% IMPORTANT: The following is UTF-8 encoded. This means that in the presence
% of non-ASCII characters, it will not work with BibTeX 0.99 or older.
% Instead, you should use an up-to-date BibTeX implementation like “bibtex8” or
% “biber”.
@PHDTHESIS{Hu:1011821,
author = {Hu, Yang},
othercontributors = {Spatschek, Robert and Sandfeld, Stefan},
title = {{M}ulti-algorithm modeling for the degradation of solid
oxide fuel cells},
school = {Rheinisch-Westfälische Technische Hochschule Aachen},
type = {Dissertation},
address = {Aachen},
publisher = {RWTH Aachen University},
reportid = {RWTH-2025-04748},
pages = {1 Online-Ressource : Illustrationen},
year = {2025},
note = {Veröffentlicht auf dem Publikationsserver der RWTH Aachen
University; Dissertation, Rheinisch-Westfälische Technische
Hochschule Aachen, 2025},
abstract = {Solid oxide fuel cells (SOFCs) have attracted much
attention in the last decades due to their unique features
and high energy conversion efficiency. A very common cathode
material (La, Sr)(Co, Fe)O3−δ (LSCF) is considered to
have a huge potential for SOFCs applications. Furthermore,
the Fe-Cr ferritic alloy Crofer 22 APU as metallic
interconnect also possesses good performance but with low
costs. However, several degradation phenomena still limit
their commercial breakthroughs, such as Cr poisoning. Many
studies have indicated that a Cr2O3 oxide layer can form on
the interconnect surface, and Cr containing gaseous species
such as CrO(OH)2 and CrO3 could be observed due to a Cr
evaporation process at the SOFC operating condition, leading
to the formation of possible compounds like Cr2O3 at the
electrolyte/cathode interface and secondary compounds like
SrCrO4 that degrade cathode performance.In this thesis, a
data-driven strategy is proposed for the phase field
modeling of Cr2O3 oxide for Fe-Cr alloy, in which the phase
field simulations incorporate thermal fluctuations. Three
relevant independent parameters in phase field modeling for
determining the nucleation behavior are identified and used
as input features for machine learning classification and
regression model training. Whereas the classification could
be utilized for a phase field modeling acceleration, the
regression model can also determine the proper Langevin
noise strength for obtaining the proper nucleation behavior
in the phase field simulations. The predicted nucleation
density of Cr2O3 is helpful for the quantification of grain
sizes and understanding oxidation and evaporation mechanisms
of Fe-Cr alloys, which is helpful to optimize the materials
composition and SOFC operating conditions. Furthermore, this
data-driven strategy for phase field parameter optimization
through machine learning can serve as a blueprint for
efficient parameter identification for mesoscale
microstructure evolution simulations.Many degradation
phenomena in SOFCs involve various chemical reactions,
sufficient thermodynamic information is the foundation of
the thermodynamic calculation for investigating degradation
mechanisms involving volatile species. In this context,
sublimation-based vapor pressure modeling offers not only a
theoretical prediction approach but also a way to validate
and extend existing thermodynamic data. Due to the
limitations of the current experimental methods for the
vapor pressure measurement of a solid, a theoretical
prediction of the vapor pressure is extremely meaningful. As
long as the sublimation requires only overcoming a single
energy barrier, the vapor pressure is described by an
Arrhenius-type sublimation function. Therefore, the main
challenge is to determine the two related parameters:
sublimation enthalpy and the constant prefactor. Here, two
different models are developed to predict the sublimation
enthalpy. In the physical model, the cohesive energy is used
to approximately calculate the sublimation enthalpy by using
first-principles DFT calculations. And the machine learning
models are based on four different machine learning
algorithms and a small in-house training database. Here, a
feature filter strategy by introducing AutoML technologies
is proposed, which is for the determination of the possible
input feature configurations. Three input configurations are
filtered from 14 possible configurations with different
dimensions for further productive predictions as being most
relevant by using the feature filter strategy. The best
extreme gradient boosting regression model possesses a good
performance, which is evaluated from statistical and
theoretical perspectives, reaching a comparable level of
accuracy to density functional theory computations and
allowing for physical interpretations of the predictions.
The prediction of the other parameter prefactor is based on
statistical mechanics, lattice dynamics and thermodynamics.
The chemical potential equilibrium is established between
the gaseous molecules and that in the solid lattice via
statistical mechanics. However, the polyatomic molecules
possess generally complex motions, a simple monoatomic
molecule approximation is proposed to simplify the
calculation of polyatomic gaseous molecules at a convincing
level of accuracy. The predicted results are compared
against thermodynamic databases, which possess high
accuracy. Furthermore, the partial pressures caused by gas
phase reactions are also explored, showing good agreement
with experimental results. Finally, the model is applied to
calculate the vapor pressure of the Cr2O3. Thus, this
theoretical prediction workflow can not only extend the
thermodynamic database but also verify the existing
thermodynamic database, which can provide a complement for
the thermodynamic analysis of the SOFC degradation
phenomena.},
cin = {525820 / 520000},
ddc = {620},
cid = {$I:(DE-82)525820_20160614$ / $I:(DE-82)520000_20140620$},
pnm = {Verbundvorhaben WirLebenSOFC: Verständnis der
Wirkzusammenhänge der Alterungsmechanismen zur
Lebensdauervorhersage von SOFCs (03SF0622B)},
pid = {G:(BMBF)03SF0622B},
typ = {PUB:(DE-HGF)11},
doi = {10.18154/RWTH-2025-04748},
url = {https://publications.rwth-aachen.de/record/1011821},
}
Gast ::