Design and development of 48V Battery Management System for Lithium ion battery
Research Area - Lithium ion battery SOC estimation, Battery management system, Battery modeling.
Dipali Dange
Guide : Dr. D. N. Sonawane
Year : 2017-2019
Mail : [email protected]
My Profile
Research Area - Lithium ion battery SOC estimation, Battery management system, Battery modeling.
Dipali Dange
Guide : Dr. D. N. Sonawane
Year : 2017-2019
Mail : [email protected]
My Profile
Abstract -
On-board Battery Management System (BMS) needs accurate SOC estimation of Li-Ion cells during runtime and drive cycle assistance in the vehicle. Lithium-ion cells have a relatively flat discharge profile indicative of the topotactic intercalation of lithium ion, this makes state of charge (SOC) estimation more complicated than just measuring the open circuit voltage of the cell at equilibrium. Today 90% SOC estimation is performed by coulomb counting way, 7% uses electrical modeling approach, & 3 % by Multiphysics modeling.
Coulomb counting method of SOC estimation is easy to implement but this technique is not that accurate due to incorrect battery current measurement and losses which produces estimation error. This technique also requires the use of look up table for calculating initial SOC which eventually utilizes large memory for the storage. Electrical modeling approach has moderate execution time of electrical model but the components are not the function of degradation and temperature thus lacks in prediction capability as battery ages. Physics based models are highly accurate, sensitive to temperature and the model parameters are function of battery degradation. But, these models are highly computationally expensive due to high computational complexity which eventually becomes difficult for hardware implementation to achieve the fast sample time for online SOC measurement.
Understanding the difficulties in online SOC estimation and the problems associated with these methods, in this research, we propose to explore a modified (hybrid) SOC estimation approach for single Li-Ion cell. We propose to have the combination of lookup table way and the electrical equivalent model which is a function of SOC and the temperature. Due to low computational complexity and low memory footprints of electrical equivalent model, the suggested approach will be feasible and robust solution for embedded implementation to achieve the fast sample time for online SOC estimation.
Experimentation is performed on the 18650 cylindrical cells, data during charging and discharging was logged and profiled. Initial SOC is modeled with experimental data by polynomial equation with ensuring least Root Mean Square Error (RMSE). Estimation result is verified under different C rate during charge and discharge cycles at different operating temperatures. We observed 93.3% accuracy in SOC estimation compared to 83.3% by conventional coulomb counting way, the work is in progress and the efforts will be targeted to further improve the SOC estimation by exploring electrical equivalent models those are function of SOC and temperature.
On-board Battery Management System (BMS) needs accurate SOC estimation of Li-Ion cells during runtime and drive cycle assistance in the vehicle. Lithium-ion cells have a relatively flat discharge profile indicative of the topotactic intercalation of lithium ion, this makes state of charge (SOC) estimation more complicated than just measuring the open circuit voltage of the cell at equilibrium. Today 90% SOC estimation is performed by coulomb counting way, 7% uses electrical modeling approach, & 3 % by Multiphysics modeling.
Coulomb counting method of SOC estimation is easy to implement but this technique is not that accurate due to incorrect battery current measurement and losses which produces estimation error. This technique also requires the use of look up table for calculating initial SOC which eventually utilizes large memory for the storage. Electrical modeling approach has moderate execution time of electrical model but the components are not the function of degradation and temperature thus lacks in prediction capability as battery ages. Physics based models are highly accurate, sensitive to temperature and the model parameters are function of battery degradation. But, these models are highly computationally expensive due to high computational complexity which eventually becomes difficult for hardware implementation to achieve the fast sample time for online SOC measurement.
Understanding the difficulties in online SOC estimation and the problems associated with these methods, in this research, we propose to explore a modified (hybrid) SOC estimation approach for single Li-Ion cell. We propose to have the combination of lookup table way and the electrical equivalent model which is a function of SOC and the temperature. Due to low computational complexity and low memory footprints of electrical equivalent model, the suggested approach will be feasible and robust solution for embedded implementation to achieve the fast sample time for online SOC estimation.
Experimentation is performed on the 18650 cylindrical cells, data during charging and discharging was logged and profiled. Initial SOC is modeled with experimental data by polynomial equation with ensuring least Root Mean Square Error (RMSE). Estimation result is verified under different C rate during charge and discharge cycles at different operating temperatures. We observed 93.3% accuracy in SOC estimation compared to 83.3% by conventional coulomb counting way, the work is in progress and the efforts will be targeted to further improve the SOC estimation by exploring electrical equivalent models those are function of SOC and temperature.
