BCAM Scientific Seminar: Optimization of the design and operation of multi-energy systems

LOCATION: Maryam Mirzakhani Seminar Room at BCAM and Online

ABSTRASCT:
Aggregated or multi-energy systems (MES), such as energy hubs, virtual power plants and microgrids, are playing a major role in the transition toward a cleaner and affordable energy system since they are able to achieve overall better technical, economic and environmental performances with respect to corresponding not integrated solutions. Because of the large variety of possible configurations and the level of complexity deriving from the high number of components involved and their interactions, several ad-hoc modeling techniques and optimization tools have been developed to address the issue of MES design and operation. In this context, the thesis work aims to develop a MILP-based optimization framework to address the problem of MES design and operation. From a mathematical point of view, such problem is essentially a two-stage stochastic program with binary variables in both stages: (1) the design stage, where the decisions on the type (which technology to install) and size (which size is required) of units are made and (2) the operation stage, where the optimal scheduling of the set of units selected in the first stage is evaluated and their commitment status (i.e. on/off) and operative loads are determined.

Traditionally, the design and operation are handled separately, leading to suboptimal solutions. Conversely, MILP-based approaches are able to tackle both stages simultaneously providing an optimality certificate on the solution found. However, the MILP methodology when applied to real-world case studies often results in large-scale problems, with a large number of variables and constraints. Therefore, in order to contain the computational burden and preserve the tractability of the optimization problem, the development of ad hoc solution techniques is required: in this sense, particularly interesting are decomposition approaches (such as Benders, Lagrangean and the bi-level decomposition) that can exploit the two-stage structure of the problem to speed-up the overall solution process

Link to the session here