Decision-making models to structure and support urban water supply problems

Abstract

Water supply systems in urban areas may become deficient due to increasing demand, which accompanies population growth but is further jeopardized by equipment aging and problems related to maintenance management as well as water rationing. Moreover, operating the system involves a huge number of issues which a decision-maker must address simultaneously. Therefore, the use of tools to aid the decision process is quite relevant in providing a better grasp of the problem and to generate a recommendation that better meets the wishes of decision-makers. This thesis demonstrates such situations in two cases, treating a water system management and a maintenance management of equipment for water distribution. The first case deals with individual decision-making and a model is developed based on Value-Focused Thinking for problem structuring phase, but requiring only partial information in the multi-attribute analysis by FITradeoff, entailing less effort to achieve the ideal alternative according the manager preferences. It yielded a deep analysis of his reasoning, which was transcribed through the objective’s hierarchy, and reached a solution to the problems of the local water supply system. However, as water resource management problems involve more than one decision-maker, and in the second case this thesis deals with group decision making, particularly Social Choice Theory. In traditional voting procedures, individuals are required to inform their preferences into a single choice or a ranking of alternatives. In this respect, the second case analyze the management of preventive maintenance of water pumps, developing a voting model which individuals’ preferences were described through reliability engineering models and aggregated through quartile-based voting, which had its properties demonstrated. Collected information from individuals may not be enough to represent their real individual wishes. Consequently, the result of the aggregation of preferences may be distorted, and not representative of group preferences. A social choice function considering the intensity of preference over the alternatives has been developed to meet this need. Therefore, this thesis also presents a new voting method in which individuals’ preferences are elicited based on an adaptation of the Simos procedure aiming to evaluate the alternatives and generate scores that are additively aggregated. Simulation and sensitivity analysis are performed, proving easy and intuitive for application to any group decision environment. Moreover, the model is axiomatically characterized and has its limitations presented in terms of Arrovian social choice theory.