Action Filename Description Size Access License Resource Version
Show more files...


Product end-of-life treatment is unavoidable, no matter how well the product is designed. European regulations becoming stringent regarding the disposal of products, emphasis on end-of-life (EOL) issues at the conceptual design stage gained in importance. The goal of this thesis is the development of a framework for the integration of multicriteria decision aid methods (MCDA) at the conceptual design stage (which is characterized by design imprecision), where the EOL of the products being designed should already be considered. To achieve this goal, two major objectives are targeted: (1) the development of a decision model and (2) development of a MCDA method, constituting the core of the decision model. For the first objective, a bottom-up approach was considered, resulting in a three-level decision model, each level ranking the following three types of alternatives: EOL options, EOL scenarios and design concepts. The model contains the definition of the criteria, of the alternatives and explains how their performances with respect to criteria are calculated. To fulfill the second objective, a new fuzzy outranking method is proposed. The method can successfully handle design imprecision. Each alternative is characterized by a number of design variables (DVs), each modeled by a fuzzy set that incorporates the decision-maker's (designer) preferences for a range of values. The performances of each alternative with respect to each criterion are functions of DVs, and consequently they are modeled by corresponding fuzzy sets. With the aid of our fuzzy outranking relation, each pair of alternatives is compared and a degree of weak preference is obtained. Fuzzy preference structures (FPS) are then constructed, consisting of a triple of more subtle fuzzy relations: fuzzy strict preference P, indifference I and incomparability J. We aggregate these relations over the set of criteria using aggregation operators, capable to model interaction between criteria. By selecting appropriate choice functions the alternatives can be ranked, where indifference and incomparability are allowed. The FPS provides an increased power of discrimination between pairs of alternatives. Besides the ability of working with imprecise performances, the method allows decisions to be taken without being constrained to aggregate performances with different measuring units (no need to "mix apples with oranges"). The theoretical developments in this thesis are implemented in a tool that can help a concurrent design team (playing also the role of decision-makers) to perform detailed analysis at the conceptual deign stage, for finding the trade-off between two dimensions: economic and environmental. The contributions of this thesis are both theoretical and methodological. The new fuzzy outranking relation captures decision-maker's (DM's) attitude, has an intuitive understanding, is simple to integrate in computation modules and its degree of accuracy can be set according to the needs. We showed, by means of various examples that there is no right method for taking good decisions, but the flexibility we introduced via a number of parameters (DM's attitude, choice functions measuring the strength and the weakness, etc.) helped understanding and justifying the results of the decision process. Aggregating single criterion preferences has been successfully performed using an aggregator capable to model the interaction between criteria, which allowed the selection of alternatives without weaknesses. The methodological contributions are related to the practical aspects of evaluating imprecise design alternatives (design concepts) having embedded EOL aspects. Unlike existing methods that considers detailed designs, the proposed decision model can be used at the early design stages. The theoretical developments allowed putting together the trio EOL process – EOL strategy – Design concept, such that one can estimate for example the recycling potential of a design concept under given circumstances or pinpoint design weaknesses through a backward analysis.