FE-modeling of Ideal Grain Growth Based On Preprocessed EBSD Data

See
discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/259683809
FE-modeling of ideal grain growth based on

preprocessed EBSD data
ARTICLE in PAMM DECEMBER 2011
DOI: 10.1002/pamm.201110227
READS
21
3 AUTHORS, INCLUDING:
Tobias Kayser
Salzgitter Mannesmann Forschung, Salzgitt
12 PUBLICATIONS 33 CITATIONS
SEE PROFILE
Available from: Tobias Kayser

Retrieved on: 13 January 2016
PAMM Proc. Appl. Math. Mech. 11, 471 472 (2011) / DOI 10.1002/pamm.201110227
FE-modeling of ideal grain growth based on preprocessed EBSD data

Slawa Gladkov1, , Tobias Kayser2 , and Bob Svendsen1
1
2
RWTH Aachen, Chair of Material Mechanics, 52062 Aachen, Germany

TU Dortmund, Institute of Mechanics, 44227 Dortmund, Germany
The purpose of this work is to show the use of experimentally measured micrograph data in the context of ideal grain growth
simulation which is modeled with a help of finite element method. In this regard the micrograph data is considered as initial
condition for a grain growth simulation. General remarks on preparation (preprocessing) of the micrographs and staggered
algorithmic formulation are presented.
c 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Introduction
There is a number of models available to simulate an ideal (normal) grain growth of polycrystals both on continuum and
discrete side. In this work the continuum phase field model developed in [4] in conjunction with numeric finite element
method is used to perform simulation of a grain growth process. To initiate the simulation there are few options available for
the choice of initial conditions: (i) random data which in a sense simulates the initially liquid phase, (ii) artificially generated
microstructures (in 2D and 3D) and (iii) real, measured preprocessed EBSD data (2D only). In this work the last option is
chosen.
The data one gets directly from the microscope (Fig. 1, left) cannot be used in the simulation because it will lead to
tremendous computational effort in the initial simulation steps due to a fact that practically each point in the micrograph have
to be associated with a certain grain orientation. This can be avoided using preprocessing stage [1, 2] where among others the
symmetries of a crystal are taken into account and grain structure is clearly visible (Fig. 1, right). Exactly this preprocessed
data is used as initial condition for the finite element simulation explained in the next sections.
Fig. 1 Steps for polishing up the micrograph data from microscop to be ready for use in finite element simulation.
Grain growth model
Based on the M. Gurtins microforce balance [3] approach one can formulatate the most general form of evolutionary partial
differential equations for the phase fields which is consistent with the second law of thermodynamics:
p
X
[B] =
=1
= 1, 2, ..., p
(1)
with certain initial and boundary (homogeneous Neumann in this case) conditions:
(x, 0) = 0 (x),
(x, t) n = 0 on ,
= 1, 2, ..., p.
(2)
k }p , {k }p ) is a free energy functional, {k }p are so called external

Here {k (x, t)}pk=1 are p phase fields, ({
k=1
k=1
k=1
p
p
microforces, [B]({k }k=1 , {k }k=1 , { k }pk=1 ) is a symmetric positive semi-definite matrix of constitutive moduli, is the
domain of problem definition with its boundary = \ and n is a unit outward normal to .
Corresponding author: e-mail svyatoslav.gladkov(AT)rwth-aachen.de, phone +49 241 80 25009

472
Section 7: Coupled problems
The particular model can be given by the proper specification of the free energy functional and the matrix of constitutive
moduli. Firstly, lets assume that the matrix of constitutive moduli is diagonal and its elements do not depend on solution (they
are constants):

1
[B] {k }pk=1 , {k }pk=1 , { k }pk=1 [B] =
(3)
L
where L > 0 = 1, 2, ..., p are positive constants and is a Kroneckers delta operator. Secondly, assume that there is
no external microforce, i.e.
= 0 = 1, 2, ..., p.
(4)
Finally, assume the following form of the free energy:

({k }pk=1 , {k }pk=1 ) =

p
p
p
p X
X
X
X
C1 2
C2 4
+
(1 )2 2 +
+ C 3

2
4
2
=1
=1
=1
(5)
=1
where C1 > 0, C2 > 0, C3 > 12 C2 are positive constants and {k }pk=1 > 0 are positive gradient energy coefficients.
Plugging all these into the kinetic equations (1) one recovers the grain growth model from [4], i.e. = 1, 2, ..., p
p
X
(1 )2 .
= L + C1 C2 3 4C3
(6)
=1
This is a system of p coupled quasilinear partial differential equations of second order in space and first order in time.
Finite element formulation
There are two possibilities to discretize the problem (6) in space: monolithic and staggered. Monolithic approach means that
the problem is being solved in the fully coupled setting. This leads to a definition in each finite element node p degrees of
freedom, what in the end will lead to enormous system matrix size and make problem numerically non-treatable. On contrary
staggered scheme means that the problem is being solved in decoupled fashion, i.e. equation by equation holding all the other
fields constant (from the previous time step or equation step). This leads to definition of only p scalar finite element fields and
sequential update of them. This scheme is employed here.
Explicit time discretization for a problem (6) leads to very small time steps. Fully implicit scheme leads to very long
computation times due to re-assembly of the system right hand side vector on each step of the Newton-Raphson procedure
and for each phase field (staggered scheme is employed). Thus semi-implicit scheme is used here. In this case only the linear
part (the "mass" matrix and discrete Laplacian which are constant for all time steps and all phase fields) is taken on the new
time step while the local nonlinear part is taken from the previous time step. This formulation allows firstly to build up system
matrix only once and secondly to avoid Newton-Raphson iterations.
Discretized in space and time the weak form of equations (6) will take a form:
Z
Z
Z
p
X
1
1

(7)
t+1
+ 1 + 32 4
t+1
2 dx
dx =
dx + 2
t
t
=1
where now t+1

are unknown solutions on the new time step, are known from the previous time step, t is the time step
and to simplify the notation it was assumed that C1 = C2 = C3 = 1 and L = 1, = 2 = 1, 2, ..., p.

Variational equation (7) is solved using Bubnov-Galerkin method where as a basis functions bi-linear finite elements are
taken. Numerical results obtained using the open source finite element library deal.II [5] will be reported in the future
works.
References
[1]
[2]
[3]
[4]
[5]
F. J. Humphreys, J. Mater. Sci., 36, 38333854 (2001).

F. J. Humphreys, J. Microsc., 213, 247256 (2004).
M. E. Gurtin, Physica D, 92(3-4), 178192 (1996).
D. Fan and L.-Q. Chen, Acta. Mater. 45, 611622 (1997).
W. Bangerth, R. Hartmann and G. Kanschat, ACM Trans. Math. Soft. 33, 24/124/27 (2007).

www.gamm-proceedings.com

FE-modeling of Ideal Grain Growth Based On Preprocessed EBSD Data

Uploaded by

Copyright:

Available Formats

FE-modeling of Ideal Grain Growth Based On Preprocessed EBSD Data

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

FE-modeling of Ideal Grain Growth Based On Preprocessed EBSD Data

Uploaded by

Copyright:

Available Formats

See

FE-modeling of ideal grain growth based on

Available from: Tobias Kayser

FE-modeling of ideal grain growth based on preprocessed EBSD data

RWTH Aachen, Chair of Material Mechanics, 52062 Aachen, Germany

Grain growth model

k }p , {k }p ) is a free energy functional, {k }p are so called external

Corresponding author: e-mail svyatoslav.gladkov(AT)rwth-aachen.de, phone +49 241 80 25009

c 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Section 7: Coupled problems

Finally, assume the following form of the free energy:

Finite element formulation

where now t+1

and to simplify the notation it was assumed that C1 = C2 = C3 = 1 and L = 1, = 2 = 1, 2, ..., p.

F. J. Humphreys, J. Mater. Sci., 36, 38333854 (2001).

c 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

You might also like