Abstract : In the early phases of a traditional design process, many decisions are often made by designers. For that purpose, they take advantage of their experience and company knowledge. These decisions are necessary in a sequential design process but may hide many embodiment solutions. Moreover, designers often use a time consuming trial-and-error mode to find a working combination of standard elements.
To overcome these difficulties, a decision support system based on constraint programming is proposed. The object of the design process set out in this work is to facilitate the embodiment design phase by avoiding a-priori decision making and searching for feasible architectures in which all the points of view of the various participants in the project are taken into account. The main stages are: i) the writing of the design problem into a Constraints Solving Problem (CSP) form, ii) the exhaustive search for the feasible architectures, iii) the exploitation and the reduction of the solutions space to help in the decision making.
A preliminary analysis of the design problem is necessary for the stages (i) and (iii). On one hand, it is essential to restrict the design problem to the only necessary and sufficient characteristics for the embodiment design phase, which we name structuring characteristics. On the other hand, it is necessary to express the design objectives and the qualification criteria of the design, in order to sort the embodiment solutions being obtained.
An systematic four step method is detailed in this thesis
Our approach is confronted with the 'classic' method of a designers group, for the same embodiment design.
This methodology is applied to the embodiment design of an automatic weight-winding system of a monumental clock. To determine the most interesting solutions, objectives and performance indicators are entered. Finally, the benefits of our approach are discussed.