Interests of genetic algorithms to select and optimize scenarios in a system design process

This paper explores the interest and the possibility to join system design and project management methods and tools. Our motivation is to prevent the obvious incompatibilities between technical objectives and socio-economical requirements in the enterprise. What we recommend is to work on a generic unique model based on the classical top down design steps, to which costs models and non-functional requirements are associated. Project management thus appears as an activity of diagnosis and optimisation, allowing to choose certain realisations between the different possible scenarios and to optimise the management by an allocation of tolerances, which is calculated for each supplier on the base of a global objective. This analysis concludes on the interest of two complementary tools; the evolutionary algorithms to arbitrate the scenarios, and the Monte-Carlo methods for the allocation of tolerances.