Abstract
The recent advancements in energy production, storage, and distribution are creating unprecedented opportunities in the field. Major consumers can benefit from the implementation of distributed energy resource networks capable of generating electricity or heating from sources, often renewable ones, in close proximity to the point of use, rather than relying on centralized generation sources from power plants. In this paper, we introduce a pioneering model designed to determine the optimal set of energy commands in a distributed energy resource network, minimizing operational costs in a time horizon. Indeed, we propose an innovative mixed-integer linear programming formulation rooted in the production-inventory models commonly employed in aggregate production planning. The system integrates diverse energy generation sources, storage facilities, and demand points, encompassing both electric and heating commodities. The optimum of the model is achieved for all analyzed instances of the test library (2 scenarios-20 instances) in an exceptionally short time, outperforming other approaches previously presented in the literature. We employed the Gurobi optimizer to solve the model, obtaining rapid responses that ensure real-time decision-making and facilitate effective control of the distributed energy resource network within a three-days' rolling horizon, as discussed in a simulated real-life application case study. Indeed, the proposed model solves in less than 1 s, enabling near-instantaneous decision-making. This swift solution time surpasses any known references in the field, effectively shifting the bottleneck in DER network operation from the decision-making process to the forecasting of demand and weather conditions. While forecasting typically requires a minimum of 15 min, our approach suggests that a reduction in this forecasting time could further enhance the control system's response time, given the model's ability to deliver optimal solutions almost immediately. The real-time availability of optimal solutions allows for the seamless incorporation of stochastic elements into the control loop via a rolling horizon process.