Stochastic optimization of photovoltaic-integrated data centers with hybrid cooling and waste heat recovery for district energy supply

A novel photovoltaic-integrated data center system combining hybrid free-mechanical cooling and waste heat recovery for district energy supply is proposed, addressing the multi-dimensional cooperation of PV generation, electricity, and cooling demands. The waste heat of data center is recovered through the air-source heat pump to produce high-temperature heat for district users. A stochastic optimization method of configurations and operation is developed to manage uncertainties in PV generation, cooling demands, and waste heat utilization. The two-layer optimization model is constructed to optimize the capacity sizing considering the energy scheduling management, in which the shift strategy computation tasks of data center is adopted for matching uncertain photovoltaic power and energy demands. A case study validates the proposed system and methods. The simulation results demonstrate that the integration of hybrid cooling and waste heat recovery effectively improves energy and economic performances of the whole system. Their integration reduces electricity consumption for cooling by 34.72 % and 15.57 %, respectively, and the corresponding total cost reduces about 3.52 % and 2.18 %. The proposed computation task management strategy, influencing power and cold demands and energy dispatch strategies of the integrated system, directs computational tasks to be executed during periods with low carbon emission factors and low grid electricity prices. In comparison to traditional method, the proposed strategies reduce total cost by 11.15 % although installation larger capacity increases investment cost.