Optimal Design of Multiwinding-Transformer-based Power Architectures in Data Center Applications

Abstract:
The increasing demand for high-performance power delivery in data centers and telecommunications is calling for high-efficiency and high-availability power conversion. In this context, this paper presents a general overview of classifications and requirements of power distribution architectures for data center applications. The power distribution architectures are mainly based on AC and DC types, which are reviewed and analyzed. Thus, to achieve the fault-tolerance and the high integration of renewable energy sources (RES) and energy storage systems (ESS), multiwinding-transformer-based (MTB) power architectures are employed to compose the new data center architectures. As a case study, a co-design optimization framework is formulated to optimal design the capacity of the photovoltaic (PV) strings and battery energy storage (BES) as well as power allocations between utility grid and BES in a 100kW data center. A long-term normal operation period is considered to evaluate the performance of systems with and without the presence of PV strings and BES.