Optimisable system-level thermal models for power electronic converters


EPSRC Centre for Power Electronics


£55,248 (total) - £50,167.38 (for Coventry)


Oxford University


Rolls-Royce & IMV Europe


Mauro S. Innocente & Daniel J. Rogers (University of Oxford)


This project is focused on the design of reliable yet efficient thermal models underpinning an optimal design framework for power electronic converters. Due to the high number of times these models must be evaluated during the optimisation process, they are required to be of low computational cost (so-called ‘optimisable’). These models are intended to be efficient representations of the complex thermal management systems found in real power converters. Accuracy is to be progressively increased by relaxing simplifying assumptions whilst ensuring a reasonable computational cost is maintained. Computational efficiency is required so that it is feasible to embed these models in the optimal design framework of the converter, whilst accuracy is paramount because thermal constraints are typically active and therefore drive the design. In order to assess their accuracy, the optimisable models will be validatedagainst computationally intensive simulations performed with commercial software.


Power electronics is a key enabling technology that provides flexibility and controllability in a broad range of electrical energy systems. Increases in the power density of power electronics systems lead to increases in their range of applications. The use of electrical power distribution and actuation in place of hydraulic or pneumatic alternatives has increased in recent years due to their simpler installation, higher accuracy and efficiency, easier control, lower energy costs and maintenance, smaller system mass and complexity, lower noise, and cleaner environment. In this regard, the design of power-dense converters is key. Underpinning their optimal design is the thermal modelling of its components. In fact, our current work shows that the thermal constraints drive the design. This project is aimed at finding out whether a compromise between computational efficiency and accuracy can be found which make the thermal models reliable and the optimisation process practicable. These ‘optimisable’ thermal models can be used as stand-alone pieces of software to investigate the thermal effects of a given design and set of specifications, or they can be embedded into the optimal design framework of the converter as constraints driving the design to investigate the level of power density that can be achieved for those specifications. The latter will also provide insight into the power densities that can be expected over the coming years.

Coventry University ranked 15 in the UK
Coventry University awarded TEF GOLD Teaching Excellence Framework
University of the year for student experience
QS Five Star Rating 2019
Coventry City of Culture 2021