With the rapid development of big data, the scale of data center is becoming larger and larger. The heat generated by the servers in the data center is also increasing, which puts forward higher requirements for the cooling of the data center, which should have strong enough heat dissipation capacity and low enough energy consumption. This paper is committed to reducing the cooling energy consumption of the data center and reducing the PUE of the system. This paper studies a server cabinet with a power of 4.8KW. The fins indirect water-cooled radiator is used to cool the chip, and the cooling tower cold source cycle is constructed outside. In order to facilitate the research, TRNSYS software is used to model the system, and the internal modules of TRNSYS software are selected to build the system.Through TRNSYS simulation and thermodynamic calculation, the relationship between the chip temperature and the server inlet water temperature and secondary water flow is studied. The system working conditions from March to October in Tianjin are studied which all change with the change of ambient temperature and relative humidity. The results shows that the optimal cabinet inlet water temperature, primary water flow, secondary water flow, system power and PUE of the system will change accordingly with the change of ambient temperature and relative humidity. In Tianjin in each month, the range of PUE is 1.032-1.091, and the variation range of system power is 0.154 kW to 0.437kW when run at the optimal conditions.