Electro-thermal Simulations of Battery Systems


Electrical model

  • Current profile as input
  • Dissipation calculated with Rs is strongly temperature dependent




Thermal model

  • Dissipation as input
  • Calculate resulting temperature distribution




Issue: long FEM simulation time

  • Example: Simulation of 1000s for one cell on 4 CPUs @ 3.4GHz with maximum time step of 1s -> 2000s








Thermal Modeling Using Model Order Reduction

  • Aim: generate a low dimensional approximation of the battery system
  • Mathematical  method (i.e., does not rely on intuition)

  • Typically: n~10000-100000; r~100
  • Results for a single pouch cell:
    • Mean temperature on electrode stack
    • Error < 0.1°C
    • FEM: 2000s (4 CPUs)
    • MOR: 5s (1 CPU) -> 1600:1 ratio




  • Model order reduction (MOR): needed to make system level simulation feasible

Simulation at system level with MOR

  • System design: need to change parameters (e.g., cooling condition)
  • MOR: system must be reduced again to take such a change into account ? not practical for system design

Simulation at system level with PMOR

  • Parametric MOR (PMOR): parameters can be changed for reduced system, no need for a new reduction
  • Enables system level design with practicable simulation times

PMOR And System Design

  • PMOR allows complex system simulation with variable parameters, e.g. for a battery system, a variable flow rate of the liquid cooling.
  • Allows consequential a fast analysis of the developed system with the current system design
  • Changes and improvements can be made fast during the development to the system, leading to a better product with a faster time to market.