Impact of Predicted Combustor Outlet Conditions on the Aerothermal Performance of Film-Cooled HPT Vanes

Abstract

Turbine inlet conditions in modern aero-engines employing lean-burn combustors are characterised by highly swirled flow and non-uniform temperature distributions. As a consequence of the lack of confidence in numerical predictions and the uncertainty of measurement campaigns, the use of wide safety margins is of common practice in the design of turbine cooling systems, thus affecting the engine performance and efficiency. Previous experiences showed how only scale-resolving approaches such as Large-eddy and Scale-adapting simulations are capable of overcoming the limitations of RANS, significantly improving the accuracy in the prediction of flow and temperature fields at the combustor outlet. However it is worth investigating the impact of such differences on the aerothermal performance of the NGVs, as to understand the limitations entailed in the current approach for their thermal design. Industrial applications in fact usually rely on 1D, circumferentially-averaged profiles of pressure, velocity and temperature at the combustor-turbine interface in conjunction with Reynolds-averaged Navier-Stokes (RANS) models. This paper describes the investigation performed on an experimental test case consisting of a combustor simulator equipped with NGVs. Three numerical modelling strategies were compared in terms of flow field and thermal loads on the film-cooled vanes: i) RANS model of the NGVs with inlet conditions obtained from a RANS simulation of the combustor; ii) RANS model of the NGVs with inlet conditions obtained from a Scale-Adaptive Simulation (SAS) of the combustor; iii) SAS model inclusive of both combustor and NGVs. The results of this study show that estimating the aerodynamics at the NGV exit does not demand particularly complex approaches, whereas the limitations of standard RANS models are highlighted again when the turbulent mixing is key. High-fidelity predictions of the conditions at the turbine entrance proved to be very beneficial to reduce discrepancies in the estimation of local adiabatic wall temperature of even 100 K. However, a further leap forward can be achieved with an integrated simulation, capable of reproducing the transport of the unsteady fluctuations generated in the combustor up into the turbine, which plays a key role in presence of film cooling. This work therefore points out how keeping the analysis of combustor and NGVs separate can lead to a significantly misleading estimation of the thermal loads and ultimately to a wrong thermal design of the cooling system.

External heat transfer on nozzle guide vanes under highly swirled combustor outlet flow

“…For brevity, only the computational domain and mesh of the NGVs are reported in Fig. 2, however, the grid characteristics and setup of boundary conditions for both simulation cases (NGVs and combined combustor-NGVs) are inherited from Cubeda et al (2018) and Andreini et al (2015Andreini et al ( , 2017.…”

“…Andreini et al (2017) have compared SAS results with the experiments carried out by Bacci et al (2018a,b) on the EU Project FACTOR ('Full Aerothermal Combustor-Turbine interac-tiOns Research') hot streak generator, which replicates a lean-burn combustor outlet distortion field. The success of SAS in reproducing the coolant/mainstream mixing process more accurately than RANS and, ultimately, the impact on nozzles thermal load and film-cooling adiabatic effectiveness has been lately described by the authors (Cubeda et al, 2018).…”

“…Impact on the NGVs' aerothermal performance The authors (Cubeda et al, 2018) have found that the aerodynamic field is well captured even by a RANS approach, i.e. without the need for an integrated Scale-adaptive simulation on the combustor-turbine module.…”

Unsteady Phenomena at the Combustor-Turbine Interface

“…This is not simply high turbulence, but coherent structures such as the swirl core, which persist over time, but vary in position. Many of these flow interactions are highly unsteady in nature, and are not picked up by steady state CFD, or even unsteady CFD if it does not have the correct time-varying inlet conditions [27]. The ordinary view of secondary flows in the NGV as being essentially steady structures is shown to be oversimplified.…”

“…Cubeda et al [27] compared different numerical methods to the experimental measurements of Bacci et al [25,24], finding that neither RANS of combustion chamber and vanes, nor SAS of the combustion chamber used as a boundary condition to steady RANS gave satisfactory results. RANS simulation gave local adiabatic temperature errors exceeding 150K.…”