| Concrete used as an engineering material offers a great deal of flexibility to a potential owner, construction contractor, and design engineer. This flexibility carries with it a burden of difficult design standardization, an increased awareness of constructibility issues unique to concrete, and potential difficulty with optimization. The number of unique combinations of independent variables in structural concrete systems grows exponentially as new variables and elements are introduced, and can easily number 1020,000 or more for a typical concrete building. For concrete structures typically found in practice, structural designers are faced with selecting good solutions from a set often too large to exhaustively search. Reinforced concrete design falls into the general category of engineering design problems in which both continuous and discrete variables are specified, highly nonlinear single or multiple objective functions and constraints can be present in a potentially highly restricted design space. Problems of this sort are very challenging for existing optimization systems. This work seeks to synthesize several optimization and search methods in a manner to effectively solve the most challenging of engineering design problems. The resulting algorithm, FEGA, is successfully tested against numerous highly constrained single and multiple objective optimization problems from the literature and against a representative set of progressively complex reinforced concrete design problems. FEGA often generates superior solutions for a wide variety of problems relative to a more traditional GA that insures feasibility through exclusion of infeasible solutions (D&barbelow;eath P&barbelow;enalty G&barbelow;enetic A&barbelow;lgorithm). The effectiveness FEGA demonstrates is tempered some by the fact that it requires 5–2000 times the number of function evaluations required by DPGA to arrive at its solutions. |
