Catapults
Catapult Network Cell and Gene Therapy Connected Places Digital Energy Systems High Value Manufacturing Medicines Discovery Offshore Renewable Energy Satellite Applications
Search

“Faster simulations, smarter design” – a research breakthrough in thermal modelling

Date: 22.10.2025

Topics: AI, Design, Power Electronics

Rows of solar panels on a grassy field use AI to maximize energy output under a clear blue sky. - CSA Catapult

Understanding how heat moves through power modules is critical to designing reliable, high-performance systems for EVs, renewable energy, and the smart grid. But detailed thermal simulations can take hours – or even days – to run.

At CSA Catapult, a new research breakthrough is changing that.

Thermal Design Engineer Xuejiao Huang recently presented the team’s findings on Enhanced Thermal Impedance Analysis at ECCE Europe, one of the leading international conferences for power electronics and energy conversion.

The paper – “Enhanced Thermal Impedance Analysis with Reduced-Order Modelling: A Fast and Efficient Computational Approach” – introduces a new Linear Time-Invariant Reduced-Order Model (LTI ROM) method that delivers highly accurate, computationally efficient thermal analysis of power modules.

This is a smarter, faster way for us to understand how heat behaves in complex electronic systems – cutting simulation times dramatically while maintaining exceptional accuracy. The approach also provides quick and reliable insights into how heat transfers between components, enabling AI-driven design and real-time thermal management for high-power applications in transport, renewables, and smart grid technologies.

We spoke with Xuejiao to get her take on what this means for the future of power electronics, how it supports UK innovation, and how businesses can benefit from this faster, more accessible approach to design.

Smiling woman with dark hair in a light sweater, photographed against a plain white background, evokes an AI portrait. - CSA Catapult

What challenge were you aiming to solve with this research?
When we design power modules, thermal performance is always one of the biggest challenges. You need to understand how heat spreads through complex materials and multiple chips inside the module. Traditionally, engineers use Finite Element Analysis (FEA) to do this, which is very accurate – but it’s also very time-consuming and computationally expensive.

That slows down the whole design process. It means fewer design variations can be tested, and it can take longer to reach the best solution. We wanted to find a way to keep the same level of accuracy but make simulations much faster and easier to integrate into AI-driven design tools.

What’s new about the approach?
We developed a Reduced-Order Model (ROM) that simplifies the thermal system without losing important detail. Essentially, it captures the dominant thermal behaviour from a full FEA simulation and turns it into a lighter, more efficient model.

The result is that thermal simulations which might take hours can now run in seconds — and still deliver accuracy within 1% of a full-scale FEA. It’s a big improvement in speed, and it allows engineers to explore more options in the same amount of time.

It also means that thermal coupling effects — how heat from one chip affects others nearby — can be captured quickly and accurately. That’s very important for modern multi-chip modules.

 

How does this research translate into benefits for businesses and partners?
The biggest advantage is time to market. Faster simulation means faster design validation, which shortens the development cycle.

It also reduces costs because you don’t need as many physical prototypes for thermal testing. And because the model is so reliable, it can even be integrated into real-time monitoring or adaptive cooling systems — allowing products to operate more efficiently in demanding environments.

For SMEs in particular, this opens up access to advanced design capability that would normally require large-scale resources. It helps level the playing field.

Close-up of an electric car charging on a city street, with AI-powered systems, trees, and a bike in the background. - CSA Catapult

What kinds of applications could this approach help accelerate?
There’s so many, but there’s some key areas that it will really impact. In the electric transport space, for example, it’ll speed up the design of silicon carbide-based inverters and power modules – the critical components for EVs and their charging systems. So this is a really big deal for the automotive sector.

For renewable energy systems too, like solar and wind converters, it will improve reliability under changing loads and variable temperatures.

Another big one is also the smart grid, where multiple power modules interact. This will change the game by allowing us to simulate and manage thermal performance across systems, making power distribution more efficient and sustainable. So it has the potential to provide a genuine contribution towards the UK’s net zero goals.

How does this work fit into CSA Catapult’s wider mission?
Our aim is to take cutting-edge research and make it practical and accessible for UK industry, supporting businesses and the UK with growth and to becoming leaders in innovation in the semiconductor space. This type of research that myself and others are working on are all examples of how we’re transforming a complex academic method into a real engineering tool that can be used by companies developing power electronics.

We’re not just improving simulations; we’re helping businesses innovate faster, reduce risk, and design for long-term sustainability. That’s how we strengthen the UK’s position in power electronics and compound semiconductor technology.


What’s next for this line of research?
We’re now in the process of exploring how to integrate the reduced-order model into AI-driven design pipelines and real-time embedded systems. The goal is to make predictive thermal management an active part of how devices operate — not just a design-stage tool.

It’s an exciting step, because it connects digital design with real-world performance. And that’s where the next generation of smart, efficient power systems will come from.

Final thoughts: Paving the way for the future

ABOUT XUEJIAO HUANG

Xuejiao received her Ph.D. in Computational Mechanics from Queen Mary University of London. She worked as a Research Associate at Lancaster University and later joined Dynex Semiconductor as a module design engineer, focusing on multiphysics design and simulation of power modules.

She now works as a thermal design engineer at CSA Catapult, with research interests in thermal management and electromagnetic analysis for power modules.

CSA Catapult’s Enhanced Thermal Impedance Analysis research marks an important milestone in power electronics innovation — one that speeds up design, supports smarter systems, and helps UK companies stay competitive in fast-moving global markets.

“Our goal is to make advanced thermal simulation faster and more accessible,” says Xuejiao.

“Because when we can test ideas in seconds instead of hours, innovation moves much faster too.”

Powering our future

Discover how we're helping industry accelerate innovation

Our work
Previous Article