Entropy Generation Rate for Profiled Endwall Design in Turbines

This thesis investigates the use of entropy generation rate as a design variable for Profiled Endwalls (PEW) to reduce secondary loss in turbines. Entropy generation rate is a measure of the local loss production in the machine, therefore any reduction in this variable leads to a reduction of the losses at their source. To that end, a numerical investigation was conducted to calculate the entropy generation rate via Computational Fluid Dynamics (CFD) by solving the Reynolds-Averaged Navier-Stokes Equations (RANS) in three dimensions, for the so-called Durham Cascade. This was part of a PEW Design System optimisation that used the entropy generation rate as a design variable in a Genetic Algorithm (GA) coupled with CFD validated against experimental measurements. After 3000 evaluations and 48000h CPU using the Hamilton High Performance Computing Service, the result was a new PEW denominated E2 that reduced the predicted entropy generation rate by 9.7% compared to the planar case (P0). An experimental campaign that consisted of axial traverses using a 5-hole probe, confirmed that the E2 reduced the stagnation pressure loss coefficient by 0.0241 compared to P0. Two loss reduction mechanisms were identified: reduced vortex interaction of the suction side horseshoe vortex (SSHV) with the pressure side horseshoe vortex (PSHV); and delayed blade suction side boundary layer separation. The first use of entropy generation rate as a design variable for iterative design optimisation has been explored and its use recommended for PEW design.