The huge amount of energy consumed in data centers has prompted researchers to continue studying how to improve their energy efficiency

and according to a US Environmental Protection Agency report

the energy consumption in data centers is doubling every five years; in contrast

the energy consumption in data centers in China have a growth rate significantly higher than the average level globally. Improving the efficiency of the refrigeration system and optimizing the data center airflow pattern will minimize the energy consumption. This paper is the second part of an investigation aiming to find a more satisfying airflow pattern in a data center and to help provide servers with a safer and more reliable operating environment. Based on the simulation described in Part I

the optimal model

including six groups of duct structures

is verified. The odd-numbered sub-modules of the data center were taken as the measured objects

changing the perforation rate of the floor

and the angle of the baffle

to obtain the six groups of reasonable airflow patterns. A comparison of the two key parameters of the front door air supply

namely

the speed and temperature

proves the reliability of the simulated airflow. The maximum relative deviation between the measured and simulated velocity is 17%

and the maximum relative deviation of temperature is 5%

which shows that the simulated temperature and velocity fields are in agreement with the measurement. Concerning the data center heat flux density and energy consumption

the air supply conditions of the cooling system are changed to achieve a comparative analysis of the temperature field under various working conditions. The return air temperature index (RTI) is employed as an evaluation indicator to find that the recommended air supply temperature range

which is 16–19 ℃. The results of this investigation can provide reference for the design of a data center using an under-floor air distribution.