Comparing MCDM methods for robust decision support in industrial energy system planning

Planning industrial energy systems requires balancing economic, environmental, and technical objectives. Multi-Criteria Decision Making (MCDM) can support this process, yet it remains unclear how strongly the chosen decision method affects the recommended system design. This study provides a systematic robustness comparison of MCDM methods for industrial energy system planning by assessing both the selected system configuration and the sizing of individual technologies under uncertain criteria weights. The analysis is based on a real industrial case study in southern Germany. About 10,000 Pareto-optimal energy system configurations were generated using a multi-objective evolutionary algorithm and a Python-based energy system model. These alternatives were evaluated with seven representative MCDM methods covering additive, multiplicative, value-based, distance-based, compromise-based, and outranking approaches. Criteria weights were derived from expert judgments using the Analytic Hierarchy Process. Robustness was assessed by perturbing the weights through Dirichlet based Monte Carlo simulation and evaluating the results using Stochastic Multicriteria Acceptability Analysis. The preferred energy systems differ only marginally across methods. All methods identify a similar technological core comprising a natural gas boiler, pellet boiler, photovoltaic system, battery storage, and thermal storage. Differences occur mainly in component sizing, especially photovoltaic and thermal storage capacities. The findings indicate that simple compensatory decision methods can provide robust and transparent recommendations. For practitioners, this means that planning efforts can focus on validating preferences, technology assumptions, and key component sizes rather than on selecting a highly complex decision method.