An experimental investigation into combined film and internal cooling of turbine blades

Abstract

Gas turbines represent a technology that is the backbone of many significant industries. Increased efficiency of gas turbines can therefore result in reduced emissions across a number of sectors within society. Operation at temperatures in excess of maximum allowable material temperatures has led to sustained improvement in turbine efficiency over the decades, requiring complex cooling systems to protect component parts. Two techniques are traditionally used within turbine blades: internal cooling and film cooling. These techniques are often used in tandem, known as combined cooling. A third technique, transpiration cooling, aims to leverage advances in additive manufacture to produce porous turbine blades to introduce the coolant via pores. This potential future technology presents an opportunity to relax design constraints on cooling systems but is still within its infancy.

Combined cooling techniques are often studied in isolation with relatively little understanding of their interaction. The aim of this thesis is to experimentally investigate combined film and internal cooling. To do this, adiabatic effectiveness (η) is the well-established non-dimensional measure of temperature given to heat transfer, used to evaluate aerodynamic performance of film cooling systems. It is most sensitive to the blowing ratio (M), the ratio of coolant momentum flux to mainstream momentum flux. The blowing ratio quantifies coolant usage for a given system. In addition to adiabatic effectiveness, the matched-Biot technique can be used to make overall effectiveness (θ) measurements. Overall effectiveness is a non-dimensional measure of temperature with heat transfer. By not only matching adiabatic conditions but also the Biot number of the system, engine-representative overall effectiveness values can be captured for combined cooling systems as opposed to isolated film cooling systems. The Biot number is the ratio of conduction through a solid to heat transfer from a fluid to the wall. It is often matched in cooling studies by selecting an appropriate material conductivity to provide engine-representative overall effectiveness measurements. Tuning of thermal conductivity therefore requires measuring thermal boundary conditions of the system for which transient data can be used when paired with numerical models to infer boundary conditions. This is known as solving the inverse problem. The utility in experimental-numerical techniques like these is their flexibility, the difficulty being sufficiently bounding the inverse problem and their sensitivity.
To collect overall effectiveness data, an open circuit wind tunnel with modular test section was fabricated to accommodate several hole geometries. Following a commissioning campaign, adiabatic effectiveness measurements were made downstream of four discrete hole cooling geometries and one transpiration geometry of varying porosity. Temperature data were collected using Infrared Thermography (IRT) to make adiabatic effectiveness measurements above a Rohacell (k=0.04W/mK) test piece. The two cylindrical geometries showed similar separation behaviour, separating at M>0.5, with the shaped holes providing improved performance. Expansion in the lateral direction did little to improve performance, separating at M>0.75, whilst combined lateral and streamwise expansion showed the greatest improvement, showing little signs of separation at M<1.5. Measurements downstream of the transpiration test piece implied that the blowing ratio, standard for discrete hole studies, was not sufficient to adequately compare porosities due to varying open area. The injection ratio (F) was instead used to remove this effect and provide fairer comparison between porosities.

Conductive effects were then considered, first by measuring thermal boundary conditions. Transient temperature data were collected using IRT for a Corian test piece (k=0.86W/mK). The data were input into two inverse models to compare inferred boundaries. The first model, Maximum Likelihood Estimate (MLE), produced highly oscillatory non-physical local distributions but fair area averaged results when compared to pre-existing correlations. The second model, the Bayesian model, produced sensible local distributions for the baseline case of the generic geometry at M=0.5, but produced results that degraded for other cases as they became further removed from the baseline. The results were found to be sensitive to the level of smoothing used within the model, an area requiring further research.

Steady state conductive results were then combined with the adiabatic measurements to infer the influence of film cooled and internal surface conditions. This was done using conductive difference, whereby overall effectiveness results were subtracted from the adiabatic measurements. This showed the dominant influence of the cooling jet in reducing the adiabatic wall temperature for attached cases but reduced influence when separated. Whilst the results provided insight into the relative contributions, the thermal conductivity was not adequately matched to provide engine representative overall effectiveness values. This was corrected for the generic geometry at M=0.5 and M=0.25 using the numerical model developed for the boundary condition measurement. The matched-Biot results showed less variation in overall effectiveness than collected data due to the increased thermal conductivity. The model was also used to derive additional parameters such as surface gradients, surface heat fluxes, and thermal shocks for the generic geometry at M=0.5 and M=0.25. These parameters were proposed to highlight the advantage of inverse model techniques in inferring quantities that would otherwise be difficult to capture.

The investigation aimed to provide insight into combined cooling effects. This was done following the established matched-Biot technique with the development of novel boundary condition measurement capabilities. Future work identified as part of the project includes investigation into smoothing optimisation to determine local distributions of thermal boundary conditions and matching of experimental Biot numbers to provide engine-representative overall effectiveness measurements with additional insights provided by the inverse model.

Date of Award	26 Jun 2024
Original language	English
Awarding Institution	University of Bath
Supervisor	Oliver Pountney (Supervisor), Hui Tang (Supervisor) & Carl Sangan (Supervisor)