Accurate battery pack modeling for automotive applications

• Modeling at the battery pack level (vs. cell level) for automotive applications.
• Using one simple model rather than aggregating hundreds for pack representation.
• Building model based on the bandwidth of the stimulus by assigning time constants.
• Model robustness is proved on four Ultralife 14.4 V, 6.8 Ah lithium-ion battery modules.
• High fidelity is proved on an A123 360 V, 21.3 kWh lithium-ion battery pack.

This paper describes an advanced battery pack modeling method for automotive applications. In contrast to the common approach of aggregating hundreds of battery cell models in series and parallel for battery pack representation, a simple yet accurate electrical analogue battery model with constant parameters is used to represent the whole battery pack. The modeling process involves only the external characteristics of the battery pack, thus detailed knowledge of the physical construction of the battery pack or the physical parameters of the battery cells are avoided. Battery experimental tests include an independent pulse charging/discharging cycle test and several performance estimates acquired from the anticipated battery application. This modeling approach is achieved by anticipating the bandwidth of the battery application and then optimizing the bandwidth of the battery pack model with this knowledge. The reported work enables a fast dynamic battery pack simulation with a new level of high fidelity. Although the scope of this work does not include temperature and lifetime effects, it is shown that the form of the model and the parameter extraction procedure can encompass these important improvements. This approach was experimentally verified on a 360 V, 21.3 kWh lithium-ion battery pack operated in a plug-in hybrid electric vehicle.