Flow-heat topology optimization of internally cooled high temperature applications using a voxelization approach for domain initialization

A method is presented for obtaining topology optimized designs for internally cooled high temperature applications, using a flexible geometry description, by means of a voxelization methodology and a novel boundary detection algorithm. A conjugate heat transfer approach is taken; the physics is described by a Stokes-Brinkman model for the flow, weakly coupled with a convection-diffusion model for the heat transfer. A practically relevant optimization formulation, consisting of a maximum temperature objective with a mass flow constraint, is used, and applied to an industrial-relevant non-trivial geometry resembling a guide vane in a gas turbine. Temperatures and velocities from the optimized design are compared with the response from a Stokes flow model with body-fitted mesh and a high-fidelity Reynolds-averaged Navier-Stokes model. A comparison of the performance from a mixed and a penalty approach for solving the flow problem is included. The voxelization approach shows good promise for handling complex design domains.