While Computational Fluid Dynamics (CFD) has been an accepted best practices tool for many years, it still amazes me how many data center managers don’t see its value. Smaller or homegrown facilities often don’t see the need either because they’re so small that the money lost to overcooling isn’t significant enough to make a change; or, “I’m already monitoring. That’s basically the same thing, right?” Wrong. At many data centers we tour – either during an introductory visit or as part of an initial assessment – we discover that many people don’t understand the difference between monitoring and CFD modeling.
The fact is, CFD modeling is much more than monitoring equipment, identifying hot/cold spots and improving efficiency. Now don’t get me wrong. All of that is important, but monitoring only tells part of the story. It only establishes a baseline for your environment. CFD’s arguably bigger purpose is prediction – serving as a springboard for making smart changes and being prepared for any type of failure situation. Planning for the future is something data centers of ALL sizes should be doing.
Computational Fluid Dynamics is a term given to the study of fluid under motion that uses numerical analysis and data structures to solve some form of the Navier-Stokes equations, or conservation of momentum. The aim of CFD analysis and modeling is to learn the effect of flow on boundaries and for flow visualization.
CFD has been around since the 1930s, beginning in the field of aerospace and F1 racing, and the electronics industry in the late 1980s. By comparison, CFD for the data center is still a new concept having been around for a little over a decade. But in that short period of time, CFD modeling has proven to be a reliable method for optimizing the energy efficiency of the data center.
More than a Crystal Ball
But improving energy efficiency is only the beginning. Failure model prediction is a key benefit of CFD.
All cooling units need to be taken offline periodically for maintenance; and occasionally they just fail. Knowing which servers will need to be shut down, or which cooling units are the most critical to the room, can be accurately and precisely predicted well in advance with CFD. This prediction capability also allows for procedures and safeguards to be developed.
CFD Failure model prediction also extends to space and capacity planning. Considering the large difference in lifecycle between onsite cooling equipment (decades) and the equipment it’s cooling (3-5 years), it stands to reason that serious problems can arise if that next generation of IT equipment no longer match the cooling characteristics of your data center. CFD modeling allows you to check airflow and temperature as a result of making physical changes to the data center, but before equipment is bought and installed. And once you’re ready to go live, CFD can be used to test any IT deployment before implementation without risking business continuity.
The uses and benefits of CFD modeling I’ve discussed here only scratch the surface. CFD solves and even prevents many issues in data center design and operations. For that reason, Data center managers should stop thinking of it as optional. It’s a necessity for ensuring future efficiency and uptime. To learn more about CEG’s vendor partner’s CFD solution 6SigmaDCX, click here.