Microstructure Towards Plastic Behavior

IAP P5/08 Progress Report 2002
PROGRESS REPORT 2002

15-03-2003
IAP-action: phase V
2002 - 2006
Project P5/08
From microstructure towards plastic behavior

of single- and multiphase materials
Partner 1: Katholieke Universiteit Leuven, departement materiaalkunde (KUL-MTM)

Partner 2: Université Catholique de Louvain, Unité de physico-chimie et ingénierie
des matériaux (UCL-PCIM)
Partner 3: Technische Universiteit Eindhoven, Mechanical Engineering (TUE)
Partner 4: Université de Liège, département MSM (ULg-MSM)
Partner 5: Vrije Universiteit Brussel, departement MEMC (VUB)
Partner 6: Katholieke Universiteit Leuven, dept. mechanica, afdeling PMA (KUL-PMA)
Partner 7: Koninklijke Militaire School, leerstoel bouwkunde (KMS)
Partner 8: Rijksuniversiteit Groningen, toegepaste fysica, afd. micromechanica (RuG)
Partner 9: Université Paris 13, LPMTM-CNRS (UP13)
1
1. SCIENTIFIC ACTIVITIES
Introduction
This report contains an overview of the main scientific activities, carried out during
the first year of the IAP P5/08 project. The results are reported per work package.
On the experimental side, materials have been prepared (WP1), forming tests (WP2),
mechanical tests (WP3) and microstructural characterization (WP4) have been started.
Several of these tests had an “explorative” character, in the sense that some test
procedures had to be established and characterization methods had to be optimized.
Most activities concerning modeling work, were related to the exchange and
integration of existing programs and/or subroutines. The first concepts for integrated
models have been established and discussed. The connection between the
experimental work and the modeling work will be optimized in 2003.
Work Package 1: Materials Preparation
1. Titanium-alloys
The main alloy chosen for the present project is Ti 6Al 4V. The first samples were
prepared by partner 7 from sheet material. This sheet was annealed after hot and cold
rolling, and has a grain size of 10 µm. Starting from this condition, thermal treatments
were carried out in order to generate different microstructures, as required for this
project:
- heating to a homogeneous temperature of respectively 850, 950 and 1000 °C
- quenched, air cooling and furnace cooling.
This material will further be used by partners 1, 4 and 7
Partner 2 has focused considerable attention on the properties of metastable beta
titanium alloys. Recently, a new alloy, Timetal LCB, Ti-6.8Mo-4.5Fe-1.5Al (in
wt.%), has been introduced. This material is expected to be used in automotive and
aerospace industry since its production cost is relatively low while its mechanical and
working properties are excellent. The alloy is beta-metastable since it preserves a
metastable bcc lattice structure (beta phase) when it is solution heat-treated above the
beta-transus and subsequently quenched. Annealing of the alloy below Tbeta then
causes the precipitation of the alpha phase in much finer form than is the case in
alfa/beta alloys like Ti-6Al-4V. This results in better mechanical properties (yield
strength and ductility). Furthermore, depending on the thermo-mechanical conditions,
the alpha phase nucleation could arise from the existence of a transition phase, omega,
allowing still finer precipitation and dispersion of the alfa phase.
2. Fine grained Aluminium-alloys

Material preparation, carried out by partner 1, starts with a slab of hot rolled
commercial purity aluminium (AA1050), with a thickness of 23mm. Blocks are cut
off, parallel to the rolling direction. Samples with a diameter of 12mm and a length of
40mm or 60mm are cut out of these blocks by machining. As a result of this
2
preparation method, the axis of these cylinders corresponds with the former rolling
direction, but the transverse direction and the normal direction are lost.
The ECAE technique (Equal Channel Angular Extrusion) was used to refine the grain
size of the samples by severe plastic deformation. Route C (sample rotation over 180°
after each pass) was chosen because of its tendency to develop a substructure after
fewer ECAE passing then the other routes.
The final step in material preparation is cutting the samples for further analysis.
Depending on the test that will be performed, different samples will be cut out. In the
first phase, all sections are transverse sections. Starting and ending piece: these
samples are preserved but not prepared for further analysis, because this part of the
material was not in a steady state deformation. For heat treatment, small samples of a
few millimeters thick are cut out. These samples also serve for microstructural
characterisation and hardness measurement. The samples are cut out of the middle of
the ECAE-material, because this is thought to be the most homogenously deformed
zone. Samples for compression tests are taken from the outer zones, after cutting of
the starting and ending pieces.
This fine grained material will in the next year be further characterized by partner 1
and used in WP6 (phenomenological modeling) by partners 1 and 4.
1 3 2 3 1
I O
end start
Figure 1 - Different zones of the ECAE sample for further analysis. (1) rejected, (2) heat
treatment, hardness measurement, microstructural characterisation, (3) compression testing.
3. IF-steels
Partner 1 has received IF steel as 65% hot rolled material with initial grain size of 100
µm. From these slabs cylindrical samples with a diameter of 12 mm and a length of 4
cm to 5 cm were machined. The ECAE technique was used to refine the grain size.
The ECAE die had an angle of 90° (corresponding to an equivalent strain of 1,15 per
pass). The processing temperature was 200°C. On one hand, samples were prepared
with an effective strain of 9.2 following route BA (rotation of +90° after steps 1 and 3
and rotation of –90° after steps 2 and 4) in order to study the grain fragmentation at
large strains. On the other hand the 4 classical ECAE routes A, BA, BC and C were
used to prepare samples in order to study the influence of the strain path followed
during severe plastic deformation.
4. TRIP-steel
After hot and cold rolling following classical practice, the heat-treatments to generate
the required multiphase microstructures were carried out on samples oriented in the
rolling direction. The times and temperatures of the intercritical annealing and of the
bainitic holding were adjusted depending on the required microstructure.
Several specimens presenting different microstructures for which key parameters,
such as the volume fraction of the phases and the austenite stability, were processed
by partner 2, following different heat-treatment conditions. This material was further
3
used by partners 2 (characterization and micro-mechanical tests), partner 1 (pre

straining experiments) and partner 6 (bending experiments).
Processing of Acicular Ferrite Multiphase Steels
Three steels containing 0.2wt.%C-0.4wt.%Si-1.5wt.%Mn and different levels of Cu,
Ni and Ti (that are supposed to facilitate the intragranular nucleation) were studied.
Heat-treatments were conducted in a dilatometer. Specimens were austenitised at
1100°C for 5 min (giving an austenite grain size of 20 µm), then cooled down at a rate
of 30°C/s, maintained at temperatures ranging from 400°C to 620°C for times
comprised between 30s and 5min. After the isothermal holding, the specimens were
quenched at 30°C/s to room temperature.
Processing of Fe-Mn Steels and Austenitic Stainless Steels
Three different steel grades were studied: the AISI 301L stainless steel (containing
6.5% Ni, 18.5% Cr, 1.5% Mn and 0.5% Si) and two Fe-Mn alloys containing virtually
no carbon, 3% Si, 3% Al and 15% or 20% Mn (thereafter named Mn15 and Mn20,
respectively). After casting, the two Fe-Mn steels were homogenised for 24h at
1100°C and then hot rolled between 1100°C and 900°C. The specimens were kept at
600°C for one hour for coiling simulation. After complete cooling, they were re-
heated to 1100°C and hot rolled in 2 steps to the final thickness of 1.5mm. The
influence of 4 different annealing conditions following hot rolling were tested : no
annealing or 1h at 900°C, 1000°C or 1100°C, respectively.
The 301L steel was received as sheets of 1.5mm in thickness. The influence of cold
rolling cycles (cold rolling 20 to 30% then 15 minutes at 900°C) on the texture
evolution was tested on the 301L
5. TWIP steels
Two partners are involved in the research on TWIP steels (partner 1 and partner 2).
Three different materials are studied, all based on Fe-Mn-Al-Si alloys.
Two of the alloys have already been mentioned in the description of the materials for
the TRIP-research (cf. paragraph 4: Mn15 and Mn20 alloys). These materials are also
integrated in the TWIP research because the martensitic transformation and
mechanical twinning are competing deformation mechanisms, depending on
experimental conditions (stability of austenite). Details of their processing have been
given above.
The third material, studied by partner 1 is a Fe-30Mn-3Al-3Si alloy that was melted in
a vacuum induction furnace and ingot cast. Plates were cut from the ingot,
homogenised in an air furnace at 1200°C (1 hour), hot rolled and air cooled. The
plates were hot rolled in 6 passes: total strain of 80% (20mm to 4mm). The hot rolling
start and end temperature were 1100°C and 800°C, respectively. The hot rolled plates
were annealed at 1100°C for 20 min and water quenched. The SFE of the material is
about 42 mJ/m2.
6. Pearlite
No material has been prepared yet.
4
Work Package 2: Forming Tests
Introduction
Bending is a process by which a metal is stressed beyond the yield strength but below
the ultimate tensile strength. In order to increase the quality and performance of
bending it is necessary to know the materials characteristics. For this purpose, the
sheet specimen must be bent with a pure couple and simultaneously the bending
moment and curvature of the specimen are measured. This measurement yields the
unit moment curve, which is the appropriate material characteristic to study the
bending process. During the first year of this project, the bending characteristics of
TRIP steel has been investigated by partner 6. The material was received from
partners 1 and 2 and the results returned to them.
Experimental
A special bending machine has been used to apply a pure moment on a test strip
(figure 2). Details about the experimental set-up can be found in: R. Aerens,
Characterisation of materials by bending, WTCM-CRIF Center for scientific and
technical research in metal manufacturing and Z. Marciniak, J. L. Duncan,
Mechanics of Sheet Forming, edited by Edward Arnold, London, 1992, ISBN 0-340-
56405-9. A test strip i.e., a steel sheet, of a definite size (57.8 mm x 57.8 mm) is
clamped with two drums. The drums are lying horizontally on a layer of ball bearings
which allows almost frictionless movement. Since very negligible friction occurs
between drums and balls, the weight of the drums has no effect on the couple
measurement.
The moment-curvature diagram is derived from measurement of load and the cross-
member displacement of the machine. During bending tests, the tensile force is
measured on-line with a load cell, and an image of the bend is recorded with a CCD
camera and an image processing system. To get an exact view of the curvature at the
core of the bend but not at the edge where anti-clastic effect occurs, the samples are
chamfered at 45°. Nevertheless a small correction factor needs to be applied in
curvature measurements. To have better contrast in imaging, the chamfered edge of
the samples was painted half an hour before the bending test. With the image
processing system, a series of points belonging to the contour of the bend is
determined. These points are then exported to a CAD system where points, not strictly
belonging to the inner and the outer radii, are deleted. Subsequently, using the method
of the least squares, the values of the radii of the two concentric circles which fits best
the two series of points are computed.
Results
Unit moment measurements
The unit moment (σ*) vs. strain (εb) curves have been determined on TRIP-steel with
a sheet thickness of 1 mm. In order to compare the unit moment of TRIP steel with
other steels, tests were performed on 304 stainless steel, St-12 and St-14 steels.
Experimental data for two repeated tests on each steel, are exhibited in figure 3. TRIP
steel distinctly shows better performance compared to other steels, but the
reproducibility of the results is less. To investigate the influence of strain rate on unit
moment, tests were performed at different strain rates. No remarkable influence of
strain rate on unit moment could be observed.
5
JACK
*
Unit Moment σ
2F
σ*= b.s
M
2
FORCE BRIDGE
SENSOR
M = ρ.F
F F
b
s/2
R
CAMERA
IMAGE PROCESSING
F F
Ri Re
s Ri + Re
R=
2
Figure 2. Diagram of the device for determining moment-strain relationship
400 1 mm thick, at 4 mm/s

TRIP steel
360 304 stainless steel
St14 steel
320 St12 steel
unit moment (N/mm )
280
2
240
200
160
120
80
40
0
0.00 0.05 0.10 0.15 0.20 0.25 0.30
εb
Figure 3. Unit moment for different steels
The experimental data were best fitted by following equation: σ = C ′′ ε n where C″ is

b b
a constant known as the strength coefficient and n is the strain-hardening exponent or
coefficient. The values of these coefficients for TRIP steel are reported in Table 1. For
comparison the coefficients were also determined for the steels reported in figure 3.
In order to correlate phase transformations with deformation, samples were bent to
different percentage of maximum strain (e.g., 25%, 50% and 75%) at a strain rate of
0.01 s-1. The unit moment curves followed a similar trend as shown in figure 3. The
phase transformations are still to be investigated.
6
Table 1: Coefficients of Unit moment for different materials
Deformation speed C″ n
mm/s
1 495.6 0.296
TRIP steel 4 476.5 0.251
(Thickness: 1 mm) 15 474.3 0.235
40 457.3 0.243
St-14-03 4 168.9 0.260 (roll direction)
(Thickness: 1 mm) 165.4 0.260 (perpendicular)
ST-12 4 178.1 0.260 (roll direction)
INOX 304 4 245.5 0.47 (roll direction)
Spring back determination

Spring back has been determined using an air-bending set-up. In general, spring back
of materials depends upon thickness, angle of bending and on internal properties. The
TRIP steel (1 mm thickness) was bent to an angle of 90° and the experimental value
of the spring back was recorded as 5.6°. For the same sheet thickness and angle of
bending, 304 stainless steel revealed a value of 4.3°.
In addition to spring back calculation, the TRIP steel sheets were bent to several
angles in order to investigate any crack generation during bending. No crack
generation could be revealed under light microscopy (up to 40X) until a bending of
40°, the maximum limit that could be reached by the system available at the PMA
Laboratory.
Conclusions
The tested TRIP-steel is a very promising material because of its very high strength.
But the strain-hardening coefficient (n) measured in this investigation is in the range
of most steels. The repeatability of bending test data is poor. TRIP-steel exhibited a
larger springback compared to stainless steel. No cracks could be generated for a
maximum bending angle of 40°.
Comments
Using the current experimental set-up, bending at larger deformation rate is not
possible due to the lack of a quicker imaging system. It would be interesting to check
the influence of larger deformation rate on unit moment. Phase transformations at
different bending angle are yet to be evaluated. Hardness measurements along cross
section of bent samples are vital. Influence of different sheet thickness can be an
added value for this research. Repeatability of tensile test data is also required to be
know.
7
Work Package 3: Mechanical Tests
1. Titanium alloys
Tensile tests were performed in a tubular furnace with inert atmosphere at different
temperatures (room temperature, 500 and 960°C) and at different strain rates (5.10-4
and 5.10-3). Room temperature tests were reproducible, but at higher temperatures
several experimental problems had, and still have, to be resolved. A comparison
between the experimental facilities of partner 7 and 1 have been carried out and
exchange of expertise between partner 4 and 7 is planned for next year.
2. Fine grained Aluminium alloys

Compression Tests (partner 1)
Until now the Instron 4505 was utilised, to perform the compression, but for future
samples, some tests will also be done on the Instron 1196, because of it’s higher
loading capacity.
• Force: max 100kN
• Speed: 0,25 mm/min
• Stroke: 15mm
• Sample diameter: +/- 11,9mm
• Sample height +/-18mm.
Reliability tests have been carried out and it seems that homogeneous deformation can
be obtained till a true strain of 0.6. More tests are needed to validate the results
between strains of 0.6 and 1.
Figure 4. Compressive stress as function of true strain for CP-Al after different ECAE passes
(Micro)Vickers Hardness measurement

To make sure that the results of the micro-hardness measurement are reliable, 16-24
measurements are taken at different positions on the sample. The normal Vickers
measurement is performed to get an overall value for the hardness. At this moment
too few data are available, because the microstructural characterisation has to be
completed first.
8
3. IF-steel
As a preliminary mechanical characterisation, the Vickers micro-hardness was
measured for the 4 classical routes after 1 pass, 2 passes, 4 passes and 8 passes (i.e. up
to an equivalent strain of 9,2) (partner 1). For each route the hardness increases with
increasing number of passes. The increase is most significant during the first pass, and
decreases during the following passes so that the hardness tends to saturate at the
highest strain level investigated.
After 8 passes the hardness for the 4 routes are significantly different, and can be
classified as follows: BA > BC > A > C.
4. TRIP-steel
Transformation Kinetics
In order to assess the influence of the stress state on the kinetics of the mechanically-
induced martensitic transformation, several mechanical tests were performed on
different multiphase steels: uniaxial tensile tests, tensile tests on DENT and notched
specimens, Marciniak tests, balanced biaxial test and simple shear tests. For each type
of tests, the austenite volume fraction has been measured as a function of the
equivalent strain using X-ray diffraction or scanning electron microscopy, coupled
with image analysis.
Influence of Heterogeneous Stress State on the TRIP Effect
A special bending machine has been designed at the PMA Division of the KUL
(partner 6) and used to apply a pure moment on a test strip (cf. work package 2).
Mechanical Tests of Highly Alloyed Steels
Tensile tests were performed on samples of the Fe-Mn steels annealed in the 4
conditions reported in the Materials Preparation section. The mechanical properties of
the Fe-Mn steels are better after the annealing treatments. It seems that a large TRIP
effect is accompanying the plastic deformation. From 50 to 75% of the initial
austenite content is indeed transformed into martensite at necking. Some Twinning-
induced plasticity seems also to take place in steel Mn20 as shown by TEM.
Steel 301L also exhibits a mechanically-induced martensitic transformation. Some
301L samples were also strained in tension at different testing speeds, together with
the in situ measurements of the temperature rise. These tests show that the mechanical
properties are strongly influenced by the deformation speed. Temperatures as high as
100°C can be reached for crosshead speeds of 25mm/min. This temperature rise
stabilises the austenite and thus reduces the transformation phenomenon. Interrupted
tensile tests were also carried out on steel 301L in order to determine the martensitic
transformation kinetics.
5. TWIP-steel
1. Fe-15Mn-3Al-3Si and Fe-20Mn-3Al-3Si: see above, “Mechanical Tests of Highly
Alloyed Steels”.
2. Fe-30Mn-3Al-3Si: The hot rolled and annealed material was used to carry out cold
rolling at room temperature. Eight different strains were applied, varying from 0,1 to
9
2,5. The rolling direction was the same as for the hot rolling and the cold rolling was
done without reversing the rolling direction.
6. Pearlite
No mechanical tests have been performed yet.
Work Package 4: Microstructural Characterization
1. Titanium
Preliminary tests on the microstructures generated for workpackage 1 were done by
partner 7, using the OIM (Orientation Imaging Microscopy) facility of partner 1. A
suitable specimen preparation procedure for OIM measurements was determined and
the different phases (α, β and even the martensitic α’) could be successfully
identified.
The metastable β-titanium alloy (Timetal LBC) has been investigated by partner 2.
The microstructure of Ti LCB consists of equiaxed β grains, 5 µm in size, with
dispersed finer grains of α, 1 µm in size. The microstructure of this alloy has also
been investigated by X ray diffraction and the results confirm the presence of the two
phases, α and β, respectively.

Microstructural characterisation of samples after different ECAE passes (cf. WP1) has
been carried out by light optical microscopy, using Barkers anodising and with SEM-
EBSD (OIM) (partner 1). DSC measurements have been attempted but up to now, no
reliable data could be obtained. Until now, the results obtained from the EBSD
measurements, seem to be reliable. Further analysis of the data is still needed. Using
some algorithm for rebuilding the grains, the gaps caused by cleaning out bad
patterns, can be filled. The grain size is about 1-3µm. A larger deformation is needed
to transform the substructure with low-angle boundaries into distinct grains with
medium to high angle boundaries. Further analysis of the misorientation between
grains and/or subgrains is still needed, together with a longitudinal section of the
sample determining if the grains are equiaxial or not.
3. IF steel
Fined grained IF steel has been chosen as one of the ‘model materials’ in this IAP
project. Before delivering material to other work packages, the samples prepared by
partner 1 by ECAE (cf. WP1), have to be investigated after different ECAE passes
and as function of different ECAE routes.
The evolution of the substructure of IF steel with strain was followed up to an
effective strain of 9,2 for route BA by partner one, using OIM. It was found that the
degree of fragmentation differs from grain to grain even up to a strain of 9.2. This
non-uniform fragmentation of the initial grains during ECAE is in agreement with the
results of Seefeldt and Van Houtte [1] who modeled the fragmentation in aluminum
during ECAE and found that for some orientations the structural fragmentation into
10
new texture elements was counteracted by the textural rotation during the
deformation, whereas for other orientations they assist each other, resulting in an
orientation dependence of fragmentation. The orientation dependence of
fragmentation will be further investigated experimentally.
From the comparison of IF steel samples processed to a strain of 9.2 by ECAE via the
4 classical routes, a significant influence of the strain path was established. Routes BA
and BC, corresponding to cross loading, produce a higher hardness and finer structure
than routes A and C which correspond to continuous and reverse loading respectively.
The classification of the hardnesses for the 4 routes after 8 passes (BA > BC > A > C)
can be interpreted in terms of the different strain paths realised by the routes. The
change of strain path from one ECAE pass to another can be quantitatively expressed
by the α parameter introduced by Schmitt:
ε&p ⋅ ε&
α=
ε&p ⋅ ε&
where ε& and ε&p are the strain rate tensors of the consecutive passes. Route A
corresponds to an approximately continuous strain path (α = 1) regardless of the die
angle. For a die angle of 90°, route C corresponds to reverse loading (α = -1) and both
routes B correspond to cross loading (α = 0). These values for α correlate with the
hardness of the 4 routes. Reverse loading during route C leads to an inferior hardness
with respect to the continuous loading during route A, because in each pass the same
slip systems are activated as in the previous pass, but in the reverse sense. The
resistance to glide is therefore smaller, and fewer dislocations need to be generated to
accomplish the applied strain. Cross loading during routes BA and BC necessitates the
activation of new slip systems, and the dislocations have to cut through the existing
dislocation pattern. Therefore more dislocations are generated, yielding a higher
hardness. This may also explain the higher fraction of transverse boundaries in the
route B samples, when compared to the route A sample. The difference between the 2
route B samples may be explained by the fact that the initial grain boundaries are
elongated during route BA, leading to an extra increase in high angle boundary
fraction. After 4n (where n is an integer) passes via route BC on the other hand, the
original volume elements are restored to their initial shape, and no extra grain
boundary area is introduced by this geometrical mechanism.
References
[1] M. Seefeldt and P. Van Houtte, Proceedings of the 2nd International Conference
on Nanomaterials by Severe Plastic Deformation, in press.
4. TRIP steel
Characterisation of Acicular Ferrite Microstructures
The microstructure of the steel grades devoted to the formation of acicular ferrite
(presented in the Materials Preparation section) was characterised by SEM after
mechanical polishing and nital etching. Acicular ferrite was found only for the steel
grade containing some Ti. The evolution of the microstructure was studied as function
of the holding temperature. At 450°C, the microstructure is fully bainitic. For
intermediate temperatures, there is a competition between intergranular and
intragranular nucleation so that the microstructures consists of a mixture of bainite
11
and acicular ferrite. Small amounts of pearlite can form on the prior austenite grain
boundaries. Finally, the decomposition of austenite into pearlite takes place at the
highest temperature (580°C).
Measurement of the Strain Partitioning by Digital Strain Mapping

Developing a micromechanical model of the TRIP-aided steels (WP 7)
requires the knowledge of the flow properties of each phase of the microstructure, i. e.
of the ferritic matrix, bainite, retained austenite and martensite. However, it is not
possible to obtain the same phases as single-phase finely grained microstructures.
Furthermore, it is more accurate to characterise in-situ the flow properties of the
different phases. For these reasons, the strain partitioning was measured directly in the
multiphase steels.
A digital image correlation technique was tested by partner 2 and applied on
SEM micrographs of TRIP-assisted multiphase steels during in-situ tensile straining.
It was shown such a kind of microstructure possesses a sufficiently random repartition
of the grey level to allow the good running of the method. Figure 5 exemplifies the
strain mapping determination in the case of a multiphase steel. The macroscopic true
strain imposed to this sample is equal to 7%.
1 0 µm
Figure 5. Typical SEM micrograph of steel H-TRIP2 and corresponding strain distribution
Characterisation of Fe-Mn and 301L Steels

The Fe-Mn steels as well as the 301L steel were examined, by light and transmission
electron microscopy. In agreement with the thermodynamic data, the Mn content of
the ferrite is about 3% lower than for the austenite. The annealing treatments also
bring about the recrystallisation of the microstructure, leading to equiaxed grains.
However, numerous stacking faults and annealing twins can be found in steel Mn20
after annealing.
5. TWIP steel
Fe-15Mn-3Al-3Si and Fe-20Mn-3Al-3Si: Before and after the annealing treatment the
microstructure was characterised optically and by transmission electron microscopy
(TEM) to reveal the grain boundaries and analyse the morphology of the different
phases. The volume fraction of the different phases was determined using XRD
measurements. As the annealing temperature increases the volume fraction of
12
austenite of a 15% Mn alloy decreases when thermodynamic equilibrium is reached.

This tendency corresponds with the measured volume fraction (phase partitioning).
The presence of annealing twins is common for low SFE fcc materials.
Figure 6. Evolution of the volume fraction of ferrite (black) and austenite (grey) with respect
to the annealing temperature (XRD).
Fe-30Mn-3Al-3Si: This material was analysed by optical microscopy and TEM. In

addition quantitative texture measurements were performed by X-ray diffraction.
Hot rolled and annealed samples have been cold rolled to different strains. Annealing
twins in the hot rolled and annealed material, mechanical twins after small cold
reductions (ε =0.2) and deformation bands after large deformation (ε = 1) have been
characterized. Some aspects of the deformation mechanisms occurring in the material
can be deduced from microstructural observations. At low strains, the sets of parallel
lines represent slip traces and mechanical twins (proven by TEM). When the strain
increases the slip and twin traces are no longer straight due to crystal lattice bending.
At very high strain levels the deformation becomes inhomogeneous as shear bands
become visible. These bands are oriented at +35° and –35° to the RD. This
microstructure is often referred to as a fish bone structure. This behavior at large
strains corresponds to a deformation mechanism common to other low SFE materials
such as brass and austenitic stainless steel in which twinning acts as a prior
deformation mechanism.
The evolution of texture during cold rolling has been studied. The brass orientation
{110}<112> spreading towards the Goss orientation {110}<001> is dominant at
every strain levels. The intensity of those orientations increases with increasing
rolling strain. At ε = ±1,0 the intensity of the Goss component no longer increases and
the rotated Goss orientation completely disappears. In addition a weak γ-fiber
develops. The intensity of the copper orientation is rather low. The brass orientation
dominates the texture of the material at every strain level. It is a very stable texture
component in rolling deformation when octahedral slip and mechanical twinning are
active deformation mechanisms. At high strain levels new orientations develop. This
probably implies that other deformation mechanisms become active which could also
be seen in the microstructure. The development of orientations that have {111} planes
parallel to the rolling plane, could correspond to twin lamellae aligning themselves
parallel to the rolling plane. These grains make continued homogeneous deformation
difficult and shear banding becomes the dominant deformation mechanism. These
shear bands could contribute to the development (increase) of the brass component.
13
Work Package 5: Interpretation and Fundamental Understanding
The main purpose of work package 5 is to summarise the new scientific knowledge
and understanding, generated by the experimentalists and to transfer this knowledge
to the modelers. Work package 5 was planned to become “active” in the project
month 18. It is clear that a large part of the experimental work carried out in the first
project year, was mainly devoted to the optimization of test procedures and
characterization methods and to the exploration and comparison of each others
experimental facilities and know-how. Nevertheless, some preliminary remarks and
conclusions about the experimental work can already be attempted.
1. Titanium alloys
Microstructural characterization of Ti alloys seems to go well and the generation of
different microstructures (e.g. equiaxed and lamellar grains) is possible. A good
cooperation between partners 1, 2 and 7 is established. Mechanical tests at room
temperature seems possible, but at higher temperature a number of problems remain
to be solved. A ‘brain storming’ between partners 1, 2, 4 and 7 about sharing facilities
and expertise is planned in March 2003 in Liège.

This part of the work has only started in September. ECAE at room temperature was
technically possible, but the substructure development was not sufficient to generate
the desired submicron material; more ECAE passes will be performed in the future.
EBSD analysis after 4 ECAE passes was successfully performed and compression
tests at room temperature on ECAE samples seems to be a reliable way for the
measurement of mechanical properties. Contacts between partner 1 and 4 about co-
ordination of mechanical tests to gather data for phenomenological modeling is
planned for the end of February.
3. IF steel
The work on submicron IF steel was already initiated before the start of the IAP
program. The ECAE technique was optimized by partner 1 as well as the EBSD
(OIM) analysis procedure. Thermal stability of this submicron material will be further
investigated and mechanical tests will be carried out and in cooperation with partner
4, phenomenological modeling will be initiated.
4. TRIP steel
A very intensive cooperation between partners 1 and 2 has been developed during the
first year of the project. Sharing know-how, samples and test facilities has become the
standard procedure. Additional cooperation with partner 6 for bending tests has also
been established. Knowledge about material preparation and about mechanical and
microstructural characterization is already well developed and co-ordination between
the development of micromechanical models and multilevel models (WP 8) and the
generation of well chosen test results has started.
14
TWIP steel
As for the TRIP steels, a close collaboration between partners 1 and 2 has been
established for the study of TWIP steels.
6. Pearlite
Due to lack of manpower (a foreign researcher engaged for this project could not
start), the work on pearlite will start in the course of 2003.
Work Package 6: Phenomenological modeling

In the frame of this project (IAP P5/08) focused on material behavior, different
phenomenological constitutive models can be applied. A very useful anisotropic law
is the well-known Hill 1948 model. Compared to isotropic models, this basic
quadratic yield surface can already improve the simulation results for a large number
of steel grades and it allows to investigate anisotropy effects. The identification of this
yield locus is based on tensile tests in different directions that provide the yield
stresses and the Lankford coefficients (ratio of thickness strain and transversal strain).
However such data already show a weakness of the model as Hill yield locus
identified by Lankford coefficients is not identical to the one identified by yield
stresses. Such a simple model proposes a compromise that will be compared to more
complex models during the IAP project. As the micro-macro models developed by
other partners (1, 3, 9) are three-dimensional ones, during this year 2002, the ULG
team (partner 4) has decided to improve the 3D Hill model available in the
LAGAMINE code:
- isotropic hardening has been linked to plastic strain (classical approach in
commercial package) and not any more to plastic work;
- analytical and perturbation stiffness matrices are now available;
- the integration scheme, the radial return more sensitive in anisotropic yield loci than
in isotropic ones have been improved;
- kinematic hardening has been implemented, it is important for material submitted to
cyclic loading or orthogonal loading, generally speaking to complex loading as IAP is
focused on.
This last point is not as simple as one could imagine. A bibliographic work has been
devoted to kinematic hardening. Different choices are possible:
- classical phenomenological models: Ziegler, Prager, Armstrong & Frederick;
- phenomenological model based on microscopic knowledge: Teodosiu & Hu .
Partner 4 analysis shows that the latter model is in fact an association of Ziegler and
Armstrong & Frederick models and that it verifies the thermodynamic frame as
established by the Chaboche & Lemaître model. The interests of Teodosiu & Hu’s
model are the following ones:
-it takes into account evolution of dislocation substructures;
-it simulates the strain hardening stagnation and the influence of the pre-strain
amount;
-it allows to model Bauschinger effect and strong path changes.
15
During this year 2002, ULG (partner 4) has coupled two kinematic hardening models
with its 3D Hill model: Armstrong & Frederick hardening and Teodosiu & Hu one.
The latter model has been developed by professor Teodosiu who belongs to the team
of Monique Gasperini (partner 9 ). Previous cooperation between Van Houtte (partner
1) and Teodosiu and between Paul van Houtte (partner 1) and ULG team (partner 4)
were quite helpful to get direct access to precise details. Fortran 90 modules were
exchanged about this Teodosiu’s model.
Two applications were analysed in Liège with LAGAMINE (partner 4):

-A one deep drawing test;
-B one cyclic shear test.
Test A concerns FEM simulations of a cylindrical cup deep drawing. Different

constitutive laws have been compared:
-TexIso texture yield locus defined by 6th order series in strain space coupled
with isotropic hardening (law provided by partner 1);
-TexMic texture yield locus defined by 6th order series in stress space coupled
with mixed hardening (Teodosiu & Hu) (law provided by partner 1);
-ANI3VH yield locus defined by 6th order series in stress space coupled with
isotropic hardening (law developed by partner 1 and 4 in previous collaboration);
-HILL yield locus in stress space coupled with isotropic hardening.
TexMic and TexIso laws were implemented in the frame of this IAP in the
LAGAMINE code by partner 4. They have been developed by partner 1 in
collaboration with partner 9. Partner 1 has provided the Fortran 90 routines to partner
4 for this implementation. A publication in NUMISHEET conference shows the
comparison of simulation results. The authors are Liège scientists (partrner 4) and E.
Hoferlin the researcher who has developed these laws during his PhD in KUL (partner
1) but who left t KUL team before 2002. Not well aware of the required explicit
mention of IAP P5/08 in the acknowledgement, this publication is however not
eligible to belong to the list of publications of IAP, but from a scientific point of view
it should be included as it is a result of the strong collaborations between IAP
partners.
Test B interests both Monique Gasperini (partner 9) and ULG (partner 4) teams. The
studied shear test is associated to an experimental device currently developed in
Liège. The test equipment will allow to simultaneously apply cyclic shear and tensile
loading. Such a device has already been developed in the Netherlands (team of prof.
Huétink, member of the follow-up committee). Another type of device with no tensile
load is also available in Paris (partner 9). These experiments are necessary to identify
the kinematic hardening models. In a first step, the current simulations performed in
ULG are just focused on simple cyclic shear test. They use LAGAMINE code with
Hill 3D yield locus coupled with Teodosiu’s hardening that allows to separate clearly
the effect of isotropic, kinematic and mixed hardening. For large shear (larger than 60
%), material axes present strong rotations and some decreasing force has appeared at
saturation. This phenomena not noticed in identical simulations performed by partner
9 is currently investigated. It is believed that this phenomena is linked to current
choice to define material axes rotation.
In the near future the interests of the different partners about phenomenological laws
for fine grained aluminium alloys and for titanium alloys will be further defined.
16
Work Package 7: Multilevel Modelling
Introduction
Multi-level models aim at integrating into a single model the relevant microstructural
features and micro-(or even nano-) mechanical phenomena that affect the response of
a material subjected to macroscopic loading. With respect to other mechanical
properties, plasticity is extremely complex to deal with as it involves lots of physical
scales from the atomistic dislocation level to the scale of a large aggregate of grains
representative of the distribution of crystallographic orientations.
There are two mains types of scale transition "building blocks" that can be used to
develop a multi-level model :
a. numerical simulation of representative volume elements (RVE)
b. semi-analytical homogenization methods for RVE
These approaches can be coupled or enhanced in order to account for more than two
physical scales, which is usually mandatory for solving macroscopic boundary value
problems on volume much larger than an RVE, e.g. structural parts undergoing
forming steps. For instance, the FE2 model of the TUE group relies on a FE
framework to solve the micro-meso and meso-macro scale transitions while the
approach at KULeuven or UCL aims at developing constitutive laws (micro-meso) to
be implemented in FE code (meso-macro). The aspect of coupling numerical methods
(typically finite element methods) and micro-macro models is the purpose of work
package 9.
Activities within this multi-level modeling workpackage can be divided in two

different types of interactions among the partners :
1. "Vertical" interactions/validations on a specific material. Vertical interactions

occur naturally as a multi-level model is always a physical model which directly
emerges from experimental characterization and which must (as for
phenomenological models) be validated on experimental results. Note also that micro-
macro models are by definition not generic (although the codes, numerical methods,
formulation can be generic).
2. "Horizontal" interactions aiming at comparing different types of models, at

coupling different models (e.g. coupling crystal plasticity and enhanced scale
transition methods), or at working together on a similar class of models (e.g. for
instance using different numerical methods). Horizontal interactions require more
time to set up. Horizontal interactions will allow to compare various micro-macro
models in order to provide mutual validation or to assess simpler models (but perhaps
more efficient) to more accurate one. Another more challenging purpose of the
horizontal interactions is to couple different models in order to extend their domain of
validity.
Of course, a significant part of this research activity, especially during the 2 first
years of the project, is undertaken in each group by working "separately" on the
development of the micro-macro model (setting up the model, writing programs to
integrate the model, solving for simple cases, ...).
17
Sub-task 7.1
Development of a multi-level model for IF steels and aluminium (single-phase, fine
grained, metallic, polycrystalline), based on dislocation mechanics and crystal
plasticity.
Progresses for this sub-task have been made at KULeuven, UCL and University of
Groningen. Strong interactions already take place between KULeuven and UCL, in
particular through the work of Dr. L. Delannay on the Lamel model.
a. Progress on Taylor-type models
1) Progresses with the original Lamel model:

The purpose of this research is to formulate a new constitutive law for polycrystalline
aggregates. This law relies on a micro-macro homogenization, and it accounts for the
interaction of adjacent grains. During the year 2002, the original LAMEL model has
been transformed in order to consider the simultaneous development of elastic and
plastic strains within each grain of the polycrystalline aggregate. The elastic-plastic
model has then been implemented as a new "user-defined rheology" in the finite
element codes Forge3 and ABAQUS.
2) Advanced Lamel Model

A new type of Lamel-model has been proposed: the Advanced Lamel Model.
The Lamel model has been quite successful in quantitatively predicting rolling
textures, especially for steel [Van Houtte, Delannay and Kalidindi, 2002]. However, it
is in principle only valid for rolling of materials with flat, elongated grains, because it
has been designed for that case. To overcome that restriction, a new theory has been
proposed: the Advanced Lamel Model (Alamel). It is based on the following
assumptions:
(i) The local strain rate l is not constant throughout a crystallite such as grain α in
figure 7; as a result, the slip rates are not constant either.
(ii) In regions close to a grain boundary (such as Region 1 in figure 7), the local
stress and strain rate (and slip rates) are heavily affected by the interaction with
the neighboring grain;
(iii) the set of crystallographic orientations which develop in the various regions
close to the grain boundaries (region 1 and similar regions along grain
boundaries BC, CD, DE, EF and FA in figure 7) can be regarded as the
contribution of the grain to the overall deformation texture of the polycrystal.
It is clear that these assumptions (especially (iii)) may still be subject to debate. At the
time being, the judgment about the validity of this assumption is based upon the
comparison of predicted deformation textures with experimentally observed ones.
18
y1
Grain β
A F
y2 Region 2
Region 1
Grain α
B E
C D
Figure 7. Schematic representation of a microstructure
The evolution of the crystallite orientations in regions 1 and 2 (fugure 7) can be

obtained with the same equations as those of the Lamel model. In practice, the
Advanced Lamel Model is implemented as follows:
- as before, the ODF of the initial texture is discretized in several thousands
orientations;
- a microstructure file describing the initial microstructure is read. It consists of a
large number of records which describe interfaces such as AB in figure 7. This
description essentially consists of the surface of the interface and the Euler
angles of the frame y1 y2 y3 with respect to the macroscopic frame. Normally,
this file is initialized to represent an equiaxed microstructure in three
dimensions. Non-equiaxed versions can be derived from it by applying a user-
defined "pre-deformation" to the elements in this file. This (real or fictitious)
pre-deformation should be described by a deformation gradient tensor F, which
has to be specified by the user.
- a large number of pairs "region 1 + region 2" (figure 3) are now created by
random selection of an interface from the microstructure file, and (for each
region) a random choice of a crystallite orientation from the discretized ODF.
- The slip rates in each "region 1 + region 2" are then calculated as described
above, leading to the lattice spins WL in each region.
- The crystal orientations of the two regions at the beginning of the next strain
increment are calculated by explicit integration, using the WL.
- The interfaces between the regions (orientation, surface) are updated by first
updating the deformation gradient tensor F (see above) by explicit integration
of the classical equation over the increment:
F&= L • F
in which L is the macroscopic velocity gradient at the beginning of the
increment. The updated tensor F is then used to obtain new values for the
orientation and the surface of the interface at the beginning of the next
increment.
- At the end of the simulation of a large deformation, the deformation texture is
assembled by taking the final crystallite orientations of all regions of all pairs,
using the surfaces as weighting factors.
As opposed to the Lamel Model or the classical RC models, the geometrical nature of
the relaxations is not related to the geometry of the prescribed deformation. In the
case of rolling, this means that the shear plane of the two relaxations need not be
19
parallel with the rolling plane. This makes the Advanced Lamel Model applicable to
any deformation mode. However, the relaxations do sometimes relax non-zero
components of the prescribed macroscopic strain rate tensor (when it is expressed in
the y1 y2 y3 frame). It was found that this can be tolerated, on condition that a single-
crystal hardening law is used (for every region separately) during the simulation.
b. Validation of deformation texture predictions by various models for IF steel and

two aluminium alloys.
The results of the models will be quantitatively compared to experimentally observed

deformation textures, and to the results of CPFEM models. The texture indices I(∆f)
of the difference ODFs will be used for the comparisons. Such texture index is the
integral of the square of the difference between the orientation distribution function
(ODF) of the experimental rolling texture and the one simulated by a model:
∫[f ]
( g ) − f exper ( g ) dg
2
I( f) = model
Table I. Texture index of difference ODFs for IF steel results.

70 % rolling reduction. Iexper = 6.7
CPFEM1 CPFEM2 FC RCP GIA Lamel1 Alamel1 Alamel2

1.22 0.90 4.65 1.40 0.76 0.59 1.08 0.65
CPFEM1: crystal plasticity finite element code developed by Bate (University of
Manchester); CPFEM2: crystal plasticity finite element code developed by
Bronkhorst, Kalidindi and Anand (MIT); FC= FC Taylor theory; RCP: Pancake
variant of the RC Taylor theory; GIA (Grain Interaction Model): a generalized
Taylor relaxed constraints model developed by Crumbach, Pomona, Wagner, Lücke
and Gottstein (RWTH Aachen); Lamel1: standard Lamel model; Alamel1:
Advanced Lamel Model, starting from equiaxed grains; Alamel2: starting from
elongated grains (as if the grains of the hot rolled microstructure already had 40%
rolling reduction) and using additional {123}<111> slip systems.
g is a crystal orientation. The integrals are taken over entire orientation space. The
value of I(∆f) is zero if the two ODFs are identical. I(∆f) can if desired be normalized
by dividing it by the texture index Iexper (integral of the square of the ODF) of the
experimental texture. The results of deformation texture simulations are sets of
weighted discrete orientations. They are converted into continuous ODF's by putting a
Gaussian distribution of the type exp(ψ/ψ0)2 with a spread of ψ0=7° upon each of the
orientations. Table I shows results obtained for an IF steel after 70% rolling. some of
these results have been described elsewhere in much greater detail (Van Houtte et al,
1999, 2002). It should be pointed out that prior to cold rolling, the IF steel was hot
rolled in the ferritic range. As a result, the as-hot rolled microstructure was not
equiaxed. It was attempted to take that into account in one of the two Alamel-
simulations.
20
Table II: Texture index of difference ODFs for 2 aluminum alloys

Material AA1200
Reduction 40% 63% 86% 95% 98%
FC 0.54 1.43 2.87 1.87 3.58
RCP 4.25 4.03 4.60 6.69 9.57
CPFEM2 0.42 0.92 1.62 1.48 2.52
VPSC 1.11 2.02 2.26 2.81 4.29
GIA 0.57 0.87 1.16 1.35 2.34
Lamel1 0.89 1.60 1.56 1.26 1.81
Alamel1 0.49 1.06 1.38 1.07 1.58
Iexper 2.59 2.45 4.51 6.87 9.23
Material AA5182
Reduction 40% 63% 86% 95% 98%
FC 0.29 1.43 2.58 2.84 4.53
RCP 1.89 0.84 2.60 3.37 3.89
CPFEM2 0.16 0.76 1.41 2.51 3.72
VPSC 0.57 1.33 1.92 2.18 2.68
GIA 0.12 0.35 0.90 1.67 2.51
Lamel1 0.46 1.15 1.66 2.41 2.83
Alamel1 0.27 1.09 1.29 1.70 2.08
Iexper 1.96 2.64 3.17 3.99 5.10
Models: see Table I, except VPSC: Visco-Plastic Self-Consistent Model (developed

by Lebensohn and Tomé, Los Alamos National Laboratory)
Table II shows results obtained in the framework of an European project. Some

results by the VPSC model, a self-consistent, have also been included
The main conclusion from these results is that the best quantitative predictions are
obtained by those models which take grain interactions between specific neighboring
grains into account. At low strains: crystal plasticity finite element models (CPFEM)
lead to the best predictions. Surprisingly, some Taylor-type models such as GIA,
Lamel and Alamel are better at high strains. the self-consistent model fails for the
AA1200 alloy. Overall (and according to these results only), the Alamel model is the
best.
c. Dislocation dynamics.
The Groningen group has had many activities during the last few years in the area of
discrete dislocation modeling. Making use of a superposition technique, developed by
Van der Giessen and Needleman in 1995, a number of boundary-value problems for
discrete dislocation plasticity have been addressed (E. van der Giessen and A.
Needleman: Micromechanics Simulations of Fracture, Annual Review of Materials
Research 32 (2002) 141-162). Particular emphasis has recently been on the role of
dislocations during crack propagation, either during monotonic or cyclic loading.
Another class of activities has been the comparison of discrete dislocation results with
non-local plasticity models, in order to assess the adequacy of such models and to
estimate the value of the length scale parameter in non-local models. While these
simulations were two-dimensional, a three-dimensional version of the discrete
21
dislocation model came available very recently (D. Weygand, L.H. Friedman, E. van
der Giessen and A. Needleman: Aspects of boundary-value problem solutions with
three-dimensional dislocation dynamics. Modelling Simul. Mater. Sci. Eng. 10 (2002)
437-468).
The link with the other activities in this sub-task (or with other sub-task) will be
initiated in the following years.
Sub-task 7.2
Development of multiphase models for α−β Ti alloys, based on crystal plasticity
Progresses related to this sub-task have been made at UCL with the development of
constitutive models for multi-elastoplastic-phases materials.
Homogenization schemes and numerical algorithms for two-phase elasto-plastic

composite materials and structures have been developed at UCL based either on an
incremental theory [Doghri and Ouaar, 2003] (see in sub-section "generic
developments") or on a deformation- based theory [Lani et al., 2003] (see description
in sub-task 7.5). These models have been validated for ellipsoidal inclusions based on
unit cell FE calculations. The application to Ti-alloys is now possible but requires the
experimental characterization of the microstructure and phase properties (partners 2
and 7).
Sub-task 7.3
Development of multiphase models for TWIP and pearlitic steel based on crystal
plasticity
A researcher will be hired in the first half of 2003 to perform this task.
Sub-task 7.4
Development of multiphase models for TWIP steel based on crystal plasticity due to
both dislocation glide and mechanical twinning, using a thermodynamical model for
assessing the tendency to mechanical twinning
This task will be started once the first results of task 7.4 are available.
Sub-task 7.5
Development of multiphase models for TRIP and TWIP and pearlitic steel based on a
thermodynamical model for phase stability and on crystal plasticity
Important progresses have been made in this sub-task owing to a joint research effort
between UCL and KULeuven (cf. part 2: organization of the network).
The transformation model of Fischer (F.D. Fischer and G. Reisner: Acta Mater. 46
(1998) 2095) was applied in the case of strain-induced martensitic transformation in
multi-phase TRIP steels. The model was only adapted to account for the multi-phase
character of the TRIP steel, since the model was developed for single-phase austenitic
steels (e.g. maraging steels). Input data of the different phase present in the material
(ferrite, austenite and martensite) were taken from experiments carried out by
22
researchers at UCL. The agreement between the calculated results and experimental
data (P. Jacques, Q. Furnémont, T. Pardoen and F. Delannay: Acta Mater. 49 (2001)
139) was rather good. It was, however, noticed that some experimental features could
not be accounted for by Fischer's model. The model of Fischer was adapted to
incorporate specific features of low-alloyed, multi-phase TRIP steels, namely burst
transformation of martensite, the occurrence of multiple martensitic variants in an
austenite grain, the influence of plastic deformation on the nucleation of martensite.
The model was implemented by Tim Van Rompaey (KULeuven), and calculations
were performed and compared with experimental data obtained by Quentin
Furnémont (UCL) on the influence of different loading conditions on the kinetics of
strain-induced martensitic transformation. Two joint publication have already been
published on that part of the project (Van Rompaey et al., 2002)
In parallel, the same transformation criterion has been introduced by Frederic Lani
(Ph. D. student at UCL) within a micro-macro code developed for two-phase elasto-
plastic materials, based on the deformation theory of plasticity (Lani et al., 2003). A
first part of this research consisted in the development of a micro-macro model for the
mechanical behavior of dual-phase elasto-plastic materials: the shape induced plastic
anisotropy has been studied by looking at the variation of the coefficient of anisotropy
as a function of straining for uniaxial loading along the transverse direction. R
changes significantly during deformation. The shape induced anisotropy is intensified
when the inclusions remain elastic, i.e. for high yield ratios. The model and the
analysis of the model has been the subject of a paper submitted to International
Journal of Plasticity (Lani et al., 2003). An experimental validation on multiphase
steels exhibiting elongated second phases has started end of 2002 in the context of a
master thesis work.
In a second step, the micromechanical criterion for the transformation inspired by
the work of Fischer has been implemented and the parameters of the model have been
identified by comparison with experimental measurements performed on a TRIP
assisted multiphase steel. Using these parameters and the version of the model
including phase transformation, we have been able to predict the overall stress-strain
curve and the evolution of the austenite volume fraction of the steel. The
transformation of the reinforcing phase has a strong effect on the global mechanical
response of the material. The predictions in terms of transformation kinetics as a
function of the overall stress triaxiality and macroscopic stress-state are in agreement
with the experimentally observed trends, i.e. increasing the stress triaxiality promotes
the martensitic transformation.
Finally, in order to improve the understanding of the transformation behaviour, a

unit-cell, finite element model of a ferrite/TRIP steel matrix surrounding a single
austenite grain, in which a martensite variant was formed, was developed jointly by
Frederic Lani (UCL) and Tim Van Rompaey (KULeuven). Using this model it is
possible to investigate the detailed stress and strain distribution in a transforming
austenite grain and its surroundings, as a function of transformation progress and the
orientation of the martensite with respect to the loading direction. Furthermore, this
model enables to study different important terms occurring in the transformation
criterion of Fischer, such as the mechanical driving force, and the elastic and plastic
accommodation of the material during martensite formation (F.D. Fischer and G.
Reisner: Acta Mater. 46 (1998) 2095). Preliminary results show that the
transformation strain has only a very limited influence on the mechanical properties of
the TRIP steel. The main hardening mechanism is the composite effect of a hard
23
phase (martensite) in a softer matrix (ferrite). An other important result indicates that
the accommodation of the material plays a very important role in the transformation
criterion. This part of the work will be intensively continued in 2003.
One important perspective for 2003 is also to extend the micro-macro models to
accommodate multi-phase elasto-plastic composites and to implement phase
transformation models within the incremental micro-macro formulation explained in
the next section, point (b).
Generic developments
Those are progresses and developments of models or methods that are useful for the
whole work package and not related to a specific subtask.
a. FE2 at TUE.
Computational homogenization technique (also called FE2 or micro-macro modeling)

is a versatile strategy to establish micro-macro structure-property relations of multi-
phase materials. The computational homogenization approach is essentially based on
the solution of nested boundary value problems, one for each scale. The thermo-
mechanical response at a macroscopic point is directly obtained through the detailed
modeling of the underlying microstructure. The computational homogenization
approach is especially beneficial when physically and/or geometrically evolving and
transforming microstructures (e.g. TRIP steels) are to be dealt with. Other
advantageous characteristics of the technique are (i) no constitutive assumptions
required on macrolevel; (ii) large deformations and rotations on the micro and
macrolevels, this feature is essential, for example, for modeling of forming processes;
(iii) arbitrarily physically non-linear and time dependent material behavior of
microstructural constituents, their interaction and transformation laws; (iv)
independent of the solution method used for the solution on the microstructural
boundary value problem. Besides the direct micro-macro analysis, the computational
homogenization approach may be used as a tool to assess microstructurally based
macroscopic closed-form constitutive relations and the predictive capability of other
homogenization methods.
During the previous period, the research activities (related to the IAP project) of
the partner group in Eindhoven University of Technology (partner 3) concentrated on
the development, implementation and evaluation of the computational
homogenization approaches (which is also a part of the Scientific Research Program
of the Netherlands Institute for Metals Research). A novel, second-order
computational homogenization procedure has been developed. The second-order
scheme is based on a proper incorporation of the gradient of the macroscopic
deformation tensor into the kinematical micro-macro framework. The most important
property of the second-order computational homogenization method is in fact that the
relevant length scale of the microstructure is directly incorporated into the description
on the macrolevel. Including the size of the microstructure allows to describe certain
phenomena that cannot be addressed by the classical computational homogenization
scheme (as well as conventional homogenization methods), such as size effects and
macroscopic localization. Several microstructural analyses show that the second-
24
order computational homogenization framework captures changes of the macroscopic

response due to variations of the microstructural size as well as variations of
macroscopic gradients of the deformation. In problems that exhibit macroscopic
localization, the microstructural length scale determines the width of the localization
band, thus providing mesh independent results. Furthermore, the second-order
framework allows the modeling of surface layer effects via the incorporation of
higher-order boundary conditions. Higher-order continuum modeling becomes
considerably easier with the use of the second-order computational homogenization
scheme because the second-order response is directly obtained from a microstructural
analysis, rather than by closed-form constitutive relations which are difficult to
formulate and which contain a large number of parameters to be identified.
The calculation time required for a coupled numerical analysis may be
substantially reduced if the optimal use is made of the inherent parallel nature of the
multi-scale framework. The computational homogenization schemes (the classical
and the second-order) have been implemented for computation on a cluster of parallel
processors. Thus, if a microstructural model describing the microstructural behavior
and evolution in a representative cell is available, it can be incorporated into the
parallel multi-scale program, allowing the micro-macro structure-property analysis.
b. Development of a micro-macro code
Homogenization schemes and numerical algorithms for two-phase elasto-plastic

composite materials and structures have been developed at UCL [Doghri and Ouaar,
2003]. A Hill-type incremental formulation enables the simulation of unloading and
cyclic loadings. It also allows to handle any rate-independent model for each phase.
The crucial issue of tangent operators was studied: elasto-plastic (or "continuum")
versus algorithmic (or "consistent"), and anisotropic versus isotropic. We apply two
methods of extraction of isotropic tangent moduli. We compare mathematically the
stiffnesses of various tangent operators. All rate equations are discretized in time
using implicit integration. We implemented two homogenization schemes: Mori-
Tanaka and a double-inclusion model, and two plasticity models: classical J2 plasticity
and Chaboche's model with nonlinear kinematic and isotropic hardenings. We
consider composites with different properties and present several discriminating
numerical simulations. In many cases, the results are validated against finite element
(FE) or experimental data. We integrated our homogenization code into the FE
program ABAQUS using a user material interface UMAT. A two-scale procedure
allows computing realistic structures made of nonlinear composite materials within
reasonable CPU time and memory usage.
Other related researches
Although the coupling between damage and plasticity in single and multi-phase
metals does not constitute a sub-task of the project, it is a key issue when considering
the limit of homogeneous deformation and the transition to fracture (i.e. the ductility
of the material, which is the purpose of work package 8). Progresses have been made
this year at UCL in the development of multi-level models accounting for void growth
in plastically deforming metals (Pardoen et al. 2002, 2003, Pardoen and Hutchinson,
2002, 2003).
25
Work Package 8: Modelling of Strain Localisation and Ductility
Introduction
The first activity of this work package has been a workshop in which the following
partners have participated:
Partner 1: Katholieke Universiteit Leuven, departement materiaalkunde (KUL-MTM)
Partner 2: Université Catholique de Louvain, Unité de physico-chimie et ingénierie
des matériaux (UCL-PCIM)
Partner 3: Technische Universiteit Eindhoven, Mechanical Engineering (TUE)
Partner 4: Université de Liège, département MSM (ULg-MSM)
Partner 9: Université Paris 13, LPMTM-CNRS (UP13)
Representatives of these partners have highlighted the background knowledge existing

in their laboratories in the field of strain localization, reported recent progress and
explained their research plans in this field. This was followed by a coordination
meeting, in which actions for future cooperation were laid out.
Background knowledge
1) Three-dimensional Localized Deformation
This is of course the most general category; typical examples are "necking" in tensile
tests and localized or diffuse necking in sheet metal deformation.
There were three points of view:
(i) ”Strain localization in strain softening rate-independent materials” (Issam Doghri,

Partner 2)
This is a quite general view from continuum mechanics. A framework is offered for a
theoretical analysis of a multitude of problems. Much work has still to be done for the
implementation of particular material behavior (for example: texture-based
anisotropy).
(ii) "Necking and ductile fracture in thin plates and round bars” (Thomas Pardoen,
Partner 2). This approach makes use of fracture mechanics, mechanical metallurgy,
and continuum mechanics. From a conceptual point of view, the approach is less
general than (i), but it is closer to particular problems, and considers crack-
propagation types of phenomena, which (i) and (iii) do not. Validated for Al alloys.
(iii) ”Prediction of forming limit diagrams taking texture and microstructure-based

anisotropy into account” (Paul Van Houtte, Partner 1) This approach is focused on
sheet metal forming. It makes use of the well-known Marciniak-Kuszynski theory for
the simulations of necks in plates. Combination with dedicated modeling of
anisotropy caused by texture and dislocation substructure is possible. The method is
successful and has already been validated using experimental results for steel sheet.
2) Two-dimensional Localized Deformation (Shear Banding)
In shear banding, the strain is zero in one of the three dimensions. It is phenomenon
that is often observed in processing of bulk materials (as opposed to sheet material).
26
There was an experimental and a theoretical contribution:
(i) ”Shear banding at several scales in aluminium alloys” (Monique Gaspérini, Partner
9). The study of shear banding in shear tests is one of the focal points of the research
activities of Partner 9. Extensive experimental results on al alloys are available,
mainly about shear banding during shear testing. Current work focuses on the physical
explanation of the observed phenomena. Cooperation with other partners more
specialized in modeling activities is sought.
(ii) "Texture-induced anisotropy as a driving force for shear banding in rolling and
torsion” (Paul Van Houtte, Partner 1). Shear bands during rolling have been studied
using metal-physics type models. The approach focused on the texture softening
effect as driving force for shear banding. This is not a model where "shear banding"
emerges as a possible deformation pattern for a given material with a given
constitutive model and under given boundary conditions. Rather, it assumes that shear
banding might occur, proposes a geometrical model for it, and then studies the
contribution of work softening or work hardening (including texture-induced effects)
to the possibility of the bands to develop.
3) Framework for the numerical simulation of plastic deformation of materials prone

to strain localization.
There was a single contribution:

"Enriched continua and homogenization methods for strain localization and
softening” (Marc Geers, Partner 3). This was an illustration of an advanced method of
modeling strain localization by using non-local models at element scale. The
constitutive model of an element is itself derived from a FE simulation of the material
behavior at a smaller scale. One could call this a "FE2"-approach.
Coordination of future activities
(i) LPMTM (Partner 9) is interested in collaboration with other groups in order to

explore sensitivity to strain localization of the other materials (in addition to Al
alloys): If steel, Pearlite and TRIP steel.
It has been agreed that the other partners will send samples of these materials to
LPMTM.
(ii) LPMTM is also interested in benchmarking of mechanical testing.

It has been agreed that in a first stage, Ulg (Partner 4) and LPMTM will cooperate to
carry out this task.
(iii) Modeling Activity:

A collaboration between LPMTM and MTM (Partner 1) has been agreed upon in
order to estimate the contribution of textural softening to the observed
softening/localization behavior in Al alloys. The results will be discussed by all
partners involved in Work Package 8 during another workshop.
(iv) Benchmarking:
It has been agreed upon that strain localization predictions in sheet metal forming will
be conducted by different methods, in order to compare the results (partner 1 and 2).
27
Work Package 9: Finite Element modeling at macroscopic scale

In order to apply FE simulations with phenomenological and multilevel constitutive
laws to experiments or forming processes of macroscopic size, a parallel version of
the LAGAMINE code has been developed by partner 4. A typical approach of a
parallel version of a FE code is the multi-domain decomposition. In this case, the
mesh is subdivided in X domains, each one being computed by one processor, then a
special assembly procedure takes care of common degrees of freedom of these
domains and provides the final results. This approach is adapted for the large numbers
of degrees of freedom (> 500 000 DOF) met in industrial applications. In the frame of
the current IAP project, basic research on material behavior is our aim, so we do not
intend studying in one FE simulation a car or a plane body. We would like to compare
different phenomenological and multilevel constitutive models applied to simple
experiments (macroscopic samples going from a few square millimeters to a few
square centimeters but still of limited size). In order to identify law effects and not to
investigate the accuracy of different finite element codes, the LAGAMINE code tries
to gather a lot of the constitutive models developed in this IAP project. As some
multilevel constitutive laws are really greedy in CPU time, a parallel version of
LAGAMINE where the loop of elements during one equilibrium iteration is divided
between different processors has been implemented. Due to the limited size of the
studied problems (<50 000 DOF), the iterative solvers are not as efficient as direct
solvers. So direct solvers have been kept, even if these ones are hard to parallelize. A
previous collaboration with the Centro Europeo de Paralelismo de Barcelona
(Universitat Politenica de Cataluna, Spain) has allowed Pierre de Montleau (post-doc
of partner 4 paid by IAP project) to implement in LAGAMINE a parallel direct solver
in the first semester of 2002. Comparisons of CPU time using one to eight processors
show the interest of the LAGAMINE parallel approach. The detailed results can be
found in 3 publications. As new partner of IAP project, partner 4 was not aware of the
importance of telling the link with IAP P5/08 in the acknowledgments, so these
publications are not eligible to belong to the list of publications of IAP. Their
references are given hereafter as they summarized interesting scientific results about
LAGAMINE parallelism:
Acceleration of finite element analysis by parallel processing, S. Moto, P. de

Montleau, A. Godinas and A. M. Habraken, The 5th Int. ESAFORM conf. On Material
Forming, ed. by M. Pietrzyk, Z. Mitura, J. Kaczmar, publishing house Akapit,
Krakow Poland, April 2002, 47-50
A parallel computing model for the acceleration of a finite element software, P. de
Montleau, J.M. Cela, S. Moto Mpong and A. Godinas, The 4th International
symposium on high performance computing, BerlinWompei, Japan, H. Zima et al.
(Eds) in Lecture Notes in Computer Science 2327, G. Goos et al. editor, Springer
Verlag May 2002.449-456.
A parallel computing model for the acceleration of a finite element software, S. Moto
Mpong, P. de Montleau, A. Godinas and A. M. Habraken, 2002 International
conference on parallel and distributed processing techniques and applications, Las
Vegas, PDPTA’02, Arabnia editor, CSREA Press, June 2002, 185-191.
With the help of other partners (1 and 9) who have transmitted FORTRAN 90
routines, partner 4 has implemented multilevel models in LAGAMINE. To be
28
clear let us describe which models were available before IAP projects and which ones
have been implemented in the IAP P5/08 frame.
Work
Models available before IAP P5/08:
ANI3VH model (PhD thesis Winters, 1996, Partner 1) Anisotropic yield locus based
on texture and crystal plasticity. This yield locus is described by a 6th order yield locus
in stress space. This analytical law is identified thanks to a minimization based on
macroscopic TAYLOR factors. The latter coefficient is a quantitative value telling the
level of slip system activities during a specific strain rate. For one single crystal the
microscopic TAYLOR factor can be computed by crystal plasticity model. In the
current approach, the Taylor-Bischop-Hill model is applied. For polycrystal behavior,
a set of representative crystal orientations is selected according texture measurements
that provide the Orientation Distribution Function, characteristic of the polycrystalline
material. Then the Taylor assumption telling that each representative crystals is
submitted to the same strain rate (the macroscopic one) is applied and one can reach
the macroscopic TAYLOR coefficient, characteristic of one texture and one strain
rate. The definition of the 6th order yield locus is based on Taylor coefficients
associated to thousands of different strain rate modes. During FEM simulation, this
model assumes that the texture does not evolve which is wrong. This yield locus is
coupled with isotropic hardening.
Minty model + isotropic hardening (DEA thesis Duchêne 2000 partner 4) Here
no analytical function describing the whole yield locus is used. The yield locus is only
known locally. In practice, 5 points of the yield locus are computed by a similar
approach as explained in ANI3VH. Then an interpolation method is used as long as
the strain rate stays in the local zone identified by these 5 points. When the strain rate
is strongly modified, 5 new points must be computed to describe a new zone of the
yield locus. Such an approach allows to take texture evolution into account during FE
simulations, as it means only to update the 5 points currently used in the local yield
locus description. This yield locus is coupled with isotropic hardening.
Models implemented in LAGAMINE during the first year of PAI P5/08:
TexIso model (PhD thesis Hoferlin 2001 partner 1) is a 6th order yield locus described
in strain rate space. The ideas are similar to ANI3VH but working in strain rate space
and not in stress rate space allows to take into account more easily texture updating.
The drawback is that the computation of stress state requires an additional
minimization to go from strain rate space to stress rate space which slows down the
computation. This yield locus is coupled with isotropic hardening.
TexMic model (PhD thesis Hoferlin 2001 Partner 1) is identical to TexIso but
coupled with the kinematic hardening of Téodosiu & Hu (partner 9).
As summarized in work package 6, the quadratic phenomenological Hill yield locus

has also been coupled with the kinematic hardening of Téodosiu & Hu during year
2002 by partner 4.
Another transfer from partner 1 to partner 4 worth to mention is also the software
ODFLAM which selects the representative set of crystals when the Orientation
29
Distribution Function is known. This software improves the accuracy of texture

intensity computed from a limited number of crystal orientations.
In their current work, Laurent Duchêne and Paolo Flores compare the different
constitutive models on cup deep drawing simulations. Different cases have been
studied: the hemispherical punch case of NUMISHEET benchmark, the experiments
performed in RDCS in the frame of a previous Region Walloon project with
cylindrical punches. Let us note that these simulations are possible only because the
parallel version of LAGAMINE has been developed, previous attempts to perform
such simulations with texture evolution had failed in 2001. Details on these
comparisons can be found in following publications,:
Texture evolution during deep-drawing processes, L. Duchêne, A. Godinas, S.

Cescotto and A. M. Habraken, Journal of Materials Processing Technology, 125-126,
2002, 110-118.
Micro-macro analysis of steel sheet behavior in finite element simulations.
Application to deep-drawing process, L. Duchêne, A. Godinas, S. Cescotto and A. M.
Habraken, The 6th European Mechanics of Materials Conference on Non-linear
Mechanics of Anisotropic Materials (EMMC6), Liège, Belgium, ed. Cescotto,
September 2002, 337-344.
Texture analysis of cylindrical cup during deep-drawing process, P. Flores, L
Duchêne., A.M. Habraken Proc of 5th Int Conf NUMISHEET 2002 , Jeju Island
Korea, ed. Y. Yang, October 2002, 605-610.
Again the link with IAP P5/08 in the acknowledgment has been forgotten for these
2002 publications, so these publications are not eligible to belong to the list of
publications of IAP, but they cover research belonging to IAP project.
A last point common to work package 9 and 10 is the development of a semi-analytic

sensitivity matrix for inverse modeling. To identify constitutive model parameters, an
inverse method is used: one set of parameters is chosen to simulate by FE some
experiments, the discrepancy between simulation and experimental results is then
analyzed to provide a new set of parameters and a new iteration begins. The evolution
of parameters is linked to an optimization algorithm. The most efficient ones require
the sensitivity matrix: the derivative of the chosen simulation result versus each
parameter. Finite differences can be applied to provide this information but it is very
slow. Analytical derivative is not possible with the used multilevel constitutive laws,
so a semi-analytical approach has been developed and implemented in LAGAMINE
code. Such a computation requires modification in the whole architecture of the FE
code and in consequence it is not straightforward. Report 37 of the current Walloon
Region research (981/3793) describes the current approach: Identification de
parameters par analyse inverse, AM Habraken, S Moto Mpong.
30
Work Package 10: Inverse Modelling
General Context
The goal of WP 10 is the development of an Inverse Method for the estimation of

parameters in generic phenomenological models. Three important components must
be developed for this purpose:
a finite element package containing the material model
an experimental set-up that delivers the information for the material parameter
estimation
an estimation algorithm.
The activities in 2002 in work package 10 were hampered by a lack of a permanent

researcher. Some kick-off meetings were organized in 2002 to prepare the work of
WP10. Meanwhile, David Lecompte (KMS) was asked to set-up preliminary tests
with the CCD-camera on aluminum test specimens.
Kick-off Meetings 2002
In 2002, three specific meetings for WP10 took place:
Meeting (KMS) 21/05/2002 (KMS/Ulg/VUB)

The purpose of this meeting was the discussion which FE-packages and which
parameter estimation platform could be used for WP 10.
Meeting (VUB) 14/06/2002 (VUB/Ulg/KMS)

Eddy Dascotte, the manager of DDS (a company that commercialize software for FE-
updating) was invited to present the possibilities of using FEMTOOLS as a platform
for the elasto-plastic model updating.
Meeting (KMS) 15/01/2003 (VUB/KMS)

The meeting was used mainly to brief Kurt De Proft about his work planning in 2003.
Kurt De Proft presented the possibilities of using the FE-software FEAP as the
numerical model in the parameter estimation procedure.
Experiments with the CCD-camera
The CCD-camera displacement field has to provide the necessary experimental data
for the elasto-plastic model parameter identification. Some important conclusions
were taken from the preliminary tests:
• good surface preparation is necessary (a combination of white powder paint
with small gray particles yielded the best results)
• curve fitting of the measured displacement field is necessary
A detailed report with a discussion about the precision and possible measurement
ranges is in preparation.
31
Task planning 2003
Task 1. CCD camera tests
Curve fitting of the displacement field, estimation of the precision of the obtained
experimental results
Task 2. FEAP simulations
The aluminum test specimens used for CCD-camera testing will be modeled in FEAP.
A simple material model (Von Mises) will be adopted in a first attempt. The type of
elements and the necessary number of elements to reach convergence will be verified.
The data interface between the CCD camera and the FE-package FEAP will be
established.
Task 3. A possible generic stress/strain curve fit
The six independent constitutive components (anisotropic assumption) of the

relaxation matrix in the 2D stress/strain relations of elasto-plastic behavior will be
curve fitted with high order polynomials (as a function of the stress level). Only
loading can be considered for this kind of curve fitting. The curve fitted model will be
in a next step be used (together with eventual other experimental results) to establish
the parameters in existing elasto-plastic phenomenological models.
The number of polynomial parameters will be about 200. This means that a
considerable number of data (strains) points in a different state of stress from the
camera will be necessary to assure observability of the unknown parameters.
32
2. ORGANISATION OF THE NETWORK
General organization
The experimental work of this IAP project is organized around 6 “model materials”.
For each material, several partners are involved, as shown in the table below. For each
material a “material coordinator” is responsible for the organization of data transfer,
cooperation between partners and contact between experimentalists and modelers.
The modeling work is mainly organized according to the work packages. In these
activities the ‘work package coordinator’ plays a crucial role. A formal link between
experiments and models is accounted for in work package 5 and is supervised by the
main promoter of the network.
Each year two “joint meetings” are organized for the whole network in order to
discuss the general progress of the work and to initiate new activities are re-orient
ongoing activities when needed.
A more detailed account of joint activities and links between partners, is given below.
W.P Task Coordinator IF TRIP TWIP Perlite Alu Ti

(MTM) (UCL) (MTM) (MTM) (MTM) (KMS)
1 Mat. Van Humbeeck 1 2 1 (2) 1 1 7
Preparation (MTM)
2 Forming Van Humbeeck 1 (4,6) 2 (1,6) 1 (4) 1 (4) 1 (6) 7 (1,4)
Tests (MTM)
3 Mechanical Van Humbeeck 1 (4,5,7,9) 2 (5,7) 1 (9) 1 (5,7,9) 1 7 (1,5)
Tests (MTM)
4 Microst. Verlinden 1 2 (1) 1 (2) 1 1 7 (1)
Charac. (MTM)
5 Interpretation Van Houtte 1 (all) 2 (all) 1 (all) 1 (all) 1 (all) 7 (all)
(MTM)
6 Phenom. Habraken 4 (1) 4 (1,2) 4 (1) 4 (1)
Models (Ulg)
7 Multilev. Pardoen 1 (3,4,8,9) 2 (1,3) 1 (2) 1 1 (3,4, 2
Model (UCL) 8,9) (1,4,7,8)
8 Modeling. Van Houtte 1 (2,3,4, 1 (2,3,
Strain (MTM) 6,8,9) 4,6,8,9
Localisation.
9 FE Models Habraken mainly 4 with the help of 1,2,3,5,6,7,8,9
(ULg)
10 Inverse Sol 5 (1,4,7) 5 5 (1,4,7) 5 (1,4,7) 5 (1,4,7)
Models (VUB) (1,2,4,7)
Material co-ordinators Partners
IF-steel: B. Verlinden (MTM) 1. KUL-MTM 6. KUL-PMA

TRIP steel: P. Jacques (UCL) 2. UCL 7. KMS
TWIP steel: P. Wollants (MTM) 3. Eindhoven 8. Groningen
Perlitic Steel: P.Van Houtte (MTM) 4. Ulg 9. Paris 13
Submicrom Al: B. Verlinden (MTM) 5. VUB
α/β Titanium: L. Rabet (KMS)
33
Joint meetings of the whole network
• Kick-off meeting (Leuven, 26 April 2002)

In this meeting the main promotor summarized of the objectives of this IAP project.
The six material coordinators gave an overview of experimental activities that had
already started and of plans for the near future. The coordinators of work packages 6
to 10 described their view on future activities and how different partners should
interact. A first inventarisation of “missing data” for effective modeling was
proposed.
• First annual meeting (including the steering committee) (KMS-Brussels, 20

January 2003)
In this meeting a short overview of the main scientific activities of 2002 was
presented by the material coordinators, work package leaders and several researchers.
Ongoing interactions between partners have been reported and new opportunities for
cooperation have been discussed. The power point presentations of this first annual
meeting are available on CD-ROM and have been distributed among the partners. The
nucleus of an “IAP-P5/08 website” has been created:
http://www.mtm.kuleuven.ac.be/Research/IUAP/P5-08.html
Meetings of subgroups
During the “Grasmech” course in February 2002, partners 2,3 and 4 (Issam Doghri,
Marc Geers and AM Habraken) have discussed and compared different
homogenisation techniques which should be used in the modeling part of this project.
8 February 2002: delegates of most partners attended the PhD defense of Bart Peeters
(Promotor P. Van Houtte, partner 1). The models developed in this PhD thesis are an
important input for the present IAP project
9 April 2002: visit of P Jacques, T Pardoen et L. Ryelandt (partner 2) in Ulg.

Æ discussion about tests to be performed in superplasticity as well as modeling
Æ presentation of a project about Titanium (Æ project defined and written by both
teams and submitted on 29 April 2002; link with part of IAP project)
23 April 2002: discussion between partners 4 (Ulg) and 7 (KMS)

Æ first information about inverse modeling applied in identification of constitutive
models,
Ædefinitions of the phenomenological and multi level constitutive laws.
Æ exchange of documents : Agrégation thesis of AM Habraken , PHd thesis of JP
Kleidermann
21 May 2002: meeting at VUB between partners 4, 5 and 7

Æ discussion on interaction between the tasks of phenomenological model and
inverse modeling, official cooperation agreement about the use of computer codes
Æ presentation by AM Habraken about the data required for phenomenological
models, the tests, the models, the inverse models already used
Æ presentation by H Sol about available inverse models
Æ discussion about the work of David Lecompte.
34
14 Juni 2002: meeting at VUB between partners 4,5 and 7

Eddy Dascotte, the manager of DDS (a company that commercializes software for
FE-updating) was invited to present the possibilities of using FEMTOOLS as a
platform for the elasto-plastic model updating.
14 July 2002: meeting at VUB by partners 4, 5 and 7 about FEM Tools and a
description of a possible platform for inverse modeling
9-12 September: presence of partners 1, 2, 3 and 4 in Liège at the EMMC6

conference. Important presentations and discussion about micro-macro models and
anisotropy with several participants.
8 October: meeting in Louvain-la- Neuve between partner 2 and partner 6. During this
meeting, the research program concerning the influence of complex loading
conditions on the mechanical behavior of TRIP steels was defined. The specimens
were processed at the PCIM and then tested by bending at the PMA. They will be
returned to the PCIM for characterization of the deformed microstructure.
Exchange of people, material and software
Ir Mathieu Vanderhasten is payed by partner 7 (KMS), but is inscribed as PhD student

at KULeuven (partner 1); promotors are Prof. Bert Verlinden (KUL) and Prof. Luc
Rabet (KMS). This PhD student spents about 50% of his time at KUL, mainly for
microstructural characterisation of samples and 50% at KMS, mainly for mechanical
tests.
Dr ir Kurt De Proft is paid by partner 7 (KMS). This researcher spends about 50% of
his time at KMS and 50% of his time in VUB, but his total work deals with the work
package 10; he is guided in a direct way by Prof H. Sol, the work package leader.
ir David Lecompte is not paid by the IAP budget, but is an employee of the Min of
Defense (partner 7 KMS). He works for the IAP project in work package 10, in the
context of the preparation of a PhD-thesis at the VUB (Promotor Prof H. Sol of
VUB). He divides his time working at VUB and at KMS.
Several meetings were and are still organized between researchers of partner 2 (UCL)
and partner 1 (KUL-MTM)in order to define the experimental procedure for the pre-
straining tests carried out in Leuven or the calculation of phase diagrams of Fe-Mn-Si-
Al alloys. TRIP steel specimens have been processed in Louvain-la-Neuve and are
currently used by partner 1. The characterization and interpretation of results are
discussed commonly.
Partner 2 (UCL) provided TRIP steel sheet for bending tests to partner 6 (KUL-
PMA); after bending these samples return to UCL for characterization
Several members of the team of partner 6 (KUL-PMA) regularly use the equipment in
the lab of partner 1 (KUL-MTM)
35
Partner 4 (Ulg) received from partner 1 (KUL-MTM) a number of software packages

or subroutines to intagrate it in their own software.
Numerous mails about Textiso and Texmic constitutive laws have been exchanged
between partner 4 (Ulg) and Dr. E Hoferlin, researcher now in OCAS but having
developed these routines in the team of Paul van Houtte (partner 1).
36
3. PUBLICATIONS
a. List of publications for each team
Partner 1 (KUL-MTM)
P. VAN HOUTTE and B. PEETERS: "Effect of deformation-induced intragranular
microstructure on plastic anisotropy and deformation textures", (Proc. ICOTOM 13),
Mat. Sci. Forum 408-412 (2002) 985-990.
A. VAN BAEL and P. VAN HOUTTE: "Convex fourth and sixth order plastic
potentials derived from crystallographic texture", Non Linear Mechanics of
Anisotropic Materials (Proc. EMMC6), S. Cescotto, Ed., Dept. MSM, Univ. Liège,
Belgium (2002), pp. 51-58.
Partner 2 (UCL)
I. DOGHRI and A. OUAAR, “Homogenization of two-phase elasto-plastic composite
materials and structures. Study of tangent operators, cyclic plasticity and numerical
algorithms”, International Journal of Solids and Structures, in press (2003)
T. PARDOEN and J.W. HUTCHINSON “Micromechanics-based model for trends in
toughness of ductile metals”, Acta Materialia, 51/1 (2003) 133-148
T. PARDOEN, D. DUMONT, A.DESCHAMPS and Y. BRECHET, “Grain boundary
versus transgranular ductile failure”, Journal of the Mechanics and Physics of Solids,
51/4 (2003) 637-665
F. LANI, Q. FURNEMONT, P.J. JACQUES, F. DELANNAY, and T. PARDOEN,
“Modèle micromécanique du comportement plastique de matériaux biphasés avec
transformation de phase : application au cas d'aciers multiphasés à effet TRIP”,
Matériaux 2002 – De la conception à la mise en oeuvre, Premier congrès
interdisciplinaire sur les matériaux (CD-ROM edited by the French Materials
Societies, ISBN 2-914279-08-6) 21-25 Oct 2002, Tours, France, CD-ROM Ref CM-
02-037
F. LANI, Q. FURNEMONT, P.J. JACQUES, F. DELANNAY and T. PARDOEN,
“Micromechanical modeling of plastic anostropy and strain induced phase
transformation in dual-elastoplastic phase materials”, EMMC6, 6th European
Mechanics of Material Conference - Non Linear Mechanics of Anisotropic Materials
(S. Cescotto ed.) 9-12 Sep 2002, Liège, Belgium, 169-177
T. PARDOEN and J.W. HUTCHINSON, “Microstructure-based model for linking
fracture toughness and microstructure in ductile metals”, ECF14, 14th European
Conference on Fracture (Neimitz A., Rokach I.V., Kocanda D., Golos K. eds.) 8-13
Sep 2002, Cracow, Poland, Vol. III, 635-644
T. PARDOEN, D. DUMONT, A. DESCHAMPS and Y. BRECHET, “Grain
boundary versus transgranular failure in aluminium alloys” ,ECF14, 14th European
Conference on Fracture (Neimitz A., Rokach I.V., Kocanda D., Golos K. eds.) 8-13
Sep 2002, Cracow, Poland, Vol. III, 625-634
37
S. GODET, P. HARLET, F. DELANNAY, and P.J. JACQUES, “Effect of hot-rolling

conditions on the tensile properties of multiphase steels exhibiting a TRIP effect”,
Steel Research, 73(6-7) (2002), pp 280-286
P.J. JACQUES, P. HARLET and F. DELANNAY, “Critical assessment of the phase
transformations occurring during the heat-treatment of TRIP-assisted multiphase
steels”, Proceedings Int. Conf. on TRIP-aided high strength ferrous alloys, B.C. De
Cooman, editor, Publ. Wissenschaftverlag Mainz GmbH, Aachen, Germany, 2002, pp
129-134
A. MERTENS, P.J. JACQUES, P. HARLET and F. DELANNAY, “On the
optimisation of the mechanical properties of two aluminium-alloyed multiphase
steels”, Proceedings Int. Conf. on TRIP-aided high strength ferrous alloys, B.C. De
Cooman, editor, Publ. Wissenschaftverlag Mainz GmbH, Aachen, Germany, 2002, pp
293-298
T. ROS-YANEZ, Y. HOUBAERT, R. PETROV and A. MERTENS,
“Characterisation of TRIP-assisted steel by atomic force microscopy and OIM”,
Proceedings Int. Conf. on TRIP-aided high strength ferrous alloys, B.C. De Cooman,
editor, Publ. Wissenschaftverlag Mainz GmbH, Aachen, Germany, 2002, pp 85-89
T. PARDOEN, Y. MARCHAL and F. DELANNAY, “Essential work of fracture
compared to fracture mechanics-towards a thickness independent plane stress
toughness”, Engineering Fracture Mechanics 69 (2002) pp 617-631
D. TIBERGHIEN, J. LAPIN, S. RYELANDT and F. DELANNAY, “On the control
of the residual porosity in iron aluminides processed by reactive squeeze-infiltration
of aluminium into a preform of steel fibres”, Material Science and Engineering A
323/1-2, (2002) pp 427-435
C. SALMON, C. COLIN, R. MOLINS and F. DELANNAY, “Strengthening of
Al/Ni-based composites by in situ growth of intermetallic particles”, Material Science
and Engineering A 334 (2002) pp 193-200
b. List of co-publications
Partner 1 (KUL-MTM) with partner 2 (UCL)

T. VAN ROMPAEY, Q. FURNEMONT, P.J. JACQUES, T. PARDOEN, B.
BLANPAIN, P. WOLLANTS, “Micromechanical modelling of TRIP steels”,
Proceedings Int. Conf. on TRIP-aided high strength ferrous alloys, B.C. De Cooman,
editor, Publ. Wissenschaftverlag Mainz GmbH, Aachen, Germany, 2002, pp 259-262
P. VAN HOUTTE, L. DELANNAY and S. R. KALIDINDI : “Comparison of two
grain interaction models for polycrystal plasticity and deformation texture prediction”,
International Journal of Plasticity 18 (2002) 359-377.
L. DELANNAY and P. VAN HOUTTE: "Influence of Local Lattice Orientation on
Grain Subdivision in 40% Cold-Rolled aluminium", Solid State Phenom. 87 (2002)
157-162.
G. LANGELAAN, L. DELANNAY, I. VERPOEST and P. VAN HOUTTE:
"Deconvolution of morphological texture from crystallographic texture data", J. Appl.
Cryst. 35 (2002) 327-337.
38
L. DELANNAY, S.R. KALIDINDI and P. VAN HOUTTE: "Quantitative prediction

of textures in aluminium cold rolled to moderate strains", Materials Science and
Engineering A336 (2002) 233-244.
G. LANGELAAN, S. RYELANDT, F. DELANNAY, J. H. ROOT and P. VAN
HOUTTE: "Determination of the morphological textures of fibres in a composite
material made from a textile of AISI 316L fibres", (Proc. ICOTOM 13), Mat. Sci.
Forum 408-412 (2002) 1389-1394.
Partner 1 (KUL-MTM) with partner 9 (Paris 13)
B. PEETERS, S.R. KALIDINDI, C. TEODOSIU, P. VAN HOUTTE and E.

AERNOUDT, “A theoretical investigation of the influence of dislocation sheets on
evolution of yield surfaces in single-phase b.c.c. polycrystals”, Journal of the
Mechanics and Physics of Solids, 50(4), 2002, 783-807.
39

Microstructure Towards Plastic Behavior

Uploaded by

Copyright:

Available Formats

Microstructure Towards Plastic Behavior

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Microstructure Towards Plastic Behavior

Uploaded by

Copyright:

Available Formats

IAP P5/08 Progress Report 2002

PROGRESS REPORT 2002

From microstructure towards plastic behavior

Partner 1: Katholieke Universiteit Leuven, departement materiaalkunde (KUL-MTM)

Work Package 1: Materials Preparation

2. Fine grained Aluminium-alloys

used by partners 2 (characterization and micro-mechanical tests), partner 1 (pre

Work Package 2: Forming Tests

Figure 2. Diagram of the device for determining moment-strain relationship

400 1 mm thick, at 4 mm/s

Figure 3. Unit moment for different steels

The experimental data were best fitted by following equation: σ = C ′′ ε n where C″ is

Table 1: Coefficients of Unit moment for different materials

Spring back determination

Work Package 3: Mechanical Tests

2. Fine grained Aluminium alloys

(Micro)Vickers Hardness measurement

Work Package 4: Microstructural Characterization

2. Fine grained Aluminium alloys

Measurement of the Strain Partitioning by Digital Strain Mapping

Characterisation of Fe-Mn and 301L Steels

austenite of a 15% Mn alloy decreases when thermodynamic equilibrium is reached.

Fe-30Mn-3Al-3Si: This material was analysed by optical microscopy and TEM. In

Work Package 5: Interpretation and Fundamental Understanding

2. Fine grained Aluminium alloys

Work Package 6: Phenomenological modeling

Two applications were analysed in Liège with LAGAMINE (partner 4):

Test A concerns FEM simulations of a cylindrical cup deep drawing. Different

Work Package 7: Multilevel Modelling

Activities within this multi-level modeling workpackage can be divided in two

1. "Vertical" interactions/validations on a specific material. Vertical interactions

2. "Horizontal" interactions aiming at comparing different types of models, at

a. Progress on Taylor-type models

1) Progresses with the original Lamel model:

2) Advanced Lamel Model

Figure 7. Schematic representation of a microstructure

The evolution of the crystallite orientations in regions 1 and 2 (fugure 7) can be

b. Validation of deformation texture predictions by various models for IF steel and

The results of the models will be quantitatively compared to experimentally observed

Table I. Texture index of difference ODFs for IF steel results.

CPFEM1 CPFEM2 FC RCP GIA Lamel1 Alamel1 Alamel2

Table II: Texture index of difference ODFs for 2 aluminum alloys

Models: see Table I, except VPSC: Visco-Plastic Self-Consistent Model (developed

Table II shows results obtained in the framework of an European project. Some

Homogenization schemes and numerical algorithms for two-phase elasto-plastic

Finally, in order to improve the understanding of the transformation behaviour, a

Computational homogenization technique (also called FE2 or micro-macro modeling)

order computational homogenization framework captures changes of the macroscopic

b. Development of a micro-macro code

Homogenization schemes and numerical algorithms for two-phase elasto-plastic

Other related researches

Work Package 8: Modelling of Strain Localisation and Ductility

Representatives of these partners have highlighted the background knowledge existing

1) Three-dimensional Localized Deformation

(i) ”Strain localization in strain softening rate-independent materials” (Issam Doghri,

(iii) ”Prediction of forming limit diagrams taking texture and microstructure-based

2) Two-dimensional Localized Deformation (Shear Banding)

There was an experimental and a theoretical contribution: