Вы находитесь на странице: 1из 23

OF

NAllIIAI
ELSEVIER Mechanics of Materials 21 (1995) 1-23

The influence of material anisotropy on transformation induced


plasticity in steel subject to martensitic transformation
F.D. Fischer, S.M. Schl6gl
Institute of Mechanics and Christian DopplerLaboratory for Micromechanicsof Materials, Universityfor Mining and Metallurgy,
Franz Josef Strafle 18, A-8700 Leoben, Austria
Received 30 June 1994; revised version received 19 October 1994

Abstract

A simplified micromechanical model adapted to martensitic transformation is applied to describe transformation


induced plasticity (TRIP) in steel. The accommodation of the transformation volume and shape change is performed
by plastification considering Hill's anisotropic yield condition. Both a uniaxial and a triaxial Ioadstress state are
investigated. A generalized TRIP-term is developed in the sense of an "extended" Greenwood and Johnson (1965,
Proc. R. Soc. London A 283, 430-422) relation.

Keywords: Transformation induced plasticity; Martensitic transformation; Orthotropic plastic anisotropy; Microme-
chanical model; Unia~ial and triaxial loading

1. Introduction

1.1. Explanation o f transformation induced plasticity (TRIP)

The author of the only existing monography entirely devoted to T R I P (Mitter 1987) describes T R I P
as: "significantly increased plasticity during a phase change. Even under an externally applied load stress
state with the corresponding equivalent stress being small in relation to the " n o r m a l " yield stress of the
material, plastic deformations occur."
This softening has its origin in the fact that during a phase transformation of a certain part of a material
(let us say, a "microregion") this microregion may change its volume a n d / o r its shape. To achieve
compatibility between the neighbouring material and the transforming microregion under consideration
the misfit must be compensated (or accommodated) by an eigenstress state which may vary within a grain
of a polycrystalline material, but at least from grain to grain. In many cases (e.g. in the case of "classical"
transformations of ,;ted during quenching) the misfit leads at least to a plastification of the neighbouring
material of the microregions, sometimes even to a plastification of the microregions themselves. It can be
easily imagined that the development of this eigenstress state is influenced by an externally applied stress
state on a certain specimen. The superposition of these two stressing or straining " m e c h a n i s m s " may
initiate plastification, even under a low external stress level. Therefore, a macroscopic plastic deforma-
tion of the specimen can be observed. This mechanical effect associated with phase transformations has

0167-6636/95/$09.50 © 1995 Elsevier Science B.V. All rights reserved


SSDI 0167-6636(94)(10070-0
2 F.D. Fischer, S.M. Schli~gl/Mechanics of Materials 21 (1995) 1-23

originally been considered as TRIP. It should be mentioned that even in the case of no plastification but
only elastic accommodation of the strain-incompatibility a global deformation of the specimen may
remain if the transformed microrcgions arc arranged spatially in such a way that their shape changes add
up to a non-zero overall strain. In the case of stress-induced thermoelastic martensitic transformation
(e.g. shapc memory alloys) the externally applied stress thoroughly affects the orientation of the
microregions. In this case, too, the term T R I P has been adopted for the non-elastic deformation of a
specimen even though this deformation is generally reversible during the reverse transformation from the
product to the parent phase.
Greenwood and Johnson (1965) were among the first researchers presenting a quantitative relation
between a [oadstress "~3 externally applied on a tension specimen, the transformation volume change za
and the final longitudinal strain e:, 3 of the specimen being ez. 3 = ±A
3 +/~A~ 3. The T R I P strain kA,~ 3 (k
is a constant discussed later) is obviously proportional to 2' 3 which means that in the case of a load-free
specimen only a final strain component ±A 3 can be observed. The Greenwood and Johnson relation was
reexamined by Leblond and co-operators (Leblond et al. 1986a,b 1989; Leblond, 1989), Mitter (1987) and
Fischer (1990) and extended to a triaxial external loadstress 2. Finite element studies have also been
performed, see, e.g. Sj6str6m et al. (1992). These researchers c~me to the common conclusion that for a
transformation with only a volume change (as it is the case for a first-order diffusional transformation as
the austenite-pearlitc transformation in steels) the transformation strain tensor is proportional to the
product A_S with S being the deviator to _2". Fischer (1992) could show that this is even correct for a
martensitic-transf~rmation with a volume-change and an additional significant shape change of the
transforming microregions.

1.2. Problem definition

Almost all investigators on T R I P consider an isotropic material. However, in many practical applica-
tions materials are used which show a certain kind of anisotropy, e.g. a texture developed by forging and
rolling. Therefore, the question has arisen how anisotropy influences the final T R I P strain. The main
goals of this paper are:
- to investigate the local stress state in a martensitically transforming microregion under the assumption
of plastic anisotropy represented by one anisotropy parameter p;
- to simulate actual experiments for uniaxial and triaxial external loading;
- to present a relation between the T R I P strain, e-rR, and the deviator S of the externally applied stress
tensor as well as a set of parameters k l ( p ) , k3(~) reflecting the aniso~-ropy. This result can be seen as
an extended Greenwood and Johnson relation for martensitic transformation in an anisotropic
material and can be used in a constitutive material law for further practical applications, e.g. in
incremental form as an extra term in addition to the "classical" plastic strain increment.

1.3. Martensitic transformation

As a representative part of the material a mesodomain is considered. All material properties and the
temperature T are assumed to be spatially constant. Equal elastic properties of the parent phase and the
product phase are considered. The mesodomain is loaded by a homogeneous load stress state _~ which
remains constant during the phase transformation. The mesodomain is now divided into micr~regions
forming subunits. The progress of martensitic transformation is thought to consist of a successive
"switch" of one microregion after the other from the parent phase to a martensite variant. The
orientation of each variant is defined by a set of Eulerian angles 6, O, ~0 which generally vary from
microregion to microregion. The distribution of the variants can be described by a certain distribution
function g(s~; 0, 6, ~P), ~ is the martensite volume fraction. If a specific microregion were assumed to be
isolated and, therefore, not constrained by its neighbouring material, the transformation deformation can
F.D. Fischer, S.M. Schl6gl / Mechanics of Materials 21 (1995) 1-23 3

be described, due to the phenomenological theory of martensite, by a "free" transformation strain tensor
_e'¢ defined with respect to a local coordinate system x', y', z',
le'~=h(t/t~)~'¢, h(O)=O,]h(1)=l
0 0 1
(1)
o o o.
-- 1

The x'-y' plane is the habit plane being the common crystallographic plane between the parent and the
product phase, h(t/t~) represents a function of the dimensionless time ? = t/t~, t is the time, t¢ the
transformation time (which is very small), 3' is the transformation shear angle (of order 0.2 for steel) and
1 2
(of the order 0.04 for steel) corresponds to the transformation volume change A, 6 = A + ~[A + (1 +
A2) sin 2 3']. The Eulerian angle 0 is the angle between the global z and the local z' axis.
A standard coordinate transformation procedure by the orthogonal tensor Q,
[cosO,:osq~-sin~cosOsinq~ -cosq, sinq~-sin~kcosOcosq~ sinq~sinO ]
__Q= s i n $ . , . i n q ~ + c o s $ c o s O s i n q ~ -sin~Osinq~+cos$cosOcosq~ -cos$sinO ,
- sin 0 sin q~ sin 0 cos q~ cos 0
(2)
gives the tensor _e',: in the global coordinate system as
e__~= Q. e_'~"QT = h(t/Q)~, ~:~= a" ~_'~"Q'r. (3a)

e_c=h(t/t,)!!c, ~c = ~c - ½t~_/. (3b)


The deviatoric part e_-c of _~c possesses the following components according to Fischer (1992):
e¢.x = ½3'(sin 2~ sin 0 cos q~ - sin 2 ~b sin 20 sin q~) + 6(sin 2 q, sin 2 0 - 3),
~.y = - ½3'('.;in 2~ sin 0 cos ~p + cos 2 ~b sin 20 sin q~) + 6(cos z ~ sin 2 0 - ½),
- ~ = ½3'(sin 20 sin q~) + ~6(3
ec. 1 l + cos 20), (4)
~.x~ = 3,(cos ~ cos O cos ¢ - sin ~ cos 20 sin ~p) + 6 sin ~ sin 2 0 ,
~/¢.yz = 3'(sin q, cos 0 cos q~ + cos ~ cos 20 sin ~o) - 6 cos ~b sin 2 0 ,

Yc.xy = 3 ' ( - cos 2~ sin O cos q~ + ½sin 2q, sin 20 sin ~p) - 6 sin 2~ sin 2 0 .
The phenomenolcgical theory does not deal with any thermodynamic condition for the transformation of
a microregion tha~ is mostly in the form of a very thin plate. In the past few years considerable research
has been devoted to this topic, for a review see Fischer et al. (1994). Without going into details, a
condition has to be formulated which expresses that the sum of a chemical and a mechanical driving
force must prevail over a certain energy barrier. The newly formed martensitic thin plate takes now such
an orientation that the mechanical driving force is maximized. This was recognized by Patel and Cohen
(1953) with respect to the global load stress _Z on a single crystal and extended, e.g. by Marketz et al.
(1995), to a polycrystal with respect to the lo~zal stress state ft. Such an orientation effect can also be
observed experimentally, as reported by Gautier and Simon (1ff87). The transformation tensor represents
a local strain incompatibility. Therefore, it must be compensated by an accommodation process both in
the neighbouring material and the microregion under consideration. This leads to plastification in the
remaining parent phase as well as, up to a certain amount, in the martensitic phase. Orientation and
4 F.D. Fischer, S.M. Schl6gl/Mechanics of Materials 21 (1995) 1-23

accommodation are strongly coupled via the mechanical driving force as mentioned above. In a series of
papers, see, e.g. Lcblond et al. (1986a,b, 1989) and Leblond (1989) only the accommodation of the
volume change was treated. Mitter (1987) was the first who pointed to a plastic accommodation of the
transformation shear as well as to the orientation effect; for a detailed treatment, see Fischer (1990,
1992). Recently, some finite element studies by Ganghoffer et al. (1991) and Marketz et al. (1994) were
published that dealt with the coupling of orientation and accommodation. However, a kinetic equation
for ~: in relation to an external stress state -2 as well as to the temperature T still has to be found.
Obviously the most difficult task in such an approach is to apply a proper relation between _2, T and the
local stress state _,2. One way leading to such an equation has been indicated recently by Obe)-aigner et al.
(1993). Of coursC, the lack of such a kinetic equation prevents satisfying relations between ~: and _eXR
even though they have been reported in the literature. This paper is based, however, on the simplifying
assumption that a loaded specimen (mesodomain) is cooled so quickly that the whole material transforms
simultaneously from the parent to the product phase.

2. The micro-macromechanical model for TRIP in an anisotropic material

2.1. Assumptions

In the following context all assumptions and preconditions are summarized on which the derivation of
the T R I P strain term -~TR is based:
- A special kind of orThotropic plastic anisotropy, the so-called transverse isotropy, is assumed with the
yield stress R l in the longitudinal (z-) direction of the specimen and the yield stress Rq in the
transverse x - y plane, see Appendix A. Such an anisotropy can often be observed in longitudinal
specimens machined out of forged rods. For the sake of simplicity the same anisotropy is assumed
both in the parent and the product phase.
- All the austenitic microregions transform into martensite simultaneously. During this process the yield
stresses R l, Rq change from Rt,a, Rq, a to Rl,m, Rq,m, the label "a" is for austenite and the label " m "
for martensite. For simplification an average yield stress R [ , Rt. a <_R [ < Rt.m, is introduced, which
will be justified later.
- Only two types of g(~:; O, $, ~0) are considered:
(a) A uniform distribution of all variants, leading to g - 1.
1
(b) A uniform distribution with respect to O, $ with ~0 = + Err in case of a uniaxial tension loadstress
and ~ = - ½rr for compression leading to extreme values of ec.z, for details see Fischer (1990, 1992).
- The strain incompatibility due to _ec is compensated only by plastification. The elastic part of the total
strain tensor is ignored. Although~ignificant straining (y ~ 0.2!) takes place, a linear decomposition of
the total strain tensor e t is assumed,

~t = ~p + ~c + 38h/- (5a)

A dot represents differentiation with respect to the dimensionless time t'.


- A Taylor-Lin assumption is introduced stating that in each microregion the same final strain state _et
can be observed. This is, of course, a simplification but it led to rather good results in the case of a-n
isotropic material, see Mitter (1987), Fischer (1990). This assumption allows one to find finally simple
analytical relations for the T R I P strain tensor. Since _et is now spatially constant in the mesodomain,
the macroscopic T R I P tensor _eTR follows as
_ETR = _E"t -- 1 ~ ! . (5b)
F.D. Fischer, S.M. Schliigl / Mechanics of Materials 21 (1995) 1-23 5

The unknown entiqt now is the TRIP strain tensor era for which a differential equation will be derived
in the next section.

2.2. Derivation o f e-rn and e r r

The derivation of the above-mentioned differential equation is based on the introduction of -~0, see
(A.3), into relation (5) and then into the yield condition (A.2). Due to a global principal stress tensor ,~
(~3 in z-direction) as load stress tensor, the tensor -~TR has the following structure:

ETR,ij = EiSij ' ~ij Kronecker 8, el + e2 + e3 = 0. (6a)


Note that the label " T R " is skipped for the components of eTR.
It follows from (5) and Appendix A that
_ _ _ _ 2
_~o = ~TR-- ~¢ = , i ( s + Xp_), (Ta)
p is an anisotropy factor, - 0.75 _<p ~ 3.0, see Appendix A, ~i >_ 0, and with (7a) for the yield condition
(A.2)
31 [1 ]
Now the components of ~ must be expressed by strain rates using (7a),
1
s + ~ps = ~-

leading to
1 1
sx+2p(s,,-s,)= 7(6- ox1,
The solution of these two equations with respect to G, sy delivers
1
Sx= [ ( ~ . - ~¢,~)(1 + 2P) + ( ~ 2 - ~o.~)2P],
(1 +
1
sy = [(~2 - ~,y)(1 + 2P) + ( e , - ec.,)2P]. (7c)
(1 +
Using _ec : _e~ = ( 582 + ~y m finally leads to
I 2,L2

= 1 N~ 1 +P 4 p N 2 - 4(1 + ] (8a)

3 "2 +e~• + f32) - 3(~,~c. x +e2ec.y


N, = ~(e, • " + ~3~c.z)+ ( 8 2 + ~7
3 2';-2
}n , (85)

N2 = [~l- ~2--(ec,.t . -- ec,y)] 2 "]- '~c,xy'2 , (8C)

N3 = 7Z,x~
-2 + 7~,y~.
.2 (8d)
The stress deviator components in each individual microregion can now be expressed by only the three
6 F.D. Fischer, S.M. Schl6gl/Mechanics of Materials 21 (1995) 1-23

unknown components kl, ~2, ~3 of -ETR and the known components of ~¢ by inserting (7c) and (8a) into
(7a): - -
Sz e3 - ec,z
--= [ ]1/2, (9a)
R/ P N2- P I - N3"
N, l+3p 4(1+3p) j

sx 1 (g,-ec.x)( 1 + }P) + ( e 2 - ec",y)~P2


- - = - - [ p ]1/2' (9b)
Rl 1 + }p NI 1 +P 4p Nz 4(1 + ½p) N3

Sy 1 (gz2-O.c,y)(l + Z p ) + ( g , - ~ , ~ ) Z p (9c)

Rt 1 + 3P
( Nl l + 30
P N3]'J2
P- --N2 4(1 + 30)
Note that sx, sy, s z depend on the individual set of Eulerian angles of each microregion via the
components of _~c and that they vary from microregion to microregion! The shear components are not
written down exi~licitely.
The final differential equations for ~1, ~2, ~3 can now be found by considering the global equilibrium.
The averages of the deviator components s x, sy, s z must be equal to the global deviator components $1,
$2, S3 which can be calculated directly from the global stress tensor -X,
- S = X - 3('~1
~ + "~2 + "~3) -/" The
average is performed with respect to the Eulerian angles as

f=o=o f2= f2,'rsg sin O dO d(p dO


~=o~q,=oz
S = (s) = (10)
= _ f'~ f2'~f2~'gsinOdOdq~d~O
O=o ~ = o ~=0
The differential equations (9a)-(9c) are of a complicated nature since the integrals over the Eulerian
angles cannot be solved exactly. All terms ~ . . . . . . . ~'¢,yz are transcendent functions of O, ~p, ~b and appear
in the numerator as well as in the denominator of (9a)-(9c) together with the unknown deviator
components kl, ~2, ~3. However, the differential equations can be reduced to algebraic equations if the
same time dependence exists for the known components e¢.x. . . . . y¢,yz and the unknown TRIP strain
components e~, e 2, e 3. Since this is a reasonable assumption, as discussed in Fischer (1990), it is applied
in the following context,
ei=h(t/Q)ei(1 ), i = 1, 2, 3. (6b)
If the relations (3b) and (6b) are inserted into (9a)-(9c), the time derivative ~fit~Q) can be eliminated.
As the yield stress R t changes from R~,. to Rt,m, we implement an average yield stress Rf. This
assumption will be checked later. The first of the three final algebraic equations now follows

3Rf(2"~3- " ~ ' - ' ~ 2 ) = e3-~' ) 3 2 - +e3~,,~)+62+ 3 2

"
1 + 3P
[( el -- e2 -- (ec,x -- ~'c,y + Yc,xy

P 2 (lla)
4(1 + ½p) (y~2x~+ Y~'Yz

The other two equations are skipped for the sake of brevity.
F.D. Fischer, S.M. Schl6gl/ Mechanics of Materials 21 (1995) 1-23 7

Explicit relations for e~, e 2, e 3, e I + e 2 + e 3 = 0, depending on 2~, Z 2, '~3 c a n n o t be found. However,


by assuming a given set of e I, e 2, e 3 the left-hand side of (11) can be calculated by numerical integration
over 0, ~0, qJ. This leads to three deviator components of S which finally indicates that the inverse
problem can be solved numerically. The corresponding stress state Z is, of course, not defined in a
unique way since the hydrostatic part of the stress tensor does not appear in the final relations (1 la), etc.

3. Simulation of tests

In the following, dimensionless stress terms Z " = 2 / R ~ are used. For all further numerical calcula-
tions, 6 = 0.04 and y = 0.2 are used being typical data for a F e - N i martensite.

3.1. The tension / compression test

3.1.1. Numerical tests


* * ]
A uniaxial loading ,~3@0 (2; = , ~ 3 / R t ), 2 : 1 = 0 leads with E T a = e 3 , e I = e 2 = - - 2 e 3 and g = 1
(uniform distribution) to a set of curves depicted in Fig. 1. A not too remarkable difference between the
curves with varying parameters t9 can be observed. It is interesting to note that a significant, almost
linear part of the e T a - - ~ ° exists for the whole range of p (as predicted 30 years ago for an isotropic
material by Greenwood and Johnson (1965)). However, for the case Z3" = 0 a certain T R I P strain e3,0
remains for t9 @ 0: This means that in the case of an anisotropic material the volume change due to
martensitic transformation is not distributed uniformly over all directions, see Fig. 2, which qualitatively
confirms experimental observations, see the discussion in Section 3.2.3. It further seems that all curves
run towards a fixed point.
• • ,, 1
If the "~-restnctlon , ~o = + ~ rr, is activated, the difference between curves for a varying anisotropy
p a r a m e t e r p is much smaller, see Fig. 3.

3.1.2. Linearization
As reported by Fi,;cher (1990) a very simple relation, eVR = g5 /(~2+ ZY 3 2w'*
"~ , could be found for p = 0
(which can be seen as an "extended" Greenwood and Johnson relation) if the almost linear part of the
curves in Figs• 1, 3 is considered. This can be done for p ~ 0, too, by developing, e.g. relation ( l l a ) into a
Taylor series at eTR = 0, then truncating it to 2 terms. Only some steps of this procedure are repeated• It
follows with ( l l a ) a n d eTR = e3, ~ / = ~ 2 " = 0

3/.m-f<.A
"~; = f ( e T R ) =2 t V~ /'
9 2
N = ZeTR -- ~-c,zeVR
9~ + t~2 + zY
3 2 1 +P 4,0 g ' c , x - e c . , ) + Yc,x, 4(1 + 1#9) (~i2.x~ + Yc,rz ,

(12a)

of
.v; - J q . . = 0 + ,.-o

= - + c,, .
(12b)
4

The averaging must be done numerically and finally leads to the relation
t~3 = eTR = e 3 , 0 ( p ) + k 3 ( P ) . ~ 3 . (13a)
8 F.D. Fischer, S.M. Schl6gl/Mechanics of Materials 21 (1995) 1-23

0,24 - ==~'p
i i,

t~
.... = p =-0,36 0,2 ~-

p=0
0,16 ..~
*-- p = 1,25

--------~ p = 3 0,12 --

i
0,08 -~-
i
I

0,04 T

i I .... d ...... I i
-1 -0,8 -0,6 -0,4 0,2 0,4 0,6 0,8 I

E*
I
I

-0,08 ~-

-0,12
g=l

/ /' /
I I -0,16
iI ii / /

-0.2

<~ 1I) i -0,24

Fig. 1. Relation between the TRIP strain erR and the dimensionless stress E ° for different values of the anisotropy parameter a,
g-~l.

e3,o(p) a n d k 3 ( p ) a r e given in T a b l e 1 b o t h for g - - 1 a n d the "~o-restriction". T h e l i n e a r i z e d r e l a t i o n is


also d e p i c t e d in Figs. 4a, 4b. A g a i n , the w e a k influence o f p on e r R for t h e "q~-restriction" can be seen.

3.1.3. Consideration o f a variable yield stress


In this case, the following a s s u m p t i o n s a r e m a d e :
- T h e t i m e v a r i a t i o n of t h e t r a n s f o r m a t i o n strain t e n s o r c o m p o n e n t s is linear, see (1),

h(t/Q)=h(?)=t, h(i)=l, 0_<i'<l.


F.D. Fischer, S.M. Schl6gl / Mechanics of Materials 21 (1995) 1-23 9

.... - p=-0,36 ~ 0,03 ] / / / /

....o - o : o 4oo21 .//~/////

~- °-3 ......j o,o,

-0,3 . . . . . . . . . o, 0,'2 0,3

,°02
Fig. 2. Demonstralion of eTR e 0 for .~" = 0.

- The yield stress changes in a linear way from Rt, a to R/,m,


R t ( t ' ) = ( 1 - i') Rt. a + [Rt. m
- g - 1 (uniform distribution of the variants).
W i t h ( 8 a ) - ( 8 c ) , (9a) a n d a n a l o g o u s l y to (12a) t h e d i f f e r e n t i a l e q u a t i o n for i T R = ~3 c a n b e w r i t t e n as

2 '~3 f'n" f2"rr / ' 2 . (~6TR--dcz)sin 0 dO d, d6


-3 (1 - "[)Rta• + iRt,---~m8rr2 = . ~ o j ~ = o j,o~,, {[~.+ ' :::] _-~i..]} ~- . (14a)

This relation is solved again numerically for fixed time-parameters i = i~K, 0 < K _< 100, and inverted to
iTR = f(.~,Rt,.Rt, m,p,[). (148)
Equation (148) is :integrated numerically again. The results are presented as full curves in Fig. 5 for
RLa = 250 N / m m z and RI, m being 500, 800, 1100 N / r a m z. The stress "~3 varies from - 2 5 0 to 250
N / m m z.

Table 1
TRIP-strain parameters e3,o, k 3
p uniform distribution "~o-restriction"
e3,o k3 e3,0 k3
- 0.75 0.006907 0.2866 0.011357 0.2201
- 0.5 0.002975 0.1837 0.005064 0.1911
- 0.36 0.001724 0.1674 - 0.002966 0.1913
0 0 0.1481 " 0 0.1942
1.25 0.002034 0.1248 0.003573 0.1974
3.0 0.002955 0.1119 0.005226 0.1964

. ~1' 21•
5 /~2_3
This agrees with the exact value g~o
10 F.D. Fischer, S.M. SchliSgl / Mechanics of Materials 21 (1995) 1-23

0,24 -[-
.~',#
i--- p =-0,36
0,2 -.
', p=O y-, #<~
/ ;' /
--*- -- p = 1,25 0,16 • -,/,?
--.-o p=3
0,12 !

0,08 -

0,04 -~

....... 1

-1 -0,6 -0,6 -0,4 -0,2 0,2 0,4 0,6 0,8


E"

-0,08

s:/;; -0,12 -
g..."~-restriction"

, / ,.,

-0,16 !
1
I

l/' i'l'
,ft,'/',; -0,2

-0,24
Fig. 3. Relation between the T R I P strain eTR and the dimensionless stress E" for different values of the anisotropy parameter p, g
correslx)nds to "p-restriction".

For the isotropic case and the linear part of the ETR -- ~"~3 curve, Fischer (1990, 1991) gave the following
relation for an average flow stress:

R [ =R/, m 1 - Rt,m in . (14c)

For the sake of comparison, this constant value R 7 is also used instead of (1 - i ) R t , a + iRt, m in the case
of an anisotropic material. The corresponding curves can be seen as dotted lines in Fig. 5. They agree
very well with those due to (14b). Simplified calculations with R[ instead of a variable flow stress are
justified as a consequence.
F.D. Fischer, S.M. Schl6gl / Mechanics of Materials 21 (1995) 1-23 11

3.1.4. Plastic work


T o get a n i d e a h o w m u c h plastic w o r k is p e r f o r m e d a n d to a large a m o u n t t u r n e d i n t o heat, t h e
specific plastic w o r k W~ c a n b e c a l c u l a t e d easily. It follows with (5), (7a) a n d (A.2)

W/=' /01~ : ~ p d t ' = f0' A s : ('+'Ps)


: 2 - 2 [ ' R 2'i di" (Td)
i c a n b e t a k e n f r o m (8a), ( 8 b ) w i t h e 3 = e.rR = - 2 e I = - - 2 e 2. W i t h t h e a b b r e v i a t i o n /~t for t h e i n t e g r a l
/dR,h d? it follows
W; = ~R I [~ETR - 2ec'=eTR
9- + 62 + 33'] 1 +P 4p - 2 --2

P -2 ~ i/2
4(1 + ½ p ) ( % 2 + 3,c.,,~) / (7e)

(a) 0,3-
I

0,25 .
t~
a .... k3 ~ • ... E3,0
'-~ 0,2
q
,'~
q.I 0,15

0,1

0,05 • g=l

0 .m__• . -- • t ............... •

-o~75 -0,25 0,25 0,75 1,25 1,75 2,25 2,75 3,25


-0,05
P

(b) o,3
£TR ---- E3,0 "J7 k 3 G "
0,25
n . k3 , ... E3,0

0,2 - L
•' . i [3
o
~ 0,15 -

0,1

0,05 g... "~-restriction"

.•

0 m U m---, • • ..... !----- l •


-0,75 -0,25 0,25 0,75 1,25 1,75 2,25 2,75 3,25
-0,05 . p

Fig. 4. (a) Parameters e30 and k 3 for the linearized relation between ~'TR and the dimensionless stress E" for different values of
the anisotropy parametel p, g -~ 1. (b) Parameters e3,0 and k 3 for the linearized relation between eTa and the dimensionless stress
E" for different values of the anisotropy parameter p, g corresponds to "p-restriction".
12 F.D. Fischer, S.M. Schl6gl / Mechanics of Materials 21 (1995) 1-23

0,24

0,2

"'-'--- ~¢ Rt,,,, = 500 N / m m ~ Rt,. = 2 5 0 N / m m 2

0,16 ~- 9=-0'36 /
t

Rt,~ = 800 N / m m 2 J
J. i
• 0,12 - An" ,t

00,

004 4

, I~ ~-'~
, - - . - ~ - _., ....... ; .... ~ ~ . . . . . . {- -.. .......

-1 -0,8 -0,6 ~,4~;~ 6 0,2 OA 0,6 0,8 ]E 1

-o04 t_ R,,.

'~ i'// -0,12 -


J

• /
~/ ,
. -o,16 i g=l
6
=

-0,2 -

-0,24

Fig. 5. Comparison of TRIP strain e.vR-2"/Rt.a-CUrves for a variable flow stress R I (full lines) in relation to those for an average
flow stress R t" ( d o t t e d lines), g =- 1.

If a linear time dependence, R t = (1 - i ) R i . a + iRt. m, is considered again, /~l can be calculated as the
weighted average

W~ was evaluated both for g = 1 and the "y-restriction" and shows almost the same behavior. W ~ / I ~
is depicted in Fig. 6 in relation to E*. With an approximative value of 0.15/~ at E ° = 0.5 the specific
plastic work W~ reaches the rather high value of ca. 100 N m / m m 3, which is approximately one third of
F.D. Fischer, S.M. Schl6gl / Mechanics of Materials 21 (1995) 1-23 13

0,3
= p=0 • - p=3 ~,

0,05 g-1

t t ! --. j 0 0 .... t J I I i
-1 -0.8 -0.6 -0,4 -0,2 0 0.2 0,4 06 0,8 1
E"
Fig. 6. D i m e n s i o n l e s s s p e c i f i c p l a s t i c w o r k W ~ / / ~ / in r e l a t i o n to t h e d i m e n s i o n l e s s s t r e s s v . fl)r d i f f e r e n t v a l u e s o f t h e a n i s o t r o p y
parameter p, g -~ 1.

the specific latent heat set free during martensitic transformation. Usually the latent heat due to a
change in specific entropy is much more important with respect to heat production than plastification.

3.2. The triaxial loading case

3.2.1. Numerical te~ts


We study the case of a triaxial Ioadstress state with the principal dimensionless stresses ,~f, 2'2", ~v3
and the principal deviatoric strains e l, e 2, e 3. As already done in Fischer (1990), a demonstration of thc
results is performed in a S(-S°3 plane with S ( , S~'. being the deviator componcnts, S ( + S 2 + S;. = 0,
corresponding to .,~(, Z2*, Z ; . g - 1 is used as orientation distribution of the martensitic variants.
Further, an average yield stress R t" is implemented, sec (14c). Since ideal plasticity is followed, only
those S ( , S ; are admissible which do not cnforce plastic flow and, thercfore, infinite deformations
before the martensitic transformation. With the yield condition (A.1) it follows
R2
3(SI. 2+SI,$3 ,+S.2). +p(4SI, 2 + 4 S ( $ 3 . + S ~ 2 ) _ R[ 2"t'a- (15)

Eq. (15) describes a rotated ellipse in the S l' - S ; plane with an origin coinciding with that of the
coordinate system. Figs. 7a and 8a demonstrate the lines e I = const., e 3 = const, in the S ( - S 3 plane for
p = - 0 . 3 6 and p =: 3.0. The corresponding graph for p = 0 can be taken from Fig. 3 (Fischer 1990).
Different load paths are marked by the corresponding lines with the numbers 1 to 7:
Path 1, .~3 ÷ I), Zl =-~2 = O, S~ = - 2 S ( ,
Path 2, Z] ~ O, "~2 = '~3 = O, S ; = - 1 / 2 S ( ,
Path 3, Z 2 ~ O, .Y,] = "~3 = O, S~ = Sl" ,
Path 4, ~'I = Y,2 = Z, ~3 = O, S 3 = - 2 S ( ,
Path 5, "~2 =~g3 ='~, ~1 = 0, S 3 = - 1 / 2 S f ,
Path 6, Z~ = ~3 = 2 , .a~2 = 0, S ; = SI*,
Path 7, "~t, Z2 ~ 0, ~3 = 0, S ; = - ~ / at- Sl*"
14 F.D. FLwher, S.M. SchliJgl / Mechanics o f MateriaL~ 21 (1995) 1-23

Figs. 7b and 8b show the corresponding admissible regions for S ( , S 3 duc to condition (15) for the s a m e
data RI.,,, RI. m as defined in Section 3.1.3.
The strongly varying influence of anisotropy (yield stress R / in the longitudinal direction, Rq in the
transversal direction, p = R t2/ R 2q - 1) can best be seen from an e l - e 3 graph, see Fig. 9, for the different
paths (Path I = Path 4, Path 2 = Path 5, Path 3 = Path 6) and two anisotropy parameters O = - 0 . 3 6 ,
p = 3.0 with g = 1.
If thc s p e c i m e n is loaded in the longitudinal direction (Path 1), no significant difference in the
isotropic case can bc found in the T R I P strain c o m p o n e n t s . H o w e v e r , for a loading in the transversal
direction (Path 2 and Path 3), the anisotropy effect can by no m e a n s be neglected.

(a)
O' ~ = 0.0 p =-0,36 0" : ¢:~ = 0,0
1' el = 0 . 0 ' 1 I -- 1":~3 =0.0t
2' ~l = 0 . 0 8 2" : ea = 0.08
3' ~1 = 0.13 ~" ~ 3 " : ~ 3 = 0.13
,t I el = 0 . 2 1 0,8 ~ - ,t,,:~3 = 0 , 2
,~/
q = 0.3 \ / 5 " : ~:~ = 0,3
5tn

-I s; 1

i ' : ~l -. v,,,-. -u, ts ~ - . • ,:~ - - 0 , 0 4


'2' : el = - 0 , 0 8 ; 2 " : ~:~ = - 0 , 0 8
3':q = -0,13 j 3":ca = -0,13
4':~1 =-0,2 -1 ~ /l":~s= -0,2
~' :e, = -O.a g = 1 :i":~3 = - 0 , 3

Fig. 7. (a) Curves of constant TRIP strains el, e3 in the deviatoric S ( - S f plane, g --- 1 a n d p = - 0 . 3 6 . The lines 1 to 7 represent
different loading paths. (b) Regions for admissible S ( , S~ in the deviatoric S l' - S ~ plane, g ~- 1 and p = - 0 . 3 6 . The different
curves belong to different flow stress values Rt, m of martensite. Flow stress Rt, a of austenite is 250 N / m m z.
F.D. Fischer, S.M. Schl6gl / Mechanics o f Materials 21 (1995) 1-23 15

(b) ~,
1
1: Rt,,~ = l l O O N / m m 2
Rt,. = 250 N / r a m 2 8
0,8 2: Rt,,~ = 800 N / r n m ~
3: R~,,~ = 50ON/turn 2
' .....
0,6 ..... ~ "- ..4 -. "-4" Ri, m = 250 N/ram 2
.J
/
/" 0.4 " " ---3 "\~

\
i'
, :./ / / 0.2 " "\
,'.(\
\, \,\
..
"',
\
', \ X \, \",\
[
• "; ...... )'" "" ..... 0 . . . . . X i' >' • -~), . ;
" '\ "\ X\
-1 -o,8 \-o,:~,,,
,,-o,2 0 0,2 0.4 0.6 0.8',,
ST
1

\. -\ '" ~-0,2 •" i


\

" -0,4
//

-0.6

p =-0,36 -0.s g=l

-1

Fig. 7 (continued).

3.2.2. L m e a r i z a t i o n
A l t h o u g h a highly nonlinear problem exists, it is again surprising that for a T R I P strain level up to
0.04 (and with some inaccuracy up to 0.08) an almost linear relation exists between the c o m p o n e n t s of
the T R I P strain and those of the dimensionless deviator. This allows one to write the following relations
using Eq. (13a) for a uniaxiai test in thc 3-direction, and using equation (13b) for the uniaxial test in the
l-direction,

s,--e,,o(p) + kl(p)E?, (13b)

keeping in mind that el, o, e3.0, kl, k 3 are functions of p:

e I = el, 0 + (2/:, -- ~ k 3 ) S~" +(k~-kOS;,


(16a)
e2 = e 2 , 0 - - ( 2 k l -- 2l k 3 ) + 2)s;' ~ 1 , 0 = /~2,0, (16b)
3k 3 (16c)
e 3 = e3, 0 + 0 SI* + ---~--S 3 .
16 F.D. FLs'cher, S.M. Schliigl / Mechanics o f Materials 21 (1995) 1-23

By introducing the stress c o m p o n e n t s E ( ,E 2 , E j instead of S ( , S~ Eqs. (16a)-(16c) can be expressed as

el = e l , o + k l E l , _ ( k 1 _ ~ k 3 ) ~ 2 1' , _ ~3k 31E *, (17a)

e2=e2.() - k , - - ~ E( +k,E;---~E.~, el.(,=e2,o, (17b)

1 -. l .. el,o (17c)
e3 = e3.0 - 7k3Y, l - ~k.~Ez + k3E3', e3'° - 2

A different grouping of the right-hand side of (17a)-(17c) leads to the following matrix relation with the

(a)
0' : (l = 0,0 p = 3 0" : ~3 = 0,0
1' : (l = 0,04 1 , 1" : (3 =.0,04
2 ' : ~ 1 = 0 . 0• 8 ~,, ~ - 2" : e3 = 0, 08
3':~1 =0,13 J 3":(3=0,13
4':(1 =0,2 1 0.8 ,~- 4" : (3 = 0, 2
5' : ~1 = 0,3 5" : e3 = 0,3

5~

-0.8 -0.6 -C 5 0,8 S~

'\\
]' : st = -(I,04 -0,8 -~- 4 ]" : e:, = - 0 , 0 4
2' : (1 = - 0 , 0 8 /; 2" : ea = - 0 , 0 8
.3' : ~1 = - 0 , 1 3 ] 3" : (a = - 0 , 1 3
;l': el = - 0 , 2 -1 _t_ 3,": e3 = - 0 , 2
5':el =-0,3 g-- 1 ,5":ca=-0,3

Fig. 8. (a) Curves of constant T R I P strains in the deviatoric S l" - S 3" plane, g --- 1 and p = 3.0. The lines 1 to 7 represent different
loading paths. (b) Regions for admissible S I ' , S [ in the deviatoric SI° - S 3. plane, g -= 1 and p = 3.0. The different curves belong to
different flow stress values R/. m of martensite. Flow stress RI. a of austenite is 250 N / r a m 2.
F.D.FLwher,S.M.Schl6gl / Mechanics of Materials 21(1995)1-23 17

(b)
1: Rt,~ = l l O O N / m m ~

R4,. = 250 N / r a m 2 0,8 2: Rt,m = 800N/rnm ~

3: Rt,~ = 500N/ram:
...... !
~ , ~,]
, 4 : Rt.,,~ 250 N / m m 2

\ ', i ( ' \ \ ] ' , , \

, ' ,x,x \ \,

\ '~\ I ',5", ',


-1 -0,8 -0,6 -o,4 -~2\\\o ',~:~ 04 06 o8
s;
1
~x \ \ \\ ' '. ', ',

~9,z\
\ \,~-,
\\ ~ ,. ,...
• ..
~ \ \ _.. ~ ,
• !
-0,4 \\ ,
\ \
',\
\,\
\.
-0,6 "\ \

p=3 -0,8 g=l

°1 --

Fig. 8 (continued).

vectors ETR = (El, F2, 6"3) , _T = (El,0, E2,0 ' E3,0), and _~.T = ( 2 ( , E~, E 3), the superscript T means
"transposed":

_6TR = _6 0 -]- k3
1

-- ~
I
1
2 --2

-
1
• _E* + ( k , - k 3 ) [1 1 i] -1
0
1
0
"_2". (18a)
2
l
2
l

If we introduce the deviator _S" analogously to __,_E', and the vector ~" = [S~ - S 2, - s? + s ; , 0]
analogously to the tensor g in Appendix A, the following relation can be written:

3 *
£Tg = go(P) + -~k3(p)S- + (k,(p) - k3(p))_S" ; (18b)

5 /~2 3 2
p=0:eo(p) =0, kl =k3 = ~V u + i T •

Now the question arises whether the relations (18a), (18b) can be extended to a general strain state
eVVR= (r x, ey, e~, e~y, exz, eye) and a general stress state _~'T = ( 2 ; , 2 ; , E~', Tx~., T~, Ty~) with T~y,
18 I.D. Fischer, S.M. Schlb'gl / Mechanics of Materials 21 (1q95) 1-23

o 0'3 ,..~-~.

p a t h - , 1,4 " - , . . , ~ ~ ,'" ~


'l l r

..... ~ "~ 0,1'. ~,-'


.... *,;/ ........................ •

t -- I ----+--~--~-~....~ , .. I ......... q
-0,3 -0,2 .......-.---~,t "~ ...... .,'~, NN~"~ "-L.3~_ 0,2 '3 0,3

_ ~- - }p=3 " i '. " " - - .

• / -0,3 -

• I p ---0,36
,L

Fig. 9. Almost linear e I, e 3 relations for different loading paths depending on the anisotropy parameter p = -0.36 and p = 3.0.

T.,~, T~.': being the dimensionless mascroscopic shear stresses on a unit cube of the anisotropic material.
It is easy to check this for Tx"~. since isotropy exists in the x - y plane. Applying Mohr's circle by
considering the load stress state T ~ and using (17a), (17b) immediately leads to the relation

3 *
e,:. = ~k3Txy + 2(k] - ka)T~;.. (19a)

Such a simple relation, however, cannot be written for Tx~, Tv~ since the x - y or x - z plane, respectively,
is now an "anisotropy plane". This can also be seen from Appendix B, relation (B.5). Since the principal
stresses due to Tx~, Ty~ are operative in directions with normal vectors of 45 ° to the global z-axis, the
angles Y],Y2,Y3 (for definition see Appendix B) are 45 °, 90 °, 45 ° leading to a yield condition of the type
]
½(1 + p X o r 2 - 0"3)2 + ~-(Or3 - - 0"])2 + ½(1 + f i X 0 . 1 - 0"3)2 with respect to the coordinate system of the princi-
pal stresses. However, this yield condition differs from (A.1) on which the former derivations were based.
For a combination of global normal and shear stresses it can again be seen from (B.5) that the transversal
isotropy disappears with respect to any one of the principal stress axes since Y~,Y2,Y3 usually are
different angles.
A direct relation between a global shear stress Tx~ and the TRIP-strain tensor components e x, e z and
exz can only be found by an analysis explained in Section 2.2, considering, however, that ( s x / R ; ) = 0,
( s : / R ; ) = 0, see (9a,b), and ( ~ x z / R t ' ) = T~'~('rxJR ; ) is not expressed in Section 2.2 but can be
evaluated with the same procedure as the other local stress tensor components s x, etc. Such an
investigation has not yet been performed up to now and will be postponed to a future e x p e r i m e n t a l / a n a -
lytical program for thin tubes under normal load and shear.
However, due to the proper relations between normal stresses and strains, see (18a), (18b), and the
KD. Fischer, S.M. Schl6gl/ Mechanics of Materials 21 (1995) 1-23 19

shear strain relation (19a), the following tensorial relation between the T R I P strain tensor _eTR and the
dimensionless stress deviator S ' can be suggested:

_ETR = _E0 +
3
~k3S
*
+ (k I - k3)_S" ' _e0 =
[00 01
o° .o o ' =
is/ s
2T,; s;-s:
o ~,o [ ½rx; ~T,;
(18c)

With respect to the: factors ~3 in connection to k 3 and 2 resp. ~1 to (k~ - k 3 ) , one may now admit that the
S ", _S" vectors can be extended to vectors with 6 components,

_S" = (S~" ,Sy" ,S~" ,T~; ,T~; ,Ty~), (20a)

8" = ( s ; - s ; , - Sx" + s;, o,-.x~,~.~z,~..


9T" 'T" 'T'), (20b)

" obviously is the analogous vector to the tensor S in the Appendix A.


This means that relation (18b) can also be used For the general load stress state. However, the vector
_% is assumed to have the components

_~T = ( ~ . 1 , 0 , E 2 . ( , , £ 3 . 0 , 0 , 0 , 0 ) , El. 0 -{- e2, 0 q- E3, 0 = 0. (20c)


Relations (18b) and (18c) can now be extended to take into account a certain volume fraction ~: of
martensite as was already done by various researchers in the past, for an overview see Denis et al. (1989).
In their incremental form they read now

-eTR = [-% + -32k3-S" + ( k , - k3)_S" ] d ~ ) ( , (18b)

~TR=[~0+3k3S' + ( k x - k 3 ) _ si* l d ~_- - J-) s . (18c)

The constants k 1, J¢3 a s well as the components of _%, _% are functions of the anisotropy factor p and are
defined above. The monotonic function q~(s¢) must-lie in the interval 0 <q~(~)< 1 for 0_<~ _< 1.
Proposals for ~p(s¢) can also be taken from Denis et al. (1989).
~TR or -~TR can now be added as an additional strain rate term to the "classical" plastic strain rate
term as it ~vas, e.g. proposed by Leblond (1986a,b)

3.2.3. S o m e comments on experiments

Only few experimental reports on the anisotropy effect on T R I P exist in the literature. Leblond (1992)
informed the authors of some interesting results on low-alloy forged cylindrical specimens with signifi-
cantly elongated grains. When the grains were orthogonal to the specimen axis the material behaved,
macroscopically speaking, isotropically during a full temperature cycle. But whenever the grains were
parallel to the axis the specimen did not recover its original length after a full temperature cycle.
However, no quarttitative data have been communicated up to now. Motivated by this observation,
tubular specimens with their axis in the direction of longitudinal forged rods of low carbon (C < 0.03%),
11% Ni-steel were subjected to a few thermal cycles without any load applied. After heating to 830°C
and holding this temperature for 10 min. they were quenched to room temperature. Brozyna (1994)
20 F.D. Fischer, S.M. Schl6gl / Mechanics of Materials 21 (1995) 1-23

indeed found an irreversible elongation of e3,0 = 0.002 per cycle. This agrees well with the calculated
strain for p ~ - 0 . 4 , see Fig. 2 with Z* = 0. Further experiments are under preparation for the same
tube type, however, under an external pressure and a longitudinal load.
Finally it is referred to the experimental work by Sattler and Wassermann (1972). They investigated
tension specimens from a plate with a pronounced rolling texture, (112)[1 l~] + (011)[21]]. The specimens
were cut out of the plate under different angles a in the rolling direction and loaded by a stress ,~ in
their length direction. However, due to a lack of information it is not possible to relate this type of
anisotropy to a specific value of p. Different angles a lcad to different combinations , ~ , ,~:, Txz due to
Mohr's circle if one defines the x direction as the rolling direction and the z axis as the width direction
of the plate.
It is interesting to note that up to an eTa-value (measured in the specimen length direction) smaller
than 0.06 a linear relation between ETR and Z can bc envisaged. This clearly points to a linear
superposition of the T R I P strain! Additionally, a certain but small stress state exists for eTR = 0. Both
effects appear in this scmianalytical study as well which, at least, indicates a qualitatively correct result!
It may be, however, of importance that specifically for Z leading already to a plastification of the
austenitc, only very few martensite variants (sometimes only 2!) exist. This means that in addition to
accommodation an orientation effect appears.
Finally it must be said that even though the research of Sattler and Wassermann (1972) certainly is of
the pioneering kind, further and better documented experimental results will be necessary to check the
influence of anisotropy on T R I P in specimens machined from rolled plates.

4. Conclusion

The influence of transverse isotropy (which is a special kind of orthotropic anisotropy found, e.g. in
forged longitudinal specimens) on the transformation strain is not very significant in the uniaxial loading
case if the specimen is loaded in the longitudinal anisotropy direction. However, in some triaxial load
stress cases as well as for a uniaxial loading in the transverse direction, the anisotropy contributes to the
T R I P strains much more significantly, (18c). As a general rule, a modified T R I P strain tensor is proposed
for this special orthotropic anisotropy which consists of three "subtensors". One "subtensor" is
"standard" and proportional to the stress deviator S ' . A further "subtensor" reflects the anisotropy in
the case of no external loading (e0). The third ten~-or (k I - k 3) _S" represents the "coupling" between
the loadstrcss state and the orth6tropic anisotropy. Of course, ifi the case of a general anisotropy the
third tensor must obviously be modified. This will require further studies, though.

Acknowledgement

The project was partially supported by the "Christian Doppler Laboratorium for Micromechanics of
Materials" which is sponsored by the former Austrian Industries, as well as by the Austrian Federal
Ministery of Science and Research concerning the project "Experimental Material Law-Verification",
project number G Z 49.809/3-24/92. These supports are gratefully acknowledged.
The authors further appreciate the discussions with Franz Marketz, Ph.D. student at the Institute of
Mechanics, University for Mining and Metallurgy, Leoben, specifically on the physics of martensitic
transformations.
Finally the authors want to express their thanks to Prof. H.P. StiJwe and Doz. B. Ortner who are
currently performing an extensive experimental program on biaxially loaded transforming specimens.
F.D. Fischer, S.M. Schliigl/ Mechanics of Materials 21 (1995) 1-23 21

Appendix A. A n i s o t r o p i c plasticity

T h e classical anisotropic yield condition is given by Hill (sec, e.g. 1950),

F(try - tr~
)2 + G(tr~ - o~) 2 + H ( ~ - try,)2 + 2L~.2 + 2M7"~ + 2NT"~y- 1 = 0

for m o n o t o n i c stressing (i.e. without a Bauschinger effect) is the starting relation. Rt is the yield stress in
the longitudinal ( z - ) direction which coincides with the specimen axis. R q is the yield stress in the x - y
plane, i.e., in the transverse direction. Due to this assumption the following equations for F, G, H can be
written:

tr~=~=0, tr.=R l F+ G =I/R~,


o ' ~ = R c, try=try=0 G +H =I/R 2,
cry=0, cry=Rq, trz=0 F +H =I/R2q,

with the result


1
1 2 1 2 1 2
F = ~R t, G = TR t, H = ---5 - ~Rt.
Rq

A pure shear test (7-xr = z, trx = try = trz = %z = ry~ = 0) delivers 2 N z 2 - 1 = 0. D u e to M o h r ' s circle this
test is equivalent to trl = +7-, tr2 = -7-, tr3 = 0, 7-12 = 7"13= 7"23 = 0 if one considers the 1-direction at an
angle of 45 ° to the x-direction. D u e to the axial symmetry with respect to the z (or 3)-axis it follows

FT"2 + GT"2 + H47" 2 - 1 = 0.


This allows to calculate 7"2 as 1 / ( 4 / R ~ - 1 / R 2) resulting in 2 N = 4 / R 2 - 1 / R 2. A similar p r o c e d u r e
allows to write

2 M = 2 / R 2 + l/R2q, 2 L = 2 / R 2 +1/R2q.

Now the yield condition can be written with p = ( R I2/ R q)


2 - 1 as

{-(try _ tr.)2+~(tr
1 z _ o'~)2+({-+p)(o',:-try)2+(3+,o)7"y~ 2 + (3 +p)7-2 + (3 + 4 p ) % y2 = R 2.
(a.1)

T h e r e are certain restrictions on the factor p. D u e to the equivalence of some slip systems, the ratio
R t / R q for a fcc single crystal can be 2 at maximum, see, e.g. Dieter (1961). Therefore, Pmax is assumed to
1
be 3. O n the other hand, for R J R q < ~ the convexity of the yield surface is lost, see, e.g. Stiiwe (1974),
leading to Pmin-- -'0"75 and, therefore, - 0 . 7 5 < p _< 3.0. T h e stress deviator tensor s is introduced as
1
tr = s + 3(tr~ + try + tr~)l. A second deviatoric tensor g_is defined as

| 27".
27". ] .
77"yz 0

T h e yield condition can be written as

e(tr) - o, - 0 . 7 5 < p < 3.0. (,,.2)


22 F.D. Fischer, S.M. Schl6gl / Mechanics o f Materials 21 (1995) 1-23

T h e plastic strain increment is found by the relation ~p =~0F/O_o-. T o achieve this, (A.1) must be
differentiated, e.g.
OF
= - - + + -

aF
-- = (3 +p)ryz,
O~'yz
2 1 2 2
note ryz = ~(~yz + ~'~)- Using the tensor s again leads to

(A.3)

The specific plastic work increment follows with (A.2) as


+

and, due to the second law of thermodynamics, to the necessary condition


,~ >__0~

Appendix B. General yielding

If we assume a principal stress o-I to be operative in the direction of a unit vector n I with the
components n T = (cos a l, cos/3 I, cos yl), with cos2al + c0s2/31 + cos23'1 = 1, standard tensor transforma-
tion leads to the corresponding stress tensor _a in the global coordinate system:
COS 2 0 ' 1 COS O' l COS /31 COS O~1 COS T1

i f = o " t cos a I cos/31 cos 2/3 cos/31 cos 3'1 (B.1)


cos oq cos Y~ cos fll cos yl cos2 Yt

Insertion of (B.1) into the anisotropic yield condition (A.1) delivers, after some algebra,
a ? [ i + p(1 - cos23q] = R t (8.2)
For an orthogonal triad n~ with the corresponding principal stresses a/, i = 1, 2, 3, relation (B.2) can be
generalized to
o'/211 + p ( 1 - cos 2 3'/)] = R 2 - (B.3)
If one turns now to the coordinate system with the basis _ni to formulate an anisotropic yield condition, it
follows immediately from (B.3) for the yield stresses o'r, i in the three orthogonal directions n i
try./= R,[1 + p ( 1 - c o s z 3'i)] - ' / 2 (B.4)
T h e coefficients F',G',H' of the yield condition
F'((r2- (r3)2 + G'((73- (rl )z + H'(orI tr2)2 = 1
can now be written as
1 1 1 1
F'=~---t2(~+pcos23',), G' = --~t ( ½ + p cos2 3"21, H' = R2(½ + P c°s2 3'3)
- - (B.5)
F.D. Fischer, S.M. Schl6gl / Mechanics of Materials 21 (1995) 1-23 23

Some of the above results were recently published in a more specified form by Pereda et al. (1993). If,
therefore, a general stress state Zx, Zy, Z~, T~v, T,~, Ty~ with the corresponding principle stresses cry, o-2,
~r3 and their cigen~ectors n t, n__:,n 3 is applied to a unit cube, the TRIP tensor _eva (18.3) cannot directly
be applied to the n_~, _n2, _n3 coordinate system since it was derived undeF the preposition F = G
or F ' = G ' .

References

Brozyna, St.A. (1994), Transformation induced plasticity of 11% Ni-steel subjected to a biaxial stress state (in German), M. Thesis,
Univ. Mining and Metallurgy, Leoben, Austria.
Dieter, CO.E. (1961), M~'chanical Metallurgy, 2nd Ed., Mccoraw-Hill Series in Materials Science and Engineering, p. 130.
Denis, S., R. Gautier ard A. Simon (1989), Modelling of the mechanical behaviour of steels during phase transformation: A review,
in: CO. Beck, S. Deais and A. Simon, eds., International Conference on Residual Stresses - ICRS2, Elsevier, London, pp.
393-398.
Fischer, F.D. (1990), A micromechanical model for transformation plasticity in steels, Acta Metall. Mater. 38, 1535-1546.
Fischer, F.D. (1992), Transformation induced plasticity in triaxial loaded steel specimens subjected to a martensitic transformation,
Eur. J. Mech. A. 11,233-244.
Fischer, F.D., M. Bervt:iller, K. Tanaka and E.R. Oberaigner (1994), Continuum mechanical aspects of phase transformations in
solids, Archit:e Appl. Mech. 64, 54-85.
Coanghoffer, J.F., S. De:3is, E. Gautier, A. Simon, K. Simonsson and S. Sj6str6m (1991), Micromechanical simulation of martensitic
transformation by finite elements, J. Phys. Colloq. (France) I, C4 77-82.
Gautier, E. and A. Simon (1987), Transformation plasticity mechanisms for martensitic transformation of ferrous alloys, in: Proc.
Inst. Conf. Sofid Ph6'se Transformation, The Institute of Metals, pp. 285-287.
Coreenw(×)d, CO.W. and R.H. Johnson (1965), The deformation of metals under small stresses during phase transformation, Proc. R.
Soc. London A 283, 403-422.
Hill, R. (1950), The Ma.'hematical Theory of Plasticity, The Clarendon Press, Oxford, pp. 317-344.
Leblond, J.B. (1989), Mathematical modelling of transformation plasticity in steels, II: Coupling with strain hardening phenomena,
Int. J. Plasticity 5, 573-591.
Leblond, J.B. (1992), plivate communication on unpublished experiments by Y. Desalos.
Leblond, J.B., Co. Mottet and J.C. Devaux (1986a), A theoretical and numerical approach to the plastic behaviour of steels during
phase transformations - I: Derivation of general relations, J. Mech. Phys. Solids 34, 411-432.
Leblond, J.B., CO. Mottet and J.C. Devaux (1986b), A theoretical and numerical approach to the plastic behaviour of steels during
phase transformations - lI: Study of classical plasticity for ideal-plastic phases, J. Mech. Phys. Solids 34, 411-432.
Leblond, J.B., J. Devattx and J.C. Devaux (1989), Mathematical modelling of transformation plasticity in steels, I: Case of ideal
plastic phases, Int. J'. Plasticity 5, 551-572.
Lippmann, H. (1974), Ans~itze und L~sungsbeispiele zur Theorie des anisotropen plastischen Flieflens, in- H.P. Stiiwe, ed.,
Mechanische Anisotrgpie, Springer Verlag, Wien, New York, pp. 257-278.
Marketz, F. and F.D. :Fischer (1993), A micromechanical approach to transformation induced plasticity (TRIP) for martensitic
transformation, in: (~llection de la Directions des Etudes et Recherches d'Electricite de France, ed., MECAMAT'93 Seminar:
Micromechanics of Materials, Editions Eyrolles, Paris, pp. 258-269.
Marketz, F. and F.D. Fischer (1994), Micromechanical modelling of stress-assisted martensitic transformation, Modelling Simul.
Mater. Sci. Eng. 2, 1017-1046.
Marketz, F. and F.D. Fischer (1995), A mesoscale study on the thermodynamic effect of stress on martensitic transformation,
Metall. Trans. A 26, 267-278.
Mitter, W. (1987), Umwandlungsplastizitiit und ihre Beriicksichtigun8 bel der Berechnung ~'on Eigenspannungen, Materialkundlich-
Technische Reihe 7, Coeb~der Borntr~iger, Berlin, Stuttgart.
Oberaigner, E.R., F.D Fischer and K. Tanaka (1993), A new micromechanical formulation of martensitic kinetics driven by
temperature a n d / o r stress, Arch. Appl. Mech. 63, 522-533.
Patel, J.R. and M. Coaen (1953), Criterion for the action of applied stress in the martensitic transformation, Acta metall. 1,
531-538.
Pereda, J.J., N. Aravas and J.L. Bassani (1993), Finite deformations of anisotropic polymers, Mech. Mater. 15, 3-20.
Sattler, H.P. and CO. Wassermann (1972), Transformation plasticity during martensitic transformation of iron with 30% Ni, J.
Less-Common Met..?8, 119-140.
Sj6str6m, S., J.F., Coanghoffer, S. Denis, E. Gautier and A. Simon (1992), FEM calculation of pearlitic transformation plasticity, in:
J. Fujiwara, T. Abe and K. Tanaka, eds., Proc. International Conference on Residual Stresses - ICRS3, Elsevier, London, pp.
1209-1214.

Вам также может понравиться