Two-stage robust planning of data center microgrid considering batch load flexibility and multi-energy complementarity

Abstract

The intrinsic coupling between dynamic batch load characteristics of data centers and robust microgrid planning presents significant challenges in coordinating economic efficiency with power supply reliability, particularly under high renewable energy penetration. This paper proposes a novel temporal flexibility modeling framework for batch workloads and establishes a two-stage robust optimization method for data center microgrid planning. Three fundamental innovations are introduced: (1) A matrix-based adjustable domain model quantifying batch load transfer capability with 2-hour delay constraints, mathematically characterizing temporal flexibility through shiftable power matrices; (2) A min–max-min robust planning architecture coordinating photovoltaic-storage-gas turbine systems, employing C&CG algorithm for hierarchical solution of capacity configuration (upper-level) and operational scheduling (lower-level) under renewable-load uncertainties; (3) Practical validation through Guizhou’s data center cluster demonstrates 14.7% reduction in capital expenditure and 23.2% improvement in load scheduling flexibility while maintaining service quality constraints. The proposed method provides a quantitative tool for optimizing infrastructure planning-operational dynamics in renewable-dominated microgrids, supporting low-carbon transition of data center clusters in Guian New Area. Comparative analysis reveals the model’s superiority over conventional robust approaches in achieving Pareto-efficient solutions between cost reduction (¥12,819/$1,789 daily operational savings) and reliability enhancement (63% gas turbine capacity optimization under 30% uncertainty scenarios).