Lakhdar Taleb François Sidoroff - A Micromechanical Modeling of The Greenwood-Johnson Mechanism in Transformation Induced Plasticity

A micromechanical modeling of the
GreenwoodJohnson mechanism in
transformation induced plasticity
Lakhdar Taleb
a,
*, Franc ois Sidoro
b
a
INSA/LMR, UMR 6138, avenue de luniversite BP 08, 76801 St Etienne du Rouvray Cedex, France
b
ECL/LTDS, UMR 5513, 36 avenue Guy de Collongue, BP 163, 69131 Ecully Cedex, France
Received in nal revised form 15 December 2002
Abstract
An experimental analysis of transformation induced plasticity (TRIP) considering bainitic
transformation in a 16MND5 steel (SA508Cl3 in ASTM norm) has been reported recently in
the International Journal of Plasticity. Its main results have shown that among the existing
models used in practical applications, the one proposed by Leblond 10 years ago seems to be
one of the models which describes the more accurately the phenomenon. It was also noticed
that this model leads to a singularity in the description of the TRIP at the beginning of the
transformation under low applied stress. This singularity is usually avoided by the introduc-
tion of an arbitrary threshold. The TRIP norm at the end of the transformation seems also
slightly overestimated by this model. The work which is presented here is devoted to an
investigation of these discrepancies from a reevaluation of the micromechanical model as
originally used by Leblond et al. A more complete formulation taking into account the elas-
ticity in both phases will be developed and solved resulting in an improved model enabling a
better description of the experimental results and removing the singularity mentioned above.
A comparison between experimental results and the predictions of the proposed model is
given.
# 2003 Elsevier Science Ltd. All rights reserved.
Keywords: A. Phase transformation; A. Thermomechanical processes; B. Pressure vessels; Transformation
plasticity; Micromechanical analysis
International Journal of Plasticity 19 (2003) 18211842
www.elsevier.com/locate/ijplas
0749-6419/03/$ - see front matter # 2003 Elsevier Science Ltd. All rights reserved.
doi:10.1016/S0749-6419(03)00020-2
* Corresponding author. Tel.: +33-2-32-95-97-65; fax: +33-2-32-95-97-10.
E-mail address: lakhdar.taleb@insa-rouen.fr (L. Taleb).
1. Introduction
Transformation induced plasticity (TRIP) can be dened as the anomalous plastic
strain observed when metallurgical transformation occurs under an external stress
much lower than the yield limit. From a technological point of view it plays an
essential role in many problems, in particular for the understanding of residual stres-
ses resulting from welding operations and for their prediction in practically signicant
cases. From an experimental point of view, it is usually characterized in a TRIP test
(called also sometimes creep test) where a constant stress is applied during transfor-
mation under prescribed cooling conditions as presented by Taleb et al. (2001).
From a microstructural point of view, two mechanisms are usually considered to
explain TRIP,
+ the Magee (1966) mechanism corresponds to the formation of selected mar-
tensitic variants resulting from the applied stress,
+ the GreenwoodJohnson (1965) mechanism corresponds to the micro-
mechanical plastic strain arising in the parent phase from the expansion of the
product phase.
The relative importance of these two mechanisms depends on the material and the
transformation under consideration. Strictly speaking, both mechanisms are gen-
erally present in diusional and diusionless transformations.
We are interested here especially in welding conditions in a 16MND5 steel
used for pressure vessels in which Magee mechanism seems not be signicant
(Grostabussiat et al., 2001). Most existing models for TRIP based on the
GreenwoodJohnson mechanism nally result for a low applied stress, in the
following equation,
E
.
tp
= k.(z).z
.
.
3
2
S
giving the TRIP rate as a function of the volume fraction of the product phase z and
the applied deviatoric stress S. This relation introduces a material parameter k
(TRIP coecient) and a normalized function z ( ) governing the TRIP kinetics.
Dierent forms have been proposed for this function z ( ) see for instance the review
of Fischer et al. (1996).
During the last years several experimental, theoretical and numerical studies have
been performed on the TRIP phenomenon (Fischer et al., 2000a, b; Cherkaoui et al.,
1998; Nagayama et al., 2000, 2001a, b; Taleb et al., 2001; Grostabussiat et al., 2001;
Coret et al., 2002).
We shall focus our attention here on the model developed by Leblond (1989) and
which, although rather old, still is one of the most widely used for practical applica-
tions and which is implemented in the nite element codes SYSWELD and ASTER.
This model is obtained from a micromechanical analysis based on the determina-
tion of the plastic strain induced in a spherical parent phase by the growth of a
spherical product phase core. This model is very crude:
1822 L. Taleb, F. Sidoro/ International Journal of Plasticity 19 (2003) 18211842
+ The assumed spherical shape, which allows an analytical solution, obviously
is not true in reality, however it may be expected that this is not essential at
the beginning of the transformation which dominates TRIP (85% of the
whole transformation plasticity takes place during the rst half of the
transformation after Leblonds model),
+ Many supplementary assumptions are made which will be discussed in the
following.
The purpose of the present paper is to discuss these approximations, evaluate their
inuence and release some of them resulting thus in a more complete model. It will
then be shown that these improvements may modify the TRIP description at the
beginning of the transformation and lead to lower TRIP value at the end of the
transformation. The obtained results will be analyzed through further experiments
and more rened nite element micromechanical analyses as those presented in
(Barbe et al., 2001a, b).
2. Leblonds model
2.1. General background
A macroscopic volume element V containing two phases is considered by Leblond
et al. (1986). The starting point is the decomposition of the local (total) strain for
each phase into elastic, thermo-metallurgical and plastic strains:
c
t
= c
e
c
thm
c
p
In fact c
thm
and c
p
should be considered as respectively the isotropic and devia-
toric parts of the non elastic strain c
t
c
e
, which generally includes thermal expan-
sion, plastic deformation and transformation deformation. This means that
following Leblond et al. (1986), the transformation strain has been included in c
thm
for its isotropic part and c
p
for its deviatoric part.
It follows from the assumed elastic homogeneity that the macroscopic elastic
strain is the average of the microscopic one as established by Hill (1967), E
e
= c
e
)
V
.
The same therefore holds for the isotropic and deviatoric parts of the non elastic
strain, E
thm
= c
thm
_ _
V
and E
p
= c
p
)
V
The macroscopic elastic strain E
e
will be obtained from the macroscopic elastic
stress S by the homogeneous elastic law. The thermo-metallurgical part E
thm
can be
observed experimentally (for an isotropic material) by varying the temperature
under no applied mechanical stress (free dilatometric test, S = 0).
As mentioned before, the macroscopic plastic strain must include in addition to
the classical plastic term, the contribution of the deviatoric transformation strain.
The plastic strain rate therefore must account for the variation of the geometric
extent of each phase, thus leading to the following expression:
E
.
p
= 1 z ( ) c
.p
1
_ _
V
1
z c
.p
2
_ _
V
2
z
.
c
p
1 2
_ _
F
(1)
L. Taleb, F. Sidoro/ International Journal of Plasticity 19 (2003) 18211842 1823
where, z is the volume fraction of the newly formed phase (2), c
.
p
i
is the microscopic
plastic strain rate tensor in the phase i, c
p
1 2
is the deviatoric part of the trans-
formation strain and c
p
1 2
_ _
F
expresses the average value of c
p
1 2
over the
transformation front F.
As discussed in Leblond et al. (1989), the last term in Eq. (1) which accounts
for the average deviatoric transformation strain disappears when there is no pre-
ferred orientation (GreenwoodJohnson mechanism). On the contrary it would be
the dominant term for the Magee mechanism. It should be noted that this term
would disappear during reaustenitization so that the resulting strain should be
recoverable upon heating. This would not be true for the other terms. This remark
provides an experimental procedure for validating the respective role of both
mechanisms.
The second term in Eq. (1) will also disappear assuming only elastic behavior for
the product phase. Eq (1) nally reduces to,
E
.
p
= 1 z ( ) c
.
p
1
_ _
V
1
(2)
Now the evolution of c
p
1
in the parent phase will result from the variations of the
loading conditions (temperature T
.
and the applied macroscopic stress
.
) and from
the transformation process.
Dependence of E
.
p
upon T
.
and
.
corresponds to the usual macroscopic thermo-
plastic behavior while transformation plasticity refers to the evolution of E
.
p
from
the variation of z at constant temperature and macroscopic stress.
E
.
p
= E
.
cp
E
.
tp
.
where,
E
.
tp
= 1 z ( )
oc
p
1
oz
_ _
V
1
z
.
(3)
where
oc
p
1
oz
can be evaluated by changing z under constant temperature and macro-
scopic stress (T
.
=
.
= 0).
Using Von Mises standard plasticity and assuming a uniform yield stress o
y
1
, this
relation is transformed into:
E
.
tp
= 1 z ( )
3
2
1
o
y
1
oc
eq
1
oz
s
1
_ _
V
1
z
.
(4)
where, o
y
1
is the yield stress of the parent phase (phase 1), s
1
is the deviatoric tensor
of the microscopic stress in the phase 1, c
eq
1
is the Von Mises equivalent microscopic
strain in the phase 1.
It should be noted that as discussed in Leblond (1989), the assumption of uniform
o
y
1
does not allow a precise description of the local non uniform hardening. How-
ever, hardening can be roughly taken into account by allowing this uniform yield
stress o
y
1
to depend on some macroscopic averaged hardening variable. This will be
further discussed in the application.
The last assumptions consist of neglecting the correlation between s
1
and
oc
eq
1
oz
and
identifying the average S
1
of s
1
in V
1
to the overall average S (respectively H4 and
H5 in Leblond et al. 1989), nally reducing Eq. (4) to
E
.
tp
= 1 z ( )
3
2
S
o
y
1
oc
eq
1
oz
_ _
V
1
z
.
(5)
It should be noted that according to Leblonds discussion, this identication of S
1
to S can only be used as a rst approximation. A numerical micromechanical ana-
lysis (Fig. 7 in Leblond et al. 1989) in fact suggests that S
1
/S is a decreasing function
of z. The following second approximations,
S
1
= 1 z ( )S or S
1
= 1 z
2
_ _
S.
appear reasonable. We shall come back later when dealing with applications.
2.2. Micromechanical model
In order to evaluate the average
oc
eq
1
oz
_ _
V
1
which appears in (5), Leblond et al. use
the micromechanical model of a spherical product phase expanding in a sphe-
rical shell of the parent phase (austenite) without external loading (see Fig. 1).
A variation oz of z corresponds to an expansion oR
2
of the product phase radius:
oz =
3R
2
2
oR
2
R
3
1
Two further assumptions are made:
Fig. 1. Geometry considered by Leblond et al. for the evaluation of
oc
eq
1
oz
_ _
V
1
.
H2a: Elastic strains are much smaller than plastic strains in the austenitic phase,
H2b: The (compressive) forces exerted by the , crust on the o core have a negli-
gible eect on the volume of the latter.
The transformation strain (c
12
) therefore appears as an imposed displacement:
ou = oR
2
V
V
= 3c
12
oR
2
on the R
2
oR
2
boundary of the austenitic shell.
V,V is the relative dierence of volume between the two phases.
The micro-incompressibility condition resulting from H2a in the austenitic phase
directly leads to
ou =
C
x
2
=
3c
12
R
2
oR
2
( )
2
oR
2
x
2
-
3c
12
R
2
2
oR
2
x
2
(6)
C is a constant.
The Von Mises equivalent strain is then easily computed and integrated over the
austenitic phase, nally resulting in:
oc
eq
1
oz
_ _
V
1
=
2c
12
1 z
.Ln z ( ) (7)
It should be noted that this model is based on the assumption that the plastic
deformation in the austenitic phase essentially results from the internal stress arising
from the expansion of the daughter phase and is not signicantly aected by the
applied external stress. This is entirely consistent with Leblonds assumption (H4 in
Leblond et al., 1989) and may be expected to remain true for small values of this
stress.
2.3. Leblonds model
Having
oc
eq
1
oz
_ _
V
1
by Eq. (7), Eq. (5) which gives the transformation plasticity rate
becomes:
E
.
tp
=
2c
12
o
y
1
.Ln(z).z
.
.
3
2
S (8)
Comparing the predictions of the above model with some test results given by
Desalos (1981), Leblond et al. concluded that at the beginning of the transforma-
tion, Eq. (8) overestimates the transformation plasticity rate. To account for this
discrepancy and also to avoid the singularity at z=0, they proposed to replace (8)
by
E
.
tp
=
0 if z 40.03
2Dc
12
o
y
1
.Ln(z).z
.
.
3
2
S if z > 0.03
_
_
_
(9)
for small applied stresses. The threshold 0.03 has been chosen in order to have a
good description of some micromechanical numerical simulations presented by
Leblond et al. (1989).
This formulation is later extended by Leblond (1989) to larger stress level in the
following expression,
E
.
tp
=
0 if z 40.03
2Dc
12
o
y
1
.Ln(z).z
.
.
3
2
.h
o
eq
o
y
_ _
S ifz > 0.03
_
_
(10)
where,
h(
o
eq
o
y
) =
1 if
o
eq
o
y
4
1
2
1 3.5(
o
eq
o
y

1
2
) if
o
eq
o
y
5
1
2
_
_
(11)
o
y
is the yield stress of the mixture austenitic+ferritic phases.
h
o
eq
o
y
_ _
has been chosen to t with the results of some micromechanical numerical
simulations.
2.4. Comparison with experimental results
In the original work of Leblond et al., this model was validated from experimental
results obtained by Desalos (1981). These tests were carried out considering bainitic
transformation in a A533 steel under applied stress which exceeds a half of the aus-
tenitic conventional yield stress (o
0.2%
) at the temperature corresponding to the
beginning of the transformation. These experimental results therefore cannot be
used for the validation of the model in the framework in which it has been devel-
oped, namely for small applied stress. Furthermore, as shown in some previous
works as in Gautier et al. (1985) and Simon et al. (1994), such a level of applied
stress can modify the kintics of the bainitic transformation. Consequently, the z(T)
function can not be taken from a free dilatometric test (free stress) as considered by
Desalos (1981).
A precise identication of the function E
tp
(z) would require a direct determination
of z from the transformation plasticity test. More details on the analyses of Desa-
loss tests are given by Taleb et al. (2001).
Experimental tests on bainitic transformation of the almost same steel under
small applied stresses have been performed and published by Cavallo (1998) and
Taleb et al. (2001). These tests allow a direct comparison with Leblonds model.
Some typical results are presented in Fig. 2(a) and (b) respectively giving the evo-
lution of the function E
tp
(z) and the ratio E
tp
z ( ),E
tp
1 ( ). The latter will be called the
normalized transformation plasticity (NTP).
In order to assess the theoretical result of Leblond et al. [Eq. (8)], in addition to
the empirical threshold proposed by the authors, z>0.03 (Leblond 0.03 in Fig. 2),
the threshold z>0.01 (Leblond 0.01 in Fig. 2) is also considered. This shows the
practical importance of the choice of such a value.
To be complete the experimental NTP is also compared to the predictions of the
models proposed respectively by Abrassart (1972) and Desalos (1981).
Considering the results of Fig. 2, it is obvious that, compared to Abrassarts and
Desaloss proposals, Leblonds model is the most representative of experimental
results. However, two main signicant discrepancies can be noted for this model: the
Fig. 2. Bainitic transformation in a 16MND5 steel under applied stress (24 MPa) equal to about a quarter
of the austenitic conventional yield stress at the beginning of the transformation (600

C). Transformation
plasticity (a) and normalized transformation plasticity (b) versus the volume fraction of the formed bai-
nite.
transformation plasticity kinetics at the beginning of the transformation seems
underestimated and the transformation plasticity norm at the end of the transfor-
mation seems overestimated. The same tendencies concerning the prediction of the
TRIP kinetics were also shown by Grostabussiat et al. (1999) and also by Videau et
al. (1995) studying TRIP phenomenon in martensitic transformation of a Fe-20Ni-
0.5C steel. Many reasons may be invoked to explain these dierences. The purpose
of the present work is a more precise formulation and resolution of the micro-
mechanical problem of Section 2.2. This extension, which relaxes some of the
restrictive assumptions made by Leblond et al. (1989), will be shown to avoid sin-
gularity at z=0.
3. Reevaluation of the micromechanical model
3.1. Presentation
The assumptions made by Leblond et al. (1989), recalled in the previous section,
can be classied in four main categories related to,
(a) the mechanism of TRIP: Magee mechanism is supposed to be not signicant,
(b) the elastoplastic behavior in the phases:
H.2: for small or moderately high applied stresses, the austenitic phase is
entirely plastic, but the a phase remains elastic or its plastic strain rate
remains always much smaller than that of the g phase. This hypothesis
has later been extended as follows:
(H.2a): Elastic strains are much smaller than plastic strains in the
austenitic phase.
(H.2b): The (compressive) forces exerted by the g crust on the a core
have a negligible eect on the volume of the latter.
H.3: Both phases are ideal-plastic and obey the Von Mises criterion and the
Prandtl-Reuss ow rule.
(c) the transition between the microscopic and the macroscopic behavior in the
phases:
H.1: the microscopic elastic compliance tensor may be equated to the mac-
roscopic overall elastic compliance tensor.
H.4: correlations between oc
eq
1
,oz and s
1
can be neglected.
H.5: For small applied stresses, the average stress deviator in phase 1 is
almost equal to the overall average stress deviator.
(d) the assumed spherical geometry for the micromechanical problem.
Our primary objective is to remove assumption H.2 and its extensions while
keeping the others.
Assumption (a) is a physical one which may be satised or not according to the
investigated material, and as mentioned before, it has been justied in our case by
the experimental results described in Grostabussiat et al. (2001).
Assumptions H.3, (c) and (d) are rough mathematical simplications allowing a
convenient treatment. It should also be noted that assumption H.3 may be partially
loosened by considering some kind of global hardening like Leblond (1989), but the
basic analysis will remain elastic perfectly plastic. The inuence of the elastic beha-
vior of both phases in fact is our main concern.
Following assumption (d), the same geometry as in Fig. 1 is considered:
The product phase is considered as a core, 0-r4R
2
=z
1/3
R
1
, which expands in the
parent phase (shell) R
2
4r4R
1
with a vanishing radial stress s
r
at the outer
boundary r=R
1
. The loading process in fact results from the expansion Dc
12
corresponding to the transformation.
The parent phase consists of an outer elastic region (A
e
) around an inner plastic
region (A
p
). The elastic-plastic boundary is at r = , R
2
4 4R
1
(see Fig. 3).
3.2. Governing equations
Due to spherical symmetry, spherical coordinates r, y and j will be used
throughout.
^the displacement is purely radial, u
r
= u r ( ), u
= u
= 0
u =
u
0
0
_
_
_
_
= c =
u
/
0 0
0 u,r 0
0 0 u,r
_
_
_
_
=
c
r
0 0
0 c
0
0 0 c
_
_
_
_
=
c
r
0 0
0 c
0
0 0 c
_
_
_
_
(12)
c
r
and c
are respectively the radial and the circumferential components of the strain
tensor.
^the spherical stress tensor will be decomposed in its spherical and deviatoric
parts r ( ) and t r ( )
Fig. 3. New geometry taken under consideration for the micromechanical analysis.
o =
o
r
0 0
0 o
0
0 0 o
_
_
_
_
=
o
r
0 0
0 o
0
0 0 o
_
_
_
_
= (r)
1 0 0
0 1 0
0 0 1
_
_
_
_
t(r)
2 0 0
0 1 0
0 0 1
_
_
_
_
Equilibrium provides a dierential equation relating r ( ) and t r ( ),
2t
/
r ( ) 6
t r ( )
r

/
r ( ) = 0 (13)
The strain and stress tensors are related through the elastic plastic constitutive
equation which is based on the strain decomposition,
c = c
e
c
thm
c
p
The elastic strain is given by Hookes law,
c
e
r
=

3K

t
j
. c
e
=

3K

t
2j
(14)
where K and j respectively denote the bulk and shear elastic moduli.
The plastic strain rate results from the yield condition and plastic evolution law.
Using Von Mises plasticity with tensile yield limit o
y
, this gives,
c
.
p
r
= c
.
p
= 0 if t [ [ -
o
y
3
.
c
.
p
r
= 2c
.
p
= lsgnt if t [ [ =
o
y
3
.
(15)
where, l = c
.
eq
50 is the plastic multiplicator and sgnt =
t
t [ [
.
Finally the thermo metallurgical strain in the two phases are related by,
c
thm
2
= c
thm
1
Dc
12
where Dc
12
is the transformation strain at the prescribed temperature. Strictly
speaking Dc
12
should depend on the temperature and therefore on z, but for con-
venience this will be neglected in the following.
1
Using the parent phase stress free conguration as reference, this nally results in,
c
thm
1
= 0. c
thm
2
= Dc
12
(16)
3.3. Elastic solution
The rst step in the analysis is the elastic solution (c
p
= 0),
1
In fact the most part of the transformation occurs in a rather small temperature range so that the
resulting variation in Dc
12
can be neglected.
c
1
= c
e
1
. c
2
= c
e
2
Dc
12
The equilibrium Eq. (13) has to be combined with Hookes law resulting in Navi-
ers equation which in our case gives,
div uu = Tr c ( ) = u
/
2
u
r
= C
where C is a constant.
The corresponding solution being,
u r ( ) = o
i
r
[
i
r
2
where o
i
and [
i
are two constants.
In the product phase 0 4r - R
2
, i = 2, the constant [
2
must vanish to avoid sin-
gularity at the origin, hence,
u = o
2
r. t
2
= 0. and
2
= 3K o
2
Dc
12
( ) (17)
The product phase is under hydrostatic stress and will therefore remain elastic
whatever the yield limit o
y
2
.
In the parent phase, on the contrary, [
1
will be present and,
u r ( ) = o
1
r
[
1
r
2
.
1
= 3Ko
1
and t
1
=
2j[
1
r
3
(18)
In the completely elastic case the three constants o
1
, o
2
and [
1
are determined
from the boundary conditions o
r
R
1
( ) = 0. and the continuity condition for u and o
r
at the boundary r = R
2
. The solution is (z =
R
2
R
1
_ _
3
),
o
1
= zDc
12
4j
4j 3K
. [
1
= zDc
12
R
3
1
3K
4j 3K
. o
2
= Dc
12
4jz 3K
4j 3K
. and
t
1
=
6Kj
4j 3K
R
3
1
r
3
Dc
12
z.
The maximum value of t is reached at the interface of the phases. This elastic solu-
tion will be valid if this maximum value is smaller than
o
y
1
3
[see Eq. (15)] therefore if,
a =
9Kj
4j 3K
2Dc
12
o
y
1
- 1 (19)
This condition only depends on the material characteristics. It follows from
this that two dierent situations may occur according to the value of the material
parameter a:
+ if a-1, transformation plasticity does not exist (in the present approximation
of course),
+ if a>1, then plasticity will occur in the inner parent phase from the very
beginning of the transformation resulting in the geometrical situation
described in Fig. 3 with a plastic part for R
2
- r - and an elastic one for
- r - R
1
if - R
1
.
In order to have an idea about the value of the parameter a, let us consider for
instance the continuous cooling transformations of a 16 MND 5 steel.
Dc
12
which appears in condition (19) is estimated by,
Dc
12
T ( ) = o
2
o
1
( ) T 20 ( ) Dc
20
C
12
where o
2
- 15 10
6
C
1
and o
1
- 23.510
6
C
1
are respectively the thermal
expansion coecients of the product and parent phases after Dupas and Waeckel
(1994). Dc
20
C
12
= 0.01 after Taleb (1998).
Table 1 gives the values of the parameter a for the ferritic, bainitic and martensitic
transformations.
T
s
is the temperature corresponding to the beginning of the transformation, after
Taleb (1998),
E T
s
( ) is the elastic modulus at T
s
, u is the Poissons coecient and o
y
1
T
s
( ) is the
austenitic yield stress at the temperature T
s
, after Dupas and Waeckel (1994).
The material parameter a is in all cases much bigger than one so that transfor-
mation plasticity will occur for any z>0.
3.4. Elastic plastic solution
Eqs. (17) and (18) remain valid respectively in the product phase (0 4r - R
2
) and
in the (assumed) elastic part of the parent phase ( 4r - R
1
) while in the plastic part
(R
2
4r - ) the equilibrium equation now completed by the yield condition
(t
1
=
o
y
1
3
). is integrated to give
1
.
t
1
=
o
y
1
3
.
1
= 2o
y
1
Ln
r
,
_ _
(20)
Table 1
Values of the parameter a at the beginning of ferritic (perlitic), bainitic and martensitic anisothermal
transformations in a 16 MND 5 steel
Transformation T
s
, (
C) E T
s
( ) (MPa) u o
y
1
T
s
( ) (MPa) Dc
12
T
s
( ) (%) a
Ferritic 750 64,000 0.3 67 0.38 6
Bainitic 565 158,500 0.3 109 0.54 12.9
Martensitic 380 177,050 0.3 155 0.70 13
The displacement u then directly follows from Hookes law (u
/
2
u
r
= Tr c ( ) =

K
),
u
1
=
2o
y
1
3K
rLn
r
,
_ _
r
3
_ _
j
r
2
(21)
The complete solution now depends on the ve constants, o
2
for the product
phase, o
1
, [
1
for the outer (elastic) part of the parent phase, ,, j for the inner plastic
part of the parent phase, and also on the plastic boundary .
These constants are determined from the 6 boundary conditions,
(a) no external load,
(b), (c) continuity of o
r
and u at r = R
2
,
(d), e) continuity of o
r
and u at r = ,
(f) continuity of t r ( ) at r = .
Strictly speaking (f) is not a boundary condition, it expresses the fact that the yield
limit is reached in the elastic zone at the elastic-plastic boundary.
+ Condition (a) substituted in (18) leads to,
o
1
=
4j
3K
[
1
R
3
1
+ Conditions (b) and (c) together with (17) and (20) result in,
Ln, = LnR
2

1
3

3
2
K
o
y
1
o
2
Dc
12
( ) and j = R
3
2
Dc
12
+ Condition (f) substituted in (18) and (20) gives,
[
1
=
o
y
1
6j
3
Conditions (d) and (e) provide respectively,
=
3Dc
12
o
y
1
6Kj
4j 3K
_ _1
3
R
2
. and
o
2
=
2o
y
1
9K
3
R
3
1
2
3
o
y
1
K
Ln

R
2
_ _
Dc
12
R
3
2
3

o
y
1
6j
Dc
12
Finally, the complete solution is given by Eqs. (20) and (21) with,
o
1
=
4j
4j 3K
Dc
12
R
3
2
R
3
1
.
o
2
=
4j
4j 3K
Dc
12
R
3
2
R
3
1
2
9
o
y
1
K
Ln
18Kj
4j 3K
Dc
12
o
y
1
_ _
4j 3K
18Kj
o
y
1

o
y
1
6j
Dc
12
.
[
1
=
3Kj
4j 3K
Dc
12
R
3
2
.
=
2Dc
12
o
y
1
9Kj
4j 3K
_ _1
3
R
2
.
j = R
3
2
Dc
12
and Ln, = LnR
2

1
3

3
2
K
o
y
1
o
2
Dc
12
( ). (22)
One important result is the fact that the ratio

R
2
is given from the coecient a
previously dened in Eq. (19) (a>1 in the plastic case). The corresponding values in
the special cases discussed above (see Table 1) are 6, 12.9 and 13 respectively at the
beginning of the ferritic, bainitic and martensitic transformations in a 16MND5
steel. This clearly shows the limitation of Leblonds assumption H2. More precisely,
for small values of z (at the beginning of the stransformation), the parent phase is
not entirely plastied. Complete plastication will occur when reaches R
1
which
corresponds to a limit value z
l
of z given by,
z
l
=
1
a
.
which is respectively equal to about 0.2 for ferritic transformation and 0.09 for bai-
nitic and martensitic transformations in a 16MND5 steel.
3.5. Plastic solution
For z > z
l
, the parent phase is completely plastic, the solution is now,
. If 0 - r 4R
2
, Eq. (17) remains valid with one constant o
2
,
. If R
2
- r 4R
1
, Eqs. (12), (14) and (13), (15) respectively lead to,
u
/
2
u
r
=

1
K
. and
1
= 2o
y
1
Ln
r
,
. t
1
=
o
y
1
3
(23)
so that by integration,
u
1
r ( ) =
2o
y
1
3K
rLn
r
,
_ _
r
3
_ _
j
r
2
(24)
with two constants , and j.
The complete solution therefore involves 3 constants namely o
2
, , and j which are
determined from the boundary conditions,
. No external force is applied,
Eq. (23) gives,
, = R
1
e
1
3
(25)
. Continuity of o
r
at r = R
2
,
Eqs. (17), (23) and (25) lead to,
o
2
= Dc
12

2o
y
1
3K
Ln
R
2
R
1
(26)
. Continuity of u at r = R
2
,
from Eqs. (17), (24), (25) and (26) we have,
j = R
3
2
Dc
12
(27)
A complete solution of the elasticplastic problem has thus be obtained.
4. Evaluation of the TRIP
The transformation plasticity may now be evaluated from Eq. (5). It should
however be noted that when deriving Eq. (2) from (1) the plastic contribution
resulting from the newly formed phase still disappears due to the hydrostatic stress
state in the product phase 0 - r 4R
2
[Eq. (17)] which induces a purely elastic
deformation.
4.1. The plastic case
According to Eq. (15) and remarking that in the parent phase, t > 0,
oc
eq
1
= 2oc
p
The plastic strain c

p
is obtained as the dierence between the total strain c
and
the elastic strain c
e
[remember that in the parent phase c

thm
= 0, see Eq. (16)].
c
is given by Eq. (12),

c
=
u
1
r
.
where u
1
is obtained from Eqs. (24), (25) and (27),
u
1
=
2o
y
1
3K
rLn
r
R
1
_ _
Dc
12
R
3
2
r
2
.
so that,
c
=
2o
y
1
3K
Ln
r
R
1
_ _
Dc
12
R
3
2
r
3
.
The elastic strain results from Hookes law [Eqs. (14)] where
1
and t
1
are deduced
from Eqs. (23) and (25),
1
= 2o
y
1
Ln
r
R
1
_ _
2
3
o
y
1
. t
1
=
o
y
1
3
so that.
c
e
=
2o
y
1
3K
Ln
r
R
1
_ _
2
9
o
y
1
K

o
y
1
6j
The plastic component c
p
therefore results as,

c
p
= Dc
12
R
3
2
r
3

2o
y
1
9K

o
y
1
6j
(28)
and,
oc
p
oz
= Dc
12
R
3
1
r
3
.
so that,
oc
eq
1
oz
= 2Dc
12
R
3
1
r
3
and.
oc
eq
1
oz
_ _
V
1
=
1
V
1
_
R
1
R
2
2Dc
12
R
3
1
r
3
4r
2
dr = 2Dc
12
4R
3
1
4
3
R
3
1
R
3
2
_ _
Ln r ( )]
R
1
R
2
=
2Dc
12
1 z
Ln z ( )
Finally, according to Eq. (5),
E
.
tp
=
2Dc
12
o
y
1
Ln z ( )z
.
3
2
S (29)
This relation coincides with Leblonds results. It follows from this analysis that
taking into account the elastic deformations in both phases does not change the nal
result. This would not be true if elasticity had been taken into account in one phase
only as noted by Taleb (1999).
For the reasons mentioned in Section 2.3., Leblond et al. have proposed to cut o
the transformation plasticity below z = 0.03 leading to Eq. (9).
4.2. The elasto plastic case
As shown in Section 3.4, when z - z
l
the parent phase is not entirely plastic. Pro-
ceeding as in Section 4.1, Eqs. (20)(22) combined with Eqs. (12), (14) and (15) lead to,
c
p
= Dc
12
R
3
2
r
3

2o
y
1
9K

o
y
1
6j
(30)
which is the same as Eq. (28) obtained in the plastic case. So that,
oc
eq
1
oz
= 2Dc
12
R
3
1
r
3
and.
oc
eq
1
oz
_ _
V
1
=
1
V
1
_
a
1
3
R
2
R
2
2Dc
12
R
3
1
r
3
4r
2
dr = 2Dc
12
4R
3
1
4
3
(R
3
1
R
3
2
)
Ln r ( )]
a
1
3
R
2
R
2
= 2Dc
12
Ln a ( )
1 z
Finally, according to Eq. (5),
E
.
tp
=
2Dc
12
o
y
1
Ln z
l
( )z
.
3
2
S (31)
So that the following new model is proposed extending properly Leblonds model
to low values of z,
E
.
tp
=
2Dc
12
o
y
1
Ln z
l
( )z
.
3
2
S if z 4z
l
2Dc
12
o
y
1
Ln z ( )z
.
3
2
S if z > z
l
_
_
with z
l
=
o
y
1
2Dc
12
4j 3K
9Kj
It follows from this that the singularity obtained for z = 0 in the original
Leblonds analysis clearly results from the unnecessary assumption H.2.
5. Application
Let us now come back to the experimental results presented by Taleb et al. (2001)
and which were already mentioned in Section 2.4. In these experiments, transfor-
mation induced plasticity was characterized by the dierence between a free dilato-
metric test (no applied external stress) under prescribed cooling conditions and a
TRIP test corresponding to a xed external stress applied during the transformation
and with the same cooling conditions.
These experiments were performed on a 16MND5 steel in French AFNOR
(equivalent to the SA508Cl3 steel in ASTM norm). After the austenitization process,
the controlled cooling rate was equal to 0.3

C/s leading to bainitic transformation
for both tests. The prescribed stress was equal to about 24 MPa for the TRIP test.
As mentioned earlier, the kinetics of the bainitic transformation can be considered
for such a level of applied stress, as being the same for the two experiments. In these
experimental conditions, the bainitic volume fraction z therefore only depends on
the temperature T and its value for a given temperature is obtained from the free
dilatometric test and using a simple linear mixture law. The TRIP strain is then
evaluated as a function of T, and then of z, as the strain dierence between the TRIP
test and the free dilatometric test. More details on these test results are given in
Taleb et al. (2001). It should be noted that this procedure is performed at constant
stress and also eliminates the thermometallurgical strain which, according to the
z(T) assumption, is the same in both tests. It therefore characterizes the transfor-
mation plasticity, as dened in (3), as the macroscopic plastic strain resulting from
the transformation at constant stress and temperature.
The results of our model are compared to the predictions of Leblonds model in
which the threshold is assumed to be equal to 0.01 (Leblond 0.01 in Fig. 4). As dis-
cussed in Section 2.1 a global averaged hardening has been taken into account by
considering the yield stress o
y
1
in Eqs. (29) and (31) as depending on the averaged
plastic strain through a linear hardening rule
o
y
1
= o
y.0
1
hE
eff
1
where, o
y.0
1
and h are respectively the initial yield stress and the strain hardening
modulus.
E
.
eff
1
=
0 if z 40.03
2Dc
12
1 z
Ln z ( )z
.
if z > 0.03
_
.
for Leblonds model and,
E
.
eff
1
=
2Dc
12
1 z
Ln z
l
( )z
.
if z 4z
l
2Dc
12
1 z
Ln z ( )z
.
if z > z
l
_
_
where, z
l
=
o
y
1
2Dc
12
4j 3K
9Kj
for the new model.
The values of the parameters o
y.0
1
, h and Dc
12
have been taken from Dupas and
Waeckel (1994).
The comparison of the theoretical and the experimental results shows that our
rened micromechanical model does not explain the fast transformation plasticity
observed experimentally at the beginning of the transformation. Furthermore a
surprising agreement with Leblonds model is observed when the threshold is chosen
equal to 0.03 (Leblond 0.03 in Fig. 4).
When this threshold is taken equal to 0.01, the new model leads to lower trans-
formation plasticity value at the end of transformation which is closer to experi-
mental result, however the latter seems still overestimated. The reason probably is
Fig. 4. Bainitic transformation under applied stress 24 MPa (about a quarter of the austenitic conven-
tional yield stress at the beginning of the transformation). Transformation induced plasticity (a) and
normalized transformation induced plasticity (b) versus the volume fraction of the formed bainite.
that as discussed in Section 2.1 and by Leblond et al. (1989), S
1
,S is rather a
decreasing function of z. The modied theoretical curve obtained by replacing S
1
by
1 z
2
_ _
S (New model : S*(1z*z) in Fig. 4) leads to a better agreement.
More discrepancy between Leblonds model and the new one is expected for a
ferritic transformation in the same steel because of z
l
is much higher.
6. Conclusion
A modied micromechanical model for Leblonds approach to TRIP has been
proposed, the main features of this model are:
(a) It allows to remove some assumptions made by Leblond et al. namely : elastic
behavior of the product phase, rigid plastic behavior of the parent phase,
(b) Leblonds model can be now properly extended to low z values,
(c) It leads to lower TRIP values at the end of transformation, which is in
agreement with experimental values.
This improved agreement is realized by accounting for both plastic and elastic
regions in the parent phase.
The analysis leads to the introduction of the dimensionless material parameter,
z
l
=
o
y
1
2Dc
12
4j 3K
9Kj
, which appears as an essential characteristic of transformation
induced plasticity.
Further experiments and more rened nite element micromechanical analyses
taking into account the recent developments on this subject, are needed and will be
performed in the future in order to conrm the proposed model.
References
Abrassart, F., 1972. Inuence des Transformations Martensitiques sur les Proprie te s Me caniques des
Alliages du Syste` me FeNiCrC. The` se de tat, Universite de Nancy I, France.
Barbe, F., Decker, L., Jeulin, D., Cailletaud, G., 2001a. Intergranular and intragranular behavior of
polycristalline aggregates. Part 2: results. Int. J. Plasticity 17, 537563.
Barbe, F., Forest, S., Cailletaud, G., 2001b. Intergranular and intragranular behavior of polycristalline
aggregates. Part 1: F.E. model. Int. J. Plasticity 17, 513536.
Cavallo, N., 1998. Contribution a` la Validation Expe rimentale de Mode` les De crivant la ZAT lors dune
Ope ration de Soudage. The` se de Doctorat, INSA de Lyon, France (URGC/Structures).
Cherkaoui, M., Berveiller, M., Sabar, H., 1998. Micromechanical modeling of martensitic transformation
induced plasticity (TRIP) in austenitic single crystals. Int. J. Plasticity 14, 597626.
Coret, M., Calloch, S., Combescure, A., 2002. Experimental study of the phase transformation plasticity
of 16MND5 low carbon steel under multiaxial loading. Int. J. Plasticity 18, 17071727.
Desalos, Y., 1981. Comportement Dilatome trique et Me canique de lAuste nite Me tastable dun Acier A
533. IRSID Report no. 95 34 94 01 MET 44.
Dupas, P., Waeckel, F., 1994. Recueil Bibliographique de Caracte ristiques Thermome caniques pour
lAcier de Cuve, les Reve tements Inoxydables et les Alliages 182 et 600. Report EDF/DER (France) HI-
74/93/097, HT-26/93/058/A.
Fischer, F.D., Sun, Q.-P., Tanaka, K., 1996. Transformation induced plasticity (TRIP). Appl. Mech. Rev.
49, 317364.
Fischer, F.D., Reisner, G., Werner, E., Tanaka, K., Cailletaud, G., Antretter, T., 2000a. A new view on
transformation induced plasticity (TRIP). Int. J. Plasticity 16, 723748.
Fischer, F.D., Antretter, T., Azzouz, F., Cailletaud, G., Pineau, A., Tanaka, K., Nagayama, K., 2000b.
The role of backstress in phase transforming steels. Arch. Mech 52, 45. 569-588.
Gautier, E., Simon, A., Beck, G., 1985. Etude du comportement me canique associe a` la transformation
martensitique dalliages de type FeNiC sollicite s en uage ou en traction. In: Mc Queen, Dickson,
J.I., Jonas, J.J., Acken, M.G. (Eds.), Proc. 7th International Conference on Strength of Metal and
Alloys (ICSMA 7). p. 369.
Greenwood, G.W., Johnson, R.H., 1965. The deformation of metals under small stresses during phase
transformations. Proc. Roy. Soc. London A283, 403.
Grostabussiat, S., Taleb, L., Jullien, J.F., Waeckel, F., 1999. Transformation Induced Plasticity for a
16MND5 steel. 3
rd
Int. Congress on Thermal Stresses. Cracow, Poland, 1317 June.
Grostabussiat, S., Taleb, L., Jullien, J.F., Sidoro, F., 2001. Transformation induced plasticity in mar-
tensitic transformation of ferrous alloys. J. Phys. IV 11, 173180.
Hill, R., 1967. The essential structure of constitutive laws of metal composites and polycristals. J. Mech.
Phys. Solids 15, 79.
Leblond, J.B., Mottet, G., Devaux, J.C., 1986. A theoretical and numerical approach to the plastic
behavior of steels during phase transformationsI. Derivation of general relations. J. Mech. Phys.
Solids 34, 395409.
Leblond, J.B., Devaux, J., Devaux, J.C., 1989. Mathematical modelling of transformation plasticity in
steelsI. Case of ideal-plastic phases. Int. J. Plasticity 5, 551572.
Leblond, J.B., 1989. Mathematical modelling of transformation plasticity in steelsII. Coupling with
strain hardening phenomena. Int. J. Plasticity 5, 573591.
Nagayama, K., Kitajima, Y., Kigami, Tanaka, K., Fischer, S., Cailletaud, F.D., 2000. Transformation
induced plasticity in maraging steel: an experimental study. Key Engineering Materials 177180, 443
448.
Nagayama, K., Terasaki, T., Tanaka, K., Fischer, F.D., Antretter, T., Cailletaud, G., Azzouz, F., 2001a.
Mechanical properties of a CrNiMoAlTi maraging steel in the process of martensitic transforma-
tion. Materials Science and Engineering A308 2537.
Nagayama, K., Terasaki, T., Goto, S., Tanaka, K., Antretter, T., Fischer, F.D., Cailletaud, G., Azzouz,
F. Backstress evolution and iso-volume fraction lines in a CrNiMoAlTi maraging steel in the pro-
cess of martensitic transformation. Materials Science and Engineering (in press).
Magee, C.L., 1966. Transformation Kinetics, Microplasticity and Ageing of Martensite in Fe3lNi. PhD
Thesis, Carnegie Institute of Technologie University, Pittsburgh, PA.
Simon, A., Denis, S., Gautier, E., 1994. Eet des sollicitations me caniques sur les transformations de
phases dans le tat solide. Aspects me tallurgiques et me caniques. Journal de Physique IV, Colloque C3
Supple ment au Journal de Physique III 4, 199.
Taleb, L., 1998. Etude du Comportement Me canique des Aciers en Cours de Transformations Me tallur-
giques. Synthe` se des Re sultats des Essais Base de Donne es pour des Transformations Uniques. Report
5c, INSA Lyon, France. EDF/DER, 174/IJ6549/IMA 277.
Taleb, L., 1999. Sur le Comportement Thermome canique des Mate riaux et des Structures Me talliques.
Habilitation a` Diriger des Recherches. INSA de Lyon. URGC/Structures. No. 99001.
Taleb, L., Cavallo, N., Waeckel, F., 2001. Experimental analysis of transformation plasticity. Int. J.
Plasticity 17, 120.
Videau, J.Ch., Cailletaud, G., Pineau, A., 1995. Experimental study of the transformation induced plas-
ticity in a CrNiMoAlTi steel. International Seminar MECAMAT95: Mechanisms and Mechanics
of Solid Phase Transformations, la Bresse, France, 1619 May.

Lakhdar Taleb François Sidoroff - A Micromechanical Modeling of The Greenwood-Johnson Mechanism in Transformation Induced Plasticity

Cargado por

Información del documento

Descripción original:

Título original

Derechos de autor

Formatos disponibles

Compartir este documento

Compartir o incrustar documentos

Opciones para compartir

¿Le pareció útil este documento?

¿Este contenido es inapropiado?

Copyright:

Formatos disponibles

Lakhdar Taleb François Sidoroff - A Micromechanical Modeling of The Greenwood-Johnson Mechanism in Transformation Induced Plasticity

Cargado por

Copyright:

Formatos disponibles

A micromechanical modeling of the

The plastic strain c

is obtained as the dierence between the total strain c

[remember that in the parent phase c

is given by Eq. (12),

therefore results as,

También podría gustarte