Distributed or centralized? Designing district-level urban energy systems by a hierarchical approach considering demand uncertainties

Abstract The optimal design of urban energy system is considered as a global challenge for improving urban sustainability, efficiency and resilience. The optimization problem is normally formulated as a mixed-integer programming model. With certain spatial and temporal resolution, the model complexity will increase rapidly when the modelling scale expands. The uncertainty of demand further makes the problem more complex. Therefore, to model large-scale urban energy systems, the trade-off between modelling resolution and computational cost has to be considered. This study introduces a hierarchical based approach to decompose the district-level problem into neighborhood-level sub-problems by clustering technique. Two technical routes are further proposed, (1) the energy hub mode adopts Graph theory techniques to obtain an optimal solution rapidly with a slight sacrifice on optimality; (2) the distributed mode enables high optimality but requires significantly high computational cost. Both two routes deal with multiple uncertainties of cooling and heating demand via stochastic programming. The proposed approach is demonstrated via a case study of a business district in Shanghai. The results indicate that modelling with demand uncertainties can lead to 15% difference on project cost from the deterministic formulation. Demand complementarity and network design turn out have critical impacts on system design and project economics. Moreover, a novel Coefficient of Variation index is proposed quantifying the demand complementarity. In general, the proposed approach is efficient and in line with the procedure of real-world infrastructure development. By such approach, the problem becomes solvable by using ordinary computers, which makes it more applicable in real-world urban developments.