.GIF)
DRAGON Grid Developed
Predictive Capability Improved for Transition Region of Turbine Vanes and Blades
Heat Transfer Data Obtained in Transition Ducts
Roughness Added to Transonic Axial Compressor Rotor
Mass Flow Removal Used to Control Supersonic Boundary Layer Separation
Interactive Design Software Developed
During the last two decades both structured and unstructured grid techniques have been developed and employed for solving computational fluid dynamics (CFD) problems. These techniques are the two mainstream approaches for dealing with situations where complex geometry imposes great constraints and difficulties in generating grids. (The CHIMERA method, which oversets grids generated separately for resolving individual geometrical or flow features, belongs to the first category.) Both approaches have strengths and weaknesses but can complement each other. Thus, we propose a hybrid grid scheme that maximizes each approach's advantages and minimizes its weaknesses.
.GIF)
As in the CHIMERA method we first divide up the physical domain by a set of high-quality, body-fitted structured grids separately generated and overlaid about a complex configuration. To eliminate any pure data manipulation in interpolating the overlapped region, which does not necessarily follow governing equations, we directly replace only the region of arbitrary grid overlapping with nonstructured grids. Thus, the term "DRAGON grid" is coined. Because the nonstructured grid region sandwiched between the structured grids is limited in size, the increase in memory and computational effort is small. The DRAGON method has three important advantages:
To demonstrate DRAGON's capability, we show various types of grid topologies used for the letters "CFD," together with the resulting DRAGON grid and the evolution of flow over these letters at two different times. Extension to three dimensions is under way. This work has been conducted in collaboration with Dr. Kai-Hsiung Kao of the Institute for Computational Mechanics in Propulsion at NASA Lewis.
Bibliography
Kao, K.-H.; and Liou, M.-S.: An Advance in Overset Grid Schemes: From CHIMERA to DRAGON Grids. 2nd Overset Composite Grid and Solution Technology Symposium, Oct. 25-28, Fort Walton Beach, FL, 1994.
Liou, M.-S; and Kao, K.-H.: Progress in Grid Generation: From CHIMERA to DRAGON Grids. Frontiers of Computational Fluid Dynamics, John Wiley, 1994.
Lewis contact: Dr. Meng-Sing Liou, (216) 433-5855
Headquarters program office: OA
Predictive Capability Improved for Transition Region of Turbine Vanes and Blades
Predicting heat transfer and aerodynamic performance for a turbine blade or vane is a formidable task. High-temperature combustion gases flow with turbulence intensities ranging from 5 to 20% over curved surfaces that experience favorable and adverse pressure gradients. This task is further complicated if a significant portion of the turbine blade or vane flow is in transition between laminar and turbulent boundary layer states. To improve prediction of turbine blade performance, a NASA/university/industry team is evaluating the effects of free-stream turbulence, convex and concave curvature, favorable and adverse pressure gradients, and wakes on transition. The team consists of researchers from NASA Lewis, NASA Ames, NASA Langley, the University of Minnesota, the University of Texas at Austin, Texas A&M, Case Western Reserve University, the University of Toledo, and Dynaflow, Inc.
Experiments on flat surfaces, curved surfaces, and airfoil shapes with and without simulated rotor wakes are being carried out in a number of facilities. Measurements from the experiments are providing benchmark data for investigating fundamental mechanisms, developing models, and checking numerical predictions.
Direct numerical simulation (DNS) analyses are being made to provide a numerical data base for modeling and investigating mechanisms. The experimental data generated in the program aid in validating the DNS results.
.GIF)
Turbulence models are being developed for the numerical prediction of transition heat transfer. The development and assessment of these models are being guided by the experimental and DNS results. Using two-equation turbulence models has in general successfully predicted bypass transition onset. DNS results have shown that a two-equation turbulence model can simulate the nonlinear disturbance growth that produces the first sign of laminar-to-turbulent transition. Single-wire measurements of the boundary layer also agree with the DNS results for disturbance growth.
Experimental measurements made in the transition region indicate the incomplete development of the cascade of energy from large to small scales, pointing to the need for a multi-time-scale k-e equation. Such a model has been developed and shows much promise for the numerical prediction of the transition region.
Some program accomplishments are as follows:
Savill, A.M.: Turbulence Model Predictions for Transition Under Free-Stream Turbulence. Presented at the RAeS Transition and Boundary Layer Conference, Cambridge, England, 1991.
Simon, F.F.: A Research Program for Improving Heat Transfer Prediction for the Laminar to Turbulent Transition Region of Turbine Vanes/Blades. NASA TM-106278, 1993.
Lewis contact: Frederick F. Simon, (216) 433-5894
Headquarters program office: OA
Heat Transfer Data Obtained in Transition Ducts
A continuing objective in gas turbine technology is higher engine efficiency. The resulting higher turbine-inlet temperatures and pressures increase the importance of knowing the gas path surface temperatures and, thus, require accurate knowledge of the heat transfer for design and code validation.
At NASA Lewis we tested several transition ducts in a transient air tunnel (atmospheric inlet and vacuum exhaust) using thermochromic (temperature sensitive) liquid crystals painted on the ducts' complex curved surfaces to produce high-resolution, heat-transfer-coefficient maps. Maps were produced at various inlet Reynolds numbers to 2,400,000; Mach numbers to 0.55; and grid-generated, high free-stream turbulence intensities to over 15%. The geometries included one square-inlet, rectangular-exit duct and three round-inlet ducts (a long superelliptic; a short superelliptic; and a short conical cornered). The figure (for the round-inlet, long superelliptic duct test) shows isotherms (constant-temperature bands produced by the liquid-crystal coating on the surface of the preheated transition duct) at various times during duct cooling. Each isotherm corresponds to a different heat-transfer-coefficient value.
The transient, liquid-crystal heat transfer technique used is based on the classic one-dimensional, semi-infinite-wall heat transfer conduction problem. A transition duct was preheated (not over 66 °ree;C (150 °ree;F)) before allowing room-temperature air to be suddenly drawn through it. The resulting isotherms on the duct surfaces were revealed by using a surface coating of thermochromic liquid crystals, which display distinctive colors at particular temperatures (typically around 38 °ree;C (100 °ree;F)). A videotape was made of the temperature and time data for all points on the duct surfaces during each test. The duct surfaces were uniformly heated by using two heating systems: an automatic-temperature-controlled heater blanket completely surrounding the test duct like an oven, and an internal hot-air loop through the inside of the test duct. The hot-air loop path was confined inside the test duct by insulated heat dams at the duct inlet and exit. A recirculating fan moved hot air into the duct inlet, through the duct, out the duct exit, through the oven, and back to the duct inlet. The temperature nonuniformity of the test duct model wall was kept very small. The heat transfer coefficients could be calculated for the color isothermal patterns produced when the temperature of the air flowing through the duct, the initial temperature of the duct wall, and the surface cooling rate measured at any location with time were known. Although the tests were run transiently, the heat transfer coefficients are for the steady-state case.
.GIF)
This transient, liquid-crystal heat transfer technique is applicable to compound curved surfaces. Because the model rather than the tunnel air is heated, large mass flows of heated air are not required-saving heating costs. The short time for the transient test (2 min), in contrast to running a steady-state tunnel for long periods of time, also saves air-generating costs. This technique should now be considered as possibly a superior alternative to thermocouples for such low-temperature tests as described here because it is nonintrusive, cheaper, and continuous in coverage.
This technique (with unheated tunnel airflow) was first presented by Jones and Hippensteele in 1987 (ref. 1). A grant with Professor C. Camci, of Pennsylvania State University, had the objective of implementing a highly automated heat-transfer-coefficient mapping technique on a digital image-processing system (PC-AT compatible using data translation boards). The digital image-processing system captures color images in the hue-saturation-intensity mode. Data from the square-inlet transition duct were also used by Camci et al. (refs. 2 to 4) in developing the hue-capturing technique for the color video image processing used in the data reduction to calculate the heat transfer coefficients. The ASME awarded reference 5 the Heat Transfer Division Best Paper of the Year for 1994.
References
Roughness Added to Transonic Axial Compressor Rotor
To study the performance deterioration of a high-speed axial compressor rotor caused by an increase in surface roughness, a coating was applied to the rotor blades that produced a surface roughness comparable to that obtained from in-service use. The coating decreased efficiency by 6 percentage points and reduced the pressure ratio across the rotor by 9% at an operating condition near the design point. To assess which portions of the airfoil were most sensitive to roughness variations, the coatings were applied to different portions of the blade surface, and aerodynamic performance measurements were acquired for each coating configuration.
The results at design speed for a constant rotor-inlet mass flow are shown in the graph. The blade leading edge and the first 50% chord of the blade suction surface (configuration D) account for about 90% of the performance degradation observed for the fully coated configuration, E. In addition, comparing coating configurations A and C indicates that the leading-edge region is very sensitive to roughness effects. However, case B shows that the pressure surface is insensitive to roughness variations.
.GIF)
To determine the flow mechanisms responsible for the observed performance deterioration, detailed measurements using laser anemometry-as well as predictions generated by a quasi-three-dimensional, Navier-Stokes flow solver, which includes a surface roughness model-were performed on the baseline blade and the fully coated configuration. The results indicate that the rough coating increases the suction-surface boundary layer upstream of the shock, significantly thickening the suction-surface boundary layer downstream of the shock/boundary layer interaction. This increase in blockage reduces the overall diffusion across the blade passage, thereby reducing the aerodynamic performance of the rotor.
Bibliography
Suder, K.L.; Chima, R.V.; Strazisar, A.J.; and Roberts, W.B.: The Effect of Adding Roughness and Thickness to a Transonic Axial Compressor Rotor. ASME Paper 94-GT-339, 1994.
Lewis contact: Kenneth L. Suder, (216) 433-5899
Headquarters program office: OA
Mass Flow Removal Used to Control Supersonic Boundary Layer Separation
Renewed interest in high-speed civil transport has brought with it concerns about the performance of mixed-compression supersonic inlet systems. Such systems present several key challenges to an inlet aerodynamicist. One of these challenges is the control of their inherent shock wave/boundary layer interactions. These interactions can create boundary layer separations, which can cause unacceptable reductions in inlet performance.
When attacking the problem of boundary layer separation caused by a reflected oblique shock wave impinging on a turbulent boundary layer, inlet aerodynamicists have consistently turned to the technique of mass flow removal. Colloquially referred to as "bleed," the technique uses a suction pressure differential across a porous surface to draw off the low-momentum air associated with the boundary layer. Typically, the porous surface is an array of holes (scaled to the boundary layer displacement thickness) machined into the flow surface. When applied to the shock wave/boundary layer interaction region, bleed controls the separation by increasing the average kinetic energy of the boundary layer, thus helping it overcome the adverse pressure gradient associated with the shock wave. Most of the data on bleed effectiveness were gathered from systems tests of mixed-compression supersonic inlets. Relatively minimal effort has been spent on the basic "why and how it works" level. Because of this imbalance, bleed application tends to be heuristic.
.GIF)
.GIF)
In an effort to provide technology tools for the development of highly efficient supersonic boundary layer bleed systems, NASA Lewis began a program in the 1- by 1-Foot Supersonic Wind Tunnel to explore the fundamental dynamics of bleed mass flow removal. The initial task for this program was to develop a parametric data base describing bleed hole efficiencies for a range of supersonic conditions (Mach 1.3, 1.6, 2.0, and 2.5). Nine configurations were evaluated; the first graph shows the results of round holes inclined 20°ree; and 90°ree; to the flow surface.
The next task was to apply one of these configurations (90°ree; holes) to a reflected oblique shock wave/boundary layer interaction typical of a mixed-compression inlet. For this effort the inviscid impingement point of the shock wave was placed at the center of the bleed configuration. The second graph shows streamwise surface static pressure plots for the interaction with maximum bleed and with no bleed-the first-order measure of goodness being the comparison to the inviscid surface static pressure profile. It is evident that bleeding the boundary layer moves the static pressure profile toward the inviscid (optimal) solution. Characterization of the bleed system dynamics included taking pitot pressure profiles throughout the interaction zone as well as applying a pressure-sensitive paint technique to the interaction region.
This initial effort helped establish a baseline for the program. Future efforts are planned that will further characterize bleed system dynamics by subjecting the bleed region to various flow fields, such as distorted boundary layers. Additional bleed hole efficiency tests aimed at characterizing typical inlet nacelle materials will be performed.
Lewis contact: Brian P. Willis, (216) 433-2176
Headquarters program office: OA
Interactive Design Software Developed
To maintain a competitive edge in the marketplace, American aerospace companies must shorten the time spent on design. In the preliminary design phase the engineer must consider many possible geometric configurations and must accurately estimate or calculate the performance of each to determine trends that lead to an optimum configuration. To assist the design engineer, NASA Lewis has developed a software tool that couples the computing power of a desktop workstation with an interactive graphical user interface (GUI) to allow interactive preliminary design of high-speed inlets.
Output from the software package, called VU-INLET, is shown in the accompanying figure. The inlet shown at the top of the figure must slow the supersonic external flow to subsonic speeds to feed a gas turbine engine at the right with a minimum of drag and with maximum pressure. The flow is turned and slowed through shock waves generated by the inlet geometry. Using VU-INLET, the designer can interactively manipulate the geometry with the GUI sliders and buttons. The software package instantly solves the appropriate flow equations (ref. 1) and displays the computed shock locations and total inlet performance. After a configuration has been developed at the design conditions, the designer can use the GUI to interactively change the free-stream Mach number, altitude, and angle of attack in order to assess off-design operation. This leads to a better, faster, and therefore less expensive design cycle. The user can save the design configuration and performance to a data file by simply pushing a button on the screen.
.GIF)
The latest version of VU-INLET can solve rectangular external or mixed-compression inlets. Because the new package models and displays most phenomena present in supersonic inlets, it can also be used as an educational tool to learn basic inlet operation and design. VU-INLET contains on-line help screens to assist the designer in both operating the tool and interpreting the results.
This tool is a preliminary attempt at using new computational technologies for interactive fluid dynamics. The software was developed in conjunction with an interactive undergraduate educational package (ref. 2) to study shock waves. Other interactive fluid dynamics packages have also been developed to study and design subsonic wind tunnels and coupled turbojet/ramjet propulsion modules and for basic turbojet analysis. All the packages are available to the public and run on either Silicon Graphics or IBM RISC 6000 machines employing the FORMS library of GUI's (ref. 3). Work is under way to port the packages to 486 PC's and Apple computers, using a different GUI package. Technical papers detailing the analysis techniques are available from the author (refs. 2 and 4).
References
Last updated 1995
Responsible NASA Official:
Gynelle.C.Steele@nasa.gov
216-433-8258
Point of contact for NASA Glenn's Research & Technology reports:
Cynthia.L.Dreibelbis@nasa.gov
216-433-2912
SGT, Inc.
Web page curator:
Nancy.L.Obryan@nasa.gov
216-433-5793
Wyle Information Systems, LLC