BattMo -- Battery Modelling Toolbox

This paper introduces BattMo, a flexible MATLAB-based finite volume toolbox for simulating diverse electrochemical battery cells that supports 3D geometries, coupled thermal and degradation models, and FAIR-compliant parameterization while enabling efficient gradient-based optimization through automatic differentiation.

The Gist

Imagine you are trying to build a better battery, but instead of building physical prototypes out of metal and chemicals (which is expensive, slow, and messy), you want to build them inside a computer. That is exactly what BattMo does.

Think of BattMo as a "Lego set for battery scientists."

Here is how it works, broken down into simple ideas:

1. The Blueprint (The "JSON" and "Graph" System)

Usually, changing a battery design in a computer program is like trying to rewrite the entire dictionary just to change one word. BattMo is different.

• The Recipe Card: You tell the software what your battery looks like and what it's made of using a simple text file (called JSON). It's like filling out a form where you say, "I want a cylindrical battery made of this specific type of lithium."
• The Flowchart: Inside the computer, the battery isn't just a blob of code; it's built like a flowchart or a computational graph. Imagine a family tree where the "parents" are the laws of physics and the "children" are the results. If you want to change how the battery heats up, you just swap out one branch of the tree without breaking the whole family. This makes it very easy to mix and match different ideas.

2. The Simulator (The "Virtual Lab")

Once you build your virtual battery, BattMo acts as a high-speed simulator.

• 3D Modeling: It doesn't just look at a flat slice of the battery; it builds a full 3D model. Whether your battery is a flat coin, a rolled-up jelly roll, or a big rectangular block, BattMo can visualize it in 3D.
• The "Heat" Factor: It doesn't just track electricity; it tracks heat, too. It simulates how the battery gets hot while charging and cools down while sitting, all at the same time.
• The "Aging" Watch: It can even predict how the battery gets old. It simulates things like a thin layer of gunk (called the SEI layer) building up on the inside, or how silicon materials swell up like a sponge when they soak up energy.

3. The "Smart Tutor" (Calibration and Optimization)

One of the hardest parts of battery science is guessing the right numbers for the materials.

• The Auto-Adjuster: BattMo has a built-in "smart tutor" that uses a technique called automatic differentiation. Imagine you are trying to tune a radio to get the clearest signal. BattMo can instantly calculate exactly which knobs to turn to get the perfect match between your computer model and real-world experiments. This saves researchers from spending weeks guessing and checking.

4. Who is this for?

• The Experts: Battery designers who want to test 50 different shapes in an hour instead of building 50 physical prototypes.
• The Students: Beginners who want to see how electricity and heat move inside a battery without needing a PhD to understand the code.
• The Developers: People who want to plug this tool into their own software workflows.

5. What makes it special?

While other tools exist (like PyBaMM), BattMo is unique because:

• It was built from the ground up to handle 3D shapes and heat together from the start.
• It is built on a foundation called MRST (a toolbox originally used for oil reservoirs), which means it is very good at solving complex math problems quickly.
• It is open and flexible. You can swap out the "engine" (the math equations) easily, just like swapping a car engine, to try new battery chemistries.

In a Nutshell

BattMo is a digital workshop where you can design, build, and test batteries in 3D. It uses a modular, block-based system that lets scientists easily swap parts, predict how batteries will age, and automatically tune their designs to match reality—all without needing to build a single physical battery in the real world.

Note: This software is currently being used in major European research projects (like HYDRA and BATMAX) to design new types of batteries for electric vehicles and energy storage, but the paper focuses on the tool itself, not on specific future products it will create.

Technical Summary

Technical Summary of BattMo: Battery Modelling Toolbox

Problem Statement
The development of high-performance electro-chemical systems, particularly Li-ion and post-Li-ion batteries, is critical for the electric energy transition. However, creating rigorous digital workflows to reduce physical prototyping and extract insights from data remains challenging. Existing physics-based models often struggle with parameter calibration, and while open-source frameworks like PyBaMM, cideMOD, and LIONSIMBA exist, they face specific limitations. For instance, PyBaMM has only recently (as of 2025) enabled 2D and 3D simulations and typically relies on an external package (PyBOP) for calibration and optimization. Similarly, cideMOD supports 3D simulations but lacks automatic differentiation, making the handling of complex nonlinear systems and design optimization difficult. There is a need for a flexible framework that natively supports fully coupled 3D electro-chemical-thermal simulations, efficient parameter calibration via adjoint methods, and a modular architecture for extending models.

Methodology
BattMo is a flexible finite volume continuum modelling framework implemented in MATLAB® (with compatibility efforts for Octave). It leverages the MATLAB Reservoir Simulation Toolbox (MRST) for meshing, solving large nonlinear systems, and visualization, utilizing fundamental robust, low-order finite volume methods and fully implicit time discretization.

• Model Architecture: The framework employs a hierarchical, graph-based design. Each model corresponds to a computational graph where nodes represent variables and directed edges represent functional relationships. This structure allows for modularity; coupling models is achieved by creating a new coupling model that contains existing models as sub-graphs, adding edges for coupling mechanisms while leaving sub-models largely unchanged.
• Input and Interoperability: Simulation inputs, including material parameters and geometric descriptions, are specified via JSON schemas. This approach adheres to the Battery Interface Ontology (BattINFO) guidelines to support semantic interoperability and FAIR principles.
• Physics and Chemistry: The core model is based on the Doyle-Fuller-Newman (DFN) approach. BattMo includes fully coupled thermal simulations and supports various degradation mechanisms, such as SEI layer growth (using Safari and Bolay models) and lithium plating. It also handles composite materials (e.g., Silicon-Graphite mixtures) and Silicon swelling.
• Solver and Optimization: The solver utilizes automatic differentiation to support adjoint computation. This capability allows for the efficient calculation of objective function derivatives with respect to all parameters, enabling gradient-based optimization routines for calibrating models against experimental data.
• Geometry: The toolbox supports 1D, 2D, and 3D geometries, including parameterized designs for cylindrical, prismatic, coin, jelly roll, and multi-pouch cells with various tab layouts.

Key Contributions

• Native 3D and Coupled Simulation: Unlike some predecessors, BattMo is designed from the foundation for coupled electro-chemical-thermal simulations in 3D without requiring external packages for core functionality.
• Adjoint-Based Calibration: The integration of automatic differentiation and adjoint computation provides an efficient method for calibrating complex battery models from experimental data, addressing a common bottleneck in physics-based modelling.
• Modular Graph-Based Design: The computational graph approach facilitates the extension and coupling of models, allowing researchers to easily modify underlying equations and integrate new degradation or material models.
• Standardized Input: The use of JSON schemas aligned with BattINFO ensures that model definitions are structured, reproducible, and interoperable.
• Ecosystem: The paper introduces the "BattMo family," which includes the MATLAB version, a Julia version (BattMo.jl), a Python wrapper (PyBattMo), and a web application (BattMoApp).

Results and Capabilities
The paper presents BattMo as a functional toolbox with a library of parameterized battery geometries and a suite of documented examples. It demonstrates the ability to:

• Simulate various battery chemistries (NMC, NCA, LFP, LNMO, SiGr) and formats (coin, pouch, cylindrical).
• Perform design optimization on geometry and parameters.
• Support standard testing protocols (CC, CV, CCCV) and time series inputs.
• Model specific phenomena including SEI growth, plating, and silicon swelling.
• Extend functionality to other electro-chemical systems, such as alkaline membrane electrolyzers and proton ceramic membrane models.

Significance and Impact
The authors position BattMo as a tool for reproducible battery cell development in research and education. Its significance lies in its ability to support both experienced designers optimizing virtual designs and beginners seeking to understand fundamental processes. The software has already been utilized in several commercial and non-commercial projects, including EU-funded initiatives such as HYDRA, BATMAX, IntelLiGent, BIG-MAP, and DigiBatt. In these contexts, it has been used to model novel LNMO cells as well as commercial LFP and NMC cells in both 1D and 3D configurations. By combining a modular framework with standardized inputs and advanced optimization capabilities, BattMo aims to reduce the reliance on physical prototyping and enhance the validity and reproducibility of battery research findings.