Begin

I finished my Ph.D. in 2019. The following is the abstract of my dissertation:

In this work, a physico-chemically motivated impedance-based and real-time capable lithium-ion battery model is introduced and parametrized according to cells from a Tesla Model S. The work is divided into three parts. In the first part, electrochemical properties of the cell are derived. The second part describes the deduction of a generic electrical and thermal model. Both models form the foundation of the so-called ISEAFrame, a simulation framework for energy storage systems. Three test cases for the simulation framework are discussed in the third part. In the first part, experiments on automotive cells of a Tesla Model S were conducted. A car module was disassembled and several hundred lithium-ion cells were extracted. A cell tear-down analysis was performed to investigate single components. From the active materials, several half-cells were constructed and electrochemically tested. It was found that one major factor affecting the relaxation behavior of the full-cell was instigated by the negative electrode. The relaxation is assumed to be caused by redistribution of lithium inside the graphitic electrode caused by potential differences between particles. Further experiments were conducted on the full-cells. 27 cells were connected in parallel and slowly charged and discharged to better understand the voltage response of parallel-connected cells. During the experiment, current and voltage of each specimen were logged. It could be shown that the change of voltage (dV/dQ) in the batteries was the main influencer for the redistribution of currents. The second part focuses on modelling the electro-thermal behavior of batteries. An implementation of a generic electrical and thermal model was derived and implemented in C++. The model is capable to be executed on real-time systems. The electrical model is an input-/output-based model. The current is used as input and voltage, heat and other internal states are used as output. The model consists of a time-variant differential-algebraic system that is linearized in each time step. The electrical network is based on lumped parameters and can be represented by a unidirectional and acyclic graph (tree). The parameters of the network can be varied according to the behavior of the battery and are dependent on state of charge, temperature as well as other states. From the tree representation, the Kirchhoffe quations are derived. These equations form the basis of a state-space model that describes physical and chemical processes by means of a voltage response. In addition to the real-time capable C++ model, a transcoder has been developed that translates a single simulation scenario from an intermediate representation into other target languages such as Matlab or Simulink. The thermal model is input-/output-driven as well. Heat is treated as the input and temperature is the output. The heat is fed in by the electrical model and conducted over a set of finite volumes that are representing the battery and its cooling. The temperature is fed back to the electrical network and thus modifying the electrical network’s elements. In the last part, three different test cases are discussed. The three test cases are exploring the capabilities of the developed simulation framework and its limitations. First of all, an in-depth look is taken at the half-cell graphite anode against lithium. Several relaxation measurements are simulated and compared to the experimental results of the first part of this work. In the second part, simulations of the full-cell are compared to experiments at different states of charge and different temperatures. Finally, a simulation of the whole Tesla Model S module is compared to experiments that have been conducted. The developed simulation framework has been published and can be accessed via https://github.com/FHust/ISEAFramework .

A newer version of the codebase with an ageing model can be found here: ISEAFramework .
The dissertation itself can be found here DOI or as a PDF