Research On Airflow Simulation In Data Center Based On Intelligent Auxiliary Fan

With the increase of data center heat density,the problem of mixed hot and cold airflow is prominent.To address this issue,it is necessary to study and optimize the airflow organization in the data center.First,a computational fluid dynamics(CFD)approach is used to model and numerically analyze the data center to investigate areas or racks where mixed hot and cold airflow could occur.The results show that the less amount of cooling gas closed to racks,and the more likely it is that the top of the cabinet will have this problem.Then,aiming at the problem area A1,which is close to the air conditioner,the intelligent auxiliary fan is introduced and its CFD is modeled.The feasibility of the model is verified by grid independency and numerical simulation.That is,the model does not produce negative effects in the original data center.The second-order design and variable analysis of the two input variables,namely the outlet air mass Q and the height H,and the maximum inlet temperature T of the cabinet are carried out.The results show that when H=0.87m,Q=0.21m3/s,the cabinet temperature meets the requirements of ASHRAE9.9TC standard.Furthermore,it's more energy-efficient.Then,under this scenario,the temperature and speed of A1-A5 on the side of the cold aisle and the entrance section of the cabinet A1 are studied and compares with the physical data of the original data center.The results show that the introduction of intelligent auxiliary fan can improve air flow organization,especially in hot and cold air flow in the upper part of the cabinet.Using the dimensionless parameters of the thermal environment assessment index,the research evaluates the operation standard of the data center and compare with the original model.The results show that after the introduction of the intelligent auxiliary fan,the SHI index of mixed hot and cold air flow becomes smaller,indicating the data central cold utilization efficiency improving.Finally,in order to enable the intelligent auxiliary fan to reach the mixed area of hot and cold air flow in advance,the temperature monitoring points at different air-conditioning outlet temperatures are simulated and analyzed by FloTHERM software.The first 60 samples are selected from the simulated 75 sample data as training data and the last 15 data as the prediction data.The RBF neural network is used to model the data in MATLAB and the data is fitted.The results show that the fitting accuracy of the latter 15 data is high with the error within 5%.Under the condition of accuracy requirements,RBF neural network can be used to predict the cabinet inlet temperature,and it can address the hot spots or airflow mixing area in advance.What's more,it solves the problem of air conditioning's slow response,low efficiency,serious waste of energy consumption in actual operation.