BattMo

BattMo is a framework for continuum modelling of electrochemical devices. We target both flexibility and efficiency. The code is open source, written in Matlab and Julia, available on GitHub.

Lithium-ion PXD simulations

• The initial development features a pseudo X-dimensional (PXD) framework for the Doyle-Fuller-Newman model of lithium-ion battery cells.
• Fully-coupled thermal simulation.
• Multi-dimensional models (1D-3D).
• Support for large 3 dimensional models
  • The difficulty of solving large systems is overcome by the use of taylored preconditioners. We use open-source AMGCL.
  • We can handle general geometries. A set of parameterized geometry is readily available, such as coin-cell, jelly roll, see here for an overview.
  • BattMo is built upon MRST

High-quality open source code

• Extensive documentation on GitHub
  • The documentation includes examples and tutorials
  • BattMo documentation
• Continuous integration for robust testing and documentation updates.

Design optimization and material characterisation using adjoint-based computation

• Support for adjoint-based optimization, which brings order of magnitude gains in the computation of the gradients or sensitivities, when the number of design parameters is large.
• Reuse existing adjoint-based optimisation module from MRST.
• Example in Inteligent project: Compute the electrode sizes and porosity that optimizes the specific energy of a cell (energy with respect to mass), given the other material properties.
• Fully differentiable grid: All the grid properties (centroids, volumes) and the resulting discrete differentiation operators (half-transmissibilitie) have AD support, meaning that they can be differentiated, with respect to node coordinates.

Flexible framework for prototyping and integration of new physics

• Built upon open-source MRST simulator.
• Native automatic differentiation support.
• Flexible 3D grid structure.
• Robust Newton solver for evolutionary partial differential equations.
• Graph-based model design
  • Model is setup as a composition of sub-models
  • A sub-model corresponds to a computational graph where the nodes are the variables and the edges the functions
• Fully differentiable models
  • Includes also fully differentiable grids
• Examples of included physical processes/models
  • Degradation mechanisms
  • Composite materials : two particle models
  • Sea-water batteries
  • Electrolysis

Julia version: BattMo.jl

• Open source code available as a julia package via GitHub.
• Order of magnitude gain in computation speed for small models.
• Uses Jutul, a shared platform with Julia resorvoir simulation tool (JutulDarcy)
• Support for fully differentiable models.

Electro-chemical systems beyond Lithium-Ion batteries

• H₂/NH₃ production electrolysis
• Physical system
  • Anion Exchange Membrane
  • Protonic membrane (high temperature, work in progress...)
  • Protonic membrane (low temperature, work in progres...)
• Physical processes
  • Electrochemical processes as in battery : Mass, energy and charge conservation with solid and electrolyte diffusion, migration.
  • SEI growth (deposition), Lithium plating, cracking, mechanics, thermal degradation, hysteresis
  • Multiphase flow in electrodes (Electrolyser and fuel cells are open systems)
• Sea water battery

Used in several EU projects

• BatMax
• BigMap
• Hydra
• Inteligent
• DigiBatt

Working directions

• Hybrid models (Physic-constrained AI models)
• More battery chemistry
• More degradation processes such Lithium plating
• Composite material
• Hysteresis
• Cracking
• Mechanics
• Thermal induced degradation