Background
Imperial College London leads a consortium of 9 universities and 13 industry partners funded by the Faraday Institution to equip industry and academia with new software tools aiding the research of batteries.
One of the tools that would greatly help researchers in this area would be the ability to easily visualise and compare the wealth of data – experimental and generated in simulations – that is being produced on a daily basis in the UK. Having access to a comprehensive data archive and a means of visualising and comparing not only the raw data, but also metadata related the exact experimental setup, configuration of the cell or battery pack, etc. will provide new insights and guide the next steps towards a breakthrough in the field.
Our Contribution
Two separate tools had been developed within the consortium to try to cover this use case, with different functionality and in different states of maturity: Liionsden and Galvanaliser.
The work done by the RSE Team built on the first one. The first step was to containerise the application to ease further development, deployment and reproducibility. Quality assurance and continuous integration tooling was put in place. The structure of the database models were rationalised and the code thoroughly documented. Tests were also added to check the different functionality. A lot of work were done on implementing robust parsers to ingest the experimental data produced by different cyclers reliably into the application. Finally, an intuitive front end was added based on Bootstrap5 and Plotly for the plots.
Outcomes
The result was a robust web application based on Django, containerised and deployed in Azure to simplify its scalability as the needs of handling more volumes of data and more users evolve. The tool can ingest experimental data in a variety of formats and, specially, handle associated metadata related to the experiments and devices under tests (eg. cell configuration, experiment type or thermal management). Plots of the main variable dependencies (eg. V-I, or V-time) are generated automatically to ease its visualisation. Other features include controlling the visibility of the data (public, only for registered users, only for certain institutions) and downloading the raw data for further processing.
The tool was released as open sourced with a BSD-3 license, and is available for contributions and further development in GitHub.
Testimonials
Prof. Greg Offer, ESE group leader:
“Working with the RSE team on this project has been part of a growing relationship, understanding the importance and benefits of working with professional software developers. This project was particularly interesting as it involved experimentalists having to communicate with the RSE team to properly define what they needed, which forced them to think really hard about what mattered and what didn’t. We expect the final product to increase our productivity and make it easier for our experimentalists to focus on what they enjoy the most, being in the lab.”