Physics-based models (4): looking forward
In the past few blogposts we have seen what physics-based models for batteries are about. First we discussed physics-based models in the general sense, next we focused on the components needed to assemble a physics-based model for batteries, and finally we reviewed the most widely used battery models. However, not everything is done and dusted in this area. The aim of this post is to discuss the current main challenges (or what we believe are the main challenges) for physics-based models.
The first challenge is parameterisation: the Doyle-Fuller-Newman model requires over 30 parameters, some of which are functions of the concentrations or temperature. The right parameter values are needed to simulate a given battery, but determining them is extremely time and resource consuming. Moreover, there are some issues like the identifiability of parameters that make the whole process extremely challenging.
Figure 1: Degradation is one of the main challenges in battery modelling. Accurate degradation models allow the prediction of the remaining useful life, which can lead to better battery design and management.
The second challenge is degradation. From the modelling point of view, degradation models exacerbate all the issues already present in electrochemical models. For example, with degradation we need to simulate longer times and they are a lot more tricky from the numerical point of view. This requires better numerical methods in terms of robustness, efficiency and memory management. This challenge builds on top of parameterisation, and how to parameterise degradation models is still an open research question.
Finally, the third challenge is knowledge transfer across teams and sectors. Being able to deploy models can really speed up the research and development of new batteries and battery systems, but is often not straightforward. The first example that springs to mind is the knowledge transfer from academia to industry, to ensure that the most recent modelling advances are used in real applications, but it also occurs between teams in the same organisation. Consider, for example, the knowledge transfer between a company's modelling and control teams.
PyBaMM already offers solutions to address these challenges. For example, PyBaMM includes state-of-the-art degradation models, and its open-source and Python-based environment makes it much easier to share models between teams. However, industry requires more specific and streamlined solutions and here is where Ionworks comes in. We will announce soon the first Ionworks products, so stay tuned!
This is our take on the main challenges in battery modelling; what do you think? Let us know in the comments!
