First and second order constraint approximations in structural optimization (Q1092724)

The use of first order constraint approximations is now a common practice in structural optimization. Second order approximations have been avoided because of the perception that the required second derivatives are expensive to calculate. Recent developments that provide inexpensive second derivatives may justify the use of second-order approximations. The present paper compares two commonly used first order approximations to the corresponding second order approximations. Truss and laminated plate problems are used to compare the accuracy of the approximation and its effect on computational efficiency in structural optimization.