THE DIFFICULTIES OF OPERATIONAL DATA CENTER SIMULATION

While airflow simulation in the guise of computational fluid dynamics (CFD) has been widely adopted in data center design, some of the most challenging scenarios present themselves in the daily operation of complex legacy facilities. The reason is simple: it is here that availability is compromised, capacity is stranded and energy efficiency is poor as a result of historic operational decisions. In order to use CFD simulation to reliably diagnose the cause and identify solutions, the simulation models must reflect reality. What exactly does it take for operational CFD predictive models to do so? This paper identifies real examples of the causes of poor agreement between model and reality. In doing so, it outlines not only what should be expected when using an operational CFD predictive model, but also identifies the predictions that are made possible as a result.