Labyrinth Seal
Labyrinth Seal
Executive summary
Performance evaluation of gas turbine labyrinth seals using computational fluid dynamicss
Report no. NLR-TP-2007-624 Author(s) B.I. Soemarwoto J.C. Kok K.M.J. de Cock A.B. Kloosterman G.A. Kool J.F.A. Versluis Report classification UNCLASSIFIED Date September 2007 Knowledge area(s) Computational Physics & Theoretical Aerodynamics Gasturbine Technology Descriptor(s) Labyrinth seal Gas turbine CFD
Problem area Increasing performance requirements on gas turbines have led to ever increasing gas temperatures and pressure ratios. Together with the resulting increase in cooling flows, this requires internal gas leakages to be minimized and controlled even further. The application of new seal designs and improved understanding of leakage flow characteristics are of particular importance in order to meet future performance goals. Labyrinth seals, despite their unavoidable leakage, remain the most common seal types in modern gas turbines. The combination of high rotational speeds, high gas
temperatures and high pressure differences often precludes the use of other seal types, and the labyrinth seal remains the only choice to meet reliability and durability standards of modern engines. Understanding the flow in labyrinth seals at engine compliant conditions is fundamentally important in developing improved seal concepts to enhance and predict component performance in gas turbines engines. Obtaining such an understanding is part of a mission to bring new gas turbine labyrinth seal concepts to a higher level of technology readiness for advanced propulsion and power systems. Simulation methods based on Computational Fluid Dynamics
This report is based on a presentation held at the ASME Turbo Expo 2007: Power for Land, Sea and Air, Montreal (Canada), 14-17 May, 2007. UNCLASSIFIED
UNCLASSIFIED
Performance evaluation of gas turbine labyrinth seals using computational fluid dynamicss
(CFD) have gained significant interest in recent years. Apart from its potential to reduce risks involved during the development process of a new seal, CFD allows one to gain a thorough insight into the flow processes inside the seal. The continuing development of CFD algorithms, facilitated by the ever increasing performance and costeffectiveness of computers, opens ample possibilities to incorporate more complete physics into the CFD modeling. Description of work The focus of the paper lies in the application of CFD to assess labyrinth seals. The ReynoldsAveraged Navier-Stokes equations are employed as the flow governing equations, with turbulence incorporated through the TNT variant of the k- turbulence model. The first objective of this paper is to perform a comparison between numerical and experimental results, involving CFD computations for a three-dimensional labyrinth seal with a honeycomb land that closely represents a tested seal model. Subsequently, the computational results facilitate a three-dimensional investigation to get a good insight into the flow phenomena in the labyrinth seal. The second objective is therefore to present a new approach for exploiting the CFD methodology to investigate the flow characteristics and sealing mechanisms in terms of losses generated in the labyrinth seals.
The third objective is to demonstrate the suitability of the approach for assessing different seal concepts. Results and conclusions The comparison between the results of the CFD computation and the experimental measurement shows a close agreement. This is a strong indication that the important flow physics responsible for the sealing mechanism have been captured by the CFD method. An analysis of the flow topology and the losses experienced along the streamlines reveals that the sealing mechanism of the flow through labyrinth seals is due to losses associated with turbulent mixing and shear which dominantly occur in the vicinity of the knife edge and to a lesser extent around the step wall. The effects of the honeycomb cells can be identified as decreased production of losses. Applicability The CFD method is considered as an analysis tool complementary to rig-testing and enables investigating the effect of new seal design features. Additionally CFD is seen as a tool to support the correct representation of test-data in semiempirical engineering models for seal design. An industrial perspective is foreseen towards an extensive exploitation of the CFD modeling capabilities for real-life design of seals.
Nationaal Lucht- en Ruimtevaartlaboratorium, National Aerospace Laboratory NLR Anthony Fokkerweg 2, 1059 CM Amsterdam, P.O. Box 90502, 1006 BM Amsterdam, The Netherlands Telephone +31 20 511 31 13, Fax +31 20 511 32 10, Web site: www.nlr.nl
UNCLASSIFIED
NLR-TP-2007-624
Performance evaluation of gas turbine labyrinth seals using computational fluid dynamicss
B.I. Soemarwoto, J.C. Kok, K.M.J. de Cock, A.B. Kloosterman, G.A. Kool and J.F.A. Versluis1
1
This report is based on a presentation held at the ASME Turbo Expo 2007: Power for Land, Sea and Air, Montreal (Canada), 14-17 May, 2007. The contents of this report may be cited on condition that full credit is given to NLR and the authors. This publication has been refereed by the Advisory Committee AEROSPACE VEHICLES. Customer Contract number Owner Division Distribution Classification of title Approved by:
Author Reviewer Managing department
National Aerospace Laboratory NLR ---National Aerospace Laboratory NLR Aerospace Vehicles Unlimited Unclassified July 2008
NLR-TP-2007-624
Summary
The paper presents an investigation on the characteristics of flow through labyrinth seals. The focus of the paper lies in the application of the Computational Fluid Dynamics (CFD) methodology. The Reynolds-Averaged Navier-Stokes equations are employed as the flow governing equations. Turbulence is incorporated through a variant of the two-equation k- turbulence model. Three test cases are considered. The first test case concerns a labyrinth seal configuration with a honeycomb land. The computational results are compared to those obtained from seal test rig measurements. The second test case addresses the same labyrinth seal where the honeycomb land is replaced by a solid smooth land. The third test case addresses the flow through a labyrinth seal with canted knives. The CFD method is considered as an analysis tool complementary to rig-testing and enables investigating the effect of new seal design features. Additionally CFD is seen as a tool to support the correct representation of test-data in semi-empirical engineering models for seal design. An industrial perspective is presented towards the exploitation of these modeling capabilities for real-life design of seals.
NLR-TP-2007-624
Contents
1 2 3 4 5 6 7 8 9
Introduction Labyrinth seal test cases CFD method and setup Experimental seal-test-rig facility Comparison of results Analysis of Case-1 results Analysis of Case-2 results Analysis of Case-3 results Industrial Perspective
7 8 9 10 11 12 14 15 16 16 17 18
NLR-TP-2007-624
Abbreviations
A Agap p pi po pt pt,ref pt,loss T Ti u W M inlet area (m2) circumferential gap area of the first knife (m2) pressure (N|m2) inlet pressure (N|m2) outlet pressure (N|m2) total pressure (N|m2) reference total pressure specified at the inlet (N|m2) total pressure loss (N|m2) temperature (K) inlet temperature (K) axial inlet velocity (m|s) mass flow (kg|s) Mach number density (kg|m3) mass flow function (kg K1/2|N|s) pressure ratio ratio of specific heats, = 1.4
NLR-TP-2007-624
NLR-TP-2007-624
1 Introduction
Increasing performance requirements on gas turbines have led to ever increasing gas temperatures and pressure ratios. Together with the resulting increase in cooling flows, this requires internal gas leakages to be minimized and controlled even further. The application of new seal designs and improved understanding of leakage flow characteristics are of particular importance in order to meet future performance goals. Labyrinth seals, despite their unavoidable leakage, remain the most common seal types in modern gas turbines. The combination of high rotational speeds, high gas temperatures and high pressure differences often precludes the use of other seal types, and the labyrinth seal remains the only choice to meet reliability and durability standards of modern engines. Understanding the flow in labyrinth seals at engine compliant conditions is fundamentally important in developing improved seal concepts to enhance and predict component performance in gas turbines engines. Obtaining such an understanding is part of a mission to bring new gas turbine labyrinth seal concepts to a higher level of technology readiness for advanced propulsion and power systems. To accomplish such a mission, three technological blocks can be identified at NLR consisting of (i) an experimental method facilitated by an advanced seal test rig, (ii) a numerical method based on Computational Fluid Dynamics (CFD), and (iii) a semi-empirical engineering model. The seal test rig (Ref. 1) has the advanced capability to simulate flows through a highly representative seal model at extreme engine conditions such as high temperatures, high rotational speeds and high pressures to which prospective advanced seals are exposed. Additionally, the test data from such a seal test rig support the enhancement and validation of engineering model and CFD methodology. Simulation methods based on CFD have gained significant interest in recent years. This trend is reflected in the literature. However, previous works were limited to simplified cases, where important geometrical features (such as a complete description of the honeycomb cells) and/or flow conditions (such as the rotation) were not included. For example, Ref. 2 presented CFD computations on a simplified rotating knife on a smooth land, i.e. without honeycomb cells. Ref. 3 presented an approach to include the effects of honeycomb cells by introducing roughness effects on the smooth land. A complete geometrical representation of honeycomb cells was considered in Ref. 4, but the configuration did not include the rotating knife. Ref. 5 proposed an approach of modeling the 3-D honeycomb cells by circumferential grooves. Refs. 6 and 7 addressed a complete configuration consisting of the knives and honeycomb cells, but did not take rotation into account. Ref. 8 considered realistic rotating knife geometries, but excluded the honeycomb cells. Apart from its potential to reduce risks involved during the development process of a new seal, CFD allows one to gain a thorough insight into the flow processes inside the seal. The continuing development of CFD algorithms, facilitated by the ever increasing performance and cost-effectiveness of computers, opens ample possibilities to incorporate more complete physics into the CFD modeling. It should be emphasized that the models and conditions tested in the NLR seal test rig facility are highly representative to real-life situations. The facility provides the mass flow value (i.e. the gas leakage) as the primary data for given conditions such as rotational speed, pressure ratio, inlet pressure and inlet temperature. Such data is not suitable for validation of a CFD method on the level of flow field properties such as velocity profiles, pressure distributions, etc., occurring inside the flow.
NLR-TP-2007-624
However, a comparison in terms of the mass flow and pressure ratio between the numerical and experimental data is still useful, provided that the same geometrical details of the seal model in the test rig are also present in the computational domain, while the underlying flow model contains important effects from compressibility, viscosity, turbulence and vortical flow structures. A close agreement between the numerical and experimental data can then be interpreted as an indication that the flow physics occurring in the test are captured by the CFD method. It should therefore increase confidence about the applicability of the method. The first objective of the paper is to perform such a comparison, involving CFD computations for a three-dimensional labyrinth seal with a honeycomb land that closely represents a tested seal model. In contrast to previous works, this paper addresses a case where the honeycomb cells and the rotation are taken into account. Subsequently, the computational results facilitate a three-dimensional investigation to get a good insight into the flow phenomena in the labyrinth seal. The second objective is therefore to present a new approach for exploiting the CFD methodology to investigate the flow characteristics and sealing mechanisms in terms of losses generated in the labyrinth seals. The third objective is to demonstrate the suitability of the approach for assessing different seal concepts. The achievement of complete technological building blocks provides a comprehensive capability not only important to the aircraft engine Original Equipment Manufacturers (OEMs) but also to companies involved in the development and production of seals. The importance lies with achieving better engine and component performance through the reduction of seal leakage at similar or reduced production cost. Accordingly, an industrial perspective towards exploiting the achieved capability is presented.
NLR-TP-2007-624
NLR-TP-2007-624
presence of the honeycomb requires a three-dimensional computational domain which strongly increases the number of grid cells compared to the two-dimensional case of Case-2. A grid refinement study results in a number of grid cells of about 10 million. This grid can sufficiently capture the important flow physics with high gradients around the knife-edge and in the region between the knife and the step. It also covers two buffer zones. One zone is upstream of the first knife to allow the swirl to develop. Another zone is downstream of the third knife to fully capture a large vortex, ensuring the consistency of the boundary condition at the outflow boundary. Figure 4 gives an impression of the grid.
10
NLR-TP-2007-624
5 Comparison of results
For comparison with the experimental data, CFD computations have been performed for the seal-test-rig conditions and measured data shown in Table 1. The computational procedure can be outlined as follows. The reference pressure and temperature are selected to be equal to the experimental values of the inlet pressure and temperature. The reference density follows from the perfect gas relation. The reference velocity u is obtained from the experimental value of the mass flow W using the relation:
W = uA
(1)
where A is the inlet area. These reference quantities are used to determine the reference Mach number and Reynolds number required as input by the flow solver, and to scale the flow variables as the flow solver operates with non-dimensional variables. For each computation, Table 1 provides: a Rotation-Per-Minute (RPM) value that determines the non-dimensional circumferential velocity on the rotating boundary, a mass flow and inlet temperature to determine the non-dimensional momentum and temperature on the inflow boundary, a pressure ratio for the specification of the non-dimensional outlet pressure on the outflow boundary. The primary output of the computation is the inlet pressure pi on the inflow boundary. The numerical values of the mass flow function can then be determined as:
W Ti pi Agap
(2)
pi po
(3)
The computational results for the three entries of Table 1 lead to Figure 6, showing a comparison between the experimental data and CFD results. Since the experiment on the test rig and the CFD computation only share the same mass flow, inlet temperature, outlet pressure and rotational velocity (i.e. the quantities appearing in the boundary conditions), the percentage shown in the figure is the maximum of the individual deviations in and . It should be noted that the test rig has an accuracy of 3.5% for the mass flow function. Also, there are uncertainties in the CFD setup due to the assumptions on the knife-edge corners and the boundary conditions. Within this measurement accuracy and CFD setup uncertainties, the
11
NLR-TP-2007-624
shown deviation of about 1% represents a close agreement between the CFD results and the experimental data.
1 2 1 p t = p 1 + M 2
The total pressure loss is defined in normalized form as:
(4)
pt ,loss = 1
pt pt ,ref
(5)
where the reference total pressure pt ,ref corresponds to the inlet condition. Figure 7 presents local views of the cavities between two adjacent knives on the mid-plane and side-plane. The geometrical difference between these planes lies in the relative position of the knife edge with respect to the honeycomb cell. On the mid-plane the knife edge is positioned below the cell-mouth, while on the side-plane the knife edge is positioned below the cell wall. For all cavities the span of the contour levels are the same, but the minimum and maximum values are adjusted locally to those occurring within the cavity. There is no significant difference in the pattern of the total pressure loss in the cavities on the different planes. Each knife edge appears to produce the same amount of losses. Only the orientation, the width and the length of the jet (and therefore the gradient of the losses), and the location with the highest gradient of the losses are different between the planes. These characteristics are determined by the effective gap orientation and width between the knife edge and the cell wall. Figure 8 shows in-plane streamlines on the mid-plane and side-plane with static pressure contours on the background. It should be noted that these streamlines are not the projection of three-dimensional streamlines. They are constructed using the radial and axial components of the velocity only. There is no significant difference in the pattern of the streamlines, except in the vicinity of the knife-edge reflecting the geometrical differences of the position of the knife edges relative to the honeycomb cells. Near the base, two clearly isolated recirculating regions can be observed in the first cavity and one recirculating region in the second cavity. This can be attributed to the different aspect ratios between the two cavities.
12
NLR-TP-2007-624
A more elaborate assessment is performed by observing the losses along three-dimensional streamlines. The streamlines are constructed by integration of the velocity vector. Figure 9 shows ten streamlines originating from a rake drawn between the two side boundaries upstream of the first knife-edge. As the flow is steady, the streamline coincides with the particle path (i.e. the streampath). The streamlines are colored with the total pressure loss. This figure shows the different paths taken by the streamlines passing through the cell-mouth and by those passing through the gap. These paths indicate that the effective distance between the knife-edge and the step may be larger than its geometrical (nominal) step distance. The nominal step distance refers to the distance between the knife-edge and the step in the case of a smooth land (Case-2). The streamline passing through the cell-mouth exhibits a stronger rotational effect. A particle on this streamline travels circumferentially farther to cover the nominal step distance. In contrast, the streamline passing through the gap exhibits a weaker rotational effect and is more axially aligned. Because of the honeycomb structure, a particle on this streamline travels a distance that is about a half-cell width larger than the nominal step distance. One of such streamlines is indicated by an arrow in Figure 9. Figure 10 shows two streamlines spanning throughout the labyrinth seal configuration. The first streamline with the path ABC, passes the first knife-edge through the gap. When passing the second knife-edge, this streamline enters and leaves the honeycomb cell. The second streamline with the path ADE passes the first knife-edge through the cell-mouth. Clearly the effect of rotation is larger on the second streamline ADE than on the first streamline ABC. The no-slip condition on the honeycomb wall (in the gap) appears to resist the flow from following the rotation of the knife-edge. It should be noted that this pattern of streamlines is repeated in the circumferential direction according to the geometrical periodicity of the honeycomb cell. The streamlines are colored with the total pressure loss (the contour levels are different than those used in Figure 9). Projecting the three-dimensional streamlines and the geometry of the labyrinth seal on an axialradial plane results in Figure 11, where curves for the total pressure loss and the turbulence kinetic energy along the streamlines have also been drawn. When passing the first knife-edge, streamline ABC experiences an abrupt increase of loss at the entrance to the gap where there is an abrupt increase of the turbulence kinetic energy. The loss increases gradually in the shear layer of the jet until it attains the level in the first cavity. Streamline ADE experiences effectively two gaps, the first one being above the first corner of the knife-edge when the streamline goes upward and the second gap is at the second corner after the streamline is bent downward. These locations are characterized by a maximum turbulence kinetic energy where a large increase of the total pressure loss is evident. Both streamlines do not experience any significant loss across the cavity. When passing through the second knife-edge, a particle on streamline ABC experiences an abrupt increase of the total pressure loss near the upstream corner of the knife-edge. It enters the honeycomb cell afterwards, where only a negligible increase of loss is observed. Finally, another significant increase of loss near the downstream corner of the knife edge is observed when the particle leaves the knife-edge. Another situation can be observed on the streamline ADE. After spanning over the cavity, the streamline attaches on the surface of the second knife, and follows the surface until it arrives at the knife-edge. The curve indicates a decrease of loss in this area. This decrease can be attributed to the effect of rotation.
13
NLR-TP-2007-624
The situation at the third knife-edge is opposite to the situation at the first knife-edge, where streamline ABC passes through the cell-mouth and streamline ADE passes through the gap. The phenomena observed around the first knife-edge can also be found here.
14
NLR-TP-2007-624
Finally, it can be observed that for the same mass flow, inlet temperature and inlet pressure (and therefore for the same mass flow function ), the labyrinth seal with a smooth land requires a larger pressure ratio of 2.2 compared to about 1.25 when the honeycomb is present, due to the larger loss of the total pressure. This finding is in line with the observation in another study (Ref. 6, 14), where in case of a small ratio between the width or clearance of the knife and the cell size a significant increase of leakage flow is expected.
15
NLR-TP-2007-624
9 Industrial Perspective
The application of CFD in combination with experimental techniques gives the possibility to obtain more detailed information and in-depth understanding of the complex leakage flow through labyrinth seals. The eventual goal of these studies is to predict the leakage flow through labyrinth seals accurately and cost effectively. During the development of new seal configurations and accompanying (upgrades of) engineering models employing a onedimensional (1-D) flow solver, the CFD method can be utilized to: Determine the sealing benefit of conceptual geometries in order to down select the most promising candidates (possibly supported by rig-testing in limited design space). Identify relevant leakage parameters for proposed geometries. Determine the sealing benefit of detailed geometries with validation in advanced test-rig under engine-like conditions. Support the correct representation of test-data into semi-empirical engineering models (e.g. by providing detailed information of loss mechanisms). Extrapolate correlations for the engineering models to a limited extent beyond the measured range of test-data. With the increasing performance of computers, a simplified 2-D CFD model could predict seal flows in less than an hour on a standard PC, while a detailed 3-D CFD model could provide simulation data in an overnight job on a mainframe computer. For the design of labyrinth seals this is too elaborate and engineering models are still preferred to maximize cost effectiveness of the analysis. One way to incorporate CFD in seal design is to use the simulation results and create a database for an engineering model (in a fashion like presented in Ref. 13). Compatibility with internal flow network solvers and the need for validated tools to reduce risk will generally limit the involvement of stand-alone CFD in seal design to specific cases. The role and added value of CFD is therefore most pronounced during the technology development phase.
10 Conclusions
Referring to the three objectives set out in the introduction, based on the obtained results for the seal configurations and flow conditions investigated, the following conclusions can be drawn. First, the comparison between the results of the CFD computation and the experimental measurement shows a close agreement. This is an indication that the important flow physics responsible for the sealing mechanism have been captured by the CFD method. Second, an analysis of the flow topology and the losses experienced along the streamlines has been performed. This approach reveals that the sealing mechanism of the flow through labyrinth seals is due to losses associated with turbulent mixing and shear which dominantly occur in the vicinity of the knife edge and to a lesser extent around the step wall. For the straight three knifeedge configuration considered, the effects of the honeycomb can be identified as decreased production of losses due to: less intense turbulent mixing around the knife-edge, preventing the formation of a coherent shear layer, mitigating the flow impingement on the step wall, inhibiting flow separations at the notch.
16
NLR-TP-2007-624
Third, from comparing two different seal concepts, the lower performance of the canted-knife seal concept with respect to the straight-knife concept can be attributed to mitigated flow impingement on the step wall, absence of flow separations from the notch, and less intense turbulent mixing and shear in the vicinity of the knife-edge. Finally, industrial perspectives have been presented to outline the importance of the CFD method, complementary to the seal test rig facility, in particular during the development of seal technology.
Acknowledgments
The investigation has been performed under the research contract number 49410N from the Netherlands Agency for Aerospace Programs. Thanks are due to Pratt & Whitney for the permission to use the geometry and data of the labyrinth seals.
17
NLR-TP-2007-624
References
1. G.A. Kool et al, Advanced seal test rig validation and operation, Paper GT2006-90915, 2006. 2. D. Rhode, S. Ko, S., and G. Morrison, Leakage optimization of labyrinth seals using a Navier-Stokes code, Tribology Transaction 37, 1994. 3. D.C.J. Allcock, P.C. Ivey, and J.R. Turner, Abradable stator gas turbine labyrinth seals Part 1: Experimental determination and CFD modelling of effective friction factors for honeycomb materials. AIAA-2002-3936 Paper, 2002. 4. G. Chochua, W. Shyy, and J. Moore, Computational modeling for honeycomb-stator gas annular seal, Intl. J. Heat and Mass Transfer 45, 2002. 5. D-C, Choi and D.L. Rhode, Development of a 2-D CFD approach for computing 3-D honeycomb labyrinth leakage, GT2003-38238 paper, 2003. 6. V. Schramm, K. Willenborg, S. Kim, and S. Wittig, Influence of a Honeycomb Facing on the Flow Through a Stepped Labyrinth Seal, Trans. ASME - A - Engineering for Gas Turbines and Power, vol. 124, pp 140-146, January 2002. 7. D. Collins et al, Numerical modelling of three dimensional honeycomb labyrinth seals employing a simplified approach, GT2006-90850 Paper, 2006. 8. R. Paolillo et al, Rotating seal rig experiments: test results and analysis modelling, GT200690957 Paper, 2006. 9. J.W. Boerstoel, A. Kassies, J.C. Kok, and S.P. Spekreijse. ENFLOW, A full-functionality system of CFD codes for industrial Euler/Navier-Stokes Flow Computations. NLR TP 96286 (presented at the 2nd Int. Symp. on Aeron. Science and Tech., Jakarta, 1996). 10. J.C. Kok. Resolving the dependence on freestream values for the k- turbulence model. AIAA Journal, 38(7), pp. 12921294, 2000. 11. J.C. Kok and S.P. Spekreijse. Efficient and accurate implementation of the k- turbulence model in the NLR multi-block Navier-Stokes system. NLR TP-2000-144 (presented at ECCOMAS 2000, Barcelona, Spain, 11-14 September, 2000). 12. F.J. Brandsma, J.C. Kok, H.S. Dol, A. Elsenaar, Leading edge vortex flow computations and comparison with DNW-HST wind tunnel data. NLR TP-2001-238 (RTO/AVT Vortex Flow Symposium, Loen, Norway, 2001). 13. D.C.J. Allcock, P.C. Ivey, and J.R. Turner, Abradable stator gas turbine labyrinth seals Part 2: Numerical modelling of different seal geometries and the construction of a second generation design tool, AIAA-2002-3937, 2002. 14. Stocker, H.L., Cox, D.M. and Holle, G.F.; Aerodynamic Performance of Conventional and Advanced Design Labyrinth Seals with Solid-Smooth, Abradable and Honeycomb Lands, NASA CR-135307, 1977.
18
NLR-TP-2007-624
Inlet pressure (kPa) pressure ratio mass flow (kg/s) speed (rpm) clearance (mm)
19
NLR-TP-2007-624
Case-1
Case-3
Figure 2 Labyrinth seal configurations (all geometries are shown on the same scale; the direction of the flow is from left to right).
20
NLR-TP-2007-624
21
NLR-TP-2007-624
22
NLR-TP-2007-624
Figure 7 Total pressure loss contours in the cavities on the mid-plane (upper) and sideplane (lower) (Case-1).
23
NLR-TP-2007-624
Figure 8 In-plane (two-dimensional) streamlines on the mid-plane (upper) and side-plane (lower) with pressure contours (Case-1).
Figure 9 Variation of the total pressure loss along the streamlines across the first knifeedge (Case-1).
24
NLR-TP-2007-624
Figure 10 Two streamlines with different paths across the knife-edge (Case-1).
Figure 11 Variation of the total pressure loss (red) and turbulence kinetic energy (green) along the streamlines (blue) following the path ABC and ADE shown in Figure 10 (Case-1).
25
NLR-TP-2007-624
Figure 13 Close-up views of the cavities with contours of the total pressure loss (Case-2).
Figure 14 Variation of the total pressure loss (red) and turbulence kinetic energy (green) along the streamline (blue) (Case-2).
26
NLR-TP-2007-624
Figure 16 Close-up views of the cavities with contours of the total pressure loss (Case-3).
Figure 17 Variation of the total pressure loss (red) and turbulence kinetic energy (green) along the streamline (blue) (Case-3).
27