Ranking of Distribution System's Redesign Scenarios Using Stochastic MCDM/A Procedure

The paper presents the original procedure of solving a multiple criteria stochastic ranking problem consisting in the evaluation of different variants of the distribution system. The problem originates from the analysis and construction of the redesign scenarios of the existing distribution system. The authors develop a computational procedure being a combination of a traditional – deterministic multiple criteria ranking method (e.g. Electre III/IV) and a classification algorithm (e.g. Bayes classifier). The proposed method is composed of six steps, including: stochastic data collection, random selection of deterministic numbers using simulation technique, solving a multiple criteria ranking problem with an application of a deterministic multiple criteria decision aiding/making (MCDM/A) method, the classification of deterministic relations between redesign scenarios (variants) to predefined classes using classification algorithm, the construction of a final ranking of redesign scenarios with an application of a spreadsheet, the recommendation of the compromise solution based on stochastic final ranking of redesign scenarios. The proposed approach is verified on the real-world analysis of the distribution system of goods which operates at the Polish electro-technical market. The results of computational experiments, including: ranking generation, classification and sensitivity analysis are demonstrated. The analysts’ final recommendation of the compromise solution selection is presented. It is based on a comprehensive analysis of the current state of the system, the perspectives of its development, decision maker's preferences, results of the computational experiments and sensitivity analysis.