Optimization of Fail-Operational Vehicle Traction Battery System Design

Abstract

Besides the driving functions, powertrain electrification and design is an important task in development of automated driving. Starting with highly automated driving (SAE Level 4), human intervention is not required anymore and therefore the requirements on fail-operability of safety-critical subsystems increase in order to maintain the safety of the vehicle and the passengers. This requires new design concepts for an electrical powertrain and its subsystems including the traction battery system. To let the vehicle automatically drive to a safe state in case of a fault, a minimum battery performance in terms of remaining energy, output voltage and power is required. Moreover, a battery design limiting the capacity degradation over the lifetime to a specified minimum can reduce the overall maintenance cost. It is thus important to quantitatively evaluate the system design in terms of performance degradation and cost already in an early phase. This paper investigates a systematic approach for the reliability estimation and optimization of fail-operational battery systems. A generic and behavioral battery model describing the dependencies between single elements is proposed, which is then further used for reliability evaluation by means of Monte Carlo simulation. This framework offers considerable advantages for modeling of complex non-linear dependencies as, e.g., for capacity degradation. The combination of stochastic failure analysis and aging mechanisms forms the basis for an overall battery design optimization with a genetic algorithm. The results for one example use case show a cost-optimal battery system design with additional cells and partly redundant E/E components.