The framework consists of two submodels on two levels. These are the electrode and the cell level. On each level, an independent coordinate system is assigned for the discretization. On the electrode level, this coordinate describes the local electrochemical processes of the cell along the x-axis of each electrochemical element, which consists of anode, cathode, separator, and current collectors, in total five computational domains. On the cell level, the cell configuration includes current collector tabs, current collectors, polymer cover, and electro-active layers as computational domains in three dimensions. Each meshed element in an electro-active layer includes all computational domains of the electrode level and is defined as a nonlinear resistor, which represents the local potential drop that stems from the local electrochemical processes. On the cell level, charge and energy conservation equations are solved in three dimensions. Both submodels are coupled by inter-level variables. The submodels and their coupling approach is provided in detail in the following sections.
2.1. Electrode Level: Electrochemical Submodel (ESM)
In the electrochemical submodel (ESM), the electrochemical processes are modeled on the electrode level as shown in
Figure 1, based on P2D models [
17,
18,
19]. The discretization in both
x and particle radius
r directions lead to high computational cost. To reduce the present complexity, two simplifications compared to a full-order P2D model are introduced. In ESM, the pore-wall fluxes
at the electrode-electrolyte interfaces
are denoted as average lumped variables over space [
20,
21]:
and it can be calculated according to
with
electrode thickness,
the local current density on the current collector flowing out of the electrode,
F the Faraday constant and
the specific area of the electrode. The average lumped variable
simplifies the lithium diffusion process in the active particles, which is usually expressed by Fick’s law:
with the boundary conditions of the surface flux at the outer boundary (
) of particles and in the particle center (
) being given as follows:
where
is the particle radius and
is the solid diffusion coefficient of the electrode. By taking into account the aforementioned boundary conditions, Equation (
3) can be approximately solved as an analytical solution of surface and average lithium concentrations [
22]. The surface concentration of lithium in each electrode is expressed as:
where
is the initial lithium concentration in the electrode particle,
N is the truncation error term number, and is defined as
in this work, and
is the
nth eigenvalue, which can be calculated as follows [
22]
The bulk concentration in the solid phases
can be evaluated by solving lumped mass balance:
It is noted that the state of charge (SOC) of the battery is also related to the bulk solid concentration. It is defined by the variation range of bulk solid concentration in cathodes [
21], as
The initial solid concentration
is set as the value when SOC = 100% and the bulk solid concentration at SOC = 0%,
can be evaluated by
The solid diffusion overpotential in each electrode
is derived as a function of surface and bulk concentrations, and is expressed as:
where
is the maximum lithium concentration in active particles, and
is the open circuit potential (OCP) in each electrode. The Butler-Volmer equation describes the charge transfer reaction rate at the electrode-electrolyte interfaces. The reaction rate is driven by the kinetic overpotential
, which is defined as the deviation of the potential in solid phase
, as well as the potential in the solution phase
, and
,
where
and
are the charge transfer coefficients,
R is the ideal gas constant, and
is the exchange current density, which is given as a function of the surface lithium concentration in solid phases and lithium concentration in the solution phase at the electrolyte-electrode interface. (For the evaluation of voltage in ECM element, all field variables, e.g.,
, in the anode are evaluated at
and in the cathode at
[
21]).
Here
is a kinetic rate coefficient,
is the lithium concentration in solution phase at the electrolyte-electrode interfaces. Assuming that
, based on Equation (
12), the kinetic overpotential
can be analytically solved [
21]:
where the term
is given by Equation (
16):
and
is the porosity of electrode
i.
For simplification, the voltage in each ESM element can be denoted as the sum of different terms, as described by Prada et al. [
21]:
The reference potential on the anode
is defined by the electric potential at the negative current collector on cell level
. The total overpotential
is defined as the sum of overpotentials of diffusion and kinetics:
The electric potential in electrolyte
is evaluated in the continuous liquid phase, which consists of 3 domains,
at the electrode level:
where
. The boundary conditions are given as follows:
and at the interfaces,
where
is the effective ionic conductivity of electrolyte in every domain,
is salt concentration in electrolyte and
is the transference number of
. With Equations (
19)–(
21),
has been calculated analytically in every domain due to the lumped reaction rate as follows [
21]:
for the anode domain,
for the separator domain, and
for the cathode domain. The effective ionic conductivities are defined by
, and
is the porosity in every domain and
is the Bruggemann coefficient.
The mass conservation of lithium salt in the electrolyte is given by:
with the corresponding boundary conditions expressed by:
Here,
is the effective diffusion coefficient of electrolyte, which is evaluated by
for every domain. The governing equations of the electrochemical submodel, all summarized parameters and material properties of the ESM are listed in
Appendix A and
Appendix B, respectively. Further equations about temperature dependence and thermal behavior are given in the following sections.
2.2. Thermal-Electric Submodel
On the cell level, a 3D thermal-electric continuum submodel is developed to solve temperature
T, and local electric potentials
and
at the current collectors and tabs. A 3D geometry of a lithium-ion pouch cell with multiple electro-active layers is considered to be illustrated in
Figure 1. It consists of 6 computational domains—two electric tabs, two electric current collectors, electro-active layers, and polymer cover. There are 40 parallel layers in the electro-active domain. The heat convection boundary conditions are set on the domains of polymer cover and tabs. Thermal and electric properties of cell components are considered to be anisotropic in the XY-plane and in Z-directions [
7]. The energy conservation for all domains yields:
where the material density of cell component
, specific heat capacity
and thermal conductivity
are attributed to every domain. The volumetric heat source in the electro-active domain
is the sum of the reaction heat
, the entropy change
and Joule heat
:
All locally volumetric heat sources are determined by the electrochemical submodel on the electrode level. The coupling values and expressions will be introduced in the following section. In present simulation scenarios, the cell surfaces are assumed to be exposed to the environment, where the convective heat transfer boundary conditions on surfaces including tabs yield:
where
is the normal unit vector,
is constant ambient temperature and
is the heat transfer coefficient between cell and the environment. The charge conservation of the cell follows Ohm’s law. In tab and current collector domains, it is expressed as:
where
is the electrical conductivity of electrode current collector (+ or −), and
is the volumetric current density flowing from electro-active layers. In tab domains,
is zero. For a discharge process, the electric boundary conditions as shown in
Figure 1 are expressed as
at the negative electrode tab, and
at the positive electrode tab, where
and
are areas of anode and cathode tabs, respectively. On the tab terminal of anode current collectors, the reference potential is defined as:
A 1D nonlinear resistor network is employed for charge conservation in electro-active domain instead of Poisson’s equation in Z-direction [
23]. The number of nonlinear resistors
is equal to the number of nodes on the XY-plane after meshing the cell (see
Figure 1). According to Gerver’s approach [
23], the local current
in the electro-active domain follows the relation:
where
and
are the local potentials at nodes on cathode and anode current collectors on the XY-plane, respectively.
is the local resistance evaluated by an independent electrochemical submodel. Using nonlinear resistors, it is possible to combine local electrochemical processes with geometry on the cell level. Therefore, the coupling of too many partial differential equations (PDEs) of electrochemical processes is avoided, and the simulation of charge conservation in 3D pouch cell configuration with a faster convergence than full-physics coupling is enabled [
23].
2.4. Model Studies
In this paper, the electrochemical submodel is validated first at various current rates with a full-order P2D model [
17], then the 3D multiphysics model is used to simulate a 12 Ah large-format pouch cell at various discharge rates under galvanostatic discharge conditions. The electrodes of the cell are composed of lithium rich NCM materials (Ni, Co, Mn) and graphite, respectively. We assume a binary electrolyte containing
in ethylene carbonate/ethyl methyl carbonate. The cell properties are presented in the
Appendix B and
Appendix C as model parameters. The cell dimensions were 99 mm width × 120 mm height × 9 mm thickness. The cell contains 40 parallel stacked electro-active layers, regarding the normal number of layers for commercial pouch cell [
12]. The multiphysics model was defined as an open system, with a convective heat exchange coefficient of 15
connected to the environment. The initial and ambient temperatures were chosen as 25
. The electrochemical submodels were implemented in MATLAB (R2015b, MathWorks, Natick, MA, USA) with SundialsTB solvers [
25]. The computation cost of electrochemical submodels is estimated by MATLAB using the function tic-toc; it can record the internal execution time for MATLAB programs. The 3D thermal-electric submodel was implemented in Ansys APDL 15.0. All simulations were executed on an 8 core I7-2600 processor with 16 GB memory.