Measuareent of firm cooling performance in a transonic signle passage model
MEASUREMENTS OF FILM COOLING PERFORMANCE IN A TRANSONIC SINGLE PASSAGE MODEL
by Paul M. Kodzwa, Jr. and John K. Eaton
Prepared with support from General Electric Aircraft Engines
Report No. TF 93
Flow Physics and Computation Division Department of Mechanical Engineering Stanford University Stanford, CA 94305-3035
c Copyright by Paul M. Kodzwa, Jr. and John K. Eaton 2005 All Rights Reserved
Abstract Film cooling is an essential technology for the development of high performance gas turbine engines. A well-designed film cooling strategy allows higher turbine inlet temperatures, improving the engine thermodynamic efficiency. A poorly designed strategy can cause high local temperature gradients, leading to component failures and costly repairs. Hence accurate prediction tools are vital for designers. With the increasing complexity of cooling designs, correlations and incremental design approaches have become outdated, signaling the urgent need for “physics-based” tools that can be coupled to standard modern computational tools, such as commercial computational fluid dynamics (CFD) codes. A glaring problem with the development of this new technology is the lack of well-resolved data with well-defined boundary conditions. Thus, a frequent problem facing model developers is elucidating if differences between experimental data and predictions are due to the experimental data, the applied model, or the applied boundary conditions. The purpose of this experiment to provide highly resolved film cooling performance and heat transfer coefficient measurements of compound angle round holes coupled with realistic gas turbine engine blade geometry and flow conditions. The ultimate goals are: 1) to develop an experimental procedure than can provide timely data for film cooling design; 2) provide full-field surface film cooling data for developing computational models in realistic flows. An experimental two-dimensional representation of the flow field between two modern, transonic turbine airfoil surfaces was used in these tests. This facility, termed as a single passage model, was carefully designed using a heuristic CFD-driven process to match that of an infinite cascade, the most common domain used for performing 2-D CFD simulations of film cooling on modern gas turbine blade geometries. By achieving this goal, the facility provided the identical flow conditions to multi-passage linear cascade, but with substantially reduced costs. Additionally, the simpler overall construction of the single passage allowed the use of steady state, constant heat flux boundary conditions which are more amenable to comparisons with standard CFD prediction techniques. Thermochromic liquid crystals (TLCs) are used to provide full-field surface temperature measurements that can subsequently be used to collect heat transfer coefficient and film cooling effectiveness data. This technique has been proven to be valuable as an evaluation and measurement tool in linear cascades and is thus implemented here. Tiny periscopes
(borescopes) are used for optical access to image the measurement surfaces. Finally, film-cooling effectiveness and heat transfer coefficient results for compound angle round holes inserted in the pressure side surface of a modern blade geometry are presented for various film-cooling flow conditions and hole geometries. This included a range of blowing conditions, density ratios and inlet turbulence ratios. The uncooled heat transfer measurements revealed two interesting results. First, the thermal boundary layer on the aft portion of the airfoil, where the flow accelerates to supersonic conditions, is unaffected by the turbulence intensity at the inlet of the passage. Additionally, these data also suggest that the heat transfer coefficient can depend on the
local surface heat flux boundary condition. This observation was supported by additional numerical and theoretical analysis. This, if true, would be an extremely important observation: it would mean that standard transient heat transfer measurement techniques for transonic flow would have an inherent error, possibly corrupting the subsequent measurements. Furthermore, it raises the importance of carefully matching numerical and experimental boundary conditions, to ensure that the accuracy of numerical models are directly tested. The measured film cooling results indicated two regimes for jet-in-crossflow interaction: one where the jet is rapidly entrained into the local boundary layer, the other where the jet blows straight through the boundary layer. It was determined that the mass flux or momentum flux rate of the jet versus the mainstream flow determines which regime the film cooling jet lies. The effect of varying density ratio and turbulence intensity on film cooling performance was found to be highly dependent on the jet regime.
Acknowledgements We wish to earnestly acknowledge our collaborators and supporters at General Electric Aircraft Engines, without whom this project would not have been possible. Dr. Frederick A. Buck, Dr. Robert Bergholz and Dr. David C. Wisler provided essential insight into the frustrating challenges that affect their business and their desire for better heat transfer prediction tools. We would also like to thank Professors M. Godfrey Mungal and Juan G. Santiago and Dr. Gorazd Medic for their valuable technical advice and expertise during various stages of this project, specifically in the development of the flow facility. Much of the work shown in this thesis would not have been possible without the consistent technical advice and effort from Dr. Christopher J. Elkins. Dr. Elkins always had the ability to appear at crucial stages in this project and bring precious sanity from chaos. Dr. Creigh Y. McNeil, Dr. Xiaohua Wu and Dr. Gregory M. Laskowski provided immeasurable technical during the design phases of this project. Their frequent frank analysis of our research was instrumental to the completion of this project. Many of the components built in the course of this experiment required an extremely high level of machining and fabrication expertise. This was amply provided by Mr. Tom Carver, Mr. Jonathan Glassman, Mr. James Hammer, Mr. Tom Hasler, Mr. Lakbhir Johal and Mr. Scott Sutton. These gentlemen spent an inordinate amount of time, well beyond what they were required, to assist a naive and inexperienced graduate student. We are deeply indebted to their blood and sweat, without which this experiment would have never left the drawing board. During the evolution of this project, several procedural and bureaucratic roadblocks were encountered. Mrs. Amy E. Osugi and Mrs. Marlene Lomuljo-Bautista were invaluable in resolving these issues, and we gratefully acknowledge their support. We wish to express my sincere gratitude and appreciation to the National Science Foundation for their award of a three-year fellowship that initially supported Paul Kodzwa’s tenure at Stanford.
List of Tables
List of Figures
Introduction to Film Cooling and Thesis Objectives . . . . . . . . . . . . .
Approaches to Film Cooling Design and Implementation . . . . . . . . . . .
Computed isentropic Mach number distributions for experimental turbine blade geometry using standard k-ε turbulence model (courtesy of Athans (2000) and Laskowski (2000)). . . . . . . . . . . . . . . . . . . . . . . . . . .
Computed isentropic Mach number distributions for experimental turbine blade geometry using Chen and Kim variant of the k-ε turbulence model (courtesy of Athans (2000) and Laskowski (2000)). . . . . . . . . . . . . . .
2.35 Examination of effect of bellmouth truncation on Mis distribution. . . . . .
2.36 Simplified 3D computational grid with applied boundary conditions. . . . .
2.34 Full and truncated computational domains with applied boundary conditions. 109
2.37 Comparison of Mis distribution at Z = 0.0 (centerline) to 2-D simulation and infinite cascade results. . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.38 Comparison of Mis distributions at Z = 0.0 (centerline), Z = −0.25 and
2.50 Flow system preceding single passage model (from Mukerji and Eaton (2002)).124 2.51 Schematic of integrated diffuser and plenum for the single passage model designed by Mukerji and Eaton (2002). . . . . . . . . . . . . . . . . . . . . .
2.52 Pictures of turbulence grid designed by DeGraaff (2000). . . . . . . . . . . .
2.53 Figure of the overall flow system, showing the exhaust system for the experiment. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.54 Orifice plate runs for measuring boundary layer bleed mass flow rates. . . .
3.11 Probable organization of the cholesteric phase (from Fergason (1966a)). . .
3.12 Variation of wavelength of maximum reflectance and molecular state as a function of temperature for a typical TLC mixture (from Anderson and Baughn (2004) and Parsley (1991a)). . . . . . . . . . . . . . . . . . . . . . .
3.13 Definition of angles φT LC,i and φT LC,s with respect to a TLC-coated surface. 185 3.14 Lighting and viewing angle effects on wavelength of maximum reflectance (derived from Fergason (1968)). . . . . . . . . . . . . . . . . . . . . . . . . .
3.15 Schematic of pressure and suction side copper calibrator pieces. . . . . . . .
3.16 Pressure side copper calibrator installed in single passage model. . . . . . .