Simulation for Better Batteries

Keyvisual image main
Visualization of a simulation depicting a cross-section of electrolyte filling of electrode structures. Image: Martin Lautenschläger

Computational methods originally developed for hydrology research and oil and gas extraction find new applications in understanding electrolyte flow physics in battery cells.

Since the Industrial Revolution, scientists and engineers have spent significant effort trying to understand how fluids behave in complex, porous spaces. To identify risks for groundwater pollution or to develop more efficient extraction methods for oil and natural gas, for example, researchers have had to determine how fluids behave beneath the Earth’s surface. Because of the difficulty in directly observing subterranean interactions, high-performance computing (HPC) simulations play an important role in understanding these complex environments, and the computational methods have continued to improve through the decades. 

Recently, researchers at the German Aerospace Center (DLR) in Ulm have begun to apply such approaches in an entirely new context: Specifically, they are using computational methods well-known from hydrology research to better understand how highly porous battery materials and cells can be filled by an electrolyte — a solution containing ions that carry an electric charge. By refining this method, researchers are gaining a better understanding of real-world conditions inside battery cells that will enable them to refine current manufacturing methods and develop new battery designs.

To that end, the team has been using HPC resources at the High-Performance Computing Center Stuttgart (HLRS). With help from the center’s flagship supercomputer, Hawk, it runs computationally demanding lattice Boltzmann simulations to study the complex fluid-solid interactions taking place inside a battery cell.

“Battery materials are quite porous, and a lot of the physics interactions happen at the surfaces of the pores at the interface between electrolyte and surrounding solid material,” said Dr.-Ing. Martin Lautenschläger, a scientist at the DLR Institute of Engineering Thermodynamics and principal investigator on the project. “Because of the complex physics, geometries, and the variety of scales we have to account for here, there aren’t many methods that can tackle this problem. However, the lattice Boltzmann method can and we’re among the first to apply this method in this context for real battery materials.”

Electrolyte filling of different electrode structures. The pressure-saturation behavior (left) is shown together with snapshots of cross sections (right) in which the electrode is depicted dark blue, gas is depicted blue, the electrolyte is depicted red, and the binder is depicted yellow. Image: Martin Lautenschläger

Using Hawk, the team has been able to modify the traditional lattice Boltzmann method, improving computational efficiency in the process. The team has accurately and efficiently simulated multi-phase fluid flow simultaneously in pores that range from nanometers to micrometers, which is relevant for batteries. It published its results in Advances in Water Resources and currently has its sights set on further developing the method to model also complex chemical and electrochemical reactions taking place inside of a battery cell.

Selective simplification

Researchers face two significant challenges when simulating fluids in motion. The first is scale, as investigators must simulate both a volume that is large enough to represent a real-world system and small-scale interactions that can influence how the fluid as a whole behaves. The second is complexity. When simulating oil extraction or the filling of a battery with electrolyte, researchers do not just simulate a liquid, but rather the interaction between the liquid and the air or other gas that occupies the pores before filling. This multiphase simulation is further complicated by the need to account for other structurally complex battery components, which often consist of a mixture of materials.

Because of the physics and geometries involved, the lattice Boltzmann method shows advantages over standard computational fluid dynamics approaches. It is based on the so-called Boltzmann equation, and treats fluids as a large collection of particles on a computational grid, or lattice. In contrast to conventional methods that are based on the equations of fluid mechanics, the lattice Boltzmann method solves numerically much simpler equations, offering a more computationally practical approach for addressing these challenges in a way that is especially favourable for HPC.

In a battery cell, the negative and positive poles called electrodes — the anode and cathode — are separated by an electrically insulating porous interlayer that prevents their direct contact. However, this separator and other materials used in batteries contain tiny, nanoscale pores that are ideally filled with electrolyte and, in turn, partially enable ion transport. When running lattice Boltzmann simulations, researchers must pay special attention to how the electrolyte behaves when passing through different pores that vary in shape and size.

Microstructures (from left to the right) of a pure electrode (black), an electrode with binder, and an electrode attached to a separator. Mescoscale pores are depicted white and materials with nanoscale pores are depicted grey. Image: Martin Lautenschläger

The team adapted a so-called “homogenization” approach from conventional fluid dynamics and transferred it to the lattice Boltzmann method for applications of multi-phase flow. Here, in a computationally efficient manner, they realistically simulate how electrolytes behave when passing through both the larger mesoscale pores and the tiny nanoscale pores, but simplify their calculations. To reduce the computational efforts, only the electrolyte flow through the larger pores is fully resolved and strictly calculated. The tiny pores are not resolved. Instead, assumptions based on observation are made about how they affect the electrolyte flow. While the resulting model does not fully account for all of the smallest-scale interactions in the system, the team sees excellent agreement with analytical solutions. The comparison with results obtained from their experimental collaborators, also at DLR, is currently ongoing.

“We rely on experiments for informing our models, because many material properties cannot be guessed,” Lautenschläger said. “Then, we can set up a system that can be reproduced by computers and use experiments to validate it. Once a model is validated, we can use the simulations to predict things that can’t be resolved in experiments. In experiment, you can measure a result, but you don’t always get a reason for that result; using simulations this can give you a better understanding. Therefore, experiment and simulation should always go hand-in-hand.”

Charging forward

The team’s initial successful simulations of electrolyte filling in a battery (published in Batteries & Supercaps) fold neatly into the DEFACTO Project, a European Union-funded initiative focused on optimizing material development and manufacturing processes for lithium-ion battery cells. The work also earned Lautenschläger praise from HLRS leadership: He was named one of three winners of the 2022 Golden Spike Awards during HLRS’s annual Results and Review Workshop.

With this proof-of-concept work complete, the team is now focused on adding additional complexity to its simulations. “The motivation goes in two directions: make common battery technologies better by optimizing manufacturing processes and start improving the early stages of development of next-generation battery technologies,” Lautenschläger said. The researchers have begun exploring how to improve current-generation batteries using more realistic simulations for manufacturing and of electrochemical processes. They are also investigating promising next-generation battery technologies such as lithium-sulfur batteries. Their goal is to optimize their design and to prevent degradation processes and unwanted side reactions to improve their long-term performance.

Regardless of whether the team focuses on further optimizing today’s battery technologies or continues to design battery cells of the future, Lautenschläger made one thing clear: “What is sure is that our need for computational resources will only increase in the years to come.”

-Eric Gedenk

Related publication

Lautenschläger MP, Weinmiller J, Kellers B, et al. 2022. Homogenized lattice Boltzmann model for simulating multi-phase flows in heterogeneous porous media. Adv Water Resour. 170: 104320.

Funding for Hawk was provided by Baden-Württemberg Ministry for Science, Research, and the Arts and the German Federal Ministry of Education and Research through the Gauss Centre for Supercomputing (GCS).