Optimal Charging Protocols for LiFEPO4 Batteries via Deterministic Dynamic Programming

Open Access
Hatzell, Kelsey Bridget
Graduate Program:
Mechanical Engineering
Master of Science
Document Type:
Master Thesis
Date of Defense:
April 12, 2012
Committee Members:
  • Hosam Kadry Fathy, Thesis Advisor
  • Lithium Ion Batteries
  • Optimal Charging
  • Deterministic Dynamic Programming
This thesis developed a single state equivalent circuit model of a Lithium Iron Nanophosphate battery and applied the model with optimal control strategies to explore an optimal charging problem. Charging and discharging open-circuit potentials, resistances, and an overall battery capacity was estimated experimentally. The model was verified against experimental data and at low charge rates the model matches experimental data <1% error. The model was used to develop optimal charging schedules in order to optimize energy throughput in a battery. The results were compared to a simple single charge rate, and two experimental trajectories. The results indicate that optimizing for energy throughput may be a trivial objective function for the given constraints. Nevertheless, the results suggest three primary trends (1.) Charging trajectories begin at low charge rates, (2.) Optimized trajectories indicate that pulses of current should be added at 30% and 70% SoC, (3.) Optimal charging protocols magnitudes differ based on initial state of charge. Future work intends to use this framework to explore a physics based model with more rigorous constraints.