Implicit and explicit sensitivities for optimization of cooled turbine blades

Implicit and explicit sensitivities for optimization of cooled turbine blades Conference

Martin, TJ, Dulikravich, GS. (2002). Implicit and explicit sensitivities for optimization of cooled turbine blades . 10.2514/6.2002-5435

The partial derivatives of field variables (temperatures, temperature gradients, thermal conductivity, heat sources. etc.) and boundary values (heat fluxes, heat transfer coefficients, heat radiation, etc.) with respect to the perturbations of the boundary conditions and geometry are very useful when performing a parametric study of a particular design. These partial derivatives are called design sensitivity coefficients. The implementation of gradient-based algorithms for inverse thermal shape design and numerical optimization require these partial derivatives as part of their operation. But the computer generation of these sensitivity coefficients can use up a great deal of computer resources and processing time. In order to make computer-automated optimization practicable in the design environment of the highly competitive turbomachinery industry, any reduction in resources, time and manpower should be attempted. Implicit differentiation of the aerodynamic, structural and thermal simulation codes can provide a significant reduction in the computer-processing time, resulting in faster turn-around times by the optimization team. It makes the use of optimization in the standard design practices of the company not only viable, but also it can reduce the design cycle time, achieve an optimal design, and make the company more competitive. © 2002 by Thomas J. Martin.