Multi-energy systems (MES) play a key role in solving many significant problems related to economic efficiency, reliability, and impacts on the environment. The multiplicity of goals pursued in the creation of MES gives rise to the problem of multi-criteria choice. The long-life cycle of MES and different development scenarios cause uncertainty in the preferences of decision makers. Focusing on these problems, the article proposes a framework for MES sizing based on multi-criteria optimization and decision-making techniques. Multi-criteria optimization is carried out to find Pareto-optimal MES configurations using the metaheuristic non-dominated sorting genetic algorithm III (NSGA-III). Multi-criteria evaluation of Pareto front alternatives under uncertainty of preferences is performed with fuzzy technique for order of preferences by similarity to ideal solution (TOPSIS). To develop MES that is the most suitable for various scenarios, a new indicator is proposed within the multi-scenario approach, calculated as the geometric mean of fuzzy TOPSIS assessments. The effectiveness of the proposed framework is demonstrated for a remote settlement located on the coast of the Sea of Japan under three scenarios. The geometric mean indicator through the multi-scenario approach identified the MES configuration most suitable for all considered scenarios (levelized cost of energy 0.21 $/kW h (within the interval 0.178–0.275), investment costs 294 289 $(43 573–535 439), CO2 emission 43 008 kg/year (3069–118 542), and unmet load 3262 kW h/year (0–24 044). Furthermore, for the problem being solved, the modified Inverted Generational Distance indicator was used to compare NSGA-III and NSGA-II algorithms. The superiority of NSGA-III over NSGA-II was confirmed (intervals of the indicator estimates are 1874–4040 and 3445–21 521, respectively).
