A unified optimization approach to design complex and sustainable configurations of single and multi-energy systems

The transition to a sustainable energy sector with net zero greenhouse gas emissions requires the planning and design of new energy system configurations that can integrate increasing shares of alternative resources to fossil fuels. In addition, the integration of different energy sources, carriers and end uses requires the appropriate combination of a large number of components for energy conversion, storage and transportation, which can lead to complex system configurations. In this context, numerical optimization methods can provide crucial support to engineers. To date, the most widespread approach to optimal design of a system is top-down. This entails the designer defining an initial, all-encompassing superstructure of the system, from which the optimal configuration is derived through an automatic procedure. However, the construction of the initial superstructure, in addition to being time-consuming, is subject to the designer’s experience. The risk is that the optimal system configuration may not be included in the superstructure. This thesis proposes a bottom-up approach for the optimal design of energy systems, which builds the final system configuration through the combination of elementary components. This combination follows a set of rules intrinsic to the optimization process, does not require the definition of initial superstructures, and is therefore independent of the designer’s experience. This significantly reduces the time of the design process, since no superstructure needs to be built. Furthermore, if the rules combining the elementary components are defined properly, all possible combinations of the elementary components can potentially be evaluated, and it is possible to identify the optimal one in an absolute sense. The bottom-up approach considered has general features and can be adapted to different types of energy systems, from the single power plant to the multi-energy system. The idea is to decompose the overall optimization problem into two levels, which are handled by an evolutionary algorithm combined with (mixed integer) linear programming. The optimization process is driven by the chosen objective function, which must be in line with the goals of a sustainable energy transition. The objectives considered include (i) minimizing the resources required to obtain a given energy product, (ii) minimizing the total costs of the system under consideration, and (iii) minimizing greenhouse gas emissions. Moreover, the bottom-up optimization approach is applied to different types of systems, namely supercritical carbon dioxide power plants as single systems, renewable energy communities and urban districts as multi-energy systems. To optimize supercritical CO2 power plants, the “SYNTHSEP” methodology is taken up and extended to supercritical systems for the first time. The application considered is waste heat recovery at the industrial level. It turns out that, given the same boundary conditions, bottom-up optimization yields cycle efficiencies up to 5 % higher (in relative terms) than the best values found in the literature (typically obtained with top-down methods). Optimization of multi-energy systems aims to find the optimal configuration of both the conversion and storage plants that meet the final demand of users and the networks that distribute the energy. In the literature, the two aspects are analysed separately or oversimplifications in terms of temporal and/or spatial resolution are applied in order to maintain acceptable computational times. Bottom-up optimization allows the overall problem to be solved accurately and in acceptable time. In fact, this thesis demonstrates that the global optimum of the problem can be found with an error below 1 %, while reducing computational time by up to 60 % compared to traditional methods. In addition, bottom-up optimization makes it possible to optimize large-scale problems that would not be solvable in acceptable time with other methods.