the last eight steels are for acid soluble amounts; the
others
are for total aluminum contents. Vi 0.26 o.1+5 0.008 0.032 0.090003
V2 0.27 0.1+50.009
0.032 0.07 0.007 V3 0.28 o1+6 0.009 0.032 0.09 0.011 V1f 0.29 0.1+5 0.009 0.032 0007 0.018 V5 0.21 0.67 0.0120.033
0.07
0.002 V6 0.19 0.67 0.011 0.032 0.07 0.007 V? 0.220.67
0.012 0.033 0.07 00011 0.19 0.66 0.011 0.033o.o8
0.015 Wi 0.23 0.52 0.013 0.037 0.09 0.001 W2 0,23 0.52 0.011 0.037 0.10 O0O1+ W3 0.23 0.52 0.012 0.01+1 0.09 0.005 W1+ 0023 0.52 0.0i30.039
0.10
0.025 W5 0.230.78
0.012 0.025 0.09 0.001 W6 0.22 0.80 0.013 0.026 0008 0.003 W7 0.20 0.80 0.012 0.025 0.08 0.007 w80.21
0078
00013
0.026o.o8
0.032 Z1 00190.67
0.0120.O8
0.01+0 0.002 Z2 0019 0.68 0.013 0.028 0.01+0 0.001 Z3o.i8
0068
0.012
0.027 0.01+0 0.013 0.190068
0.013
0.027 0.01+0 0.01+1+Z5
0.27 0.50 0.017 0.01+10.057
0.003
Z60.27
0.51
0.017 0.01+2 0.058 0.002 Z7 0.27 0.1+9 0.017 0.01+2 0.058 0.009 z8 0.27 0.50 0.018 0.01+2 0.058 0.030Heat Chemical Composition, per cent
plate thickness0 Four specimens from each plate were broken
at each temperature, and the Charpy transition temperature
was taken as the temperature at which the average curve crossed
the 12 ftC-lb level0 Tear test specimens were also taken parallel
to the rolling direction0 Four tear specimens were broken at
temperatures 100F apart throughout the transition zone0 This
practice permitted the tear test transition temperature to be
defined by three different criteria0 These definitions for transition temperature are
l The highest temperature at which one or more of four
specimens exhibits a fracture area with less than 50 per cent shear texture0 This is the definition used by Kahn and Imbembo(5D0
2 The temperature corresponding to 50 per cent shear
texture when average percentages of shear texture
in fractured surfaces are plotted against testing
temperature0 This definition is used by some steel companies
30 The temperature at which the probability of brittle specimens is 05 when brittle specimens are defined
as those having less than 50 per cent shear texture
on the fractured surface. The reasons for suggesting this criterion are discussed in Part I of this report0
The results of the notched bar tests are summarized in
- 5
TABLE 8
TRkNSITIOJ TPEATLJFS
IN N0TCH BAR TESTSWi W2 W3 1700 1750 1750 1700 l7 +2 15
80
90 100 8080
85 91+ 68 8]87
98
70W5
1700-8
110
103 108 W6 1725 J+ 110 101+ 105 W7 1800 -1+3 80 70 71 W8 172537
80 51+57
Wi 1650 -31f 50 1+5 1+0 W2 1650 -17 1+o W3 1650-33
70 1+8 1+9 1650 31+ 70 5859
1650
0
3036
35
W6 1650 -1+3 1+0 1+6 W7 1650 -63 60 50 53 W81650
-62
30 20 23 Wi 1850 -10 8073
61+ W21850
-io
60
67
55
W31850
-38
80
58 63 w1+i85o
-31
70
71
6 W61850
-17
60
66
65
W71850
J+9
60
59 57 w81850
J+o
39
LfO Wi2050
i8
6e
35
1+0 W22050
-11+ 7068
65
W32050
-25
6057
55
2050
.22
70
63 582050
-13
100
62
75
W62050
-5
70
7366
W72050
-37
70
59 61+ W82050
JO
70
50
53 12 ft-lb KeyholeHeat Rolling Charpy Transi-.
6=
TABLE 8
(C0NTINIJ)
Test
Transition Tern
°F
12 f t-1b Keyhole
Navy Tear
Heat
Rolling
Charpy Transi=
50% 50No0
Teinp,°F
tion Temp,°F
Kahn Brittle Fracture Brittle Tests
Vi
1950
*2290
90
90
V21965
+28120
105
105
V31850
+11100
91
931990
+27120
106
110
V52000
70
655
V61980
060
5050
V72000
=21
70
66
67
v8
1990
=16
60
59Vi
1650
O-11F70
71
6 V21650
=190
80
80
V31650
i6
80
76
731650
=20
60
66
65
V51650
=23 5038
36
V61650
30
30
30
29
V71650
C-50 1+0 31+30
V81650
61+30
27
26
Vi
1850
+2100
91
90
V21850
.9
90
91
90
V31850
=5
80
86
85
V'+1850
=5
100
88
85
V51850
10
50 1+335
V61850
27
1+0 1+1+ 1+1 V71850
70
51 V81850
3Lf80
51+ 58Vi
2050
=3110
102
103
V22050
+890
95
95
V32050
+8100
97
95
Vii-2050
1110
101
100
V52050
2390
5'l
59 V62050
=18
60
61
55
V72050
=22
60
63 61+ V82050
=31
50 53 1+7Heat
No0Rolling
Temp,°F
12 ft-lb Keyhole
Charpy
Transi-tion Temp,°F
Navy Tear Tet Transition. TeDIL, °F
5O 50
Kahn
Brittle Fracture
Brittle Tests
21
1820
-18
60
66
65
'Z21810
-10
60
5157
Z31830
-16
70
57
55
1830
-37
50
39
38
25
1980
-1
90
89
90
26
1950
+2110
96
100
27
1855
+990
89
90
z8
1890
0
80
7375
Zl
1650
-37
50
622
1650
-31
60
5662
231650
-38
50
39
1650
-53
30
21
19
25
1650
-11+90
75
76
26
1650
-16
90
87
83 Z71650
-2
60
60
62
28
1650
-11f60
51 51 Z11850
-30
60
6361
£2
1850
-16
80
71
75
23
1850
-21
70
59
59
1850
=35 60 5250
1850
-10
90
86
85
26
1850
-7
90
87
90
Z71850
.12
80
70
71
Z81850
-io
80
67
6)+21
2050
-12
100
9392
22
2050
-
100
88
89
Z32050
-
90
82
80
21+2050
-17
90
80
83
25
2050
+8110
108
107
Z62050
+990
91
90
Z7z8
2050
2050
+21.18
loo
80
97
8381
99
it can be mentioned that twenty specimens of each steel were
usually tested0
Before discussing the relationships among the transition
temperatures listed in Table 8, the uncertainties involved in
experimentally determined transition temperatures will be
considered Q
EXPECTED UNCERTAINTY IN TER TEST TRANSITION TEMPERATURES The data on Type A heats in Table '-e- of Part I of this
report can be used for estimating the uncertainty attached to
tear test transition temperatures defined on different bases0
This can be done for the Kahn definition and for p =
probability of brittle fractures at the transition temperature0
Fig0
10 is a plot on probability paper of data obtainedfrom 130 specimens of Type A steel0 Four specimens were tested at 50°F; groups ranging from 8 to 1e.l specimens were tested at the other temperatures0 The trend line and 9 per cent con fidence limits were determined by probit analysis0 The tem
perature corresponding to a probability of brittle fracture
p = 005 is 73°F0 This is the transition temperature for that criterion0 The 95' per cent confidence limits for this transi tion temperature are 73
52 0F
or678
and7802°F,
based ontesting 130 specimens0
If the transition temperature were to be determined by
o 20 40 60 8O
I-
100 120 140 160 -9-99 90 70 50 30 IOProbability
ofBrittle
Behavior,
per cent
FIGURE IO.
PROBABILITY
OF BRITTLE
BEHAVIOR IN NAVY TEAR TESTS
ON TYPE A STEELS AT VARIOUS
TEMPERATURESO-22383
95 per cent
limit
confidence Trendby probit
line determinedanalysis
z,,
95per cent confidence
limit
lo
e farther apart0 They would be increased by a factor
or 255
This means that by using the probability of 20brittle fracture p 0 to define the transition
tempera-ture,
95
determinations out of 100 would lie within a 2605°F(2q55 x
10+°F) interval0 The expected distribution is illus=trated by the lower chart in
Fig0 110
Similar deductions can be made about the uncertainty of
the Kahn transition temperature by considering the probabilities
of encountering brittle or ductile specimens when testing groups
of four specimens at five temperatures0 It would not always be necessary to test
20
specimens because the testing temperaturesare chosen according to the sequence in which brittle specimens
are encountered0 The Kahn transition temperature is the high est temperature at which at least one specimen of four
iS
brittleand 10°F below the temperature at which four specimens are
ductile0 Hence, it can be defined by as few as five specimens0 Since it is usually based on fewer observations the Kahn transi tion temperature would be expected to have wider limits of un= certainty0
Table
9
lists the probabilities of encountering brittle specimens when testing groups of four samples at various tem= peratures0 They are based on the trend line inFig0
10 Thelast two columns give the probabilities of encountering four
40
30
20
Kahn - Im be mb o
criteri on
95 per cent confidence
50 60 70
-11-80 90
loo
HOTransition
Temperature, F
FIGURE II. COMPARISON OF DISPERSION 0F TEAR-TEST TRANSITION
TEMPERATURES BASED ON TESTING 20 SAMPLES
O-22384
Usingbrittle
pO.5
probability
offracture
for criterion
95confidence
per cent
-4)40
3°
20lo
o 50 60 70 80 90loo
l'o
TABLE 90 PROBABILITIES CALCULATED ON BASIS OF FIGURE lo FOR PREDICTING BEHAVIOR OF GROUPS OF FOUR TEAR
TEST SPECIMENS OF TYPE A STEEL TESTED AT VARIOUS
TEMPERATURE S Testing Temp, °F Probabilitie s First Specimen Brittle Second Specimen Brittle Third Specimen Brittle Fourth Specimen Brittle All Four Ductile At Least 1 of -Brittle
60
O76
ol82+
00+38
0OlO5
0O033
09967
70
055
075
0O9O9
0O95
o0o+o8O959
80
O35
02275
0l+78
O09+1
01785
o82I5'90
Ol9
001
53 9012+6
Ol01O
O,+3O3O%97
loo
oo8
o 0736oo677
oo623
O716+
02836
13
in, a group of four0 From these values, the Kahn transition
temperatures to be expected when testing Type A steel can
be deduced0 If testing is started at 600F, for example, the
probability of
setting the Kahn
transitiontemperature at 800F
depends on the probability of finding at least one brittle
specimen out of four at 80°F or lower temperatures and on the
probability of testing four ductile specimens at 90°F0 Accord
Ing to Table
9,
the probability of the Kahn transition tempera-ture being 80°F is:0997 x 0959 z o82l5 x 0+3
o338
Similar calculations for the other temperatures lead to theupper chart shown I
Fig0
11 This chart. shows that the 95per cent confidence limits for the Kahn
transition
tempera ture covers a range of i-+°F0 It Is a wider range than that for thelower
chart because It depends onInformation
providedby eight specimens or less
even when more are tested0The frequency charts in
Fig0
li show that when 20 speci mens are available the transition temperature can be determinedmore precisely
using
the probability criterion0 The chartsindicate that the Kahn criterion is less desirable for
research
purposes because it leads to wider uncertainty limits0
Perhaps It should be mentioned that the data in Table 9
suggest that the Kahn transition temperature may he Influenced
starting tests at low
temperatures leads to lower values of
the Kahn transition temperature0 This can be illustrated bycalculations from the data in Table 90 If tests are started at 600 or 70°F, the probabilities are
O338
and O321 that
the Kahn
transition
tempe'atures would be 3QQ and 900F, re= speetively0 When starting tests at 110°F, the probabilityis 022+ that 800F and 0362 that 90°F would be chosen by the
Kahn criterion0 Biasing the Kahn transition temperature
toward the direction from which
it is approached occurs because
of the way in which it is defined0CORBELATION ANALYSIS OF TEAR TEST DATA
The relationships among the transition
temperatures listed in Table 2 were examined by standard methods ofstatistical
analysis0 The results of the correlation analyses are .summarized in Table l0
It is apparent that transition temperatures, defineJ. on the basis of probability of brittle fracture p
05 and on
the basis of 50 per cent brittle texture9 are in
closest
agreement0
Fig0
12 illustrates this correlation0 The slopeof the trend line is l01+ arid the correlation coefficient
is O986
The scatter from the trend line is small, and the standard erroriS
only 30700F0
These data show that transi tion temperatures defined on the basis of 50per cent brtt1e
C o LI, C) 1L 120 loO
- 60
C 4)o
o
LI) 40 20-15-/
/J/
/y7;
Low carbon High carbon heats O heats (_l,
0.
Or,
/
B..
..
/
.,
/
o,
/
./
/
P/.
0/
/
/
/
/
/
.S/
,
/
0
.- co" 0
/
.,
..
/
/
e.,'
/
-/
/
,
o 20 40 60 80 lOO 120 Transition Temperature, F(50 Per Cent Brittle Tests)
Fl GURE 12. COMPA RISON OF TEAR-TEST TRANSITI ON - TEMPERATURE CRITER IA
TABLE
loo
CORRELATION BETWEEN TRANSITION TEMPERATURESESTABLISHED BY DIFFERENT CRITERIA IN TESTE
ON95 MATERIALSO FOUR SPECIMENS WERE TESTED AT
EACH TEMPERATURE OF INTERESTO
X2
X3
t t rbu t e s
=160=
Attribu
X1 12
f t1b keyhole Charpy transition temperature
Kahns definition, tear test transition temperature
50 per cent brittle fracture texture, tear test transitIon temperature
O5 probability of brIttle fracture, tear test
transition temperature
Regression EQuaiofl
Qf.ii&
X1) X2 Xl, X3 Xl, X+ X2, X3
X3X
X2 o08o10 xl +8736
X3O855 X1
+8i6i
XiO85 X1
+813
X3 = O8859
X2 +l95
X
O939
X2i85
= 1O])+ X3ii8
1l temperatures în degrees Fahrenheit
Standard Correlation Error of stimate, O 723
1531b F
O791
13007°F0769
1403°
F O 91980F
O 9+37032°F
o 986 3070° Fl7
p = 05 for brittle specimens0 In either case, data obtained
from all specimens tested contribute to establishing the transi-O tion temperature0 As explained in the previous section, this
establishes transition temperatures with smaller limits of
uncertainty
The correlation between the Kahn transition temperature
and the temperature at which the fracture surfaces average
5O per cent brittle texture is shown in Fig0
l3
There is considerably more scatter from the trend line than in Fig0 l2Table 10 shows that the correlation coefficient is lower and
the standard error is higher, 80+°F Much of the scatter
from the trend line in
Fig0
13 is attributed to the wide uncertainty limits for the Kahn transitIon temperature0 Four points all on the high side, fall outside the two sigma limits
on the chart0 This behavior suggests that the Kahn criterion occasionally sets the transition temperature too high0 It
appears that erroneous ratings by this criterion are more
likely to be on the conservative side0
CORRELATION OF OHARPY AND TEAR TEST TRANSITION TEMPERATURES
Both Charpy and tear tests are used to evaluate the
susceptibility of steels to brittle fractures0 Since both tests employ notches and are used for the saine purpose, it is
natural to seek factors useful for estimating transition tem
peratures for one type of test from data obtained by the
120 100 tL Q) 80 E D 40 20
-i8-/
/
/
/
/
/
/
/
Low carbon heats
OHigh carbon heats
/'
O
/
/
O.'
/
/
/
)(
/
/
/
..sc.»./
/
/
/
/
/
Q,"
/
cfo D...
/
S((
/
/
..
/
/ ..
Oo
2o-O/
/
/
/
/
/
/
/
/
/
/
/
o 20 40 60 80 lOO 120Transition Tern pera ture, F
(50 Per Cent Brittle
FractureFIGURE 13. COMPARISON 0F TEARTEST TRANSITIONTEMPERATURE CRITERIA 0
223 86
i 9
other procedure0 Therefore, the correlation betwen keyhole
Charpy and tear test transition temperatures determined for
these 95' steels was examined0
Fig0
l+ and 15' show considerable scatter occurs 'whentear test transition temperatures are plotted against Charpy
transition temperatures0 Table 10 shows that the standard
errors of estimate range around 1+°F and the correlation
coefficients around O76 The regression equations indicate
that the difference in Charpy and tear test transition tem
peratures is larger for steels with the lower transition
temperatures. Defining the tear test transition temperature
by Kahns criterion gives the poorest correlation with Charpy
transition temperatures. However, using the other two criteria
for defining tear test transition temperatures does not
sij-icantly improve the correlation with Charpy results.
The correlation analyses indicate that one-third of the
estimates of Charpy transition temperatures from tear test data are likely to be in error by 13°F or more0 Data for
these steels illustrate the danger of serious errors in
estimated transition temperatures based on data obtained in
other types of tests0 These data suggest that Charpy and tear tests do not measure the same type of' transitions.
o 40 20 o > o 20
u
a) o >' a) -40 -o 9-('J 60-20-/
/
Charpy versus Kahn transitions
Low carbon heats O
High carbon heats S
/.
/
/
/
.
5/
___/
/
S/
.
1..
.
.
/
/
/
I.
0.
o
0.
0
.
s
o
/
.8/
/
/
O'
'O
0/
/
o
/
o
/
28
/
/
0 20 40 60 80 lOO 20Tear TestTransition
Temperature, F
KahnCriterion)
FIGURE 14. POOR CORRELATION BETWEEN TRANSITION TEMPERATURES
ESTAB-LISHED
BY CHARPY AND TEAR TESTS USING THE
40 C o C o t-> Q. 20
o
0 or-40
M-20 cJ-
60 _oi -80o 20 40 60 80 lOO 120/
/
/
/
/
/
Chorpy versusbrittle
fracture
Low carbon High carbon 50 per heats heats cent0
/
/
2e
/
/
e
t
/
o/
/
¡s
/
.
c5
5/
/
/
/
0/0
,5
/
0,
O/
.
aDoec!
Oc.
o.
O?
e
0/
/
/0
/
Ip
/
.
oo
o
00,. 0
0,00
/
/
/
/
/
/
/
o
00
0/
0/
2'
/
/
0,
0
/
/
Tear Test
Transition Temperature, F(50 Per Cent Brittle Fracture)
FIGURE 15. POOR CORRELATION BETWEEN CHARPY TRANSITION TEMPERATURES
AND THAT SET BY TEAR TESTS USING THE 50 PER CENT
BRITTLE TEXTURE CRITERIONREFERENCES
ll3 At end of Part I of this report
l+ Frazier,
R0 IL, Boulger, F0 W0,and Long,
C0 H0 'Influence of Silicon and Aluminum on the Properties of HotRolled Stee1, Sixth Progress Re
port, Ship Structure Coimnittee Report, Serial