For years researchers at the Department of Energy’s (DOE’s) Pacific Northwest National Laboratory (PNNL) have been developing tools to accelerate the materials discovery and development of new energy storage technologies, including those that can predict the performance of the batteries systems for long-term grid services. 

With a new physics-based simulation tool, EZBattery Model, it now takes less than a second to predict the performance of redox flow batteries and its variants.  

The ability to accurately and quickly predict how redox flow batteries will behave based on laboratory-scale experiments has far ranging implications in the field of energy storage. Experimenting on new batteries has traditionally been a time-consuming process, requiring long-term testing over many charge-discharge cycles to better understand a battery’s lifespan and a great deal of trial and error to find optimal combination of materials.

With this challenge in mind, PNNL researchers sought out to build a tool to help researchers more easily optimize battery designs and electrolyte chemistries without the need for extensive physical prototyping or weeks of experimentation.

“This new model is like a superpower for energy storage researchers aiming to accelerate the development of energy storage technologies,” said Jie Bao, research engineer at PNNL and lead on the project. “To get the best performing long duration energy storage system that will power our homes and the grid, we need to be able iterate quickly and efficiently.”

EZBattery Model is publicly available on GitHub and is accompanied by detailed Jupyter Notebook tutorials for users. The model itself is packaged in an easy-to-use software featuring standard input and output formats, automated model calibration and validation, and flexible control on the start and stop operating time and state of charge.

The model handles the three main architectures of flow batteries, including fully liquid inorganic or organic electrolyte flow systems, hybrid systems (which entail some solid deposition and desorption during the battery's operation), and redox-targeting/redox-shuttling flow batteries that utilizes the reaction between solid particles and redox-mediators in the tank to enhance the energy storage density.

What makes the EZBattery Model particularly powerful is its integrated analytical solutions to chemical species convection, diffusion, and reaction equation in micro-channels, simulating both individual battery cells and large-scale systems for long duration energy storage, and handling both constant current and varying power operations.

This flexibility helps in tailoring simulations to specific real-world battery designs.

Traditionally, the multi-physics numerical models require a few hours or even days to complete a simulation for large cells or stacks. The EZBattery Model can do it in less than one second, while maintaining results that are comparable to those from computationally expensive models, such as COMSOL.

Because EZBattery Model relies on physics and not a machine learning model, it doesn’t require any training datasets to work. Additionally, it can actually generate its own data to train other machine learning models, which Bao says can help address the challenge of data scarcity, providing abundant and diverse data samples that cover a wide range of operating conditions and scenarios. Although machine learning models can run simulations much faster after they are trained, they cannot provide physics-based explanations of the results.

“The physics-based approach lends itself to easily integrate and capture intricate details of battery behavior, like solid deposition during the charging process,” said Bao. “For instance, when zinc ions become zinc solid metal, the model can accurately capture this transformation and estimate corresponding battery performance—something a machine learning model cannot do due to lack of associated training data.”

Work on EZBattery Model was born out of PNNL’s Energy Storage Materials Initiative and has since expanded with support from the Department of Energy’s Rapid Operational Validation Initiative.

The EZ Battery Model team includes Jie Bao, Yunxiang Chen, Zhijie Xu, Yucheng Fu, Peiyuan Gao, Chao Zeng, Amanda Howard, Ayoub El Bendali, Tiffany Louie, Grace Yuan, Alvin Liu, Qixuan Jiang, Panos Stinis, Wei Wang, and Vincent Sprenkle.