Computational fluid dynamics (CFD) modeling is an effective way to get an accurate picture of the airflow in your data center and predict the effect of any changes you plan to make. But historically, CFD packages have been costly to purchase and required extensive expertise to operate.
That is changing, according to Paul Bemis, president of Applied Math Modeling, Inc., which makes CoolSim, a CFD modeling tool. Bemis hosted a session at the recent Data Center World conference in Orlando on how to integrate CFD into your data center infrastructure management (DCIM) tool.
Using predictive CFD modeling to understand the effect of planned changes is well-understood technology that’s been used for about 40 years in automotive industry, Bemis said. More recently applied to data centers, the technology can be used to accurately predict the maximum IT load a data center can handle, the optimal configuration of equipment from a cooling and energy consumption perspective, disaster planning scenarios and more.
Early implementations of the technology were expensive, on the order of $20,000 per user per year, and required extensive training, Bemis said. That put the technology out of reach for most companies that wanted to just use it occasionally in their data center.
But, as is the case with most any technology, CFD tools have become less expensive and easier to use over time. Cloud-based offerings, in particular, make CFD more affordable. CoolSim, for example is available under subscriptions that allow for just occasional use, but still include training and support. And customers need only a standard PC to run the client software, he said. With GUI-based interfaces, the tools are also becoming easier to use.
CFD calculations will bring more accuracy to a DCIM tool, enabling users to predict with confidence the best place to add another 30kW of IT capacity, for example. It can also help with disaster preparedness, answering questions like, “What happens if a CRAC or CRAH fails?” And CFD can help answer questions around energy efficiency, to help users optimize their PUE.
DCIM is also a good fit for CFD because the DCIM tool will have the lion’s share of the configuration data that you need to load into the CFD tool, from the location and size of IT loads, racks, cooling units, PDUs and so forth.
Bemis also went over a number of questions to ask when choosing a CFD modeling tool, the answers to which I’m sure point to something that looks a lot like CoolSim, but they all seem to make sense:
- Is it reasonably priced for occasional use?
- Can it import data from any DCIM tool?
- Can it perform what-if scenarios?
- Is it simple enough to be self-taught? It should be self-paced in terms of any training requirements, recognizing that people don’t have time for lots of training anymore.
All of this put me in mind of a post I wrote earlier this year on EcoStream, a CFD tool that’s one of the modules available for the Schneider Electric StruxureWare DCIM package. Schneider Electric partners also have access to EcoStream – at no charge – and can run analyses for you using the tool.
No matter what route you take, it does appear that CFD is becoming more accessible to the data center masses, and that is a welcome development.