Multiscale modelling of components and devices

High-performance porous electrodes are essential for advancing electrochemical energy devices, such as fuel cells and batteries. The current focus on physical attributes in electrode design falls short of achieving optimal electrochemical performance. To address this limitation, this research introduces "π learning," a novel deep learning framework that prioritises and predicts electrode performance by combining advanced neural networks and physics modelling.

The development of high-performance porous electrodes is critical for the advancement of electrochemical energy devices. While current electrode design strategies often prioritise physical characteristics, the ultimate goal is to optimise electrochemical performance.

Achieving this objective has been challenging, as existing data-driven approaches struggle to accurately capture the complex, high-dimensional features within multiphase microstructures that profoundly impact performance.

The "∏ learning" framework

In response to this challenge, this research introduces "∏ learning," a pioneering deep learning framework. "∏ learning" combines the power of physics-informed generative adversarial neural networks (GANs) with convolutional neural networks (CNNs). This integration, along with advanced multi-physics and multi-scale modelling, enables the creation of microstructures that are both performance-informed and predictive of overall electrode performance.

Applications and implications

The potential of "∏ learning" is showcased through its application in electrode design for solid oxide fuel cells. By embracing this framework, it becomes possible to unlock a performance-driven approach to electrode design, offering significant promise for the development of high-performance porous electrodes in advanced electrochemical energy devices.

This research marks a significant step toward enhancing the efficiency and effectiveness of energy conversion processes in fuel cells, batteries, and similar technologies.


Niu, Z, Zhao, W, Wu, B, Wang, H, Lin, W-F, Pinfield, VJ, Xuan, J (2023) π Learning: A Performance‐Informed Framework for Microstructural Electrode DesignAdvanced Energy Materials, 13, 2300244, ISSN: 1614-6832. DOI: 10.1002/aenm.202300244.

Chen, L-N, Wang, S-H, Zhang, P-Y, Chen, Z-X, Lin, X, Yang, H-J, Sheng, T, Lin, W-F, Tian, N, Sun, S-G, Zhou, Z-Y (2021) Ru nanoparticles supported on partially reduced TiO2 as highly efficient catalyst for hydrogen evolutionNano Energy, 88, pp.106211-106211, ISSN: 2211-2855. DOI: 10.1016/j.nanoen.2021.106211.