Study On The Thermal Performance And Energy Saving Of A Loop Thermosyphon System Used In Data Center

With the development of information technology and the booming requirements of industrial intellectualization,the number and scale of data centers are increasing in China,resulting in a rapid increase in energy consumption.In 2018,the electrical consumption of China’s data centers reached 160.8 billion k Wh,and the cooling system accounts for more than 30% of the total data center energy consumption.Therefore,it is important to reduce the energy consumption of cooling systems to improve the energy efficiency of data centers.A loop thermosyphon system can overcomes the defects of the traditional cooling system,such as cooling capacity losses during the air supply process and inefficient utilization of natural cooling sources,which has a good application prospect.This paper aims to improve the thermal performance and energy saving potential of a loop thermosyphon system to prompt its application,using experiment and simulation methods to solve the difficulties such as:(1)the heat transfer law of the loop thermosyphon is unclear and the thermal performance is difficult to predict under non-uniform conditions;(2)the performance comparison study of the loop thermosyphon filled with different refrigerants is insufficient;(3)the simplified model of the loop thermosyphon with both accuracy and computation speed is lacking;(4)the research on the energy saving effect and economic potential of loop thermosyphon systems are limit.Firstly,an experimental platform based on an enthalpy difference laboratory is set up,and the thermal characteristics of a loop thermosyphon are experiment al experimentally investigated under non-uniform conditions(including fan failure conditions and non-uniform servers’ arrangements).The results show that the lower filling ratio in the optimal range,the better anti-failure capability of the loop thermosyphon;the positions of faulty fans have little influence on the total thermal performance with an optimal filling ratio;the inhomogeneity of the evaporator inlet air temperature caused by the servers’ arrangement has a certain influence on the optimal filling ratio,but it did not affect the steady-state operating performance;The servers’ arrangement had little influence on the start-up times under high(6k W)and medium(4k W)heating powers but had a significant influence under low(2k W)heating power.Secondly,a one-dimensional steady-state model of the tested loop thermosyphon is established based on the Effectiveness and Number of Transfer Unit method,which is suitable for predicting the thermal performance under non-uniform conditions.By comparing experimental data,heat transfer correlation selection and model validation are carried out.The established model ultimately used Shah boiling heat transfer correlation,and the mean absolute deviations of its predicted results for refrigeration volume,refrigerant pressure and mass flow are less than 7.3% under uniform and non-uniform conditions.Based on this model,the effect of different emergency methods on the thermal performance under fan failure conditions are studied.The results show that the airflow rate backup of extractor fans is one of the effective methods to deal with the fan failure of a loop thermosyphon.Thirdly,a R32 loop thermosyphon is experimentally investigated for its performance.R22 and R134 a are also used for comparison.The res ults show that the optimal filling ratio of R22,R134 a and R32 is the same(37.3%).At the optimal filling ratio,the operating performance of the three refrigerants is compared and evaluated by their flow stability,cooling capacity and energy efficiency ratio.The free cooling potentials of the three refrigerants are analyzed and compared by using a simplified simulation model of a data center.The results show that the flow stability of R32 is obviously better than those of R22 and R134 a when cold water temperatures are 10-22℃.The cooling capacity and energy efficiency ratio of R32 is 5.4-15.6% higher than those of R22 and R134 a.R32 has the best free cooling potential among the three refrigerants.Finally,a simplified model of a R32 loop thermosyphon is developed by the established steady-state model,which has high prediction accuracy and calculation speed.Based on this model,the comprehensive energy consumption models of the cooling system in use(an air-cooled computer room air-conditioning system)and two energy saving retrofit plans(a water-cooled computer room air-conditioning system and a loop thermosyphon system using water-side economizers)are established by taking a typical data center as a case.Compared with the cooling system in use,the energy saving rates of the water-cooled computer room air-conditioning system and the loop thermosyphon system are 27.6-47.3% and 53.4-63.6%,respectively;the effect of different operating parameters on the energy consumption of these systems is also also simulated.According to the economic analysis,the dynamic payback periods of cooling system modifications using the water-cooled computer room air-conditioning system and the loop thermosyphon system are 4.19-4.69 years and 1.70-3.62 years respectively under different operation stages of the data center.In addition,the set points of operation parameters are optimized by the genetic algorithm.It is found that the optimization efficiency is restricted by operating parameters and performance of cooling system equipment.The longer the wet bulb temperature meets the maximum free cooling temperature,the higher the optimization efficiency of genetic algorithm in a city.