Bi-level optimization and sustainability assessment of data center integrated energy system based on emergy theory

With the rapid development of the digital economy, the energy demands and environmental pressures of data centers have become increasingly prominent. Traditional optimization and evaluation methods based on single indicators are no longer sufficient to comprehensively reflect the system's economic viability, environmental benefits, and resource utilization efficiency. This paper, based on emergy theory, converts the inputs of various energies, resources, equipment, and labor within the system into solar emergy, thereby establishing a cross-dimensional comprehensive evaluation framework. For the first time, the Emergy Sustainability Index (ESI) is introduced as the optimization objective of the data center integrated energy system (DC-IES). A bi-level optimization model is proposed in this study. In the upper level, a genetic algorithm is employed to optimize equipment capacity configuration to maximize the ESI. In the lower layer, typical daily operations are addressed by formulating a mathematical programming model that maximizes the renewable emergy input, thereby achieving a coordinated optimization of system capacity and operational scheduling. The TOPSIS method is used to select the optimal solution from the multi-objective Pareto curve, and the approach is validated through a case study of an actual data center. The results show that the ESI-based optimization case achieves an ESI value of 0.1518. Compared to the case optimized for both annual cost and carbon emissions, this solution yields a 296 % increase in ESI (from 0.0382 to 0.1518), a 61 % reduction in carbon emissions, and a 51 % increase in annual costs. Despite the moderate economic trade-off, it significantly improves system sustainability and environmental performance. Finally, the paper summarizes the current advantages and limitations of applying emergy theory in analyzing and optimizing DC-IES, thereby providing new theoretical directions for future research.