Implementation of High FST Turbulent Kinetic Energy Diffusion Model in OpenFOAM Code for Simulation of Heat Transfer over a Gas Turbine Blade

Open Access
Chittlangia, Vedant Rajiv
Graduate Program:
Mechanical Engineering
Master of Science
Document Type:
Master Thesis
Date of Defense:
November 15, 2017
Committee Members:
  • Savas Yavuzkurt, Thesis Advisor
  • turbulence
  • high FST
  • OpenFOAM
  • C3X
  • turbine blade
  • heat transfer
  • low reynolds number model
  • Launder Sharma model
  • diffusion
Several low-Reynolds number turbulence models are compared for their performance to predict the effect of free-stream turbulence on skin friction coefficient and Stanton number. Initial efforts were made to implement a turbulent kinetic energy diffusion model in FLUENT but due to unavailability of the source code the focus was shifted to OpenFOAM, an open-source CFD software. Effect of initial values of free-stream parameters is studied on a flat plate. Launder-Sharma model is modified to account for the effect of free-stream turbulence and length scale by incorporating variable cμ and an additional diffusion term in turbulent kinetic energy transport equation. Comprehensive study is carried for better predictions of skin friction coefficient, Stanton number and turbulent kinetic energy profile on a flat plate for different initial turbulent intensities of 1%, 6.53% and 25.7%. The new model is validated for its better performance on a flat plate to predict Stanton number, skin friction coefficient and turbulent kinetic energy profile with close match to the data points. Prediction of Stanton number shows an improvement of 20% as compared to baseline Launder-Sharma (LS) model at highest turbulent intensity of 25.7% with corresponding 15 % and 38 % improvement in skin friction coefficient and peak turbulent kinetic energy values respectively. The application of the new model is extended by implementing it on C3X stationary turbine vane cascade at low (0.5%) and high (20%) turbulent intensities. Prediction for Nusselt number shows an improvement as compared to other low Reynolds number models. At lower intensity, Nusselt number is predicted similar to the baseline LS model but the suction side measurements shows an improvement of over 20 %. In the real turbine like scenario of high initial turbulent intensity, Nusselt improved by over 8 % on the pressure side and showed 50 % improvement on suction side along with correct capture of the trend. It seems that the additional diffusion term enables the model to capture the physics of high freestream turbulence effects by improving kinetic energy diffusion from the freestream to the near wall region and variable cμ prevents higher values of turbulent viscosity near the wall.