Shape optimization using adjoint variable method for reducing pressure drag and its application to ITBL

Shinohara, Kazunori; Okuda, Hiroshi*; Ito, Satoshi*; Nakajima, Norihiro; Ida, Masato

Keisan Kogaku Koenkai Rombunshu, 10(2), p.629 - 632, 2005/05

To obtain optimal designed shape effectively, a parallel computing method of determining Armijo's line-search step size of the adjoint variable method is proposed. The adjoint valiable method is based on the Lagrange multiplier method (a conditional variational principle), and consists of the state equation, the adjoint equation and the sensitivity equation. The equations for decreasing the fluid drag of shaped surface under a constant volume condition are formulated. To solve the equations effectively by the steepest descent method, the parallel algorithm that finds the Armijo's line-search step size is implemented by using ITBL (the grid system of Japan's shared super computers attached to the network). Using this parallel method, the calculation cost can be reduced.