A Metamodelling Optimization Approach to a Wing Spar Design

Optimization is present in almost every engineering activity. There are several different approaches depending on the type of problem and its related complexity. A search for the optimum solution might reveal itself hard. When the output cannot be expressed by a function of the inputs, analytical optimization methods are not suitable. Another problem arises when there is the possibility of local minimums existence. The search for a global minimum can be pursued using global optimization approaches. On this paper a metamodeling global optimization approach is explained and applied to a real engineering designing problem. The algorithm is programmed in MATLAB and calls for an FEM software to run the models. The solution is compared to one attained from a FEM commercial software package. Final results show that the metamodeling method outputs a better solution and increases the guarantee of having achieved a global minimum. The algorithm here developed has direct application to problems that deal with FEM analyses with all design variables perfectly bounded. The optimization procedure is fully autonomous.