Designing HVAC systems for modern data centres: lessons from the field
Hysopt engineer Finn Hansenne shares lessons from active data centre HVAC projects, including redundancy, chilled water systems, transient analysis and simulation-driven design.
As AI workloads grow and rack densities increase, data centres are placing new demands on HVAC engineering.
Cooling systems need to be more resilient, more efficient and more predictable. For HVAC engineers, that means moving beyond standard design assumptions and getting a much clearer view of how chilled water systems behave under real operating conditions.
We sat down with Finn Hansenne, Customer Success Manager and Energy Engineer at Hysopt. Finn has been supporting several active data centre projects, helping engineering teams model, simulate and validate complex hydronic cooling systems.
We asked him what makes data centre HVAC different, why simulation is becoming more important and what engineering teams should be thinking about next.
What makes HVAC design for data centres different from a typical building?
Finn:
The biggest difference is that failure is simply not an option.
In a normal commercial building, you're usually balancing comfort, energy efficiency and cost. Of course those things still matter in a data centre, but the starting point is different. You are designing around reliability first.
Cooling continuity is mission-critical. If temperatures in the data hall rise too quickly, or if the system does not respond as expected during a failure or maintenance scenario, the consequences are much bigger than a comfort complaint.
So every engineering decision carries more weight. Redundancy, response times, flow paths, controls, pump behaviour — it all needs to be thought through much more carefully.
What is changing in the data centre market right now?
Finn:
AI is definitely changing the conversation.
Rack densities are increasing, cooling loads are becoming more concentrated and the traditional approach to data centre cooling is being stretched. You still see a lot of air cooling with hot aisle and cold aisle configurations, but there is also more discussion around liquid cooling, CDUs, heat exchangers and hybrid cooling strategies.
At the same time, operators are under pressure to reduce energy use and improve sustainability. So engineers are being asked to design systems that are extremely reliable, but also efficient.
That is not an easy balance. You want enough redundancy and safety margin, but you also do not want to blindly oversize everything and create unnecessary energy use or hydraulic complexity.
Where do engineers face the biggest challenges?
Finn:
A lot of it comes down to complexity.
Data centre HVAC systems often include redundant production units, redundant end units, multiple operating modes and looped or meshed hydraulic layouts. That is very different from a more standard commercial HVAC system.
With a conventional system, the calculation path is usually more straightforward. In a data centre, you need to think about what happens if part of the system is unavailable, how flow is rerouted, how pumps respond, and how the system behaves during different failure or maintenance scenarios.
The challenge is not only sizing the system. The bigger question is: how will this system actually behave once it is operating?
That is where things become interesting from an engineering point of view.
Why is simulation becoming more important for data centre HVAC?
Finn:
Static calculations are still important. Nobody is saying they disappear.
But in data centres, they only tell you part of the story. You’re not just asking, “Is the system sized correctly?” You also need to ask, “What happens if something goes wrong?”
That’s a big part of it. These are critical data environments, so the hydraulic installation needs to be de-risked. Failure mode analysis becomes part of the wider risk assessment. You need to understand how the system behaves if equipment is unavailable, if a section of the system is isolated, or if loads change quickly.
For example, how does the chilled water system respond when load changes? How quickly does flow stabilise? Are the pumps and valves doing what you expect? Are your conservative design choices actually helping, or are they adding unnecessary complexity?
That is where simulation becomes really useful.
It gives engineers a way to test scenarios, look at transient response and understand hydraulic behaviour before the system is built.
What kind of data centre questions can Hysopt help engineers answer?
Finn:
The value is mainly in giving engineers more confidence before construction.
That can mean comparing different cooling concepts, validating pump and plant selections, checking hydraulic behaviour or running scenario-based simulations to understand how the system responds.
In some projects, the question might be: "What happens if we close this section of the loop?" In another project, it might be: "How does the system respond if a production unit is unavailable?" Or: "Can we validate a conservative pump selection with simulation instead of only relying on manual assumptions?"
Those are exactly the types of questions where a physics-based model can help.
Hysopt is not replacing every specialist tool used in data centre engineering. That is not the point. The point is to give HVAC engineers a better way to understand and validate the hydronic system behaviour.
How does this differ from CFD or other data centre cooling tools?
Finn:
CFD and hydronic simulation answer different questions.
CFD is very useful when you want to understand air movement in the data hall — things like airflow patterns, hot spots, temperature distribution, hot aisle and cold aisle performance.
Hysopt is focused on the water-side HVAC system. So we are looking at chilled water loops, pumps, valves, heat exchangers, coils, flow rates, pressure losses, control behaviour and the connected plant.
For a data centre, both views can be valuable. The air side tells you what is happening in the data hall. The hydronic side helps you understand how the cooling infrastructure responds and whether the system can support those cooling demands reliably.
The important thing is knowing which tool answers which engineering question.
Are data centre projects harder to model than other HVAC projects?
Finn:
Yes, definitely.
The redundancy requirements make everything more complex. You can have N+1 strategies, 2N concepts, looped hydraulic networks, different flow paths and several possible operating scenarios.
That means you need to be very clear about your modelling assumptions. You cannot just build a model and assume one calculation tells the full story.
For some static design questions, especially around fully looped hydraulic networks, there can still be manual steps and workarounds. That is why it is important to be transparent. But for simulation, transient analysis and scenario validation, Hysopt can bring a lot of value when the model is built correctly and the design information is available.
That is also why experienced engineering input matters. The software helps, but the engineering logic behind the model is still critical.
What have you learned from working on active data centre projects?
Finn:
One big lesson is that data centre teams are not just looking for calculations. They are looking for confidence.
They want to know whether their design will behave as expected. They want to understand risks earlier. They want to be able to defend decisions with evidence.
Another lesson is that redundancy changes everything. It affects pipe layouts, pump selections, control strategies and how you think about operating scenarios. You cannot treat it like a normal building with a bit more cooling capacity.
And then the third lesson is that communication is really important. These projects involve consultants, contractors, operators, controls specialists and sometimes several specialist tools. A clear simulation model can help everyone have a better technical conversation.
Where do you see data centre HVAC engineering going next?
Finn:
I think the industry is looking for better ways to objectify complex HVAC systems.
Right now, a lot of decisions still depend on separate calculations, assumptions and manual interpretation. But data centre cooling systems are becoming too complex for that to be enough.
Engineers need reliable tools that can help them design and analyse multiple flow scenarios more easily. Not just one design point, but different operating modes, failure scenarios, maintenance conditions and load changes.
That is where I think the industry is heading: more simulation-driven engineering, more scenario-based validation and more evidence behind the decisions being made.
The engineers who can clearly show how a system behaves — not just calculate that it should work — will have a real advantage.
Final thoughts
Data centre cooling is becoming one of the most demanding areas of HVAC engineering.
Higher rack densities, AI workloads, liquid cooling interfaces, redundancy requirements and energy pressures are all pushing teams to think differently about system design and validation.
At Hysopt, we are supporting engineering teams on confidential data centre projects, helping them use physics-based simulation to better understand hydronic system behaviour, reduce design risk and make more confident engineering decisions before construction.
For Finn, that is where the real value sits.
"In the end, it is about confidence," he says. "If you can understand how the system behaves before it is built, you are in a much stronger position as an engineer."
READ ALSO
The State of HVAC 2026
Discover the 6 key HVAC trends for 2026 in this e-book packed with data-driven insights and actions to help you stay ahead in the changing market.
Download your copy today and see what no HVAC engineer can afford to ignore in 2026.
Ready to validate HVAC performance before construction?
Use Hysopt to simulate hydronic systems, compare design scenarios and reduce oversizing risk.
Explore more
AI for HVAC Engineers: Why Engineering Expertise Matters More Than Generic AI
Artificial intelligence is changing engineering software—but for HVAC engineers, the real value isn't AI itself. It's having trusted engineering expertise available exactly when you need it. Discover how Hysopt DeltaT combines decades of HVAC engineering knowledge with AI to help engineers find answers faster, work with greater confidence and stay focused on engineering.
Hysopt Accelerates International Growth and Appoints Brian Wimberley as Chief Marketing Officer
As demand for simulation-driven HVAC engineering continues to grow, Hysopt is expanding internationally, advancing its platform, and strengthening its leadership team with the appointment of Brian Wimberley as Chief Marketing Officer.
Hysopt welcomes Belimo as a supported manufacturer
Hysopt expands its portfolio of supported HVAC manufacturers with Belimo. The first Belimo components are now available within Hysopt, with additional product families already in preparation.
.png)
.png)