Prediction of Lapse Rate in Low Pressure Turbines With and Without Modeling of the Laminar-Turbulent Transition

Abstract

Although turbo-machinery main stream flows are predominantly turbulent, the low pressure turbine airfoil surface boundary layer may be either laminar or turbulent. When boundary layer flow is laminar and passes through a zone of adverse pressure gradient, bypass or separation transition can occur via the Tollmien-Schlichting or Kelvin-Helmholtz instabilities. As the gas turbine’s low pressure turbine operating condition changes from sea level take-off to the altitude cruise, Reynolds number is significantly lowered and the turbine’s performance loss increases significantly. This fall-off in performance characteristic is known as lapse rate. Ability to accurately model such phenomenon is a prerequisite for reliable loss prediction and essential for improving low pressure turbine designs. Establishing such capability requires the validation and evaluation of existing low Reynolds number turbulence models, with laminar-turbulent transition modeling capability, against test cases with measured data. This paper summarizes the results of evaluating and validating two 3D viscous “RANS” Reynolds-Averaged Navier-Stokes programs for two test cases with test data. The first test case is the ERCOFTAC’ flat plate with and without pressure gradient, and the second is a Honeywell three-and-half-stage low pressure turbine with available test data at high and low Reynolds number operations. In addition to evaluating the CFD codes against test data, the flat plate test cases were used to establish the meshing and modeling best practice for each code before performing the validation for the Honeywell multistage low pressure turbine. The RANS CFD programs are Numeca’s Fine Turbo and ANSYS/CFX. Numeca’s Fine Turbo employs a two-equation K-ε turbulence model without laminar-turbulent transition modeling capability and the one-equation Spallart-Allmaras turbulence model with laminar-turbulent transition modeling capability. The ANSYS/CFX, on the other hand, employs a two-equation K-ω turbulence model (AKA SST or shear stress transport) with ability to model laminar-turbulent transition. Predictions of the CFD codes are compared with test data and the impact of modeling the laminar-turbulent transition on the prediction accuracy is assessed and presented. Both CFD codes are commercially available and the evaluation presented here is based on users’ prospective that targets the applicability of such predictive tools in the turbine design process.