Hot Workability: Review

Review 158
Hot Workability
by C. M. Sellars and W. J. MeG. Tegart
Introduction
",BSO·C
70
Hot workability relates to the ability of a metal or alloy to be
130
deformed under conditions of high temperature (> O·6Tm, 5·0
where T m is the melting temperature in degrees Kelvin) and 100 t sec-'

relatively high strain rates (10-1 to 103 S-1). The two charac-
teristics that govern hot workability are strength and duc-
tility. In a previous review,1 the relationship between strength 80
and structure under hot-working conditions was examined F!NISHING

4·0
with particular reference to deformation to large strains
leading to essentially steady-state conditions. 38
Such conditions are achieved only in a limited number of
hot-working operations, e.g. extrusion or planetary rolling, ~
where continuous, roughly isothermal deformation condi- z
tions exist. The major quantities of metals and alloys are hot 4
oc 3·0
l-
20
worked under interrupted, non-isothermal deformation ll)
conditions, e.g. forging or rolling, where the incremental ",1050· C

~oc
UJ
strains are smaller and may not reach those required to I- 7. 2·6
EXTRUSION
achieve steady-state conditions. The distinction between !Z ",1150·C 20
these two cases is shown in Fig. 1, where extrusion is com-
UJ
....J
;$
t,..., 25 sec-'
pared with a typical hot-strip-mill schedule. As discussed 5 2·0
c
previously, there is a complicated interaction between dy- w
namic and static softening processes and this interaction 15
ROUGHING FOLLOWED BY
becomes particularly important under interrupted, non- COOLING IN
isothermal deformation conditions. It seems useful to sum- IV 15 SEC TO
..., 650· C FOR
n1arize the situation with respect to softening processes, as COILING
1·0
shown in Table I.
TABLE I
Possible Softening Processes Associated with Hot Working
100
TIME, SEC
Group Example Dynamic Static
Fig. 1 Typical deformation/time schedules for fxtrusion of steel
A AI, a-Fe, Recovery Recovery fol- and for rolling of 230 mm (9 in) steel slab to strip. (Mean equiv-
ferritic alloys (all strains) lowed by re- alent true strain rates. ~,S-l are given for each pass.)
crystallization
These differing processes lead to differing structures,
B Cu, Ni, y-Fe, Recovery Very limited
austenitic (small strains) recovery fol- strengths, and ductilities. The present review complements
alloys lowed by re- the previous one. in that the main attention will be directed
Recrystalli- crystallization to the non-steady-state conditions achieved with small strains,
zation (large which particularly affect Group B materials, and the duc-
strains)
tility aspects of hot working.
The most reliable method of determining hot workability
is to process materials under plant conditions where the
variables inherent in the material, e.g. composition, size,
C. M. Sellars, BMet, PhD, FIM is a Senior Lecturer in the Department of shape, and ingot structure, and the variables inherent in the
Metal~urgy, University of Sheffield. W. J. MeG. Tegart, MSc, PhD,
FIM, IS Research Manager, Melbourne Research Laboratories, Broken process, e.g. rate of strain, stress system, lubrication, and
Hill Pty. Co., Ltd., Melbourne, Australia. temperature, are simultaneously covered. However, such
INTERNATIONAL METALLURGICAL REVIEWS 1972, Vol. 17
1
2 Sellars and Tegart: Hot Workability
methods are expensive and the obvious advantages of 80
laboratory tests, e.g. ease of checking different casts, ease of III
Q.
determining optimum conditions for new materials, close

control of variables, and possibility of relating structure and
properties, have led to the development of a number of tests. lJ) lJ)
Some of these are simulative, e.g. rolling, extrusion, forging, lJ)
w 300 ~
~ 4.0 a::
and have the virtues that they incorporate factors such as lJ) In
friction which are difficult to isolate on an experimental W E.. m = 4.65 sec-1
::J 200 w
c:: ::J
scale, and are generally on a scale representative of plant f- a::
f-
conditions. Others are basic, e.g. tension, compression, 0·2 0·4. 0·6 0'8
TRUE STRAIN (c)
torsion, and bending, and have the virtue that variables can
be more readily controlled because of their smaller scale. BO
The advantages and disadvantages of these tests have been
reviewed by numerous workers2-12 and will be discussed
III
Q.
C
0·50 500 :t
0 0·30
0·20
~z
only in relation to strength and ductility measurements in 400 ~
this review. lJ)
0'10
lJ)
ffi 0·05 lJ)
W
0::
f- 300 c::
Strength lJ) 4.0 f-
lJ)
Measurement of Flow Stress W

::J
a::
L!.I
::J
200 a::
As indicated above, stress/strain behaviour under conditions f-
f-
appropriate to hot working can be measured in tension, 200 0·5 1·0 1'5 2'0
compression, or torsion. Each method has its advantages and DIAMETER / HEIGHT (Do/Ho)
limitations. Fig. 2 Effect of diameter: height ratio on compressive stress/strain
Tension behaviour of tool steel deformed at 1000 C and a mean strain 0
rate of456 S-l. (After Samanta.35)

Tests have been performed in tension over a range of
strain rates13-17 up to 103 S-l. When constant cross-head
speed is used, the true strain rate decreases slightly during
homogeneous deformation and then rises rapidly as necking ... (1)
occurs. This rise in strain rate during necking results in an
anomalous rise in flow stress and so restricts the usefulness
of the data obtained to strains less than that to the onset of where D and H are the instantaneous values of diameter and
necking, e.g. e I"V0'2-0'3 for steels.15 The advantage of simple height, respectively.21,34,35,40For plane-strain compression
calculation of stress from normal tensile tests is thus out-
weighed by the severe limitation in strain range. If special jj _
-
a/(1 + 2H
(LB) ... (2)
techniques are used to maintain the true strain rate constant,
flow-stress data can be obtained to higher strains (I"V 0'8).14
In this case, the results are in close agreement with those
obtained in torsion. where a/ is the flow stress under plane-strain conditions and
B is the breadth of the too1.31,41,42If specimens are unlubri-
Compression cated, or high coefficients of friction exist, more complex
Compression testing does not suffer the same strain limi- corrections must be used.35,40,41
tation. Tests have been performed to true strains of 0·7 An alternative method of correction, based on that of
or more using cam plastometers giving constant cross-head Cook and Larke,43 has been used by Samanta,35,36 Fig. 2
speed18,19 or constant true strain rates in the range E: I"V0'1- clearly illustrates the magnitude that frictional effects can have
300 S-l for axisymmetric compression of cylindrical speci- on jj in axisymmetric compression. It is, however, not always
mens,18,20-29 or for plane-strain compression of strip speci- clear from reported results whether or not corrections for
mens.30-32 Compression of cylindrical specimens has also frictional effects have been made.
been studied at slower strain rates (E: tv 5 x 10-3 - 0·7 S -1) Typical equivalent tensile stress (cr)/equivalent tensile strain
using standard testing machines or hydraulic presses,33-36 (e:) curves obtained in compression are shown in Figs. 3 and
and at fast strain rates (E: tv 70-2300 S-l) using instrumented 4. When plane-strain conditions have been employed, the
drop hammers16,35-39 or a modified Hopkinson pressure- stresses (cr/), strains (e') and strain rates (E:') have been con-
bar.35(a) In the latter case, Samanta35 has shown that verted to the equivalent tensile value as cr = y3cr' /2, e:= 2e' I
nearly constant true strain rate can be obtained over most of y3, and ~ = 2 E:'/Y3.
the strain range by suitably matching the mass of the tup
or projectile with the specimen geometry. Torsion
Specimens are generally lubricated with oil, graphite, or Torsion of solid cylindrical specimens is an increasingly
other solid lubricant19,20,30,31,35,36or various types of popular method of hot-workability testing, as high strains can
glass,19,20,21,24,34 depending on the temperature range being be attained at constant surface strain rate and quantitative
studied. Though this permits sliding between the specimens and measures of both flow stress and ductility can be obtained.
tool surfaces, frictional constraints still occur. The co- Also, programmed deformation to simulate rolling or for-
efficient of friction (fL) is generally considered to be low, 0·1, ging schedules can be applied relatively simply.44-47
but may vary from 0·02 to 0·4 depending on experimental Tests have been performed with specimens axially uncon-
conditions.31-35 When the coefficient of friction is low, the strained, in which case" significant changes in length occur
average pressure ([J) measured for axisymmetric compression with increasing strain.9,48-50 -More commonly, specin1ens
at any strain can be corrected to the equivalent tensile flow are axially constrained so that specimen geometry remains
stress (a) by the relationship constant, but axial stresses are developed.49,51-56In the latter
Sellars and Tegart: Hot Workability 3
EXTRUSION RATIO
1 i 'i i ~ 1p ~ EXTRUSION RATIO~ 1'0
o 20 1a 6'0 SO Js 210 • in ' 6'0 9'0 sis

ROLLING REDUCTION IN THICKNESS I % R LUNG R DUCTtON IN TH!CKN %
35
: 00 70
I 400 10 (19)
/(1055- C) _.- COMPRESSION. ~
__ COMPRESS10l\t23) COMM.-
; Eo sec-1
............ TORSION (41) PURITY
30
I ,
,/"100 =-=~:~:~~llg~\:~~
GP COMPRESSION (75)
- ..-
-----
TORSION(49)
PLANE STRAIN~ SUPER-
COMPRESSIOtJlO PURl TY 1
60
i ,.'5 -- TORSION(64) 200 -- TORSION (45)
.i! .....--.. 40 _.- TENSION (16)

I,, ,/
, I III
f ! 0-
~20 II N
..t
0
C">
o
1/ / ...-- .... 9 z
If ::E
/I :
!{oo ~
Il)
1/4 (fl
til
'I.IiIJ?~l(L-__
~\ ,.s 150 -
0::
(fl
(fl ...-
(fl
W
,j
II ,I
1i10 2·5
0:::
...-
(/) lJJ
;:)
Iii W
;:)
.-
0:::
I " ,.,
.-
0::: ...-
:z
~
...-
:z
UJ
.....•
.-
z
UJ
~
5
..J 0
~
5
15
100
~ w
o 5
0
UJ w
1·3)( 10-1
10
50 (b)
o
2·2)( 10- o 0·5 1·0 1·5 2·0
EQUIVALENT TRUE STRAIN (E:)
1" )( 10-3 (a)
Fig. 3 Typical stress/strain curves over a range of strain rates for:
0·5 1·0 1-5 2·0 2·5 (a) 18/8 stainless steel at 1100 C and (b) commercial and 0
EQUIVALENT TRUE STRAIN (E:) superpurity aluminium at 400 C. 0
~'''.
25 .I
. "- ·'.AI -5Mg
/
50 1100· C i -PLANE STRAlrJ
COMPRESSION
3O
)
t = 117 sec-1
- COMPRESSIO~21)i = 100 sec-1
.-. COMPRESSIO~'9t = 30-60 sec-1
150
._. TORSIO~14,64k = 1·1 sec-1

300 E
.•...•...
z
0 ::E
"'E
..•.•.•.
z
::E ~ AI - 2Y4 t-1g
:AI-Cu-Si-Mg
AI - 5·7 Zn
~
(/) (/)
§ (/)
100
(fl
W
w
(fl
(/)
.-
0:::
(/)
AI-4'2 Cu
0:::
•....
(fl
En 45
200 li:!
•.... lJJ
"..,._'-'-' AI-Si-Mg w
;:) / ;:)
En25 (/)
0::: 0:::
Med.C ,...- •....
Low C w 10
lJJ
;:)
0::
•....
._._. High C
_.-._._._._._._ 1B:4:1
;:)
0::
•.... .-
:z
lJJ .I
"",.-'-'- Al-1·4 Mn
~
I-
Z
...-
20 0--.__.__.__.__ •....
z
-J
~ ._._.-. C.P.AI
S.P. AI 50
~
z
W w 5 / :5
a
..J f/· ·- ._._._._.-.-._. -._. -._._1!i:.~_. _._ ..J 0 .I
~ ~ W LJJ
5 100 5
0
lJJ
i /_ . 0
w
/ '-.- -._._.- -'- _._._qJl?...f_._.
10
l~··-;·~~,=..-·
#
-"="';':;~:=: :.=:~-;.:....-!1:-S--:-:..:....-::.:.
O~C
?"...._._.- '-._._._._. -'-'-'-' ~fr. _._
(b)
0
(a) 0·5 1-0 1-5 2·0 2'5
EQUIVALENT TRUE STRAIN (E:)
Fig. 4 Effect of composition on stress/strain curves for: (a) steels

at 1100 C and (b) aluminium alloys at 400 C.
0 0
12 In the derivation of the equivalent tensile flow stress
80
TJTm = 0·7 ... (6)

10 70
£= 0·5 sec-1
1Il by means of the von Mises criterion, the longitudinal stress
Co 60
may generally be neglected without introducing significant
~ errors. Identical results can be obtained from fixed- and free-
50 "E end tests if dimensional changes in the latter case are small. 9
~ Z
L
The conversion of shear stresses and strains to their equiva-
U'l
U'l 4.0 lent values using the von Mises criterion and ideal-work con-
UJ
a:: :E
tJ V)
V)
cept, as above, has been generally accepted and gives reason-
w
Cu 30 a:: able agreement between results obtained in torsion and in
a:: l-
<{
UJ
V) tension or compression at high temperatures.1,14,48 Typical
:r: a::
V>
F~-Sj
20 <
w
equivalent-stress/equivalent-strain curves for torsion are
J:
V) shown to strains of 2-5 in Figs. 3 and 4, and much higher
Al
10 strains can be studied in ductile materials. However, most
Pb existing torsion machines have an upper strain rate limit of
o E ~ 10 S-1, though this could be extended by using specimens
+"50 +10 of smaller gauge-length : radius ratio than is customary.
Fe-C
Machines capable of much higher strain rates (~ ~ 103 S-1)
1Il
Co
Ni +7-5 E have only been employed to a limited extent for high-tem-
~ +1·0 ~~ perature work.48,61
+5
~ ~ Stress/Strain Behaviour
V> +·5 Fe-Si
(Jl
UJ +2·5 ~ The stress/strain curves in Figs. 3 and 4 indicate that the
0= Cu •...
0::
same general behaviour of a rise in stress to a maximum
ti V>
Pb value, followed by a decrease to a steady-state flow stress
~ with increasing strain, is found in a wide range of materials.
Al x
« The magnitude of the peak in stress depends on the operative
-2·5
A. 10 12 restoration process1 or on the occurrence of structural

EQUIVALENT TRUE STRAIN (Eo) instabilities,9,55,62,63 and the strain at which it occurs varies
Fig. 5 Dependence of shear stress and mean axial stress on equi- widely. In general, the strain to the peak stress increases with
valent true strain in fixed-end torsion tests at high temperatures. increasing strain rate (Fig. 3) and mayor may not exceed the
(After Hardwick and Tegart.49) strains studied in compression tests and encountered in
normal rolling passes (see scales at top of Fig. 3), depending
on the material.
case the constant geometry simplifies the conversion of
The results of different workers on similar materials show
torque (r) to surface shear stress (-r) as57
considerable discrepancies in the absolute values of flow
stress obtained at a given strain rate (Fig. 3). These dis-
-r =
1 [
21ta3 3 r + e dr. dr]
d a + e de · .. (3) crepancies are just as great when similar testing methods are
employed as when different ones have been used and indicate
that material factors such as composition and structure
where a is the specimen radius and e and are the angle of e before testing are of gr~at importance. Fig. 4 shows the mag-
twist per unit length and its time derivative, respectively. nitude of alloying effects on iron and aluminium. Significant
The surface shear strain (y) and shear strain rate (y) are re- effects can also result from differences in structure before
lated to e and as e deformation and these will be considered later.
Fig. 3 also clearly illustrates that there is a systematic
y = as · .. (4) increase in strength with increasing strain rate over many
y = ae · .. (5) orders of magnitude and that 'quasi-static' or 'dynamic'
strain rates, frequently referred to in the literature, occur at
If free-end testing is used, corrections may be made for the arbitrary positions within the spectrum. A distinction can,
change in specimen dimensions. 48 however, be made in that at low strain rates ( < ,..., 0'01-0'1 s-1,
The cause of the axial stresses is still uncertain, though they depending on material properties and testing procedure)
are probably related to anisotropy and texture development nearly isothermal conditions of deformation are attained,
in the specimens.58,59 According to Nadai60 the axial stress and a true steady-state stress level is found at high strains,
is a minimum at the specimen surface and rises to a maxi- whereas at higher strain rates the deformation conditions
mum at the axis; however, there is evidence from studies of become nearly adiabatic. In this case temperature increases
solid and tubular specimens of different geometry55 that the with strain and the flow stress decreases continuously after
variation in stress with radius is small, so an average axial the initial peak. The temperature rise calculated from the
stress «(J A) can meaningfully be obtained. The dependence total work done is in reasonable agreement with measure-
of l' and (JA on equivalent tensile strain (E = Y/V3) for a num- ments made on high-strain-rate tests18 and can be of
ber of materials tested in the hot-working range is shown in considerable magnitude, e.g. in a range of steels deformed
Fig. 5. The wide variation in form of the axial-stress curve is in compression at E ,..., 430 S-1 and an initial tempera-
borne out by other workers, as is the relatively low stress ture of 1055° C (1328 K), the temperature rise at a
level. Generally, the ratio crA/-r is <0'25, though it may be as strain of 0·5 was 30-40 deg36 and increased almost
high as 0-4 under certain conditions.49 linearly with strain. In torsion, incipient melting has been
Sellars and Tegart : Hot Workability 5
observed in ductile steels tested at an initial temperature of which reduces to equation (9) at low stresses (lXa < 0'8)
1100 C (1373 K).64 At intermediate strain rates, the actual
0
and to equation (10) at high stresses (IXo- > 1'2), with ~ = IXn',
temperature rise occurring with deformation will depend on was proposed70 for .correlation of data over a wide strain-
the method of testing or forming, as heat-flow conditions are rate range at high temperatures. This relationship has been
of importance.65-68 found to fit data on AI,35(a),70,78 CU,35(a),70,79 Ni,70 Ti,80
U,80 Zr,80 and various ferritic70,34,55 and austenitic alloys55,70
Flow Stress/Strain Rate/Temperature Relationships with values of IX and n' that are independent of temperature
(e.g. Fig. 6(c)). .
When stress/strain behaviour is examined over a range of To correlate data at different temperatures Zener and
strain rates and temperatures, it becomes cumbersome to
Hollomon 81 proposed that
report the results as complete series of stress/strain curves,
so this has been done relatively infrequentiy.l4,19,21,30,64
a* = feE exp Q/RT) = feZ) ... (12)
More commonly, a few typical curves are shown and charac-
teristic values of flow stress are reported as functions of
where T is absolute temperature, R is the gas constant, and
temperature and strain rate. The characteristic stresses usually
Q is an activation energy. An alternative concept of Mac-
taken from tension or compression tests are those at some
Gregor and Fisher82 is that of a 'velocity-nl0dified tempera-
fixed strain level (ae); whereas from torsion tests the maxi-
ture' (Tv) such that
mum stress (am) or, for tests at lower strain rates where
conditions are nearly isothermal, the 'steady-state stress (o-s)
is reported. ae and am are not exactly equivalent, as the strain ... (13)
to the maximum stress varies systematically with testing
conditions (Fig. 3), but analogous relationships with tem-
perature and strain rate have been found. where T is absolute temperature, and K and Eo are constants.
In the remainder of this section, a* will be used to repre- Both relationships have been applied successfully but the
sent ae, am, or as. former offers the advantage that the activation energy may
The most common relationships used to correlate data at be related to that for the rate-controlling restoration process. 1
constant temperature are the power law, or the semi-loga- Constant values of activation energy, independent of tem-
rithmic formula of Ludwik.69 perature, have been obtained for a number of pure metals
and simple alloys deformed at temperatures > 0·5 Tm.l,80
· .. (7) This would not be expected to apply to lower temperatures,
or to more complex alloys, in which precipitation processes
or · .. (8) could occur within the temperature range of interest. Also,
for high-strain-rate tests, when adiabatic conditions are
As pointed out by Sellars and Tegart,70 these equations are approached, there is some ambiguity as to whether the
exactly analogous to the relationships obtained from creep initial temperature or a corrected temperature based on the
data at low and high stresses, respectively work done should be employed in these correlations.
Combination of equation '(12) with equation (11) by re-
· .. (9) placing E by Z allows linear correlation of data obtained
over a wide range of strain rates and a range of high tem-
and ... (10) peratures1,70,78 (e.g. Fig. 7). If equations (7)-(10) are used
in combination with (12), equally satisfactory correlation of
where 0-0, N, B, A, n, and ~ are constants. data can be obtained,1,78,8o,83 ifit is recognized that the stress
If flow-stress data are obtained over a relatively narrow functions will be non-linear outside the restricted ranges
range of strain rate « 2 orders of magnitude) it can be mentioned earlier. The advantage of such correlations is that
difficult to distinguish whether equation (7) or (8) best fits a large number of results can be concisely reported and
results at different temperatures20 and various authors have subsequently used to obtain values of a* at any combination
favoured equation (7)2°,25-30,64,71-77 or equation (8)23,35
of strain rate and temperature within the experimental range
depending on the material and experimental conditions of Z.
employed. For more complex alloys, neither equation may
be entirely satisfactory. 21 Stress/Strain Relationships
If equation (7) is applied, N (1/n in equation (9)) is found
The relationships discussed above restrict the application
to depend on temperature, being low below about half the
of data to strain conditions equivalent to those under which
melting temperature and then rising to a value of 0·2 at
I"oJ
high temperature.20,25-30,64,73-77 This apparent temperature- a* was obtained. Because of this, there have been numerous
attempts to fit simple algebraic functions to the stress/strain
dependence is illustrated by the broken lines in Fig. 6. In the
curves obtained at constant temperature and strain rate.
correlations of ae:, N varies slightly with E, whereas ao in-
creases systematically. 20,25-30 Some of these equations, together with references to the
originator and/or to workers who have used the equation to
When equation (8) is applied, B (~ in equation (10)) also
exhibits a dependence on temperature and strain.20,23,35 correlate high-temperature stress/strain behaviour, are:
However, it was noted70 that, if equation (7) is applied only Ref.
to data below a critical stress level, a constant value of N (n), a = KEm · .. (14) 23, 25-29, 84
independent of temperature, is obtained (Fig. 6(a) data a = ao + BEm · .. (15) 23,69
below 55 MN/m2 (8000 Ibf/in2)) and, similarly, if equation
I"oJ a = A (B + E)m · .. (16) 23, 85, 86
(8) is applied to the higher-stress data only, a constant value a = A + BIn E · .. (17) 35
of B (~) is observed (Fig. 6(b )). An empirical relationship a = A - (A-B) exp (- CE) · .. (18) 23,87
a = ao +
B [1- exp (- Cs)]m · .. (19) 88
••. (11) a = A [1- exp (- Csmn)]1/n · .. (20) 89
TRUE STRESS I MN / m2 TRUE STRESS, MN / m2
50 100 150 200
/6
50 100 150
0·25 % C STEEL
0·25 % C STEEL 'i7. o
'V 1200· C
01100· C
6. 1000· C
o· 900· C
j~ I
UJ
~
0:
/I(
Jj ~ o
I} j
ill
II
o
o
I I
illI 0
/
/ J{ /
/ (a) (b)
5 10 20 30
25
10 15 20
TRUE STRESS, 103 psi TRUE STRESS, 103 psi
0·25 % C STEEL
U
QI
III
•.. 10-1
W
~0:
z o
<i t-
0:
t-
~o 'V SLOPE 4·6
(/) 0<:.= 1·1 x 10-1. p.s i.
a.
x (0·76 N/m2)
eu 1010
.u) Q = 73 k cal/mol..
(354 kJ/mol )
N
(c)
0·2 0·4 0·6 0·8 1 4 8 10 0·2 0·4 0·6 0·8 1 4 8 10

sinh a( 0" sinh oc cr
Fig. 6 Strength data for an 0-25 % C steel plotted to demonstrate: Fig. 7 Correlation of strength data of Fig. 6 using temperature-
(a) power relationship; (b) exponential relationship; (c) sinh compensated strain-rate parameter.
relationship between stress and strain rate. (After Sellars and
Tegart,70)
where A, B, C, K, (jo, m, and n are constants dependent on
strain rate and temperature.
It is clear from the number of different equations proposed
that none is entirely satisfactory for all materials and testing
conditions. Hockett23 has attempted to fit his data on alu-
•...
minium to equations (14), (15), (16), and (18) and concluded
§ 3 20 ~
that (14) provided the best fit. The more complicated equa-
(/)
tions (15) and (16), which allow for the flow stress to have a (/)
W
(/)
0::
real value at zero strain, in fact gave negative values of the tii 2 -CONST. E.
15 ~
t;
constants under certain conditions. Although equation (14)has w
---- CON ST. CROSSHEAD SPEED
the merit of simplicity, examination of Figs. 3 and 4, indi- ::>

0::
I-
10 ~
l-
cates that its applicability must be limited to strains less than
those to the maximum flow stress. In the case of aluminium
at normal working strain rates, this strain exceeds that usually
given in a single rolling or forging pass; however, this is not o
12
\ necessarily true for other materials.
To fit data to higher strains, the exponential forms of "i~ 10
equations (18)-(20) are more appropriate, as these lead to
a steady-state stress at high strains. Note that equations (19)
and (20) reduce to (15) and (14), respectively, at small strains. --------------
However, even these equations will be a good fit only for
40 0·2 0·4. 0·6 0·8
materials and deformation conditions that do not give a
TRUE STRAIN (t)
marked peak in flow stress.
While the limitation of algebraic equations describing Fig.8 Comparison o/stress/strain curves/or lead at 100 C deformed 0
in compression at constant true strain rate and constant cross-

stress/strain behaviour must be recognized, it is advantageous head speed. (After Loizou and Sims.I8)
to apply them to experimental data so that constants can be
reported from which flow stress or work done for any strain Such mean stresses have been reported for a range of steels. 97
can be subsequently computed. This is particularly true if This procedure neglects any effect that the strain-rate profile
these equations can be combined with those for strain rate in the working process may have on the flow stress.
and temperature-dependence discussed earlier. For example Little work appears to have been done on the effect of
K in equation (14) (or A in equation (20») can be replaced by variations in strain rate during deformation, though simula-
(j* in (7)-(11) and combined with (12) to give equations of tion of strain-rate profiles encountered, for example, in
the form: roIling95,98 would be feasible using a cam plastometer
Comparison of data on lead obtained at constant strain rate
(j = (BZ) lIn Em ... (21) and at a constant speed of compression18 (Fig. 8) indicates
that even a variation of a factor of 2 in strain rate has a
Em significant effect on the stress/strain curve. In this case the
(j = -In BZ •.. (22) mean strain rate in the test at constant cross-head speed is
~
equal at a strain of o· 74 to the constant strain rate in the other
Em
test. At this strain the flow stresses are nearly identical, but
(j = - sinh -1 (BZ)l/n ... (23) the mean flow stresses obtained by integration are 21·0 and
~ 22·4 MN{m2 (3050 and 3250 Ibf/in2), respectively-a difference
of 6%. Larger differences in mean flow stress would be
f'OoJ
Alternative constitutive equations have also been propo- anticipated if a strain-rate history as in a rolling pass or in
sed.35,39,90-92 In each case the material properties over a extrusion were compared with a constant-strain-rate test.
wide range of testing conditions can be expressed by a few More serious uncertainties in the application of data to
material constants which permit a reasonable approximation working processes arise from differences in microstructure
to the stress/strain behaviour to be determined for any com- of the material before the hot-working operation and before
bination of strain rate and temperature within the experi- testing. Most mechanical tests have been performed on pre-
mental range. The use of such equations is of value as, in the viously wrought material, either annealed at constant tem-
calculation of working forces in hot-working operations, perature, or heated direct to the test temperature. Neither
it is improbable that experimental data will be available for procedure simulates reheating before hot-working opera-
the exact deformation conditions of interest. However, as tions and so marked differences in grain size or distribution
discussed below, it is also necessary to consider the earlier of second phase may be obtained.
thermomechanical history of the material in the hot-working Increasing grain size leads to a significant reduction in
operation before applying derived flow-stress data. resistance to deformation, 37 ,45 (e.g. Fig. 9). Overheating
before deformation may, however, lead to other structural
changes as well as grain growth, and in a range of carbon and
A,plication of Flow-Stress Data low-alloy steels either an increase or a decrease in strength
Theories of metalworking operations generally assume ~hat may occur.99,100 Also, the properties of as-cast structures
under hot-working conditions a metal behaves as an ideal may differ significantly from those of the wrought material.
plastic solid. 86, 93, 96 In calculating working forces it is there- The few studies on strength of cast materials101,l02 indicate
fore usual first to establish a mean strain rate for the process. that they generally have a lower resistance to deformation,
A mean flow stress is then obtained for the required tempera- as expected from the coarse grain structure. The effect of
ture by integration, over the strain of interest, of the stress{ segregation and non-equilibrium phases in as-cast materials
strain curve determined at an equivalent constant strain rate. is uncertain.
~
60
400
/---i!i~~~;;;;
N
\1\
E
97 70 E.. sec-1 t"0-
350
~ 4-0~
~
50
/ ......-
§ §
/' ! i~' i6./ ,/". /'/.
300
(/)
(/) I·k· 17' .//
1Il
i!-l!.I/~<'// 30 ~
0- N
E
UJ
0::: 4 / ilk!, .. ./ UI
r/
t" 40 t- 0:::
~- ...•...•...•• (/)
ii!!i/~/9 t-
/ DELAY TIME (/)
,
250 ~
Ib I lJJ
(j) /f ;; :::>
0::: 2
SEC
1·B ~/. 20 ~
(/) t-
2·5 I -il 0:::
.I
200 UJ 3 . "/ t-
~ 30 " GRAIN SIZE 0::: t.
(/) 1~/I THERMAL TREATMENT IN 100 1Jo3 t-
(/)
t-
z
4 3·95
I
',I B·5 t-
5
w , 1 25 MIN BOO"C 5·6 ~ 6 405 i
z
lJJ
::> II 150 UJ ;:£ 10 -J
~ 20 , 2 20 MIN 900· 25 MIN BOO·

:::>
0::: 5
7 490 i ;:£
II
It- 3 20 MIN 1000· 25 MIN BOO· 100 - 240
100
t- 0
UJ
8
9
1314-
4100 i;
5
0
UJ
,14--4 20 MIN 1200· 25 MIN BOO· 1000-2000 j
10
a
o 0·2 0·4 0·6 DB
50
EQUIVALENT TRUE STRAIN (~)
Fig. 10 Effect of delay time on stress/strain behaviour of commercial-

0·1 0·2 0·3 0·4 0·5
purity aluminium deformed in plane-strain compression at an
TRUE STRAIN (E.)
equivalent strain rate of 3·8 S-l and at 410 C. (After Dunstan 0
and Evans.32)
~
Fig. 9 Effect of grain size on stress/strain curves of 0·5 % C steel .1
deformed at 800 C in compression using a drop hammer. 0
-'1 ..~ 600
(After Fink et al.37) . ]
i u
200
lJJ
0::: CJ
:::> QI
\1\
100 100~ lei 500
z 0::: 100 "'
lJJ lJJ
U
0:::
Q.
ai ~~
lJJ
lJJ t-
0-
400
BO IJ)"'
1n
~0::: / ....-~t'
<{
:I:
--- PROGRAMMED DEFORMATION I
,
I
60 ~ \1\ I
:::>
<
0-
, I
~
,,
4 -------- ISOTHERMAL TESTS ON
I
0::: ANNEALED MATERIAL
lJJ
~
0- IJ)
I
(/)
4Ot-
lJJ
IJ)
lJJ
0::: €. =2·3 sec-1 ,
I l:3
0:::
~
t-
IJ)
I ln
20
~ UJ
:::>
0:::
lJJ
:::>
0:::
~ t- t-
20 lJJ
IJ)
t-
Z
,.,---- .•••••• t-
Z
<{
~
lJJ
,
I
I ---_
~
,,
lJJ
0::: ;:£ ;:£
U
lJJ 5 10 5
o 0
lJJ ~
DELAY TIME, sec 600· C 560· C 530·C 495· C 460· C 432· C 410· C
Fig. 11 Effect of delay time on restoration index (after Dunstan

and Evans32) and on room-temperature hardness (after Farag 2 4.
et al.45) of aluminium. EQUIVALENT TRUE STRAIN (E:)
Fig. 12 Deformation temperature/time schedule and resulting
stress/strain behaviour of superpurity aluminium deformed in
torsion at an equivalent strain rate of 2·3 S-l. (After Farag
\1\
0-
et al.45)
t"
12
~
(/) .. heating
The influence of microstructure can be even more pro-
(/) 10
UJ
0::: (/) nounced in rolling or forging operations when the total
t-
(/) 60 ~ deformation is obtained by repeatedly applying relatively
ln
UJ
:::>
0:::
t-
t
temperature-
~
small strain increments (E < 0'7) in successive deformation
passes (e.g. Fig. 1).
t-
controlled 50 t-
z cooling t-
UJ Z
-J
;:£ 4
~ Flow Stress during Multi-Pass Deformation
5 ~'O ;:£
~ 20 5
@ On the second and subsequent passes, the initial structure
600 650 700 750 800
of the material is determined by the earlier deformation
TEMPERATURE, ·c schedule. In this respect, the interpass times are as important
as the deformation produced by each pass, since they deter-
Fig. 13 Effect of continuous heating or cooling on the steady-state
flow stress of vacuum-melted iron deformed in torsion at an mine the extent to which static recrystallization can occur
equivalent strain rate of 1·5 x 10-3 S-l. (After Glover.1l2) between passes.
Despite its importance, recrystallization after high-tem- 14.
perature deformation has received relatively little attention. 90
13
However, published data on several materials show that the c= 0·9 AT EACH €. N
WITH 10 sec REST E

rate of recrystallization increases with increase in deformation ~
and holding temperature14,45,47,103-112and strain rate14,103- 750· C 80 i
112 and with decrease in prior grain size.l°6,112,113In alu- •.. 11
(/)
(/)
(/)
(/)
minium and its alloys, relatively long times are required for UJ
70
lJJ
c::
to-
~ 10 (/)
complete recrystallization to occur,45,47 and it is unlikely (/)
that recrystallization takes place between passes during normal ~ 9 60

lJJ
::J
c::
c::
rolling schedules, except at the highest working tempera- to-
to-
tures.46 On the other hand, copper and its alloys105,114and to-

Z
mild and low-alloy steels in the austenite range104,l07,111 50
w
...J
recrystallize rapidly after deformation, so complete recry- ~

5
0
stallization between passes is probable over most of the 4.0
w
normal range of hot-working temperatures. 107 This will

not be the case if low finishing temperatures are employed107 10°
or if alloying elements that retard recrystallization are present, EQUIVALENT STRAIN RATE, sec-1
e.g. niobium in low-carbon steels.1ll Fig. 14 Effect of increasing or decreasing strain rate on the .flow
The effect of different holding times after prior deformation stress of copper deformed in torsion at 750 C. (After Bromley. 79)
0
on subsequent stress/strain behaviour at the same temperature

is illustrated for aluminium in Fig. 10. It can be seen that after
short holding times the flow stress rises rapidly towards the observation is consistent with the much longer recrystalli-
stress level expected if deformation had been continuous, and zation times in the former materia}ll1 if these result in
only after very long holding times does the behaviour approach an effect resembling that in Fig. 12.
that of the initial annealed material. The change in stress/ Similar effects of cooling during deformation and the
strain behaviour was assessed quantitatively as a ratio of converse effect of heating have been observed during torsional
differences in work done32 and this restoration index is deformation of vacuum-melted iron in the lX-range.112 In this
shown as a function of delay time in Fig. 11. The authors case slower strain-rate deformation was carried out continu-
represented the dependence as a linear function (broken ously (i.e. infinitely short rest periods, so that there was no
line), but comparison of the line drawn through the points opportunity for static recrystallization at any temperature)
(solid line) with data on the decrease in hardness after de- with temperature changing at 50 degC/min. The resulting
I'V
formation under reasonably similar conditions of super- flow stress vs. temperature curves are shown in Fig. 13 and
purity aluminium,45 indicates that initial softening by re- compared with the steady-state flow stress expected from
covery followed by recrystallization is taking place. In isothermal tests on initially annealed materials. This indicates
austenitic steels recovery appears to be less significant as a that the effects of changing temperature are cumulative with
softening process after deformation.14,111 Only when almost strain, as found in programmed tests on aluminium45,47 and
complete recrystallization has occurred during the delay copper79 when static recrystallization does not intervene.
time is the stress/strain behaviour approximately the same as Analogous effects on steady-state flow stress have been
that of the initial annealed materials, and even then system- observed on changing the strain rate during rest periods in
atic differences would be expected as the recrystallized isothermal deformation of copper,79 as shown in Fig. 14.
grain size cepends systematically on the prior deformation Since both an increase in strain rate and a decrease in tempera-
conditions.14,104,106,107,112,113,115 ture may occur in practice, e.g. in continuous rolling of strip,
In hot-working operations successive passes usually take their effects may be additive in successive passes, leading to
place at continually decreasing temperatures,44,46,64,67though even larger effects on stress/strain behaviour than those
in continuous rolling of rods and shapes the temperatures reported above, if static recrystallization does not occur.
may rise during the later passes. The effect of decreasing It· is fortunate, from the viewpoint of working forces
temperature during interrupted deformation on the stress/ encountered in practice, that flow stresses reported from
strain behaviour is illustrated for superpurity aluminium in isothermal, constant-strain-rate tests on annealed material
0
Fig. 12. At 500 C (723 K) (and above) the stress/strain appear generally to overestimate those that may be appro-
curve obtained by programmed deformation is similar to that priate to multi-pass deformation. However, it is clear that the
for initially annealed material, as complete recrystallization effects of thermomechanical history deserve much further
has occurred in the I'V 35 s delay period between successive study both to enable more realistic working forces to be
deformation passes. At lower temperatures recrystallization calculated and to understand the evolution of structures
is incomplete or does not occur, so that successive cycles of produced by different hot-working operations.
deformation take place in material that retains the worked
structure introduced by the prior deformation at higher
temperature. As indicated by the curves at I'V
0
400 C (673 K), Ductility
this results in much lower flow stresses than expected from The basic problem in ductility is to explain why, for a given
initially annealed material. Similar effects would be anti- material, different strains are possible in different working
cipated in steels at temperatures low in the austenite processes before fracture occurs. One proposal to explain this
range and there is some evidence for this in the simulated behaviour is the application of a workability equation, as1l7,118
rolling schedules of Rossard and Blain.44 Also Baird and
Preston116 have noted that the specific roll pressures obtained workability = 11 (material) x 12 (process, friction) ... (24)
in continuous rolling of a C-Mn-Nb steel are only slightly
higher than those of C-Mn steel, whereas NbC precipitation where 11 is a function of the basic ductility of the metal and
would be expected to cause significant strengthening. This 12 that
of the external factors which modify the basic ductility.
12 takes account of effects of friction and changes in geometry
and also depends to some extent on the material, since the
geometrical effects may become more pronounced at the high
strains reached with more ductile materials than at the low
strains achievable with more brittle materials.
Such an equation then poses the question 'Does a material
have an intrinsic ductility?' Some authors maintain that such
a material parameter exists, 117,118while others are un- ~w
certain.6,119,120,121A major difficulty in hot-working research u
is the use of differing techniques for measuring ductility. a:::
w
0.
While some of these make possible the ranking of materials
rr,'" 60
in order of difficulty of processing, generalizations from the w
~
results of different tests are often misleading. w
~<t
Recent studies122-124suggest that the accumulation of in- is
ternal damage which leads to fracture is closely related to the
deformation and restoration processes operative during hot
working. Such studies could lead to a clearer understanding a
t=
of 11 but they must be linked to the external factors covered u
::>
by 12 to give a complete picture. farr, 40
In this section we consider, first, test techniques and re-

sults, then criteria of fracture, followed by a summary of the
experimental evidence available on material parameters.
Finally, we examine the situation where ductility is modified
by surface effects.
20
900 1100 1300 1500
Measurement of Ductility TEMPERATURE, ·c
The two most widely used tests to measure ductility are Fig. 15 Ductility of a 0'14C, 0-568i, Q·65Mn, 1Q·3Ni, 18Cr steel
tension and torsion, since both give quantitative data on be- deformed in tension at two strain rates. (After Guessier and
Castro. 52)
haviours varying from completely brittle to completely
ductile. Compression and bend tests have limited use in
assessing dl,1ctilityof relatively brittle materials or in assessing
the propensity to cracking during hot working. As noted
earlier for strength data, each method has its advantages and
limitations. necking. As Rossard17 has shown, for hot deformation such
a correction would involve a complicated sensitivity to stress
For increased strain rate, his analysis shows that the uniform
Tension elongation is increased. As noted earlier, special tech-
Tests have been performed in tension over a wide range of niques14,17 must be adopted to study necking because of the
strain rates52,121,125-129but the most useful data are derived increased strain rate associated with the reduction in section.
from tests at rates characteristic of the hot-working opera- The data obtained in hot tensile tests have been used both
tion, since there is evidence that ductility depends critically to determine optimum hot-working conditions and to grade
upon rate, particularly in materials where transformation materials for a particular working process. Thus, in Fig. 15,
or precipitation occurs during testing52 (Fig. 15). the ductility passes through a minimum around 1050° C
One measure of ductility is the percentage elongation. (1323 K) and then rises to a maximum value at '" 1350° C
However, since this is composed of a uniform elongation (1623 K) after which it falls precipitously to zero ductility in
before necking and a local elongation associated with necking, only 75-100 deg. This so-called 'hot-shortness' is associated
it is a difficult measure to apply when comparing different with localized melting of segregated regions and defines the
materials with varying contributions from uniform and upper limit of the working range. The lower limit is set by the
local elongation. By using a short gauge-length, it is possible critical ductility for successful working in a given process.
to ensure that the major contribution is from local elonga- Thus, Nicholson et al.l21 found that, for austenitic stainless
tion.129 steels, at least 40% reduction in diameter (E '" 1'0) is neces-
Another measure, preferred by most workers, is the per- sary in laboratory tensile tests at '" 3-11 S-1 for successful
centage reduction in area at fracture. This parameter is ob- rolling and rotary piercing during works' processing. On this
viously related to the onset of fracture. However, as Cock- basis the working range in Fig. 15 is 1150-1350° C (1423-
croft and Latham130 have pointed out, the component of 1623 K).
hydrostatic tensile stress developed progressively during The same criterion appears to hold for valve steels and
necking increases the probability of fracture. Thus, a 'negative free-machining steels tested in a laboratory impact-tensile
feedback' effect occurs which tends to prevent large ductility machine at '" 800 s-1.129 For complex materials, such as
values being obtained. Percentage reduction in area may high-temperature alloys, successful working has been achieved
therefore show little discrimination between very ductile even though laboratory tests gave lower ductilities (Fig. 29,
metals. A method of determining true ductility, i.e. the duc- p. 19).
tility that would be observed if necking did'not occur, has been Variations in strain rate can alter the relative values of
proposed by Cockcroft and Latham for cold-working opera- measured tensile ductility and thus the critical ductility at a
tions. This depends on either artificially altering the neck given temperature, though the same general form of ductility /
shape at fracture or calculating the stress developed during temperature relationship remains (Fig. 15).
Torsion
18
An extensive review of the test has been given by Robbins
et al.8 They point out that, for meaningful results, specimens
with a reduced gauge-section must be used, so that deforma-
tion is confined to a volume where the temperature can be
held constant. Such tests have been performed over a wide 15
range of strain rates. 8, 9,10,16,47,49-55,68,70,99,131-138 In
contrast to tension, where deformation at high strains
becomes localized in the necked region, very large torsional
strains can be applied with essentially constant specimen w
~ 12
geometry up to fracture. t>
Ductility is measured as number of turns to failure or <(
0:::
l.L.
surface strain to failure. Ductility data in these terms have
been reported by many workers with the main emphasis
on ferrous alloys and extensive reference lists have been z 9
given by various reviewers.2,8,13 Since the strain rate is con- ~
0:::
I-
stant during the test, the time to rupture is readily calculated If)
and has been used for correlation of ductility data,122 as I-

Z
discussed later. ~
~ 6
In torsion, materials undergo larger strains before fracture 5
o
than in tension. This allows closer simulation of the con- UJ
ditions during practical hot-working operations where the

total strain may be large (Fig. 1). However, the greater
capacity a material may have for deformation in torsion can
at times present a problem. Thus, it is hard to place any
fundamental significance on the number of twists to fracture
when the number is very large; for example, it is difficult to
differentiate between a specimen that twists 600 times to
fracture (E 360) and one that twists 400 times to fracture
f'V 800 1000
(E "'" 240), except to say that both are ductile. With less ductile TEMPERATURE, •C
materials, the number of twists to fracture is reproducible Fig. 16 Effect of axial tensile and compressive stress on fracture
within close limits and can be used as a quantitative measure st~ain of hot-rolled Inconel 600 tested at ~ = 2·5 S-l. (After
of ductility. DIeter etal.l34)
The measured values of ductility are dependent on speci- • no applied axial stress
men geometry. Thus, for specimens with a constant length: • 50 % yield stress applied in compression
diameter ratio, Reynolds and Tegart133 found that the number .•. 50 % yield stress applied in tension
of twists to failure increased with increasing specimen dia-
Dragan suggests that EO is a measure of the intrinsic ductility
meter. Robbins et al.8 suggest that this result reflects the
of a material but further work is necessary to confirm the
sensitivity of the torsion test to surface modification by
validity of equation (25).
oxidation, since the relative surface:volume ratios are de-
The large strains before fracture, together with the rela-
creasing with increasing diameter.
tively constant geometry during testing, means that adiabatic
As noted earlier, axial stresses are developed if specimens
heating can be significant. Thus, for a low-carbon steel
are restrained during hot torsion. Though their effect on the
equivalent tensile flow stress is small, such stresses have a
Robbins et ale 8report a temperature rise from 750 to 1200 0
C
(1023 to 1473 K) after a strain E "'" 50 and similar effects have
significant influence on ductility. This has been demonstrated
been noted by other workers.51,68,133As a result of this
by external application of longitudinal tension or compression
during torsion.l34,135,136 Since cracks are present almost heating the ductility/temperature curve for torsion of a given
material tends to be shifted to lower temperatures than for
from the beginning of deformation, the torsion test essentially
tension of the same material, e.g. the maximum occurs
measures crack propagation. Compressive stresses have a
,...,100 deg lower for steels. Further, Nicholson et ale 121have
markedly greater effect than tensile stresses of the same
suggested that the apparent decrease in peak-ductility tem-
magnitude, as shown in Fig. 16 for (JA/-r = 0·5. Dragan135
perature with increased strain rate can be accounted for if
has applied a range of (J A/ -r in compression to a low-carbon
suitable correction is made for adiabatic heating, as illu-
steel and suggests that ductility increases approximately as
strated 'in Fig. 17. This also shows that ductility can be
dependent on strain rate, as noted by other workers.51,124
... (25) As with tension, the data obtained in hot-torsion tests have
been used both to grade materials and to assess the optimum
temperature for a particular working process. Thus, in Fig.
where ET is the strain at fracture on application of longi- 17, the ductility passes through a maxin1um around 1150-
1200 C (1423-1473 K) for the uncorrected data and this
0
tudinal compressive stress (J A; EO is the strain at fracture in

the absence of longitudinal stress; C is a material constant corresponds to the optimum temperature range for rotary
0
("'" 2·5-3 for carbon steel at 900 C (1173 K) but clearly of piercing.51,133,137It is ironical that the success of the hot-
the order of 10-20 for Inconel 600* from the data of Fig. 16). torsion test in predicting optimum temperatures for hot
piercing arises from the combined effects of its two potential
disadvantages, i.e. a complex stress system, together with
* Registered trade mark of Henry Wiggin and Co., Ltd. marked adiabatic heating in a small volume.
1400
6 sec -1 (corrected
6 sec-1
\/-'" 1 \
1300
u
t!" 15 / \ 0
uJ / \
/
/
\
\ ~0::
/ \ ::> 1200
/ 4
0:
/ IJJ
/
I/''''
/ , ~
0-
IJJ
t-
I / ' 1100
I /
z / /
<i / /~O.3 sec-1 (corrected)
0:: 10
til II
I
1000
t-
Z o 0·2 0·4 0·6 O'S 1·0 1-2
~ CARBON CONTENT, %
~
5 Fig. 18 Comparison of optimum ductility temperatures determined
o
IJJ from hot-torsion tests at e "'"
0·5 S-l with operating temperatures
for piercing, rolling, and forging of carbon steels. (After
Zidek. 120)
5 - - - torsion o tube mill (Mannesmann)
900 1100 1200 1300 1400 • tube mill (Stiefel) f:., general rolling mill
TEMPERATURE, •C .•. slabbing mill x forge
Fig. 17 Effect of strain rate on temperature-dependence of fracture
strain in torsion of stainless steel (AISI 321). The influence of
adiabatic heating is shown by the corrected curves. (After
Nicholson et al.l21)
Stress Criterion
For yielding, two criteria have been proposed, namely,
maximum effective stress based on distortion energy (von
Mises) and the maximum effective stress based on maximum
The correlation between the temperature of maximum shear stress (Tresca). The former is generally considered to
ductility in hot-torsion tests and the practical temperatures give the better agreement with experiment. The success of
for a variety of working processes for carbon steels are shown the von Mises criterion for yielding has encouraged several
in Fig. 18. In general, this confirms that the temperature of workersl40-142 to apply it to fracture but the experimental
the maximum corresponds closely to the optimum tempera- evidence for creep rupture indicates that it cannot explain
ture for hot piercing and is slightly higher than the optimum the results satisfactorily.
temperature for forging and extrusion. As with tension, The important role of tensile stresses in fracture was first
the lower limit of temperature is set by the critical ductility noted by Siebel,145 who pointed out that cracking in metal-
for successful working in a given process. Thus, for forging working is associated with induced tensile stresses, even in
and rolling of steels, Langerweger and Trenkler138 found processes such as forging which are predominantly compres-
that e > 3 was necessar) in laboratory tests at 1 S-l to avoid sive. Some experimental evidence is provided by the work of
cracking during works' processing. Similarly, Reynolds and WellsI46 but the idea has not been seriously developed and
Tegart133 report that, in laboratory tests at 1 S-l on steels, has been repeated without elaboration by more recent wor-
e > 3-4 is required for successful extrusion and e > 18 for kers.93 The importance of tensile stresses, however, is in-
successful rotary piercing during works' processing. For directly confirmed by the large increase in ductility when
complex alloys, such as tool steels, successful working has materials are deformed under hydrostatic pressure.I47-149
been achieved even though laboratory tests gave lower Some elegant experiments on multiaxial creep of a variety of
ductilities (Fig. 30, p. 19). materials by Johnson and his co-workers140-142 indicate
that the maximum tensile stress controls the time to fracture
Criteria of Fracture whatever the type of fracture. If true for all materials and
In view of the diffeling techniques used to assess ductility stress conditions, this would lead to a very simple criterion
and the vast amount of data that is available for different for fracture. However, in working operations, it is likely
test geometries, it would clearly be desirable to establish that both shear and tensile stresses play a part, since there is
some criterion of fracture that would make possible a ration- evidence that localized plastic flow by shear is required to
alization of the data. Implicit in any form of ductility deter- initiate cracks which are then propagated by tensile stresses.150
mination is the idea that the order of rating obtained for a
given group of metals will not be affected by a change in the Strain Criterion
stress system at fracture. General experience supports this Some workers151 have suggested that the fracture criterion
assumption and data provided by Cockcroft and Latham130 should be based on total strain1l7,137 but difficulties arise
and by Hodierne48 for fracture strains of 3-5 indicate a since the total strains vary markedly in different processes.
similar ranking for tension- and torsion-test results. Soviet workers152 have proposed a uniform workability
Such a criterion must be a general one related to defor- index using a criterion based on octahedral shear strain,
mation in a multi axial stress system and several approaches but there was considerable scatter in their results from
to the problem have been made, mostly with respect to creep. different materials and doubts have been expressed by later
Reviews of this topic are given by several workers.I:39-144 workers6,120 as to the applicability of such a criterion.
TABLE II
Comparison of Parameters Derived from Strength and Rupture Data55,83,122,154
Strength Data Rupture Data

Material Q Q Effect of
n' kJ/ kcalf 11' kJ/ kcal/ Increasing Strain
mol mol mol mol Rate on Ductility
Nickel 5·7 234 56 5·1 222 53 Small increase
Nickel-
20% Iron 5·7 393 94 6·3 368 88 None
Iron-23 % Chromium-
17% Nickel 5·3 402 96 4·9 335 80 Increase
Iron--26% Chromium-
10% Nickel 5·4 418 100 5·4 418 100 None
Plastic- Work Criterion and 8 S-1. At the lower strain rate, though there is some
Another possible assumption is that any criterion of scatter in the results, the calculated values tend to support
fracture should be based on some combination of stress and equation (27), since the average values for aluminium, copper,
strain rather than on either of these quantities separately. and nickel are 1450 ± 550, 2620 ± 310, and 1720 ± 700
There are indications that the total plastic work to fracture MN/m2 (210 ± 80, 380 ± 45, and 250 ± 100 x 103Ibf/in2).
may be an important factor. However, Cockcroft and However, at the higher strain rate, the effects of adiabatic
Latham130 have shown that such a criterion does not fit heating become significant and there is a much greater
experimental tensile results. scatter, so that the average values for aluminium, copper, and
In view of the importance of the maximum tensile stress, nickel are 1520 ± 1030, 5100 ± 2420, and 2620 ± 1660
as noted earlier, any criterion of ductile fracture should take MN/m2 (220 ±: 150, 740 ± 350, and 380 ± 240 x 103
this into account. Cockcroft and Latham130 have thus pro- Ibf/in2). Overall, the results suggest that equation (27) could
posed that fracture occurs in a ductile material when, for a be a reasonable criterion for hot working as well as for cold
given temperature and strain rate, the quantity working but clearly further data are required to test it more
rigorously for hot working.
Sf
f Q"x(~)dO = constant ... (26)

Time-to-Rupture Criterion
For creep rupture, one criterion153 is:
o
tr Es = constant ... (28)
where
cr equivalent stress where Es is the steady-state creep rate and tr is the time to
ax non-dimensional stress-concentration factor re- rupture. Assuming that this also applies to high strain rates,
a- presenting the effect of the highest tensile stress, then from equations (11) and (12)
ax
cf = fracture strain tr = AI (sinh ria) -nl exp (Q/RT) ... (29)
E = equivalent strain
Such an expression reduces to a power relationship at low
For tension, ax is calculated for the necked region, while
stresses and to an exponential relationship at high stresses, as
for torsion the criterion becomes:
discussed earlier. Relationships of similar form have been
a- _ proposed by Oding et al.144 for creep rupture. Tegart122
J
r-C,!f
"l3 dE = constant ... (27) showed that equation (29) satisfactorily correlated the frac-
ture data obtained from hot-torsion tests on a series of nickel-
o iron alloys over a wide range of temperatures and strain rates.
where In addition, similar values of Q and n' were derived from both
C,!f torsional fracture strain expressed as an effective the strength and rupture correlations, as shown in Table II.
strain. The use of a temperature-compensated parameter, as in
Cockcroft and Latham successfully applied the criterion to equation (12), allowed correlation of data at different tem-
cold working but did not test it for hot working. The present peratures.
authors have tested equation (27) using stress/strain curves The correl3.tions of equations (11), (12), and (29) have also
for aluminium, copper, and nickel derived from hot-torsion been applied by Miiller55 to rupture data obtained from hot-
data over the range 0·5-0·8 Trn and at strain rates of '" 1 torsion tests on a series of nickel-chromium stainless steels.
TRUE PEAK STRESS, MN/m2 TRUE PEAK STRESS t MN/m2
30 50 100 200 50 100 150 200
101.
103 103
u u
ClI ClI
III
III
W
a:: ~
0::
:=>
...J :=>
102 102
...J
~ <{
U.
0
I-- 0
I--
W
1: w
i= 1:
i=
10' 101
(a) (b)
4 6 a 10 20 30 40 50 10 15 20 25 30 35
TRUE PEAK STRESS, 103 psi TRUE PEAK STRESS, 103 psi
TRUE PEAK STRESS, MN/m2

50 100 200
u
~ 100% 1 o 900· C
01000· C
().110o·C
o 120o·C
3
10
,,
",
"-
~ "-
\
'\
W '\
0::: 2
3 10 "- '\
~ "-
"- ,
~
w
" '\
'\
1:
i=
1+10% oc ",
,, ~.0
103 \
'\
\
w \~
a::: \
::> \
~ \ ¢
ffic.. 10 \
I: \
w \
t- \
\
4 10 15 20 30 4.0
sinh ex: cr TRUE PEAK STRESS, 103 psi
Fig. 19 Rupture data derived from hot-torsion tests on a 23 % Cr, Fig. 20 Comparison between experimental and calculated values
17 % Ni (l00 % austenite) stainless steel plotted to demonstrate: of temperature-compensated time to rupture for single- and two-
(a) power relationship; (b) exponential relationship; (c) sinh phase stainless steels. Activation energies used for ealculati.n
relationship. (Data from MUller.55) were derived from strength data. (After Gittins and Sellars. 161)
Figs. 19 (a)-(e) show the data for the single-phase austenitic of the grain-size change due to dynamic recrystallization
23 % Cr, 170/0 Ni alloy plotted to demonstr~te the fit through an expression of the form83
to a power, exponential, and sinh relationshIp, respec-
tively. The values of nand Q derived from the strength and d = K (j -4/3 ... (31)
rupture correlations for this alloy, together with those for a
two-phase (austenite + 10% ferrite) alloy, are shown in If the strain rate of equation (30) is expressed in terms of a
Table II. For the two-phase alloy, as with the nickel alloys, power relationship with stress (this is a rea~onable appr~xi-
the values derived from strength and rupture data are similar, mati on to the sinh relationship, as noted earher) and equatIon
but for the single-phase alloy the Q value from rupture data (31) is substituted into equation (30), then
is lower than that derived from the strength data. This dis-
crepancy results from the fact that ductility in this alloy is
dependent on strain rate, as noted in Table II. While the t, = [811: Y ilL - V)f/2 AN a- (n - 0.67) exp (QfRT) ..• (32)
correlation implied in equation (29) thus holds for some cases,
there are situations, as with the single-phase alloys, where the This equation gives the theoretical rupture times for spe~i-
use of the Q value derived from strength data does not mens having a grain size comparable to that developed In
correlate the data on a single line but on a series of straight the steady-state region of deformation, assuming continuous
lines (Fig. 20). wedge-crack growth. For pure nickel, equation (32) predicts
Further data are required to test the validity of equation rupture times 2-3 orders of magnitude lower than those found
(29) but the results to date are encouraging enough to suggest
experimentally. The same analysis applied to the nickel-
that a fracture criterion based on time to rupture should be 20% iron alloy gives reasonable predictions of observ~d
applicable to hot-working data. Such a criterion allows f<;>r rupture times at the lowest temperatures, where the maten~l
correlation of ductility data over a range of stresses, straIn
fails at low strains, hut at higher temperatures the analysIs
rates, and temperatures in a more general fashion than any
again underestimates the experimental !imes to rupture: .
of the other proposed criteria.
The model of Williams158 has been reInterpreted by GIttIns
and Sellars161 to include the stress- and grain-size-dependence
of grain-boundary sliding and their analysis leads to an
Fracture Modes during Hot Working expression of the form
The macroscopic criteria discussed in the previous section
take no account of the differences in fracture modes that can
result from different deformation mechanisms in different Ir = ~, ~A" 0- (n '+ 0.33)exp (Q/ RT) ... (33)
materials. Relatively few systematic studies of fracture modes
have been reported122 and the main emphasis has been on
where K' is a constant computed from experimental data
ferrous materials. The observations suggest that fracture-
on stress- and grain-size-dependence of sliding and n' is the
initiation processes during hot working are similar to those
stress exponent of the grain-boundary sliding rate. The results
in creep but that fracture-propagation processes during hot
of calculations based on equation (33) are shown by the
working are markedly modified by the concurrent structural
broken lines in Fig. 20. As with the earlier analysis, the
changes outlined in Table I.
calculated lines still fall below the experimental lines, though
Thus, the initial stress developed for an applied strain rate
the discrepancy is now much smaller.
is sufficient to form cracks at the original grain boundaries,
This difference between predicted and experimental time
both at triple points and at irregularities developed in the
to rupture has been interpreted123 as indicating that crack
boundaries. These cracks can then grow under the applied
growth is not a continuous process but is interrupted and
stress, either by vacancy diffusion along grain boundaries or
delayed by the interaction of the dynamic softening process
by grain-boundary sliding. Conditions of low ductility
with the fracture process. In the case of materials where
correspond to the propagation and coalescence of these
recrystallization is the softening process, conditions of high
cracks to give intergranular rupture along essentially the
ductility correspond to recrystallization or the migration of
original grain boundaries.
the original grain boundaries, thus isolating the original
Models for the growth of triple-point cracks in creep
have been proposed by various workers.l55-158 That of cracks and preventing further immediate growth. Continued
growth occurs by 'capturing' a moving grain boundary for
Williams158 is based on crack growth due to grain-boundary
a sufficient time to facilitate lengthening of the crack. Linkage
sliding and, while it represents an approximation since it
of cracks leads to failure. Ductility under these conditions
neglects the stress- and grain-size-dependence of the sliding
may be governed by the rate of grain-boundary migration.
process, it nevertheless describes reasonably accurately the
In the case of materials where recovery is the softening
experimentally observed rate of crack growth during creep
process, the original boundaries tend to lose their identity
of austenitic steels over a range of temperatures and stresses.
159,160The theoretical relationship is because of the formation of subgrains and the crack-propa-
gation process is slowed down and eventually halted, since
there is no sliding to promote further growth. Fracture in
. [Ge y8It (l - V)] 1/2
... (30) such materials tends to occur as the result of the linking of
Ir Es = 'tJ-d2
internal voids that can be nucleated at inclusions and these
materials exhibit variable ductilities which are related to
where Ge = grain-edge length ('" d), y = effective fracture inclusion content. In such cases the time-to-rupture cor-
surface energy, d = grain size, 'tJ- = shear modulus, and v = relation discussed earlier would not be expected to apply.
Poisson's ratio. Thus, while the correlation of equations (11) and (12) could
Since triple-point cracks were observed in their studies of be applied to the strength data for a ferritic chromium-nickel
hot torsion of nickel-iron alloys, Luton and Tegart123 ex- alloy, the rupture data for the same alloy give a poor cor-
tended equation (30) to cover this situation by taking account relation using equation (29).
Lo
I->
''''' 500
t /
~ ~ (a)
/ (NO FRACTURE FOR O'B %
Ul
~ // AND 1·02 % C ALLOYS)
cr:
~ 30
~
~u- iOO
tJ<l: LABORATORY ALLOYS.
Ae, 50
cr: o
u.. t-
0 Z
t-
4:
~ 10
COMMERCIAL STEELS.
t-
Z
Ul
...J
~
5
o
w 1
0·2 0·4 0·6 G8 1·0
CARBON CONTENT, WT:%
Fe3 C CONTENT AT 650·C, VOL.-%

°700 BOO 900 1000 1100 1200 1300
TEMPERATURE, ·c g 500
5 10 15
Ul
Fig. 21 Variation of ductility with temperature for a high-oxygen cr:
~ ( b)
Swedish iron tested in torsion at an equivalent strain rate of 650· C
t3<l:
o· 5 S-l. (After Reynolds and Tegart.133) cr:
u..
100
0
t-
/LABORATORY ALLOYS.
Z
~cr:
TABLE III t-
CJ) 10
t-
Z
Ul
Ductility of Different Crystal Structures165 ...J
~ COMMERCIAL STEELS.
5
@
0·2 0·4 0·6 1·2
Metal Structure £,S-l Temp. E,! CARBON CONTENT,
TfTm
Fig. 22 Influence of carbon on the ductility of iron tested in torsion
Zinc hcp 0·08 0·7 1 at an equivalent strain rate of 0·03 S-l: (a) at 920 C (y ra,nge); 0
Magnesium
Beryllium
hcp
hcp
0·08
0·05
0·7
0·7
2·4
0·4
(b) at 650 C (ferrite +
cementite range). (After Robbms et
0
01.166)
Titanium hcp 3·5 0·58 1
bee 3·5 0·60 210
Iron bee 0·5 0·63 28
fcc 0·5 0·67 2·3
fcc 0·5 0·8 18 60
Aluminium fcc 0·5 0·7 50
Nickel fcc 0·5 0·7 23
Copper fcc 0·5 0·7 21
Lead fcc 0·5 0·7 70
50
I
t
I:I'
'uJ
TABLE IV
Ductility of Zinc AJIoysl74
II
--0'6Tm
Ductility (f:'f) at: ---- 0·7 Tm
--- 0·8Tm
Material Grain Size, 210 C 0
280 C 0
350 C
0
mm (483K) (553 K) (623 K) , I

Zinc 0·13 1·0 1·4 1·4
Zinc-O·24%
titanium 0·001 4·5 6·0 2'0 I
/1 ,
Material Parameters Affecting Ductility

10
;'
)j I
~ //
The subject of hot workability, particularly of ferrous ....• /
materials, has been actively studied by many investigators o 0 20 40 60
since the early years of this century and as a consequence % NICKEL
there is a wealth of information on the subject. Much of this Fig. 23 Temperature-dependence of ductility for a series 0/ copper-
is, however, in a form that cannot be readily translated to nickel alloys tested in torsion at an equivalent stram rate of
situations outside that of the writer's immediate interest and 0·6 S-l. (After Sellars and Tegart.70)
relatively few attempts have been made critically to review 300
the available data. Notable exceptions are the excellent

reviews of factors affecting hot workability of ferrous materi-
als by Nicholson,2 Portevin,3 and Contractor and Morgan ;12
while Cook and Davis162 have reviewed the influence of com-
position on the workability of copper alloys. General sum-
maries of the topic have been published by Tegart.119,122 ~
In this section we consider some basic material parameters ~
affecting ductility as expressed by the function 11 in the w
c::
200
workability equation (24) (p. 9). These are: crystal structure :::>
I-
u
and allotropic transformations; solid-solution alloying; 4:
c::
I.J..
duplex structures; cast structures; inclusions; reheating
practice. 0
I-
Z
<i
c::
I-
Crystal Structure and Allotropic Transformation (f)
I- 100
Different crystal structures have markedly different duc- z
W
tilities.l50,163,164Thus, in contrast to fcc and bcc metals and -J
~
alloys, which exhibit good ductility at high temperatures, 5
@
hcp metals and alloys display relatively poor ductility, as
evidenced by the hot-torsion data in Tables III and IV. A
change in crystal structure is reflected in a change in ductility,
as shown for the a(bcc) -+ y (fcc) transformation in o
0--
iron (see also Fig. 21) and even more strikingly for the
400 500 600
a(hcp) -+ ~(bcc) transformation in titanium.
TEMPERATURE, C
In iron, the relative ductilities of the two phases are sensi- 0
tive to variations in impurity content68,112,132,133,166,167 but Fig. 24 Temperature-dependence of ductility of aluminium-mag-

the general pattern of Fig. 21 is maintained. One interesting nesium alloys deformed in torsion at an equivalent strain rate
of2'3 S-l. (After Cotner and Tegart,47)
conclusion from studies of pure irons and low-carbon steels
(Fig. 22) is that these materials exhibit ductilities around • pure Al DAl-l%Mg f:.AI-2%Mg OAI-5%Mg
700° C (973 K) comparable to those in the austenite range. allows substantial crack propagation to occur under the higher
Since the strengths at these temperatures are only slightly stress required to deform such alloys before recrystallization
higher than in the austenite, it is feasible to roll low-carbon operates to slow crack propagation.
steels at 600-700° C (873-973K) with less oxidation and lower In the case of materials where recovery is the softening
fuel costs than at the usual temperatures of 1100-1200° C process ductility also decreases with alloying, though the
(1373-1473 K). An additional benefit from such a practice fact that solutes exert a smaller influence on dynamic recovery
is the introduction of controlled substructures to improve than on recrystallization tends to make the effect less pro-
low-temperature mechanical properties. Its feasibility has nounced. As Fig. 24 shows, the addition of magnesium to
been demonstrated168 but the accurate control of temperature aluminium reduces ductility, the most marked drop occurring
required makes the process uneconomical in production. almost at the solubility limit.47,7o Similar effects have been
In titanium and its alloys, forging, rolling, and extrusion observed for aluminium-silver and aluminium-copper
are carried out in the ~-range, since it has been found that alloys.169
cracking occurs if the temperature drops into the a-range.
Duplex Structures
The presence of a duplex structure normally leads to
Solid-Solution Alloying reduced ductility. A good example68,112,132,133,166,167 is the
Solute elements can exert marked effects up to high tem- lowered ductility of irons and low-carbon steels when tested
peratures under hot-working conditions.119 Their general in the (rJ. + y) range. The two-phase structure is much less
effect is to reduce ductility, particularly at high alloy concen- ductile than either of the component phases and it has been
trations, but in some cases ductility can be enhanced by suggested133 that the so-called 'red shortness' of iron is
small alloy additions. Thus, in the case of iron, the addition caused by the extended (a + y) region produced by the
of carbon increases ductility in the austenite range up to presence of impurity elements.
1·1% C.9,154,166Fig. 22(a) shows that the effect is nlore The ductility of a two-phase alloy is dependent upon the
marked in pure alloys based on electrolytic vacuum-melted volume fractions of the two phases present and their relative
iron.l66 In the case of an iron-25% nickel alloy, White131 ductilities. As shown in Fig. 22(b), the addition of a small
has found that 0'005% boron increases ductility, while amount of carbon first raises the ductility of rJ.-ironand fur-
0'15% copper or 0'10% tin decreases ductility. ther additions > 0·1 % decrease ductility. Concurrently, the
The influence of solid-solution alloying has been related proportion of cementite increases (from ~ 1·3 vol.-% at
to the structural changes during deformation and it has 0·1 wt.-% C to 17 vol.-% at 1·1 wt.-% C). Robbins et al.166
been suggested that the effect is more pronounced in materials suggest that the initial increase may be associated with grain
where recrystallization is the softening process (cf. Table I). refinement due to the presence of fine, dispersed particles.
Thus, copper-nickel alloys which exhibit complete solid Several workers55,101,170,171have studied the ductility of
solubility show a marked drop in ductility to a very low level chromium-nickel steels consisting of austenite + 8-ferrite
above ,...,20% added solute70 (Fig. 23). This arises from and have shown that there is a ductility trough at ~ 30%
retardation of recrystallization during deformation, which a-ferrite over a range of temperatures (Fig. 25).
Fracture is initiated by interphase boundary cracking, 16
which is particularly severe at '" 30CYo o-ferrite, since most

of the deformation is concentrated in the austenite. 55 Nichol-
son and Moore1?! have reported that steels containing 14
only 10-20% a-ferrite present problems in t.ot rolling and that

a large proportion of product is rejected because of excessive X 1200'C
'shell' defects. ..:; 12 t:::. 1100'C
The influence of composition on ductility can often be o 1000' C
UJ
rationalized in terms of the proportion of second phase 0::
::> + 900' C
present in a given material. In the case of forging a titanium- t-
~ 10
12'5% tin-2'5% aluminium alloy that had a composition 0:::
U.
close to the ~-phase boundary, Reynolds172 found that the o

t-
addition of small amounts of alloying elements had marked z B

effects on the hot workability. Thus, additions of carbon, 4:0::
lead, silicon, 8nd zirconium raised the minimum satisfactory tn
upsetting temperature by 1ODdeg C (K), whereas the remaining t-
~ 6
additions either had no effect in spite of increased solute --J
content or reduced the minimum upsetting temperature by ~5

o
up to 100 deg. However, their individual effects were rational- UJ
4
ized in terms of t he proportion of ~ phase present in the
structure, as shown in Fig. 26. Similarly, Ward and Helli- ~
t:s.
well173 have shown that the hot workability of two-phase
copper-nickel-zinc alloys containing lead and manganese ~o
can be related to the proportions of (X and ~ phases present. ------9-+
Relatively small changes in composition gave marked differ-
ences in the proportions of phases and these were reflected in 25 50 75 100
the ductility results obtained by hot torsion (Fig. 27). Ex-
d - FERRITE t PER CENT.
trusion experiments confirmed that the optimum hot work-
ability occurred at 750-800° C (1023-1073 K) and that the Fig. 25 Dependence of ductility of wrought iron-chromium-nickel
alloys on 8-ferrite content. Alloys tested in torsion at an equiva-
differing phase proportions had a significant effect on hot lent strain rate of 0·02 S-l. (After Muller.55)
workability.
The degree and nature of dispersion of the second phase
is clearly an important factor. If the second phase occurs as 1050r-----,...------'""I'.-------or--.--_
a grain-boundary film or as coarse grain-boundary particles,
then fracture can be preferentially initiated at the boundaries,
leading to low ductilities. In contrast, as noted earlier, there
is some evidence that a dispersion of fine coherent particles ec
which refines grain size can improve ductility by restricting t-
Z
UJ
crack propagation. As shown in Table IV, a zinc-titanium U'l
UJ
alloy that contains a fine dispersion of ZnTi15 particles has g: 1000I- ONO ADDITION -
a finer grain size and increased ductility compared with zinc. UJ
U'l
4:
::x:
a..
CQ.
-;f!.
Cast Structures 0
U"l
Most of the experimental results on hot workability have ::x:

u
been obtained on wrought material but in practice the major I
:3:
950 -
problems in hot working arise during the primary breakdown ~ eTa
of the as-cast ingot. Information on this topic is limited, U eCu
mainly because of the difficulty in producing laboratory ingots L
UJ
that are equivalent to the much larger industrial ingots with 0::
::>
regard to segregation and structural features and because of 4
0:: eV
the high costs involved in large-scale experiments on industrial UJ w.
ingots.
a..
L
UJ
t-
900
Mn.
eMo
Fe e
-
The limited evidence available indicates that hot work-
ability of as-cast structures is best studied by hot tensile
tests, since torsion tests rapidly destroy the original grain
structure.175 Longitudinal tensile tests show that the columnar
regions of a steel ingot are generally more ductile than the
equiaxed regions176,179 (Fig. 28), while results for titanium 850
I I I
900 1000 1100

are similar. However, cases are reported where the reverse
MINIMUM UPSETTING TEMPERATU~E)' C
behaviour has been observed.l7S,lso Such differences are
thought to be due to differences in the degree of impurity [Courtesy Iron Steel Inst.
segregation. Fig. 26 Relation between proportion of ~ phase and minimum
satisfactory upsetting temperature for a titanium-12' 5 % tin-
The melting practice is important in determining the hot 2·5 % aluminium alloy with minor additions of alloying elements.
workability of as-cast materials and the development of new (After Reynolds.172)
f-!
z
w
u 80
0:::
W
0..
W••
0:: 60
::J ELECTROSLAG - REFINED
t- :>-
u t-
W
0:::
~
::J
«
0:::
lJ..
02
olD
8~
\
~
1~
u 3
~
0::: 4.0
lJ..
«
w
0:: ."",---.
« ---- ---------r---~--
I \
.......
z lJ.. / \
---- ///\
~ 2 (71%0<)
0 / \
ALLOY 22
~
0:::
(f)
z
0
20
~§ \
I
t= I
~ u
::J 0«
00
0:::
VACUUM - ARC- REMELTED
I
z
W
-l
0
w a:~0:: I
1
g , 0:: 0
lJ..
I
:3
o 1000 1100 1200
w
TEMPERATURE, 'c
Fig. 29 Effect of melting practice on ductility of Udimet-700 alloy
o 700 800 900 tested in tension. (After Kelley.182)
600
TEMPERATURE, 'C
Fig. 27 Effect of a-phase content on the ductility of leaded copper-

14 % Ni-Zn-Mn alloys tested in torsion. (After Ward and
Helliwell.173)
t 1-0
I IoU ELECTROSLAG -
REMELTED
W
0:::
::>
t-
u
«
0::
lJ..
~ VACUUM -
-- AS FORGED 0-5 .MELTED
z
---- AS CAST ~
0::
t-
(f)
80
FORGED
t-
Z
~
g
5
0
w
1050 1100 1150 1200 1250

.-;
z
w TEMPERATURE, • c
U
0:: 60 Fig. 30. Effect of electroslag refining on ductility of as-cast 18 %

w
0.. tungsten-4 % chromium-l % vanadium tool steel. Torsion tests
u? carried out at an equivalent strain rate of 0·13 S-I. (After
0::
::J
Coward.102)
t-
U
«
0::
melting practices has in a number of cases led to an im-
lJ..
provement.181-184 Thus, vacuum-arc remelting and electro-
~ slag refining are being increasingly used to obtain refined
0::
W 4.0
grades by remelting materials produced by conventional
t-
W
~« melting techniques. In the case of nickel-base high-tempera-
0 ture alloys, marked improvements in yield have been reported
~ as a result of electroslag refining.182 These have been attri-
z buted to improved hot workability as shown in Fig. 29 for
0
i=
u
the alloy Udimet 700 (Ni 52'7, Cr 15, Mo 5'3, Co 18'5, Ti 3'5,
=>
0 AI 4'3, C 0·07%). Vacuum-arc remelting leads to smaller
W
0::: improvements,183 because of a more limited working range
20
(Fig. 29).
Another case in which electroslag refining improves hot
workability occurs with tool steels,102 where the carbide
phases can be uniformly distributed to give better workability
(Fig. 30). The usual hot working range for such a material
is 1100-1180° C (1373-1543 K) with a preferred temperature
of 1150° C (1423 K) but with electroslag refining the range is
o extended to 1240° C (1513 K).
900 1100 1300 1500
TEMPERATURE 1 • C Hot working of cast structures leads to increased ductility,
Fig. 28 Influence of ingot structure and of prior hot working on as documented by a number of workers for both ferrous138.
ductility of 18 % chromium, 11 % nickel steel at an equivalent 177,178,180,185,186and non-ferrous187,188 materials. An ex-
strain rate 0/12 s-1. (After Decroix et al.178) ample is illustrated in Fig. 28. This increase in ductility
80
0-6 00
)(
r-: 0
z
0'4 UJ
u x x x
70
a:: 0
UJ 0
0..
0-2
x.•.• ·0
b 0
~ 60
0
x
~
°o'1L Al203 Ca-aluminates z x
0
t=
u
~
0
50
UJ
3MgO.Al203 MgO.Al203 c:::
0'2~ AO. 8203
•...
>-
0·1 ( Spinel type)
>-
t- X
::i ~crea: MgO.Al2QI ::i 0
iii o i=
u 4.0 0
« ;:)
x
:::: 0
0:: 0
x.
~
LtJ
u
b
o z
1-0 30
lL.
o 0-1 0·2 0-3 0·4 0·5
o VOLUME FRACTION OF OXIDES + SULPHIDES, PER CENT~
>< 0-5 Fig. 33 Relationship between volume fraction of inclusions and hot
LlJ
o ductility of notched tensile specimens of cast low-carbon steels
~ tested at 1200° C. (After Cottingham.l98)
.• tops x middles • penultimate bottoms
o o extreme bottoms
140
0·06'" C, 0·39". Mn 0·028".SJ
1·0 ~ ----, MnS

" ' ....
120
Me X=Te '-.... X=Se ~ /0'04"'C, 0·16".Mn, 0·020".5
O.S~ ----- (Mn,Me) (S,X) -----
~ ~ / 0·03". C, 0·10'loMn, 0·021'1oS
100
-·C ~z
UJ
0·03',. C, 0·08'10 Mn, 0·019'105
u
[Courtesy Iron Steel Inst. BO 0·03'10 C, 0·04''.Mn,0·018'105
c:::
Fig. 31 Influence of temperature on the plastic deformation of W
0..
different types of inclusion, as compared to steel. (After
Kiessling.190) •..
z 60
0
i=
<t:
C:>
Z
B8 9
UJ
40
x x O·OS',.C 0·11".Mn , o-O'a",S

~z
J
20
UJ
x
84-
U
0::
UJ xx
0.. 0 o
0 700 BOO 900 1000 1100 1200 1300 1400
x 0
UJ X 0 0
~
0::
t-
80 x x x
)(
xo
x 0 TEMPERATURE 1 "C
U X 0
<t
a::
X
0 0
0 0 Fig. 34 Effect of Mn : S ratio on hot ductility of low-carbon steels
LL x 0
0
000
tested in tension at a strain rate of 100 S-l. (After Canard.125)
X 00
Xx
t-
<t: 8
76 x x 0
00
B 0
0 arises from several factors:
0:: X
UJ
t-
X
X
X
0 0
0
0
0 (a) decrease in chemical heterogeneity, which in turn
UJ 0
:L x 0 0 Q:)oo 0 decreases the amounts of non-equilibrium second phases
xX
<t: 0
0 x 0 that may be present;
00 0
72 x
l1..
0
x
0 0 0 (b) recrystallization leading to grain refinement, which in
00 0 0
z
0 0
turn gives higher ductility;
GROUP A GROUP B
}::
u (A I -TREATED) lSi - KILLED) (c) increase in directional heterogeneity of second-phase
;:) 0
0 x 0 particles, giving an increase in longitudinal ductility.
UJ 68
a:: 0
The first of these seems to be the predominant factor.
x o Inclusions
64 The deleterious effect of certain inclusions on hot ductility
o 0·005 0-01 0-015 0·02 0·025
OXYGEN 1 PER CENT.

is well known. However, the effects are very dependent upon
the inclusion composition and distribution, as well as the
[Courtesy Iron Steel Inst. stress system imposed during the working operation. The
Fig. 32 Effect of oxygen content upon hot ductility of wrought
carbon-manganese steels tested at 1100° C. (After Nicholson initiation of fracture at inclusions is dependent upon the
et al.121) deformation characteristics of the inclusion relative to the
60 of these internal cavities, leading ultimately tot"fracture.
t-= HEATED DIRECTLY TO However, as Cottingham198 has shown, the mode of initiation
Z TEST TEMPERATURE
UJ
U
••.••..•.•..•.. and rate of propagation are dependent on the stress system
applied to the material. When large hydrostatic tensile stresses
are present, as in a notched tensile test, rapid coalescence of
cavities occurs, leading to a marked effect on ductility.
The influence of sulphur, oxygen, and manganese on the
ductility of ferrous materials can be rationalized in terms of
the change in nature, shape, and distribution of inclusions
with change in composition. 68,167 ,198-203Fig. 32 shows the
~
UJ influence of oxygen content on tensile ductility of carbon-
I-
UJ
~ 20
manganese steels.121 The marked effect of deoxidation
(5 practice is clearly seen; the Group- B steels contain silicate
inclusions, while the Group-A steels contain complex silicate/
z
o
alumina inclusions. Increasing the volume fraction of inclu-
B::;l sions leads to lower ductility, since coalescence of cavities is
o
UJ easier.l38 This is illustrated in Fig. 33 for oxide and sulphide
inclusions in low-carbon steels tested in notched tension.l98
c:::
o 1300 1500
900 1100
Since the sulphur content is relatively uniform within a
TEMPERATURE, •C
cast, the major reduction in ductility with increased volume
Fig. 35 Influence of prior overheating on hot ductility of a 17 % fraction is attributable to oxides. Marked increases in duc-
chromium-II % nickel-2 % molybdenum steel containing 0·6 % tility resulting from removal of oxides from iron by zone
Cu and 0·3 % Ti. Tested in tension at a strain rate of 5 S-l.
(After Castro and Poussardin.170) refining have been reported by Keane et al.68,167 Increased
sulphur contents also give rise to reduced ductility in low-
matrix. If a material containing inclusions is plastically de- carbon steels.l99,200At lower sulphur contents this is pre-
formed, one or more of the following cases will arise: sumably due to the increase in volume fraction of sulphides
(a) the inclusion will fracture; but at high sulphur contents the formation of low-melting-
(b) the inclusion will yield and deform plastically, main- point iron sulphide films at grain boundaries can lead to
taining continuity with the matrix; 'hot-shortness'. 201 The addition of manganese improves
(c) the matrix will partially tear away from the inclusion ductiJity,125,201,202 presumably by altering the nature of the
producing a cavity. sulphide inclusions. As shown in Fig. 34, a Mn : S ratio of
The effect of inclusion composition on deformability at > 4 : 1 is needed to obtain adequate ductility in the critical
high temperatures has been documented by Malkiewicz and temperature range of 950-1150° C (1223-1423 K).l25 The
Rudnik,189 who defined an index of deformability vas: beneficial effect of manganese is reduced by an increase in
the oxygen content of steels and, in the case of rimmed steels,
2 In (b/a) Scholz and Dautzenberg203 report that hot-shortness is
... (34)
v = 3 In (Ao/A1)
. . [(%S) + (%O)J .
developed If the ratIo (%Mn)' IS > 0·8, where (%S)
where a and b are the lengths of the minor and major axes of
the deformed inclusion, Ao is the initial cross-sectional area and (%0) are the sulphur and oxygen contents of the rim.
of the ingot, and At is the final cross-section of the product.
The value of v can change from 0 for undeformed inclusions Reheating Practice
to 1 for inclusions that deform equally with the matrix, to With single-phase metals and alloys, the reheating practice
> 1 for inclusions that deform more than the matrix. Fig. 31 is relatively unimportant, provided that the material is heated
shows the temperature-dependence of the index of deforma- uniformly throughout and that excessive grain growth and
bility for some inclusions in stee1.190The constitution and oxidation do not occur. However, in complex materials, and
size of inclusions in steel are very dependent upon deoxidation particularly with cast m1.terials, the reheating practice can
practice and, in some cases, multiphase inclusions are formed. have a marked effect on ductility.
Such variations lead to different results being reported by Thus, in alloy steels containing nlassive carbide or nitride
different workers.l91-196 If the requirement is for improved phases, the prolonged heating at high temperatures required
workability under severe stress conditions such as hot for solution of these phases leads to excessive grain growth
piercing, the deformability of inclusions is of critical impor- and reduced ductility during subsequent hot working. A
tance197,198and hot working must be carried out in the tem- further complication is that, during deformation, reprecipi-
perature range where the inclusion can deform plastically to tation of these phases can occur, leading also to decreased
maintain continuity with the matrix. ductility. This reprecipitation is more marked for high
The differences in the deformabilities of sulphides, silicates, degrees of supersaturation and is therefore greater for higher
and oxides shown in Fig. 31 are reflected in the mode of solution temperatures170 (Fig. 35). Plant practice is to strike
initiation of fracture. Thus, for sulphides, which fall in case a balance by heating such materials to as high a temperature
(b), cavity formation is limited to the inclusion/matrix inter- as possible to give lowest possible working loads and yet
face and the initial cavity is much smaller than the inclusion. ensure a minimum loss of ductility.
In the case of silicates, whose deformability is temperature- The correct combination of reheating temperature and
dependent, case (a) occurs at lower temperatures and case soaking time is important in carbon and low-alloy steels
(c) at higher temperatures, leading in both situations to containing aluminium and nitrogen, since the presence of
cavity sizes equal to and generally greater than the inclusion. AlN out of solution leads to forging difficulties. Sidey204
For oxides, which fall in case (a), the cavities are generally notes that cracking of steels containing AIN can be prevented
larger than for silicates. Propagation occurs by coalescence only by working at strain rates 2-5 orders of magnitude greater
20
AI AND N IN ~EASING AI N Ugine for alloy steels.170,178Soaking conditions are simulated
SOLUTION I
AT '-20-;:O'-C---J-
by overheating for a short period at a very high temperature
llJ
gj 15
~-
u
«
a::
l.L x
r
l
4r\
I
)(
INCREASING
AT 1'OO'C
INCREA,SING
AT 1000' C
Al
AI N
~ x 1200' C
~
2=0
3
4
1b%g: ~
=
=
0
0·0021
0·0024
j % N as Al N
and specimens are then tested at a lower temperature similar
to that of reheating. The short period at the very high tempera-
ture is considered to be equivalent to a long time at a lower
temperature. In general the ductility/temperature curves for
such overheated steels lie below those for the steels heated
z o 0 5 =O· 0036
directly to test temperature and the temperature range of
=< . 6 =O' 0074 maximum ductility is shifted to higher temperatures. This
0:: 7 ;: O· 0076
I-
(J) o effect can be very marked if pronounced carbide reprecipi-
I-
Z tation occurs, as shown in Fig. 35. A similar technique
~
~ ~-~f.
should be applicable to non-ferrous materials but has not
~ been reported in the literature.
:5
a
w Reheating practice can exert a marked influence on surface
o
o 4
! I
68\
, \
quality.171,208-213For example, oxidation may cause roughen-
12 14
Ks (AI sol x N sol) x 10-4
ing of the surface, grain-boundary oxide penetration, de-
carburization, or enrichment in residual elements. Control
Fig. 36 Hot-torsion test results on forged En 36 (3'5 % Ni, 1 % of surface quality may become an overriding consideration
Cr) steel with increasing aluminium content, showing the varia-
tion of ductility with solubility product at K~ various tempera- in determining the reheating practice. The effects of reheating
tures. Tested at an equivalent strain rate of 0·13 S-1. (After practice on surface ductility have been studied using thin
Erasmus.205) tensile specimens to simulate surface layers212 but the best
means of discriminating between differing reheating practices
is the bend test.
than those satisfactory for AI-free steels; presumably, this pre- An important practical problem arising from incorrect re-
vents cooling, and hence precipitation, during working. heating practice is surface enrichment in, residual elements
Fig. 36 shows the marked effect of AIN out of solution on such as copper and tin in carbon steels due to preferential oxi-
the ductility of an En 36 (3'5% Ni, 1% Cr) stee1.205These dation of iron.21l,214,215.Eventually, the solubility limit of
effects are greater in such low-alloy steels than in carbon steels, these elements in austenite is exceeded and precipitation occurs
since the solution temperature for AIN is raised by the as a liquid phase that penetrates down the grain boundaries,
presence of nickel. Further work by Erasmus206 showed that causing surface break-up and a characteristic 'spilly' defect on
these problems could be overcome by using VN for grain- the final product. Hewitt and Meadows211 have described how
size control in low-alloy forgings. control of reheating practice can minimize this type of defect.
Reheating problems arising from chemical inhomogeneity The addition of nickel can also help to reduce the severity of
of aluminium alloys have been discussed by Kasz and Varley. this defect in carbon steels, though the mechanism has not yet
207In commercially pure aluminium slabs cast by the semi- been resolved. One theory is that nickel increases the solu-
continuous process, solidification occurs so that an excess bility of copper in austenite214-216 and another is that the
of silicon is retained in solid solution and the peritectic alloying of nickel with the copper-rich phase raises its melting
reactions resulting in the conversion of FeAl3 to cx(Fe-Si) point above the oxidation temperature. Recent work217
and then to ~(Fe-Si) proceed to a relatively small extent. indicates that the situation is a complex one in which
Thus, on reheating for rolling and also during the actual copper-enriched regions are occluded into the scale layer.
hot-rolling operation, silicon is precipitated and much of The nature of the atmosphere is important in this process,
the tough FeAlz and cx(Fe-Si) is converted to the more since Rolls and Preece218 have shown that nickel in carbon
brittle ~(Fe-Si) phase, leading to fabrication problems. steels can lead to surface defects following reheating in a
In aluminium-11% manganese alloys, after semi-continu- high-sulphur atmosphere, owing to the formation of nickel
ous casting, an excess of manganese is retained in solid sulphide and its subsequent penetration of the grain
solution. Though some precipitation takes place during boundaries. Similar considerations apply to reheating of
normal reheating, an additional treatment of 12 h at I'V
nickel alloys. 213
0
570 C (843 K) is required before rolling to reduce fabrication
difficulties. With other alloys, notably aluminium-magnesium,
homogenization may also be desirable to remove coring and Acknowledgements
improve ductility. In these cases, the wide range of solid This review was written during a year's leave of absence of
solution at the eutectic temperature lelds to marked grain- one of the authors (C.M.S.) from the University of Sheffield.
boundary segregation. This period was spent at the BHP Melbourne Research
The difficulty of satisfactory laboratory simulation of long Laboratories and the authors wish to acknowledge the co-
reheating times arises partly from oxidation problems and operation of the University and the financial support from
partly from the large scale of segregation in practice. One The Broken Hill Proprietary Co. Ltd., that made possible
technique which overcomes this difficulty is that used at this interchange between academic and industrial research.
References
1. J. J. Jonas, C. M. Sellars, and W. J. MeG. Tegart, Met. Rev., 6. V. Paur, Hutnicki Listy, 1965, 20, (5), 331; (trans. BISITS
1969, 14, (130), 1. No. 5814).
2. A. Nicholson, Iron and Steel, 1964,37, 290, 363. 7. H. J. Henning and F. W. Boulger, 'Mechanical Working of
3. P. A. Portevin, Rev. Mh., 1962, 59, 915, 1043. Steel l' (Met. Soc. Conf. 21), p. 107. 1964: New York
4. H. P. Stiiwe, Z. Metal/kunde, 1965, 56, 633. (Gordon and Breach).
5. P. Moore, 'Deformation under Hot-Working Conditions'
(Special Rep. No. 108), p. 103. 1968: London (Iron Steel 8. J. L. Robbins, H. Wagenaar, O. C. Shepard, and O. D. Sherby,
Inst.). J. Mater., 1967,2,271.
9. G. Wallquist and J. C. Carlen, Jernkontorets Ann .• 1959, 63. G. A. Redfern and C. M. Sellars, ibid., 1970,208,576.
143, (1), 1; (trans. BISITS No. 1495). 64. C. Rossard and P. Blain, Rev. Met., 1958, 55, 573.
10. H. J. McQueen, J. Metals, 1968,20,31. 65. W. Johnson and H. Kudo, Internat. J. Mech. Sci., 1960, 1, 175.
11. H. J. McQueen and J. J. Jonas, paper presented at AIME 66. F. Hollander, 'Mathematical Models in Metallurgical Process
Fall Meeting, Session on 'The Relation between Theory and Development' (Special Rep. No. 123), p. 46.1969: London
Practice in Metal Forming' (Cleveland, Oct., 1970). (Iron Steel Inst.).
12. G. P. Contractor and W. A. Morgan, Metal Treatment, 1959, 67. B. F. Bradley, W. A. Cockett, and D. Peel, ibid., p. 79.
26,65, 107. 68. D. M. Keane, C. M. Sellars, and W. J. MeG. Tegart, Ref. 5,
13. A. Nadai and M. J. Manjoine, Proc. Amer. Soc. Test. Mat., p.21.
1940,40,822; J. Appl. Mech., 1941,8, A77. 69. P. Ludwik, 'Elemente der technologischen Mechanik'. 1909:
14. C. Rossard and P. Blain, Mem. Sci. Rev. Met., 1959, 56, 285. Berlin (Springer).
15. G. W. Wallquist, Jernkontorets Ann., 1962, 146, (2),53; (trans. 70. C. M. Sellars and W. J. MeG. Tegart, Mem. Sci. Rev. Met.,
BISITS No. 4028). 1966, 63, 731.
16. K. Inoue, Tetsu-to-Hagane Overseas, 1962,2, (2), 62. 71. O. Schulte Forschungsber Land. Nordrhein-Westfalen,
17. C. Rossard, Rev. Mh., 1966, 63, 225. 1966, 1190; (trans. BISITS 6629).
18. N. Loizou and R. B. Sims, J. Mech. Phys. Solids. 1953,1,234. 72. S. Sakui, T. Nakamura, et al., Tetsu-to-Hagane, 1962,48,28,
19. R. R. Arnold and R. J. Parker, J. Inst. Metals, 1959-60, 88, 1441; 1963,49,996.
255. 73. L. D. Sokolov, Russ. Met. and Min., 1963, (3), 93.
20. J. F. Alder and V. A. Phillips, ibid., 1954-55, 83, 80. 74. L. D. Sokolov, Izvest. VUZ Chernaya Met., 1963, 6 (8), 93;
J
21. P. M. Cook, 'Conference on Properties of Materials at High (10), 62.

Rates of Strain,' p. 86, 1957: London (Inst. Mech. Eng.) 75. L. D. Sokolov, ibid., 1964, 7, (10), 59; (trans. Internat. Nickel
22. J. E. Hockett, Proc. Amer. Soc. Test. Mat .• 1959, 59, 1309. Co. (Mond), Ltd., M. 5607).
23. J. E. Hockett, Trans. Met. Soc. AIME, 1967, 239, 969. 76. L. D. Sokolov, ibid., 1965, 8, (4), 91; (trans. Henry Brutcher
24. O. Kienzle and H. BUhler, Z. Metallkunde, 1964,55,668. No. 6685).
25. H. BUhler and H. W. Wagener, Bander, Bleche. Rohre, 1965, 77. J. J. Jonas, H. J. McQueen, and W. A. Wong, Ref. 5, p. 49.
6,625, 667, 677. 78. W. A. Wong and J. J. Jonas, Trans. Met. Soc. AIME, 1968,
26. H. Buhler and H. W. Wagener, ibid, 1966, 7,467. 242,2271.
27. H. Buhler and H. W. Wagener, ibid., p. 648. 79. R. Bromley, Ph.D. Thesis, Univ. Sheffield, 1969.
28. H. BUhler and H. W. Wagener, Z. Metallkunde, 1966,57, 825 80. J. J. Jonas and J. P. Immarigeon, Z. Metallkunde, 1969, 60,
29. H. BUhler and H. W. Wagener, ibid., 1967,58, 136. 227.
30. J. A. Bailey and A. R. E. Singer, J. Inst. Metals, 1963-64, 81. C. Zener and J. Hollomon, J. Appl. Phys .• 1944, 39, 163.
92, 288, 404. 82 C. W. MacGregor and J. C. Fisher, Trans. Amer. Soc. Mech.
31. J. A. Bailey and A. R. E. Singer, ibid., p. 378. Eng., 1945,67, A217; 1946,68, All.
32. G. R. Dunstan and R. W. Evans, Metallurgia, 1969, 79. 96; 83. M. J. Luton and C. M. Sellars, Acta Met., 1969, 17, 1033.
see also J. Inst. Metals, 1971, 99,4. 84. O. Hoffman and G. Sachs, 'Introduction to the Theory of
33. G. Wallquist, Jernkontorets Ann., 1961, 145, (5), 231; (trans. Plasticity for Engineers', p. 45. 1953: New York and Lon-
BISITS No. 5810). don (McGraw-Hill).
34. J. L. Uvira and J. J. Jonas, Trans. Met. Soc. AIME, 1968, 85. H. W. Swift, J. Mech. Phys. Solids, 1952, 1, 1.
242, 1619. 86. W. Johnson and P. B. Mellor, 'Plasticity for Mechanical,
35. S. K. Samanta, Ref. 5, p. 122. Engineers' pp. 15, 194. 1962: New York and London (Van
35(a) S. K. Samanta, J. Mech. Phys. Solids, 1971,19,117. Nostrand).
36. S. K. Samanta, Internat. J. Mech. Sci., 1968, 10, 614. 87. E. Voce, J. Inst. Metals, 1948, 74, 536, 760; Engineer, 1953
37. K. Fink, W. Lueg and G. Burger, Arch. Eisenhiittenwesen, 195, 53; Metallurgia, 1953, 51, 219.
1955,26,655. 88. J. P. Sah, G. J. Richardson, and C. M. Sellars, J. Australian
38. 1. Gokyu and J. Kihara, Trans. Iron Steel Inst. Japan, 1966, Inst. Metals, 1969, 14, 292.
6, (4), 196. 89. G. Glover and C. M. Sellars, to be published.
39. R. A. C. Slater, W. Johnson, and S. Y. Aku, Internat. J. 90. J. H. Hollomon and J. D. Lubahn, Phys. Rev., 1946, 70, 755.
Mech. Sci., 1968, 10, 169. 91. J. D. Lubahn, J. Appl. Mech., 1947, 14, A229.
40. W. Schroeder and D. A. Webster, J. Appl. Mech., 1949, 16, 92. N. Perrone, ibid., 1966,33,210.
289. 93. G. E. Dieter, 'Mechanical Metallurgy'. 1961: New York and
41. J. M. Alexander, J. Mech. Phys. Solids, 1955, 3, 323. London (McGraw-Hill).
42. T. Takahashi and J. M. Alexander, J. Inst. Metals, 1961-62, 94. W. Johnson and H. Kudo, 'The Mechanics of Metal Extrusion'.
90,72. 1962: Manchester (Univ. Press). ,
43. M. Cook and E. C. Larke, ibid., 1945, 71, 371. 95. E. C. Larke, 'The Rolling of Strip, Sheet, and Plate', 2nd edn.
44. C. Rossard and P. Blain, Rev. Met., 1962, 59, 223; 'Flat- 1963: London (Chapman and Hall).
Rolled Products III, (Met. Soc. Conf. 16), p. 3. 1962: New 96. G. W. Rowe, 'An Introduction to the Principles of Metal-
York (Gordon and Breach). working'. 1965: London (Edward Arnold).
45. M. M. Farag, C. M. Sellars, and W. J. MeG. Tegart, Ref. 5, 97. P. M. Cook and A. W. McCrum, 'The Calculation of Load
p.60. and Torque in Hot Flat Rolling'. 1958, Supp!. 1960:
46. T. B. Vaughan, Ref. 5, p. 68. London (Brit. Iron Steel Research Assoc.).
47. J. Cotner and W. J. MeG. Tegart, J. Inst. Metals, 1969, 97, 98. Z. Wusatowski, Neue Hiitte, 1958, 3, 494; J. Iron Steel Inst.,
73. 1966, 204, 727.
48. F. A. Hodierne, ibid., 1962-63,91,267. 99. V. Hasek, Tech. Digest. 1967,6,372.
49. D. Hardwick and W. J. MeG. Tegart, Mem. Sci. Rev. Met., 100. C. Rossard, Rev. Met., 1968, 65, 181; (trans. BISITS No.
1961,58,869; (trans. BISITS No. 2995). 643~).
50. H. W. Wagenaar, Ref. 5, p. 38. 101. J. H. Decroix, A. M. Neveu, and R. J. Castro, Ref. 5, p. 135.
51. D. E. R. Hughes, J. Iron SteeIInst .• 1952, 170,214. 102. M. D. Coward, PhD. Thesis, Univ. Sheffield, 1970.
52. A. Guessier and R. J. Castro, Rev. Met., 1958, 55, 1023. 103. C. Rossard and P. Blain, Mem. Sci. Rev. Met., 1960, 57, 173.
53. F. Morozumi, Tetsu-to-Hagane, 1961, 47, 1457. 104. S. Gorczyca, ibid., 1960, 57, 153.
54. J. Langerweger and H. Trenkler, Berg-u. Huttenmann 105. R. Leguet, D. Whitwham, and J. Herenguel, ibid., 1962, 59,
Monatsh, 1967, 112,20; (trans. BISITS No. 5593). 649.
55. T. L. F. Muller, Ph.D. Thesis, Univ. Sheffield, 1967. 106. A. T. English and W. A. Backofen, Trans. Met. Soc. AIME,
56. T. L. F. Muller, Ref. 5, p. 131. 1964, 230, 396.
57. D. S. Fields and W. A. Backofen, Proc. Amer. Soc. Test. 107. R. A. Grange, 'Fundamentals of Deformation Processing'
Mat., 1957,57, 1259. p. 299. 1964: Syracuse (Univ. Press).
58. P. J. Regenet and H. P. Stuwe, Z. Metallkunde, 1963, 54,275. 108. A. P. Gulyayev and A. S. Shigarev, Phys. Metals Metallogr.,
59. G. J. Richardson and H. P. Stiiwe, Ref. 5, p. 131. 1964, 18, (2), 73.
60. A. Nadai, 'Plasticity', p. 216. 1931: New York and London 109. A. Younger and M. G. Cockcroft, Metal Treatment, 1964,31,
(McGraw-Hill). 296.
61. M. Itihara, Tohoku Univ. Technol. Rep., 1936-38, 12, 63. 110. J. L. Uvira, PhD Thesis, McGill Univ., 1967.
62. J. L. Robbins, O. C. Shepard, and O. D. Sherby, J. Iron Steel 111. G. A. Wilber, J. R. Bell, J. H. Bucher, and W. J. Childs,
Inst., 1964,202,804. Trans. Met. Soc. AIME, 1968, 242, 2305.
112. G. Glover, PhD Thesis, Univ. Sheffield, 1969. 164. E .M. Savitsky, 'The Influence of Temperature on the Mechani-
113. H. Borchers, R. Werner, and H. Zeilinger, Arch. Eisenhiitten
M
cal Properties of Metals and Alloys' (trans. edited by
wesen, 1969,40,437. O. D. Sherby). 1962: Stanford (Univ. Press).
114. H. J. McQueen, Proc. Internat. Conf. on 'The Strength of 165. G. P. Lewis and W. J. MeG. Tegart, J. Inst. Metals, 1963-64,
Metals and Alloys' (Tokyo, 1967), Trans. Japan Inst. 92,249.
Metals, 1968, 9, Suppl., 170. 166. J. Robbins, O. C. Shephard, and O. D. Sherby, Trans. Amer.
115. R. Priestner, C. C. Earley, and J. H. Rendall, J. Iron Steel Soc. Metals, 1967, 60,205.
Inst., 1968, 206, 1262. 167. D. M. Keane, C. M. Sellars, and W. J. MeG. Tegart, Mem.
116. J. D. Baird and R. R. Preston, unpublished report presented Sci. Rev. Met., 1968, 65, 255.
at Conference on 'Hot Workability of Steel'. 1969: London 168. O. D. Sherby, private communication.
(Iron Steel Inst.). 169. M. M. Farag, PhD Thesis, Univ. Sheffield, 1964.
117. D. J. Latham, PhD Thesis, Univ. Birmingham, 1963. 170. R. Castro and R. Poussardin, Centre Doc. Siderurgique, Circ.
U8. D. J. Latham and M. J. Cockcroft, Nat. Eng. Lab. Rep. d'Information Technique, 1962, 19, (2), 505.
(216), 1966. 171. A. Nicholson and T. W. Moore, Ref. 119, p. 46.
119. W. J. MeG. Tegart, 'Reheating for Hot Working' (Special 172. K. A. Reynolds, Ref. 5, p. 107.
Rep. No. 111), p. 26. 1968; London (Iron Steel Inst.). 173. D. M. Ward and B. J. Helliwell, J. Inst. Metals, 1970, 98, 199.
120. M. Zidek, Sbornik VSB Ostrave (Hutnicka), 1965, 11, 687. 174. G. P. Lewis, M. Met. Thesis, Univ. Sheffield, 1963.
121. A. Nicholson, D. Smith, and P. Shaw, Ref. 5, p. 161. 175. P. A. Portevin and C. Rossard, Rev. Met., 1968,65, 131.
122. W. J. MeG. Tegart, 'Ductility', p. 133, 1968: Metals Park, 176. T Yamane and G. Mirna, Tetsu to Hagane, 1959, 45, 21.
Ohio (Amer. Soc. Metals). 177. B. F. Vakhtanov, Kuznechno-Shtampov, 1963, (9), 9.
123. M. J. Luton and W. J. MeG. Tegart, Metal Sci. J., 1969, 3, 178. J. H. Decroix, A. M. Neveu, and R. J. Castro, Ref. 5, p. 135.
142. 179. B. Crawshaw and A. D. Choudhury, 'Solidification of Metals'
124. F. E. White and C. Rossard, Ref. 5, p. 14. (Publ. No. 110), p. 377. 1967: London (Iron Steel Inst.)
125. R. Canard, Rev. Mit., 1946,43, 156. 180. J. Bellot and M. Hugo, Rev. Met., 1966, 63, 7.
126. E. F. Nippes and W. Savage, Weld. J. Research Suppl., 1955, 181. J. W. Pridgeon, M. L. Pochon, R. T. Gross, and V. T. Sharma,
34, 183s. 'Vacuum Metallurgy Conference' (Miami Beach, 1968)
127. J. Dauvergne, M. Pelabon, and J. Ivernel, Rev. Met., 1955, p.525. 1968: New York (Amer. Vacuum Soc).
52,85. 182. E. W. Kelley, Rep. US Air Force Mat. Lab. (-TR-69-114),
128. C. D. Harnish, L. P. Myers, and S. E. Tyson, Metal Progr., 1969.
1965,87, (May), 86. 183. C. F. Elliott, D. C. Evans, R. L. Kennedy, and C. R. Manning,
129. A. Gittins, C. M. Sellars, and L. G. Northway, BHP Tech. Metals Eng. Quart. 1970, 10, (2), 46.
Bull., 1970, 14, 6; J. Mat. (in the press). 184. W. E. Duckworth and G. Hoyle, 'Electro-Slag Refining'.
130. M",,! G. Cockcroft and D. J. Latham, J. Inst. Metals, 1968, 96, 1969: London (Chapman and Hall).
33. 185. A. Gueussier and R. Castro, Metal Treatment, 1959,26, 323,
131. F. E. White, Mem. Sci. Rev. Met., 1966, 63, 991. 361.
132. J. L. Robbins, O. D. Sherby, and O. C. Shepard, J. Iron Steel 186. E. F. Nippes, W. F. Savage, and G. Grotke, Weld. Research
Inst., 1961"199, 175. Council Bull., 1957, 33, (Feb.).
133. R. A. Reynolds and W, J. MeG. Tegart, ibid., 1962,200, 1042. 187. J. D. Rawson, Ref. 119, p. 65.
134. G. E. Dieter, J. V. Mullin, and E. Shapiro, Ref. 5, p. 7. 188. E. A. Leech, P. Gregory, and R. Eborall, J. Inst. Metals,
135. I. Dragan, MSc Thesis, Univ. Aston in Birmingham, 1967. 1954-55, 83, 343.
136. E. Shapiro and G. E. Dieter, Met. Trans., 1970, 1, 1711. 189. T. Malkiewicz and S. Rudnik, J. Iron Steel Inst., 1963, 201,
137. D. E. R. Hughes, J. Inst. Metals, 1954-55,83,548. 33.
138. J. Langerweger and H. Trenkler, Berg.-it. Hiittenmann. 190. R. Kiessling, 'Non-Metallic Inclusions in Steel, Part III'
Monatsh., 1968, 113, 104. (Publ. No. 115). 1968: London (Iron Steel Inst.).
139. I. Finnie and W. R. Heller, 'Creep of Engineering Materials', 191. M. Wahlster, Freiberger Forschungsh. Met., 1962, [B],(62),47.
p. 168. 1959: New York and London (McGraw-Hill). 192. F. B. Pickering, J. Iron Steel Inst., 1958, 188, 148.
140. A. E. Johnson, Met. Rev., 1960, 5, (51), 447. 193. S. Rudnik, ibid., 1966,204,374.
141. A. E. Johnson, J. Henderson, and B. Khan, 'Complex Stress 194. P. J. H. Maunder and J. A. Charles, ibid., 1968, 206, 705.
Creep, Relaxation, and Fracture of Metallic Alloys'. 1962: 195. I. Uchiyama and M. Sumita, Trans. Nat. Research Inst.
Edinburgh, (HMSO). Japan, 1965, 7, 233.
142. A. E. Johnson and B. Khan, Nat. Eng. Lab. Rep. (108), 1965. 196. J. A. Charles and 1. Uchiyama, J. Iron Steel Inst., 1969, 207,
143. M. J. Manjoine, 'Fracture of Engineering Materials', p. 73. 979.
1964: Metals Park, Ohio (Anler. Soc. Metals). 197. E. J. Paliwoda and W. Wilson, 'Mechanical Working of Steel
144. 1. A. Oding, V. S. Ivanova, V. V. Burduksky, and V. N. I' (Met. Soc. Conf. Vol. 21), p. 126. 1964: New York
Geminov, 'Creep and Stress Relaxation in Metals' (English (Gordon and Breach).
trans.), p. 145. 1965: Edinburgh and London (Oliver and 198. D. M. Cottingham, Ref. 5, p. 145.
Boyd). 199. S. Y. Ogawa, T. B. King, and N. J. Grant, Trans. NIet. Soc.
145. E. Siebel, Steel, 1933, 93, (17), 37. AIME, 1962,224, 12.
146. W. H. Wells, Trans. Amer. Soc. Metals, 1940,28,424. 200. C. T. Anderson, R. W. Kimball, and F. R. Cattoil', Trans.
147. P. W. Bridgman. 'Studies in Large Plastic Flow and Fracture'. Amer. Inst. Min. Met. Eng., 1953, 197, 525.
1952: New York and London (McGraw-Hill). C. T. Anderson, V. V. Donaldson, R. W. Kimball, and F. R.
148. H. Ll. D. Pugh, Metal Treatment, 1960,27, 189. Cattoir, ibid., 1954, 200, 835.
149. H. Ll. D. Pugh, and A. H. Low, J. Inst. Metals, 1964-65,93, 201. A. Josefsson, J. Koenemann, and G. Lagerberg, J. Iron Steel
201. Inst., 1959, 191, 240.
150. W. J. MeG. Tegart, 'Elements of Mechanical Metallurgy'. 202. M. Endrle, Hutnicke Listy, 1967, 22, 28.
1967: New York, &c. (Macmillan), 203. W. Scholz and N. Dautzenberg, Arch. Eisenhuttenwesen,
151. J. Datsko and C. T. Young, Trans. Amer. Soc. Mech. Eng., 1967, 38, 915.
1960, (B), 82, 309. 204. M. P. Sidey, Iron and Steel, 1967, (May), 168; (June), 269.
152. S. I. Gubkin, 'Deformability of Metals': 1955: Moscow 205. L. A. Erasmus, J. Iron Steel Inst., 1964,202, 32.
(Izdatel. Akad. Nauk SSSR). 206. L. A. Erasmus, ibid., p. 128.
153. F. Garofalo, 'Fundamentals of Creep and Creep Rupture in 207. F. Kasz and P. C. Varley, J. Inst. Metals, 1949-50, 67, 405.
Metals', 1965: New York, &c. (Macmillan). 208. K. Sachs and C. W. Tuck, Ref. 119, p. 1.
154. A. Gittins, private communication. 209. P. Hancock, ibid., p. 18.
155. A. N. Stroh, Advances in Physics, 1957, 6,418. 210. K. Sachs and G. T. F. Jay, ibid., p. 36.
156. D. McLean, J. Inst. Metals, 1956-57,85,468. 211. J. Hewitt and J. Meadows, ibid., p. 55.
157. E. Smith and J. T. Barnby, Metal Sci. J., 1967, 1, 56. 212. D. G. Harris, ibid., p. 61.
158. J. A. Williams, Phil. Mag., 1967, 15, 1289. 213. W. Merrett and R. J. Courtney, ibid., p. 118.
159. J. S. Waddington, ibid., 1968, 17, 51. 214. D. A. Melford, J. Iron Steel Inst., 1962, 200, 290.
160. A. Gittins, J. Mater. Sci., 1970, 5, 223. 215. A. Nicholson and J. D. Murray, ibid., 1965,203, 1007.
161. A. Gittins and C. M. Sellars, Metal Sci. J. to be published. 216. W. J. M. Salter, ibid., 1966, 204, 478.
162. M. Cook and E. Davis, J. Inst. Metals, 1949-50, 76, 501. 217. G. L. Fisher, ibid., 1969, 207, 1010.
163. O. D. Sherby, Metals Eng. Quart., 1962, 2, (2),3. 218. R. Rolls and A. Preece, Metal Treatment 1960, 27, 139.
© The Institute of Metals. 1972.

Hot Workability: Review

Uploaded by

Document Informationclick to expand document information1

Document Informationclick to expand document information

Copyright:

Available Formats

Hot Workability: Review

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Hot Workability: Review

Uploaded by

Copyright:

Available Formats

Review 158

where T m is the melting temperature in degrees Kelvin) and 100 t sec-'

and structure under hot-working conditions was examined F!NISHING

conditions, e.g. forging or rolling, where the incremental ",1050· C

determining optimum conditions for new materials, close

Measurement of Flow Stress W

rate of456 S-l. (After Samanta.35)

o 20 1a 6'0 SO Js 210 • in ' 6'0 9'0 sis

i ,.'5 -- TORSION(64) 200 -- TORSION (45)

.i! .....--.. 40 _.- TENSION (16)

EQUIVALENT TRUE STRAIN (E:) superpurity aluminium at 400 C. 0

._. TORSIO~14,64k = 1·1 sec-1

Fig. 4 Effect of composition on stress/strain curves for: (a) steels

TJTm = 0·7 ... (6)

A. 10 12 restoration process1 or on the occurrence of structural

0·2 0·4 0·6 0·8 1 4 8 10 0·2 0·4 0·6 0·8 1 4 8 10

the merit of simplicity, examination of Figs. 3 and 4, indi- ::>

in compression at constant true strain rate and constant cross-

/ DELAY TIME (/)

~ 20 , 2 20 MIN 900· 25 MIN BOO·

Fig. 10 Effect of delay time on stress/strain behaviour of commercial-

Fig. 11 Effect of delay time on restoration index (after Dunstan

WITH 10 sec REST E

that recrystallization takes place between passes during normal ~ 9 60

tures.46 On the other hand, copper and its alloys105,114and to-

recrystallize rapidly after deformation, so complete recry- ~

normal range of hot-working temperatures. 107 This will

on subsequent stress/strain behaviour at the same temperature

In this section we consider, first, test techniques and re-

and has been used for correlation of ductility data,122 as I-

ditions during practical hot-working operations where the

tudinal compressive stress (J A; EO is the strain at fracture in

Comparison of Parameters Derived from Strength and Rupture Data55,83,122,154

Strength Data Rupture Data

f Q"x(~)dO = constant ... (26)

TRUE PEAK STRESS, MN/m2

Fe3 C CONTENT AT 650·C, VOL.-%

mm (483K) (553 K) (623 K) , I

Material Parameters Affecting Ductility

the available data. Notable exceptions are the excellent

Crystal Structure and Allotropic Transformation (f)

tive to variations in impurity content68,112,132,133,166,167 but Fig. 24 Temperature-dependence of ductility of aluminium-mag-

which is particularly severe at '" 30CYo o-ferrite, since most

only 10-20% a-ferrite present problems in t.ot rolling and that

close to the ~-phase boundary, Reynolds172 found that the o

addition of small amounts of alloying elements had marked z B

content or reduced the minimum upsetting temperature by ~5

Most of the experimental results on hot workability have ::x:

900 1000 1100

Fig. 27 Effect of a-phase content on the ductility of leaded copper-

1050 1100 1150 1200 1250

0:: 60 Fig. 30. Effect of electroslag refining on ductility of as-cast 18 %

1·0 ~ ----, MnS

x x O·OS',.C 0·11".Mn , o-O'a",S

OXYGEN 1 PER CENT.

21. P. M. Cook, 'Conference on Properties of Materials at High (10), 62.

You might also like