Model Reduction for Multiscale Lithium-Ion Battery Simulation

Mario Ohlberger,Stephan Rave,Felix Schindler

DOI: https://doi.org/10.1007/978-3-319-39929-4_31

2016-02-29

Abstract:In this contribution we are concerned with efficient model reduction for multiscale problems arising in lithium-ion battery modeling with spatially resolved porous electrodes. We present new results on the application of the reduced basis method to the resulting instationary 3D battery model that involves strong non-linearities due to Buttler-Volmer kinetics. Empirical operator interpolation is used to efficiently deal with this issue. Furthermore, we present the localized reduced basis multiscale method for parabolic problems applied to a thermal model of batteries with resolved porous electrodes. Numerical experiments are given that demonstrate the reduction capabilities of the presented approaches for these real world applications.

Numerical Analysis

What problem does this paper attempt to address?

The problem that this paper attempts to solve is: **Efficient model order reduction in multi - scale simulation of lithium - ion batteries**.

Specifically, the continuous modeling of lithium - ion batteries leads to complex, coupled nonlinear partial differential equation systems, which describe the reaction - diffusion - transport processes in multi - scale and multi - phase pore structures. Such models result in huge time - dependent discrete systems and require a large amount of computational resources even for a single simulation run. When it comes to parameter studies, design optimization or optimal control, multiple forward simulation runs are required, and the material or state parameters are different for each run, which makes the traditional high - dimensional models almost impossible to implement.

To solve this problem, the author introduced the following methods:
1. **Reduced Basis Method based on POD - Greedy**: Through Galerkin projection onto the POD space and combined with Empirical Operator Interpolation to deal with the strong nonlinear problems caused by Butler - Volmer kinetics.
2. **Localized Reduced Basis Multiscale Method (LRBMS)**: Applied to the heat conduction model of the battery to reduce the computational complexity and improve efficiency.

### Overview of the main content:

1. **Nonlinear order reduction of the micro - scale battery model**:
- The author shows how to apply the reduced basis method to the three - dimensional battery model, especially the nonlinear problems related to the Butler - Volmer kinetics at the electrolyte - electrode interface.
- Use empirical operator interpolation to efficiently handle these nonlinear problems.

2. **Application of LRBMS in the heat conduction model**:
- The author proposes a localized reduced basis multiscale method for handling the battery heat conduction model with an analytic pore electrode.
- The effectiveness of this method is verified by numerical experiments, especially in multi - scale and high - resolution geometric structures.

### Presentation of mathematical formulas:

- Butler - Volmer kinetics equation:
\[
j = 2k\sqrt{c_e c_s (c_{\text{max}} - c_s)} \sinh\left(\frac{\phi_s - \phi_e - U_0\left(\frac{c_s}{c_{\text{max}}}\right)}{2RT/F}\right), \quad N = \frac{j}{F}
\]
where:
- \( c_e \) and \( c_s \) represent the Li - ion concentrations on the electrolyte side and the electrode side respectively.
- \( \phi_e \) and \( \phi_s \) represent the electric potentials on the electrolyte side and the electrode side respectively.
- \( c_{\text{max}} \) represents the maximum Li - ion concentration.
- \( k \) is the rate constant.
- \( U_0(s) \) is the open - circuit potential.

- Heat conduction equation:
\[
\frac{\partial T}{\partial t} - \nabla \cdot (\mathbf{D} \nabla T) = Q
\]
where:
- \( T \) is the temperature.
- \( \mathbf{D} \) is the space - dependent thermal conductivity tensor.
- \( Q \) is the heat source term.

### Conclusion:

Through the above methods, the author shows how to effectively reduce the computational complexity in the multi - scale simulation of lithium - ion batteries, thus making large - scale parameter studies, design optimization and optimal control possible.