Research On Dynamic Allocation Strategies And Energy Saving Methods For Elevated Floor Data Center

This paper presents a room-level computational fluid dynamics(CFD)model of an elevated data center configuration using auxiliary tiles and fans.Auxiliary tiles are flexible perforated floor tiles that rationally distribute the cold air entering the cold aisle and improve the local cold air flow in the cold aisle.The fan is a combined unit with an axial flow fan.The cold air is redistributed to the underfloor ventilation system,which increases local volume flow.To our knowledge,no previous studies have numerically simulated data centers using auxiliary tile fans.The research content of this article mainly includes the following parts:First,introduce a dynamic allocation modeling strategy in the data center.Develop a computational fluid dynamics(CFD)model in FloTHERM software for three configurations of a data center computer room:a reference case with passive tiles,a reference case with auxiliary tile tiles,and finally with active tiles Configuration.These physical-based numerical models can be used to optimize design parameters to prevent hot spots without overheating the entire data center.Fan units can have a significant effect on ventilation pressure,which affects the total flow delivered to the cold aisle and the percentage of room space leaked from the ventilation space.For a given computer room air conditioner(CRAC)blower speed,an increase in the active fan speed will increase the overall floor tile air flow rate and reduce cold air leakage in the ventilation room.Second,apply FloTHERM software and related knowledge of thermodynamics and fluid mechanics to perform model simulation analysis and cooling performance evaluation of the dynamic allocation modeling strategy.The huge potential of data centers based on the laws of thermodynamics includes the ability to locally change cooling resources and understand how each variable affects the state of the data center.In order to meet the ability to change cooling resources locally,the actuators of interest in this study included auxiliary tile opening rates and variable air flow from the fan.Third,Optimization of design parameters and optimization of maximum temperature based on multivariate analysis and least squares method.A two-factor analysis of variance was used to analyze the significant impact of the auxiliary tile opening rate,fan flow,and the interaction between the two variables on the thermal management of the data center.Apply the principle of least squares to perform a non-linear fitting on the maximum temperature of the rack inlet,find a fitting equation,evaluate the fitting function according to the fitting evaluation index,and finally determine the equation and the fitted surface;use the simulated annealing algorithm to find the fitting The optimal value of the equation,the initial parameters are determined based on experience and actual conditions,the optimization iteration analysis is performed in MATLAB software,and the output results are brought into FloTHERM software for reverse verification.The verification results show that the optimization iteration strategy is accurate.Fourth,Use the thermal environment evaluation index to evaluate the cooling performance of the data center,and analyze the energy consumption of the cooling system in the data center.SHI,RHI,RCI,RTI,and ? evaluation indicators based on the first law of thermodynamics are selected to evaluate the thermal management of room and rack levels in the data center to characterize the room and rack-level thermal management levels of the data center.After the dynamic allocation strategy is adopted,the energy consumption analysis of the data center refrigeration system is performed to characterize the energy conservation of the data center refrigeration system.The data center operates 24 hours a day,365 days a year,and the energy consumption optimization of the refrigeration system will bring great energy savings to the data center.