Modern gas turbine combustion systems are characterised by enhanced air/fuel mixing and this results in more diffused and redistributed hot streaks leading to higher thermal loads established on the vane surface and endwalls. Thus, a detailed aero-thermal characterization of the near-wall region has become crucial both for the analysis of turbine performances and for the subsequent design of special features, such as the use of advanced materials and/or novel efficient cooling concepts. In order to investigate complex combustor-turbine flow interactions high fidelity experimental facilities and numerical tools are required.
In particular, detailed prediction of heat transfer at the gas turbine hot walls is computationally very demanding since it requires fully resolved boundary layers and very fine computational meshes. The present work investigates the capabilities of computationally less demanding near wall treatments (thermal wall functions) to predict heat transfer in gas turbine first stage vanes. This paper first summarizes the recent progress in the implementation of heat transfer capabilities into the CFD solver TBLOCK, by describing the near-wall treatments for forced thermal convection adopted in combination with standard turbulence models and aerodynamic wall functions. In order to assess the capabilities of the flow solver TBLOCK to predict heat transfer under engine realistic conditions an experimental cascade is then modelled numerically. The test case is the new linear cascade built at Oxford’s Osney Thermofluids Laboratory to simulate combustor-vane interactions in gas turbines for power generation; it consists of four first stage vane passages downstream a contracting inlet duct divided in two by a transition splitter that acts as the wall between two can combustors. Both aerodynamic and heat transfer numerical results are compared with available experimental data.
Numerical predictions show good agreement with experiments and well reproduce the aero-thermal influence of the combustor wall, showing the reliability of standard CFD tools in simulating these flow regimes without demanding CPU costs.
