Modeling & Simulation, Testing & Validation (MSTV)

Reduced-Order Modeling Method for Fatigue Life Predictions of Hybrid Electric Vehicle Batteries

by Sung-Kwon Hong; Bogdan I. Epureanu; Matthew P. Castanier


The goal of this work is to develop an efficient numerical modeling method for the structural dynamic response of hybrid electric vehicle (HEV) batteries in order to support fatigue life predictions. The dynamics of HEV battery packs are known to feature very high modal density in many frequency bands. The high modal density combined with small, random structural variations among the cells (which are unavoidable in practice) can lead to drastic changes in the structural dynamics. Therefore, it may be important to perform probabilistic simulations of the structural dynamic response with cell-to-cell parameter variations in order to accurately predict the fatigue life of a battery pack. However, the computational time for obtaining forced response results for just a single sample of parameter variations with a finite element model can be on the order of a day. One approach to overcome this challenge is to generate parametric reduced-order models (PROMs). The novel approach is based on two key assumptions. First, it is assumed that the mode shapes of a battery pack (with parametric variations in the cells) can be represented by a linear combination of the mode shapes of the nominal system (with identical cells). Second, it is assumed that the frame holding each cell has vibratory motion. PROMs are validated numerically with full-order finite element models by comparing forced response predictions. The new PROMs are able to predict the dynamics of battery packs 1,000 to 10,000 times faster than full-order finite element models while maintaining accuracy. For the few initial cases considered, small cell-to-cell parameter variations are found to lead to an increase of up to 60% in the vibration amplitude of a battery cell, which could have a significant impact on fatigue life.