A Lumped-Parameter Thermal Model for System Level Simulation of Hybrid Vehicles

Commercial vehicle electrification is a path to meeting stringent fuel consumption and emission targets. Considering extended operations in different applications, the powertrain components need to be selected and sized together with the cooling system to ensure consistent performance and longer product life. In order to maintain the interactions between subsystems in the optimization routines, multi-physics models should be utilized. A simplified thermal network model is developed in this work for hybrid powertrain cooling systems based on the analogy between heat transfer and current flow equations. The assumption of a lumped system is valid for these components as the conductive heat transfer is much higher than convective heat transfer (Biot number <0.1). The energy equations are solved to calculate component temperatures in state space form, whereas, in literatures, detailed object-oriented 1-D models are mainly used for this purpose. This approach reduces simulation setup time and computational effort without having a significant impact to model fidelity. In the thermal model, powertrain components are treated as heat exchangers along with cooling system components like radiator and pumps. Heat input from each component is derived from drive cycle analysis for transient simulations, and rated power is considered for steady state simulations. Thermal inertia and heat flow to the fluid are modelled as equivalent capacitance and resistances, respectively. The model results have been validated against high fidelity 1-D GT Power simulations for multiple thermal management schemes and a comparative analysis will be presented in the full paper. The simplified modelling approach eliminates the need for using detailed thermal models for system level entitlement studies. It has been effectively used to estimate thermal behaviors of powertrain components in different cooling system architectures, and the computational time is reduced by approximately 100 times compared to GT-Power simulations.