A high-performance model solver for “in-the-loop” battery simulations

Battery cell and pack simulation has become an important field for complex vehicle and energy systems simulations, as well as for model based control of such systems. The quality and performance of simulation, besides the quality of the model itself, is greatly determined by the characteristics of the model solver. Speed, accuracy, ability for quick computation of sensitivities and low memory footprint are some of the solver characteristics required for successful deployment. Different simulation time steps among the system components as well as different simulation concepts, such as continuous time or discrete event time, as is the case for power supplies simulation, favor a simulation environment where the battery subsystem is implemented “in-the-loop”, running as a co-simulation slave process. General purpose model solvers, by their own nature, impose too much overhead or lack essential functions. In this paper we will present “dtSolve”, a model solver written for physics based battery model simulations and we will demonstrate its performance in “real-time” complex battery pack multi-physics simulations. dtSolve implements Automatic Differentiation (AD), adaptive time steps, Differential Algebraic Equation (DAE) formulation for transient solutions, sparse matrix techniques, a space optimized DAE solver and an optimization code, all of them tightly coupled into an object oriented design platform.