UCLA

As an essential element of sustainable energy technologies, electrochemical energy storage makes a significant contribution to the development of many industry fields such as consumer electronics and electric vehicles. Meanwhile, the complicated and evolving application markets raise new challenges to electrochemical energy storage systems in all-rounded performance matrices including energy density, operating conditions, durability and so on. There is a great amount of relevant research in various scales and directions, e.g., material development, electrochemical analysis and reliability testing. Modeling and simulation, together with rational material/system design, explores the internal processes for performance matrices of electrochemical systems, characterizes key improvement achieved by the implementation of design, and predicts future values/tendencies of significant variables based on learned patterns for system monitoring. This dissertation could be outlined with the following three parts:1. Explaining transient responses of proton exchange membrane fuel cells (with tungsten oxide addition to anode). An equivalent circuit model is employed to characterize dynamic responses of PEMFCs under transient operations and the improved power performance achieved by a system design of integrating a WO3 layer with anodes. To explain what is going on within the fuel cell system and the contribution of the system design, a mathematical model simulating gas transport within the gas diffusion layer attributes voltage undershoot under transient operations to unbalance between the demand of gas at the catalyst layer and the supply at gas flow channel (through the gas diffusion layer). Moreover, it is also observed that the addition of WO3 layers increases the capacity of the double layer, buffers the change of reaction current density, mitigates the transient gas demand-supply unbalance, and thus improves power performance under transient operations. 2. Explaining rate/cycling performances of lithium-ion batteries (with metal-organic frameworks addition to electrolyte). Lithium ion transference number plays an important role in research on high-rate/cycling performances of lithium-ion batteries. Firstly, to show how lithium ion transference number might be improved via rational design of MOF additions, heterogeneous geometries (pattern/density/location) to the separator/electrolytes are discussed. Secondly, to explore how lithium ion transference number affects rate performances of lithium ion batteries, a mathematical model is studied for half-cells to indicate the links among lithium ion transference number, electrolyte salt concentration gradient, concentration polarization and utilizable capacity. Thirdly, to explore how lithium ion transference number affects cycling performances of lithium ion batteries, mechanisms of solid-electrolyte interphase layer formation is introduced to the mathematical model for a full-cell, which discusses the overpotential of parasitic SEI reactions and loss of cyclable lithium under stable and dynamic cycling tests. 3. Predicting the remaining useful lives of lithium-ion batteries under cycling charge/discharge operations. Efficient and accurate remaining useful life prediction is a key factor for reliable and safe usage of lithium-ion batteries. A long short-term memory recurrent neural network model, by extracting features of discharge capacity variation under specific voltages due to capacity degradation, is trained to learn from sequential data of discharge capacities at various cycles and voltages and to work as a cycle life predictor for battery cells cycled under different operating conditions. Using experimental data of first 60 - 80 cycles, the model can achieve promising prediction accuracy on test sets of about 80 samples. Overall, this dissertation applies models for proton exchange membrane fuel cell and lithium-ion battery with different techniques to simulate internal processes occurring in a component/the entire system, and explains the reason why adopted design strategies mitigate certain types of performance bottlenecks or utilizes big data to learning from time series and make predictions on remaining useful lives. With these works, some inspirations might be provided for understanding methods to enhance transient and/or high-rate performances of electrochemical energy storage systems.