2024-07-10 –, Function (4.1)
Lithium-ion battery models are highly nonlinear systems of stiff, partial differential-algebraic equations (PDAEs). It is challenging to robustly solve the system of equations over a range of inputs and parameters – many commercial and open-source battery modeling tools are prone to sporadic failure stemming from issues in setting up, initializing, and solving these systems. This talk discusses these pitfalls and the solutions implemented in the JuliaSimBatteries.jl package.
JuliaSimBatteries is an advanced lithium-ion battery simulation tool integrating sophisticated electrochemical, thermal, and degradation physics. Utilizing the Doyle Fuller Newman (DFN) model, it can predict a battery's entire lifetime with fast charging 150,000 times faster than real time. The number of connected batteries is scalable from one cell to packs of thousands using electrochemical models. Scientific Machine Learning (SciML) enables the discovery of hidden governing laws from data, such as degradation and low-temperature behavior. Characterize material properties and propose battery designs using the parameter estimation and optimization tools in JuliaSim.
Building accurate models are essential for understanding, optimizing, and designing batteries. Physically accurate battery models are computationally expensive and difficult to solve robustly. JuliaSimBatteries is more than 100 times faster than other battery modeling tools while solving the same physics, thanks to the speed of the Julia programming language. Bring your battery workflow to the next level to solve challenging problems:
- Pack modeling – JuliaSimBatteries is performant and enables the predictive power of electrochemical models for large-scale battery packs.
- Uncertainty quantification – Uncertainty is inherent in battery modeling. JuliaSimBatteries helps mitigate and understand the root causes of uncertainty with JuliaSim Model Optimizer.
- Fast charging – Built for robust and efficient simulations, even at the extreme operating conditions of fast-charge.
- Degradation – Predict battery lifetime and health with SEI capacity fade models.
- Discover hidden physics – Combine physics from our battery models and your data to discover hidden governing laws using SciML tools.
- Lifetime prediction – Estimate a battery's entire lifetime with fast charging in under a minute with the DFN model.
I am a software engineer at JuliaHub with a background in theoretical and computational physics.
https://github.com/SebastianM-C