Single-Particle Modeling and Experimental Parameter Identification for a Lithium-Cobalt-Oxide Battery Cell

Open Access
Hake, Alison Elizabeth
Area of Honors:
Mechanical Engineering
Bachelor of Science
Document Type:
Thesis Supervisors:
  • Hosam Kadry Fathy, Thesis Supervisor
  • Hosam Kadry Fathy, Honors Advisor
  • Christopher Rahn, Faculty Reader
  • Battery modeling
  • batteries
  • single-particle model
  • parameter identification
  • battery experimentation
This work explores the problems of determining the parameters of an 18650 lithium-cobalt-oxide battery cell and developing a single-particle diffusion model of the cell. Various battery chemistries exist in the category of lithium-ion cells. The differences in composition chemistry result in differing electrochemical and thermal parameters. Therefore, the ability to determine the parameters for any battery composition is important for the purpose of creating accurate models of the batteries. The models can then be used to study the applications of the battery cells, such as their use in electric vehicle battery packs. The literature includes battery modeling techniques of varying complexity with respect to both model order and parameter identifiability. Two major categories include equivalent circuit models and electrochemical models. The existing battery models in the literature, whether they are equivalent circuit or electrochemical, allow researchers to choose a model that best suits their needs. An equivalent circuit model, which has fewer parameters, is inherently simpler than electrochemical models such as the Doyle-Fuller-Newman or single-particle models. The choice of battery model consistently requires a trade-off between model complexity and model accuracy, is highly problem-dependent. The focus of this thesis is on two problems. First, the thesis uses the finite difference method to develop a single-particle model of a lithium-ion battery in Matlab. The model uses Fick’s law to represent anode- and cathode-side solid-phase diffusion dynamics. The finite difference method makes it possible to discretize this partial differential equation (PDE) in both time and space. The model neglects solution-phase diffusion dynamics, and linearizes the Butler-Volmer equation. The model’s parameters are obtained from the literature, and its computational behavior is examined for different values of key parameters (e.g., solid-phase diffusivity and solid particle radius). Second, the thesis presents a battery charge/discharge experiment on 18650 lithium-cobalt-oxide cells. The experiment is used for obtaining estimates of these cells’ parameters, as well as statistical distributions of these estimates. Together, these two contributions make it possible to parameterize the single-particle model from experimental data: an important potential future contribution. In conclusion, this project succeeds in creating a foundation and framework not only for future model development, but also for parameter identification experimentation. Performing battery cycling tests is a valuable skill in itself, but the understanding gained from testing cells to obtain specific parameters is essential to the study of lithium-ion batteries. Altogether, this thesis provides an introductory discussion of battery modeling, simulation, and experimentation and how these fundamental elements of battery study can be combined and expanded further.