Model Distillation for Batteries

Derive control-ready ECMs using physics-based models to save months of lab work.

Battery models are essential for performance, safety, and BMS control. This paper shows how to cut months of lab work by distilling SPMe’s into fast ECMs.

Simply fill out the form to receive the Model Distillation whitepaper

What you’ll learn

  • How ECMs capture SoC, SoE, and thermal losses for BMS control

  • Why traditional parameterisation is slow, and how to replace it with virtual data

  • Generate a physics-based model on fewer pulse tests than the empirical model and achieve mean RMSE accuracy within 55 mV and 1.5°C

  • Extending to aged states and new designs

Who should read this whitepaper

  • Battery & system engineers, BMS developers, simulation leads, and R&D teams looking to accelerate modelling, validation, and lifecycle control.

Different models for different jobs: DFN, SPMe, ECM.

<55 mV

mean RMSE voltage error

<1.5°C

mean RMSE temperature error

Validated

across SoC, current, and temperature windows

Proven accuracy, ready for control

  • <55 mV voltage error across SoC and current ranges

  • ≤1.5 °C thermal accuracy across operating windows

  • Validated against empirical ECMs under real test conditions

  • Fast, lightweight models ready for BMS and pack-level simulation

Frequently Asked Questions

  • An ECM is a simplified RC network that reproduces a cell’s voltage and thermal response to current. It’s lightweight, fast to simulate, and widely used for battery management systems (BMS) and pack-level simulations.

  • The DFN model represents several electrode particles (e.g. 5) representing gradients across the electrode thickness, giving maximum fidelity but high computational cost.


    The SPMe simplifies this to one average particle per electrode, retaining insight to key internal states, such as anode surface potential, at much lower runtime.

  • Yes. By introducing loss of lithium inventory (LLI) and loss of active material (LAM) into the SPMe before distillation, the resulting ECMs embed ageing effects. This makes them useful for SoH tracking and lifecycle-robust control.

  • Yes. You can create a virtual cell in Breathe Design and export these designs to Breathe Model, an SPMe. Then, you can generate the data in simulation to arrive at an ECM for a cell all before hardware exists. This enables virtual prototyping for new chemistries and designs.

  • Open Circuit Voltage (OCV) curves — including charge/discharge hysteresis — can be adopted directly from the SPMe models. This avoids slow experimental tests while still producing realistic inputs for the ECM.

  • Yes, our Breathe Model free trial of the Molicel P45B includes a MATLAB notebook that you can use to generate an ECM model from, with full control over the breakpoints it is parameterised for.