Date of Graduation

12-2025

Document Type

Thesis

Degree Name

Master of Science in Mechanical Engineering (MSME)

Degree Level

Graduate

Department

Mechanical Engineering

Advisor/Mentor

Walters, D.

Committee Member

Leylek, Jim

Second Committee Member

Millet, Paul

Keywords

Computational Fluid Dynamics; Film Cooling; Hybrid RANS-LES; Jet-in-Crossflow; Turbulence Model Validation; Turbulence Modeling

Abstract

Computational prediction of the flow physics associated with gas turbine film cooling is important for ensuring effective heat transfer design. These flow fields are highly turbulent and anisotropic, making the accuracy and efficiency of the chosen turbulence model an important aspect of performing high quality computational fluid dynamics (CFD) based analysis. Current industry standard Reynolds-averaged Navier-Stokes (RANS) turbulence models such as RANS k-ω SST, while computationally efficient, are subject to relatively high levels of uncertainty in their ability to predict these turbulent flow fields. In contrast, high fidelity methods such as Large Eddy Simulation (LES) are computationally very expensive, while displaying relatively lower levels of uncertainty in their results. A compromise between these two methods can be found in hybrid RANS-LES methods, which have been studied in great length in the open literature. These hybrid methods employ RANS modeling in areas with relatively low levels of turbulent fluctuations, and LES in areas with relatively high levels of turbulent fluctuations. Hybrid models do have drawbacks associated with their use and typically require more careful attention from the user to ensure accurate results are obtained. Therefore, the development, improvement, and validation of hybrid models is an area of active research. The goal of this research is to validate a particular hybrid RANS-LES model called the dynamic hybrid RANS-LES (DHRL) turbulence modeling framework for use in gas turbine film cooling applications. Due to the complex, nonlinear, anisotropic, and unsteady turbulence characteristics of film cooling, which is characterized by the canonical jet-in-crossflow, validation is important to lend confidence in the use of the DHRL model for film cooling flowfield prediction. Validation is achieved through replicating previously published, high quality experimental data, and assessing the model’s prediction variable field prediction capabilities. First, in section 2, a normal jet-in-crossflow with negligible temperature difference is investigated to validate the DHRL model for velocity and turbulent kinetic energy variable field prediction. Next, the potential improvement of the DHRL model’s jet-in-crossflow prediction capabilities through the addition of LES resolution in the upstream boundary layer is investigated in section 3, using the same experimental data as the original jet-in-crossflow study. With the DHRL model validated for the simple jet-in-crossflow test case, section 4 investigates the DHRL model’s prediction of adiabatic effectiveness by replicating the engine realistic parameters found in a published experimental study of gas turbine hot section film cooling. For the jet-in-crossflow test case, the DHRL model results are compared to an industry standard RANS k-ω SST model, as well as another hybrid model called the improved delayed detached eddy simulation (IDDES) model. For the film cooling test case, comparisons are again made with the RANS k-ω SST model. Across all test cases, the DHRL model shows better agreement with experimental results than conventional RANS modeling.

Download

Share

COinS