Aerodynamic Redesigning of an Industrial Gas Turbine

This paper deals with the aerodynamic redesigning of a four-stage heavy-duty gas turbine. Traditional design tools, such as through-flow methods, as well as more sophisticated tools, such as three-dimensional RANS computations, were applied in subsequent steps according to a given hierarchical criterion. Each design or analysis tool was coupled with modern optimization techniques, and the overall redesign procedure relies on a neural-network-based approach aimed at maximizing the turbine's power output while satisfying geometrical and mechanical constraints. A detailed description of the redesign procedure is provided, and the aerodynamic characteristics of the optimized geometry are discussed and compared to the original ones.