This paper presents a coupled aerodynamic and thermal study of computer-automated design and optimization of internally cooled turbine blades. The turbine blade, thermal barrier coating, coolant passages and struts were developed from a set of design variables, including β-splines for the coolant wall thickness distribution. The turbine inlet temperature, mass flow rate, and coolant wall roughness were also incorporated into the design variable set. The maximum temperature in the metal blade was enforced with equality or inequality constraint functions. Because the coolant flow rate was a design variable, this function could not be explicitly minimized. Instead, three different thermal objective functions were studied: uniform temperature, heat flux extremum, and minimum coolant ejection temperature. Results have shown that it is possible to increase or maintain high turbine inlet temperatures while decreasing the turbine blade coolant requirements. A new constrained hybrid optimization algorithm was developed and used to modify the turbine blade designs until an optimum design was found. This evolutionary optimization package incorporated four popular algorithms (steepest descent, genetic, simplex, and simulated annealing) with automatic switching among them. A computational heat conduction analysis, using the boundary element method (BEM), was iteratively coupled to an unstructured finite volume Reynolds-averaged Navier-Stokes CFD analysis for turbulent hot gas flow. A quasi-one-dimensional system with heat addition and friction was iteratively coupled to the BEM heat conduction via heat flux for the simulation of the airflow in the serpentine coolant passages. This quasi-one-dimensional system yielded correlations for the heat convection coefficients on the coolant passage walls. The coolant passage pressure loss was one of the quantities arising from the quasi-one-dimensional analysis.

This content is only available via PDF.