MEASUREMENTS AND PREDICTIONS OF THE FLOW DISTRIBUTION THROUGH PERFORATED TILES IN RAISED-FLOOR DATA CENTERS

摘要

Data centers are used to house data processing equipment such as servers, mainframes, and storage systems. The equipment, which dissipates a significant amount of heat, needs to be maintained at acceptable temperatures for its reliable operation. A typical cooling arrangement consists of installing the equipment on a raised floor and using several air-conditioning (A/C) units to force cool air into the space under the raised floor. Whereas most of the raised floor is impermeable, perforated tiles are installed at desired locations to provide cool air at the inlets of the data processing equipment. The distribution of the air flow through the perforated tiles located throughout the data center is governed by the size of the plenum space under the raised floor, the presence of flow obstructions such as cables and pipes in that space, the locations and flow rates of the A/C units, the layout of the perforated tiles, and the tile open area (or their flow resistance). The complex flow in the plenum space under the raised floor sets up a pressure distribution, which controls the flow through the perforated tiles. The purpose of this paper is to propose a computational predictive model for the problem described and to provide benchmark measurements to test the validity of the model. The measurements were conducted on an actual data center at IBM, Poughkeepsie, New York. It contained two A/C units, which could be activated independently. A large number of different layouts of the perforated tiles were created and the flow rates through them were measured with either one or both A/C units operating. The flow resistance of the perforated tiles was also determined by measuring the corresponding pressure drop. The measured flow distributions were compared with the predictions of a computational model, which calculated the pressure distribution in the plenum space and the corresponding flow rates through the perforated tiles. The model was correctly able to predict the flow nonuniformity for different tile layouts and different operations scenarios for the A/C units. Finally, the model was used to study the effect of tile open area and plenum depth on the flow distribution through the perforated tiles. The results provide guidance for choosing these parameters for achieving the desired flow rates through the tiles.