Management and predictions of operational changes and growth in mission critical facilities

Abstract

The main challenge in understanding the cooling performance in a legacy data center is the invisible transport medium (air). This emphasizes the need for smart and meticulous measurement techniques. However, the nature of measurements is finite (e.g. top, middle and bottom) and the variation between often non-linear, therefore to capture the gradients between measured points a validated CFD simulation is needed. In this paper a brief description of an experimental characterization of a new data center lab is discussed. Airflow and temperature measurements are utilized to understand the facility's performance at different operational stages, until reaching the designed capacity. Since the facility houses a wide range of different IT equipment (servers, switches, storage and blades), it is important to understand the airflow demand of each. To do that, each type of IT was tested separately and flow characteristics were obtained (i.e. free delivery, critical pressure and flow curves). In the second part, all the characterization data is integrated via compact models into a full CFD simulation. The measured points are used for validation and the full field of air temperature and flow is resolved. Both the measurements and simulation data will now be used to answer important design, deployment and operational change questions.