Dynamic constraint satisfaction in a bridge design system

Abstract

Constraint satisfaction with numerical constraints is a fundamental problem in automated design. Traditional constraint satisfaction procedures are designed for problems where there is one constant set of constraints. In design, it is often necessary to solve a dynamic constraint satisfaction problem (DCSP) where the set of applicable constraints depends on design choices. Finding a solution in DCSP requires searching among different consistent sets of constraints to find one which has an acceptable numerical solution. In this paper, an algorithm is described which supports the propagation of numerical intervals in such a dynamic environment.