Research On Air-cooling Characteristics Of A Raised-floor Data Center Based On Wind Turbine Technology

This paper investigated the use of the potential in-door wind energy in a raised-floor data center to present a significant and energy-efficient methodology and to improve the air-cooling characteristics.A new endeavor was regarded by implementing a wind turbine configuration placed in the under-floor plenum.Part of the over-provisioned airflow in the under-floor plenum can be utilized to accomplish the power generation in situations of considerably over-cooled data centers.In short,the application of the wind turbine in the under-floor plenum is a valuable cooling program to improve the energy efficiency of the data center.Firstly,the under-floor flow field and the flow rates of the perforated tiles were considered to explore the parametric features of the wind turbine using Computational Fluid Dynamics(CFD).The rotor of the wind turbine was modeled by using the air resistance producing resistive force governed by the thrust coefficient(CT).The impacts of the thrust coefficients and the distances of the rotor to the Computer Room Air Conditioner(CRAC)air supply have been obtained.Consequently,in comparison of the positioning of the rotor,the differently implemented thrust coefficients especially contributed more distinctly to the airflow patterns through the perforated tiles while the cooling performance can still be effectively maintained.The results of the present research demonstrated that the addition of the wind turbine in the under-floor plenum may yield a valid cooling solution to enhance the energy efficiency of data centers.Secondly,in order to study the rationality of trying to implement the wind turbine in the data centers,the effects of the wind turbine on the air volume and the distribution of airflow velocity and pressure for the impinging jet flow were analyzed with the theory of turbulent impinging jet.The air volume was monitored to observe the influence of the air resistance on the airflow in the under-floor plenum using a monitor face below the air resistance.The distribution of the airflow velocity and pressure at different height levels of the XZ plane and the non-dimensional velocity and pressure distribution were analyzed in zone 2,and the values of the airflow velocity in the X direction were investigated using the monitoring points at the side of the air resistance.Consequently,there are not apparent differences in the airflow distribution by comparing the based model with case 2.In addition,the differences were controlled in the acceptable range,and the application of the air resistance promoted the momentum of the surrounding airflow as well as eliminated some airflow stagnation points.In the future,the wind turbine will be a machine that can accelerate the current motion in the under-floor plenum.Finally,the prediction of the hot-spot temperature at the outlet side of the cabinets was completed by establishing neural network model.The traditional CFD simulation methods need to spend a great sea of time to obtain the computed results,but the neural network model can also get similar results in a very short space of time.In this paper,the BP neural network was utilized to predict the hot-spot temperature of the row A of the cabinets,but it isn't convenient to analyze the highest hot-spot temperature because of too much output data.Therefore,using the genetic algorithm to optimize the BP neural network to predict the highest hot-spot temperature of the cabinets,thus the air supply volumes of the fan units inside the cabinets were changed to achieve the optimal air supply scheme.And the neural network model can ensure the accurate prediction of the cabinets hot-spot temperature and greatly accelerate the solving process.The best cooling project of the data center was obtained through optimizing the design parameters.