Achieving Quantitatively Accurate Li-S Models Through Rigorous Validation

Thematic block:


Other authors:



This poster represents a paper we recently submitted. The purpose of work is to introduce both a novel zero-dimensional model for Lithium Sulfur cells as well as an empirical model validation technique.

Part of the motivation for looking at model validation procedures is that we have come across several shortcomings in the literature. For example, the highly influential Kumaresan model only charged when some of the parameters were changed significantly. Unfortunately, only about 35% of the models in the literature provided any evidence that the model could both charge and discharge without significant modifications, despite this well-known and long-established problem with Li-S models. Moreover, because we are interested in model predictive power, it is essential to know if a paper provides sufficient evidence that the model can predict out-of-sample data. Unfortunately, a clear distinction between in-sample and out-of-sample data, and therefore clear out-of-sample predictions against experiment data, were made in only 18% of published work.

Our proposed novel zero-dimensional model has been tested with clearly defined in-sample and out-of-sample data. As such, we find strong quantitative evidence that our model can predict out-of-sample electrolyte resistance data. Moreover, these predictions hold over a range of temperatures. As part of the validation procedure, we have also shown model predictions for typically observed Li-S cell operational loads. For example, different discharging currents lead to different levels of over-potentials and changes in cell capacity. Our model captures the former behavior but not the latter. Addressing this model deficiency will be part of future work.

We believe that providing clear evidence of the model performance, both good and bad, will allow researchers to have an accurate understanding of the model. We hope to push the modelling community towards this more open approach to ensure model development efforts can be directed appropriately and with the correct level of confidence in any model.

Would you like to contact this author?
We are happy to forward your request / feedback.

[su_button url="" target="blank" background="#98c219" color="#ffffff" size="4" radius="0" icon="icon: envelope-open" icon_color="#fff"]EMAIL TO THE AUTHOR[/su_button]

ist eröffnet

Wir freuen uns auf Ihre Einreichung
bis zum 31. Oktober 2022