On trade-offs between computational complexity and accuracy of electrochemistry-based battery models
Accepted, presentation at the IEEE Conference on Decision and Control (Nice, France) in December 2019 with following open-access manuscript and data repository.
Parameterization of electrochemical-thermal models show proper simulation results with low deviations to the experimentally measured constant current charge and discharge, as well as dynamic driving-cycle battery behaviour of the INR18650-MJ1 project cell (TUM).
Successful implementation and proof of real-time calculability of three different model-order reduced, electrochemical-thermal simulation models on a microcontroller (TUM).
Protocol for a generic characterization profile of current developed by Siemens to speed up process of equivalent circuit battery model from test measurement or from electrochemical model has been used with electrochemical model from TUM & TUE. Then the equivalent circuit model has been implemented in virtual test bench.
Two virtual test benches of eVan have been created by Siemens based on data from VOLTIA, powertrain measurement from TUE and equivalent battery cell model from TUM validated electrochemical model. They have been tested on different simulation environment (MiL, SiL & HiL). Demo with functional BMS connection has also been tested.
SUMMARY OF RESULTS
Electrochemical models have been developed and validated with test measurement
Model reduction approaches have been investigated and applied to embed electrochemical model in controller
Workflow have been refined to allow easier battery equivalent circuit model parametrization
Virtual tests bench models have been developed for MiL, SiL & HiL approach based on inputs from different WPs.
Real -Time implementation of MOR electrochemical-thermal simulation models on microcontroller.