Trial / Demo
Pua Ching Tian has vast experience within Sustainability and Energy Efficiency on MEP systems, as well as MEP modelling in HVAC cooling systems for commercial, pharmaceutical, modern warehouse, data centre and healthcare industries.
In this blog, he shares his expertise on the rapid shift toward liquid cooling in data centres and why traditional energy modelling practices are struggling to keep pace. These insights are also featured in the IES whitepaper: De-risking High-Performance Data Centres with Dynamic Simulation.
With data centres across Southeast Asia and beyond racing to accommodate denser, more demanding AI and GPU workloads, the pressure to rethink cooling infrastructure has never been greater. Yet for many engineering teams, the modelling approaches underpinning those decisions have not kept up with the pace of hardware change. Pua draws on his deep MEP expertise to explain why the industry's shift to liquid cooling demands a fundamentally more rigorous approach to energy analysis – and what's at stake if that gap isn't closed.
A new standard of cooling
Based in Southeast Asia, it’s clear to me that liquid cooling is swiftly emerging as the standard for data centres. I’ve watched as the transition toward an 80/20 liquid-to-air cooling ratio has accelerated rapidly, driven by increasing rack densities, climate factors, and sustainability demands. However, my main concern is that the rapid adoption of this hardware is currently outpacing the traditional energy modelling practices that many engineering teams still rely on.
Accuracy is key
Many data centre designers still depend on basic static calculations for heat loads and sizing of cooling equipment, which are incapable of generating an accurate and effective hourly annualized Power Usage Effectiveness (PUE). And as facilities strive for higher efficiency and resilience, it is increasingly critical to perform hourly, full-year analyses to understand complex thermal dynamics, integrated air and liquid cooling PUE including Coolant Distribution Unit (CDU) performance, elevated chilled-water temperatures, and cooling plant interactions. I think that holistic, physics-based dynamic modelling is now necessary to validate environmental claims and this transition is further accelerated by regional sustainability standards and the pursuit of green certifications like LEED, which require traceable, climate-specific data.
Hybrid cooling systems, in particular, must be rigorously evaluated against local humidity levels, temperature fluctuations, and peak load behaviours. Ultimately, I believe holistic, physics-based dynamic modelling is now necessary to validate environmental claims and guarantee system reliability under real-world operational and climatic conditions.
Get even more insight
Pua’s expertise is just part of a comprehensive analysis of how advanced modelling approaches are enabling data centre operators to build facilities that are more resilient, efficient, and prepared for the demands of AI infrastructure.
To learn more about dynamic simulation and how it addresses the key challenges facing modern data centres, download the full IES whitepaper: De-risking High-Performance Data Centres with Dynamic Simulation.
