Abstract
Shape annealing is a design technique that combines grammatical transformation with simulated annealing to generate optimally directed discrete structures. The primary advantage of shape annealing over other discrete methods is the introduction of new joints and members, using a shape grammar, to enable simultaneous optimization of topology, shape, and sizing. This approach makes the exploration of truss topologies that are not constrained to a pre-defined grid possible. While details of the method have been presented (Reddy and Cagan, 1995; Shea et al., 1997), this paper provides an overview of the current implementation and validation through a comparison of results for least-weight design problems to both grid-based solutions and previously reported shape annealing solutions. These comparisons illustrate the capability of shape annealing in generating optimal designs and the advantages of not restricting topology optimization to a grid-based design universe. Comparing new results to those previously reported shows improvement in solution quality and extensions for practical multiobjective design. Finally, solutions to a dome problem illustrate the inherent ease in extending the method to three-dimensional design and multiple loading conditions. These examples demonstrate that shape annealing is effective for exploring topology design of truss structures for both least-weight and multiobjective design problems.
