TRUSS DIMENSIONING WITH AN UNCERTAINTY REDUCTION PARADIGM
Constraint Programming (CP) over reals are sophisticated evolutions of interval analysis well appropriate to model and propagate uncertainties between design and performance variables during the preliminary design stage. The resulting approximate design space is easy to represent and may bring a better comprehension of the often complex coupling between variables and of the nature of trade-offs. This article details the scenario of an interactive preliminary dimensioning of a truss structure in a CP environment. The interactive nature of communication between the designer and its emerging design, the flexibility for hypothetical reasoning and the synthesis capabilities (back-propagation from performances to design parameters) are revealed.