Research

The rapid growth of AI workloads, cloud services, and edge computing has driven data center power demand sharply upward. To meet efficiency, reliability, and modularity targets, data centers are increasingly adopting medium-frequency solid-state transformers (SST) and MVDC architectures to replace legacy 50/60 Hz infrastructure. These SSTs employ medium-frequency operation to reduce transformer size and loss, but steep switching edges from wide-bandgap converters introduce wideband electrical stress on transformer windings and insulation systems. In addition to compact isolation, SSTs enable flexible delivery of multiple voltage levels (e.g., MVDC, 400 V DC, and 48 V DC) required by modern data-center architectures, supporting heterogeneous server loads and future rack-level DC distribution. Traditional low-frequency models fail to capture critical high-frequency parasitic, leading to underestimated insulation stress, common-mode noise, and partial-discharge risk, which directly affects uptime and reliability.

This project develops wideband (10 kHz to 50 MHz) transformer models integrating electromagnetic field simulation, parasitic extraction, and behavioral circuit representation. The models will be validated experimentally under representative converter waveforms. Outcomes include physics-based modelling tools and design guidelines for medium-frequency high-voltage transformers optimized for data center environments, improving efficiency, reducing EMI, and extending service life. 

This project focused on developing a high-frequency model of a 60 kW PMSM used in the Toyota Prius electric vehicle, validated up to 50 MHz to capture machine behavior under fast-switching SiC converters. The work introduced a method to represent frequency-dependent parameters directly in time-domain simulations to enable accurate representation of high-frequency effects in electric drives. This research received the 2nd Prize Paper Award (2021) from the IEEE Industrial Drives Committee and was published in two IEEE Transactions journals. This work was funded by the Engineering and Physical Sciences Research Council (EPSRC) and carried out at The University of Sheffield, UK.

In addition, high-frequency models were developed and deployed for axial-flux machines for Rolls-Royce, as well as for the CTD platform, extending the framework to multiple machine topologies and industrial applications.

This project focused on developing a passive filtering solution to suppress peak voltage stress within machine windings, significantly improving electric drive reliability. The proposed design includes retrofit solutions for machines already operating in the field. Experimental results demonstrated effective mitigation of voltage peaks with minimal additional losses, that conventional sine-wave and dv/dt filters cannot attenuate. These findings were published in one IEEE Transactions journal. This work was funded by the Engineering and Physical Sciences Research Council (EPSRC) and carried out at The University of Sheffield, UK.

This project aims to repurpose abandoned oil wells in the United States as gravity energy storage systems, based on a concept developed by Renewell Energy. The approach delivers two key benefits. First, it helps mitigate methane emissions from inactive wells, supporting sustainability goals. Second, it provides a new revenue stream for well owners by transforming legacy infrastructure into clean energy assets.

This project was focused on developing an efficient electric drivetrain for energy conversion with minimal losses. The study achieved an efficiency improvement from approximately 75% to up to 85% by eliminating the gearbox and introducing a direct-drive architecture, addressing mechanical losses that exceed 50%. This work was funded by ARPA-E and carried out at the National Renewable Energy Laboratory.