A “Who’s Who” Guide to Battery Modeling Software in 2026

part 2

Towards the end of January, the Volta Foundation released their 2025 Annual Battery Report. It’s a serious 767 page adventure collated by over 120 battery professionals. Two of those contributors, Daniel Cogswell and Andrew Weng, today bring you a ‘Who’s who’ of battery modelling, as a crash course intro to help those who might not know where to start.

---

In part two of this article, we share some observations about the capabilities and ideal use cases for several common battery simulation software packages.

First we need to start by setting some ground rules.

We are not trying to write a comprehensive review of all battery-related software. As this year’s Volta Battery Report highlights, there is an ever-growing list currently with dozens of software tools. It would be difficult for an individual, or even a small team, to become experts in all of these; battery simulation is complicated and the tools have high learning curves.

Instead, our hope is to highlight the breadth of available software. The software we discuss were chosen because they are tools that we have experience using in a significant capacity for research in academia and industry. We have tried to present the software roughly in order from lowest to highest on the software pyramid, starting from L1 (see Part I).

Figure taken from the 2025 Battery Report by Volta Foundation.

Finally, our goal is not to rank the software or pick favorites. Instead we want to identify the best use cases for each piece of software, so that readers who have little or no experience with battery simulation will appreciate the breadth of tools available and be able to reach for the best tool for their needs.

Simscape Battery (MATLAB^®/Simulink^®): A system engineer’s friend

Simscape Battery includes MATLAB objects, functions, apps, and a library of dynamic battery cell models that facilitate and automate the creation of parameterized and simulation-ready Simscape models.

There are two types of practicing engineers: those who happily use MATLAB, and those who begrudgingly use MATLAB but wish they can switch to a more “modern” modeling stack such as Python. The truth is that MATLAB is still used everywhere in engineering, including at universities. MATLAB forms the backbone of many systems design and control projects, especially in the automotive industry. It makes sense to build battery models in MATLAB if you want your model to plug in easily to existing industry system models. It is true that other programming languages, such as Python, support a wider array of modern development tools such as VS Code, Cursor, Copilot, and Jupyter Notebooks. To their credit, Mathworks, the company behind MATLAB, is trying its best to keep up with recent offerings such as MATLAB Live Editor, MATLAB Online, and MATLAB Copilot.

As of 2024, Mathworks has released a tool called Simscape Battery which can be used to create simulation-ready battery models. The main strengths: direct integration with Simscape, a popular software for developing control algorithms. The library also supports both equivalent circuit battery models (ECM) and a version of the Single Particle Model with electrolyte dynamics (SPMe). With Simscape integration, it is also relatively straightforward to co-simulate electrical and thermal dynamics at the battery pack level. Finally, one of Simscape’s key capabilities is support for C++ code generation, making it easy to deploy and test control algorithms in real-time embedded devices.

There are also some limitations with Simscape Battery. For one, the underlying battery models are close-sourced: without access to the underlying code that defines the model, it becomes harder to customize their battery models. While some developers have written standalone physics-based models in MATLAB (LIONSIMBA, DEARLIBS), these models require significant work to integrate with Simscape Battery. Finite element modeling (FEM) compatibility is also limited; while Mathworks provides some FEM battery module examples, they are implemented as part of the Partial Differential Equations (PDE) Toolbox and it is not straightforward to integrate these models with the rest of the battery modeling ecosystem. Finally, running cycle life simulations is challenging due to a lack of support for simulating cycle-by-cycle experiments and the lack of customizable battery lifetime models.

Takeaway: MATLAB is still the go-to platform to develop battery models and control algorithms for battery management systems (BMS). However, in 2026 there are probably better tools for modeling battery degradation or multi-scale electrochemical devices.

Dyad Batteries: High performance computing meets battery simulation

Dyad Batteries (formerly JuliaSimBatteries) implements fast physics-based battery simulations. It can simulate 100 days of cycling in 60 seconds.

Julia is a programming language that was developed at MIT for high-speed scientific computing. It is a just-in-time compiled language that provides native support for parallel computation, vectorization, and hardware optimization. Similarly to Python, a large ecosystem of packaged libraries and tools have been developed by the open source community.

Dyad Batteries (formerly JuliaSimBatteries) is an implementation of the physics-based Doyle-Fuller-Newman (DFN) model developed in the Julia high-performance-computing ecosystem. Available with a license from JuliaHub, a software-as-a-service (SaaS) provider for Julia applications, it is fully compatible with Julia’s machine learning and optimization libraries. Simulations can be run via JuliaHub’s Cloud Platform, or directly as Julia code. The primary advantage of Dyad Batteries is its speed - it can solve a DFN problem on the timescale of milliseconds, which is orders of magnitude faster than other software offerings.

While this blazing speed might not matter for a battery researcher who is interested in running an occasional simulation, it is critical for anyone who needs to run a large number of simulations for degradation modeling, parameter estimation, uncertainty quantification, or machine learning. Dyad Batteries is compatible with SciML, a library for scientific machine learning that can be used for physics-informed model discovery. An intriguing use case of this pairing is to develop models that learn battery degradation modes directly from data rather than requiring user-defined degradation models.

This speed however does come with a cost. The learning curve to write efficient code in Julia is high, and development in Dyad Batteries requires significant programming expertise. Tasks like embedding neural networks inside the DFN equations are cutting-edge research, and as a result are not well-documented. The training process often involves a fair amount of trial and error.

Takeaway: Dyad Batteries is a tool for experienced programmers who are interested in applying optimization or machine learning on top of customized DFN models. It offers the fastest numerical implementation of DFN and the ability to embed neural networks within models to learn missing physics. SaaS is supported by JuliaHub.

Python Battery Mathematical Modeling (PyBaMM): Open-source success story

Interactive visualization of battery simulation solutions in PyBaMM.

PyBaMM is an open-source battery simulator written in Python and maintained by a research group at the University of Oxford. It has expanded over several years to include numerous lead-acid, lithium-ion, and equivalent circuit models. It has a significant user base spanning industry and academia and receives regular updates with fixes and new features. This stands in contrast to the prototypical academic research code, which is developed by students to solve a narrow set of problems, is generally poorly documented and difficult to use, and then falls into disrepair after they graduate.

PyBaMM primarily focuses on solving variants of the DFN battery model in 1D, making it a popular choice among academic research groups looking to develop new physics-based battery models. In addition to offering a collection of standard models and a library of battery materials, it provides many extensions to the DFN model for simulating degradation. New models from literature are regularly implemented and distributed through the PyBaMM GitHub repository much more rapidly than is possible with commercial software. PyBaMM also provides an interface for specifying and solving differential equations, making it a great choice for modeling new physics.

Another advantage of PyBaMM is its convenient programming interface for specifying complex battery cycling protocols, which many of the commercial battery simulators lack. Simulation results are organized into data structures that make it straightforward to access specific cycles, or even steps within a cycle. When this is combined with the mature data analysis tools in the Python ecosystem, PyBaMM becomes a powerful tool for parameter estimation and time-series analysis of experimental cycling data.

Since PyBaMM began as a 1D battery simulator formulated to simulate a small isothermal patch of electrode, it lacks the sophistication of the finite element simulators to simulate complex heat flow and fluid dynamics in 3D geometries, limiting its applicability to thermal or mechanical analysis of cells, packs, or modules, battery control systems, and flow batteries. The commercial FEM software packages have several decades head start solving these types of problems.

Takeaway: PyBaMM is an excellent open-source battery simulator that is ideally suited for testing the latest physics-based battery models, developing your own new models, or fitting models to experimental data. Basic Python programming is required.

AVL CRUISE™ M: Beyond automotive systems engineering

Cruise M battery module simulation with custom mechanical models.

AVL is an Austrian global automotive company that specializes in R&D, technology development, simulation, and testing. Their tools for simulating automotive systems were originally developed for the internal combustion engine, and support for electric vehicles and battery simulation was added after the software was already mature. This differs from other tools which began as battery simulators and then branched out into applications. Consequently, AVL’s offerings are less known outside the automotive industry.

AVL’s battery simulations capabilities come in two different flavors: a systems engineering approach in AVL CRUISE™ M, and a finite-element approach in AVL FIRE™ M. Unlike MATLAB and Simulink, CRUISE M can natively simulate discretized cylindrical and pouch batteries and packs in 3D, a computational compromise between the speed of the 1D DFN model and the accuracy of a full finite-element approach.

The systems modeling approach offers unique advantages for lithium-ion battery simulations. When combined with AVL’s automotive capabilities, it enables simulation of lithium-ion cells, modules, and packs fully integrated into electric vehicle powertrains. With the capability to generate vehicle drive profiles, CRUISE M can be used to test the performance of a battery pack in different types of vehicles under realistic use cases, or even to study pack degradation.

Models are defined via block diagrams with inputs and outputs for each, which is more intuitive than writing lines of code, and less cumbersome than COMSOL’s nested menu system. CRUISE M provides several battery-specific wizards for fitting model parameters from experimental data such as half-cells and EIS. It is straightforward to load an experimental time series and fit a battery model using several different built-in controllers and optimizers. CRUISE M also contains a built-in job queue for scheduling and running simulations, a feature that is not offered by any of the other software we have used.

While CRUISE M itself is polished software, there are a couple rough edges around the lithium-ion battery model. As with some other software offerings, it currently lacks the concept of a battery cycle counter (although it is planned for the future). Another challenge for battery model parameter optimization is that the software only allows properties to be either constants or lookup tables. While lookup tables are good for incorporating experimental data, many published battery models represent parameters as polynomial equations. These equations must first be converted to tabular form before they can be used in a simulation, making it difficult to run an optimization over the parameters in the polynomial.

Takeaway: AVL CRUISE™ M offers the unique capability to run standard and customized physics-based battery models within the context of automotive systems simulations, making it a great choice for simulating real-world performance of EV modules and packs. It additionally offers discretized geometries as a computational compromise between lumped models and full finite element.

COMSOL Multiphysics^®: A general-purpose finite element workhorse

COMSOL simulation of a simplified physics-based electrochemical model of a lithium-ion battery pack, combined with a 3D heat transfer analysis.

COMSOL Multiphysics^® is a commercial finite element simulator that is commonly used to solve a variety of coupled multiphysics problems. It has a 30-year history of use throughout industry and academic research. COMSOL first introduced the Battery Design Module in 2010, and in its present form is capable of simulating a wide variety of battery chemistries at the microstructural, cell (e.g., DFN or lumped models), and pack scales. New features and models from literature are added regularly. COMSOL is typically run from its graphical user interface which provides a comprehensive, but at times complicated, tree of menus for defining models, adding materials, coupling physics, adjusting solver settings, and plotting results.

COMSOL is ideal for running standard DFN battery simulations in complicated geometries or when coupled to additional physics. For example, imagine designing a new type of flow battery. With COMSOL, you could load the 3D geometry from a CAD file, specify the electrodes and fluids from a database of materials, and easily couple the complex electrolyte fluid dynamics. Another good use case is simulating the heating of pouch or cylindrical cells during cycling, or examining their mechanical rigidity. Individual cells can even be assembled into packs to simulate the heat exchange between cells for the purpose of designing cooling systems.

Overall, COMSOL is a great choice for simulating spatial effects and coupled physics. However, if your needs can be addressed by solving a DFN problem in 1D, then COMSOL may be overkill. While COMSOL can solve 1D problems, it is not suitable for developing new particle physics or degradation models; PyBAMM is a better choice for this. It is also currently a tedious process to implement battery cycling protocols in COMSOL (although the newest release, version 6.4, addresses this). Although COMSOL has an optimizer for parameter estimation, it does not have the ability to load or fit experimental time-series data which presents a challenge when trying to fit battery cycling data. This typically needs to be done outside of COMSOL.

Takeaway: COMSOL is great for solving 3D battery models, incorporating fluid dynamics into battery models, or investigating thermal or mechanical effects in realistic geometries. If these tasks can be accomplished with standard battery models and do not require fitting time-series data, then COMSOL is a great choice.

Software-as-a-Service Platforms: IonWorks, About:Energy, Dyad Batteries

Screenshots of the web interfaces for Dyad Batteries, About:Energy (VOLTT), and IonWorks.

Software development has historically followed a sequence of development, testing, and release. In this traditional approach, users download and install a mature copy of the software to run on their own computing hardware. A different method of distributing software, referred to as software-as-a-service (SaaS), moves the software to the cloud where the developer manages both the software and hardware. Customers often interact with the software through a web-based interface and pay a subscription fee for services. In the battery industry, SaaS is becoming a more common software distribution approach, with companies including Ionworks, About:Energy, and Dyad Batteries (formerly JuliaSimBatteries) offering physics-based modeling services.

There are several advantages to running battery simulations in the cloud. The web-based user interface is often simpler and more intuitive because the developer can expose just the necessary settings to the user. Programming skills or detailed knowledge of complicated software is not required. SaaS also allows the developer to quickly release new or customized features to users because the developer also maintains the computing infrastructure. They can also develop a proprietary feature for one customer without that feature being distributed to all customers. Finally, both Ionworks and About:Energy provide interfaces for uploading battery data and cycling protocols, and offer services for fitting data and parameterizing models in the cloud. As we have discussed, standalone battery simulation software tends to provide limited ability to process experimental data, often requiring user coding.

SaaS battery modeling also has a few downsides. Because simulations are run in the cloud, users do not control the computing infrastructure which places limits on customizability and deployment. For certain projects, uploading confidential data to a cloud server might present security concerns. Some platforms now offer single-tenant or on-premise deployment options to address these needs.

SaaS platforms vary widely in their transparency: some provide full visibility into equations and parameters, while others operate as black boxes. Users with a software background might also prefer to access detailed simulation settings, or even modify the simulation source code itself. Interfacing with a simulation through a webpage might be seen as obfuscating important details, and unattractive for computational researchers looking to develop new models.

The use of Python as a programming language presents a challenge for developers looking to monetize tools built on PyBaMM, including Ionworks and About:Energy. Python is an interpreted language, meaning that code cannot be compiled into a binary executable that can be sold to customers. Instead, the source code itself must either be given away or hidden behind a web interface. In this regard, it will be interesting to see if SaaS is a product that R&D scientists are asking for, or if it is instead a route to monetize Python code.

Takeaway: Software as a service provides users with the gentlest interface for running and analyzing battery simulations, while at the same time offering rapid deployment of customized physics models and parameterization of data in the cloud. However, users do not have access to source code which can obscure details, and some may have security concerns with cloud computing.

Concluding remarks

We leave readers with two takeaways. First, battery simulation software is foundational to the future success of the battery industry. Simulation is a tool that greatly benefits fundamental research, product design, and systems engineering efforts with the use of physics-based models for digital experimentation. Second, those wishing to pick up battery simulation software in 2026 will be faced with the tyranny of choice, with plenty of options to choose from.

As we have reviewed, many different kinds of simulation tools are used throughout the battery industry for a variety of purposes such as materials, cell, and pack design. Some of these tools were developed specifically for batteries, while others tools were established first in other fields. Some come from academia, others from industry. They range from those that require software programming expertise to those that are run from websites. They are all designed for different use cases.

Now that you know who’s who in battery modeling software, choose the right tool for your next battery simulation job with confidence.

---

🌞 Thanks for reading!

📧 For tips, feedback, or inquiries - reach out

📣 For newsletter sponsorships - click here

🌐 Follow us on Twitter, LinkedIn, and our website