Вы находитесь на странице: 1из 11

AIAA/CIRA 13th International Space Planes and Hypersonics Systems and Technologies

AIAA 2005-3286

Investigation of Boundary Layer Bleed for Improving Scramjet Isolator Performance


Chung-Jen Tam, AFRL/PRAS/Taitech,Inc, Wright-Patterson Air Force Base, OH 45433 and Dean Eklund, Robert Behdadnia and Thomas Jackson AFRL/PRA, Wright-Patterson Air Force Base, OH 45433

Controlling shock-wave/turbulent-boundary-layer interactions is one of the most important roles of a scramjet isolator, particularly in the effort to prevent inlet unstart. Th ree-dimensional, steady-state numerical simulations were performed to improve the performance of a rectangular isolator using a bleeding technique. Ten different bleed slot locations and orientations were tested to determine their effectiveness in holding the shock train in the isolator, or pushing the shock system further downstream. The numerical results w ere based on a back-pressure condition imposed at the isolator exit. In addition, two exit-pressure conditions were set for the bleed slots to determine their impact on the isolator performance. The results show that bleeding out the low-momentum flows near the isolator corners is more effective than removing the boundary layer at the centerplane of the isolator. Numerical simulations were performed to complement experimental work to be done at the AFRL/PRA Test Cell 19 facility. Based on the CFD results and exp erimental resou rces, three different bleed slots on the isolator side wall will be tested in the wind tunnel. Eventually, the numerical and experimental results will be compared to determine the effectiveness of the bleed slots in improving the scram jet isolator perform ance. I. INTRODUCTION Shock-wave/turbulent-boundary-layer interactions occur in a wide variety of aerospace applications. O f particular interest is the shock/boundary-layer interaction found in the scramjet-engine isolator, which is designed to contain the pressure rise generated by heat release within the combustor. Scramjet operation at the lower flight Ma ch num ber envelope is characterized by dual-mode combu stion. W here a significant portion of the heat release occurs in both subsonic and supersonic regimes. In dual-mode operation the heat release is sufficient to cause thermal choking of the incoming supersonic air strea m, thereby generating a shock train upstream of the heat release and typically upstream of the fuel injection. The pre-combustion shock train aids flame stabilization by 1) increasing the static temperature and pressure of the air stream, thereby reducing the ignition delay and 2) decelerating the flow, thereby increasing the residence time. However, if this shock train interacts with the inlet flow field, significant flow spillage or even inlet unstart could resu lt. To prevent the pressure rise associated with the precombustion shock train from affecting the flow through the engine inlet, a constant (or nearly-constant) area duct called an isolator is inserted betw een the inlet and the combustor of the scram jet engine. The length of the isolator is determined by the need to contain the combustion-induced pressure rise over the entire flight trajectory. At higher flight Mach numbers, the engine flow field has a higher enthalpy and Mach number; therefore, the heat release within the combustor is insufficient to choke the incoming stream. Consequently, no upstream interaction occurs, i.e . no pre-combustion shock train is generated in the isolator. Under these conditions the isolator contributes only viscous drag and weight to system performance. An active control system for repositioning shock waves within the isolator at low flight Mach numbers should yield a shorter isolator design, resulting in reduced system drag and weight. Isolator length reduction could relieve combustor design constraints, resulting in higher propulsive efficiency, especially at higher Mach numbers.
Research Scientist, Taitech Inc., 1430 Oak Court, Suite 301, Beavercreek, OH 45430, Senior Member AIAA. Aerospace Engineer, AFRL/PRAT, WPAFB, OH 45433, Senior Member AIAA. Research Chemist, AFRL/PRAS, WPAFB, OH 45433, Member ACS. Deputy of Science, AFRL/PRA, WPAFB, OH 45433, Senior Member AIAA.

This material is declared a work of the U.S. Government and is not subject to copyright protection in the United States.

Manipulations of the boundary layer have been studied extensively in attempts to reduce the pressure loss associated with shock/turbulent-boundary-layer interactions. Passive control methods, such as suction,1,2,3 blowing, 2 and geometric modifications4 can affect the boundary layer. For example, generated turbulence enhances the momentum transfer in the boundary layer while delaying boundary layer separation.5 Various device shapes, such as grooves, slots,6 and slits combined with passive cavities7 have been used to modify the shock boundary-layer interaction through the generation of streamw ise vortices. While these methods effectively reduce wave drag, they do result in a large increase in viscous drag, usually leading to an increase in total drag. Bumps are effective at design conditions; however, if bumps are to be useful over a range of flight conditions, some form of variable geometry will be required.4 Computational8 and experimental9 results show that bleed slots can be used to start over-contracted inlets. In the work described here, numerical simulations were performed to study the effectiveness of bleed slots in the scramjet isolator to control the shock-wave/turbulent boundary-layer interactions. II. NUMERICAL APPROACH All simulations were performed using the VU LCAN Navier-Stokes code. 1 0 The numerical code solves the conservation equations for calorically perfect or mixtures of thermally perfect gases with a cell-centered finite volume scheme. The equ ation set can be integrated in a fully elliptic fashion, or by using a parabolic (space-marching) option. The inviscid fluxes are evaluated using central differences, Roes flux difference method,1 1 or a low-diffusion flux vector split scheme.1 2 Viscous fluxes are evaluated with standard second-order accurate central differences. A variety of implicit and explicit time integration strategies are available for advancing the solution in time. A variety of one-equation and twoequation turbulence models are available to describe the turbulent velocity field. All calculations were 3-D, steady-state solutions using the low diffusion flux vector split scheme of Edwards. 1 2 The MUSCL parameter (6) was chosen to be 1/3 to minimize truncation error. The Van Leer flux limiter w as employed to enforce Total Variation Diminishing (TVD). Each solution was advanced in time w ith a diagonalized approximate factorization scheme using a CFL number from 1-3. T he M enter-BSL model 1 3 was used to model the effects of turbulence. The numerical solutions were considered to be converged based on the residual history and the steadiness of mass flow rate before and after the bleed slots. The mass flow rate should not change, i.e ., constant along the isolator before and after the bleed slots. Initial/Boundary Conditions The numerica l results are based on steady-state solutions and compared to the AFRL /PRA T est Cell 19 test configuration. The Mach number at the nozzle exit is M 4 = 2 with T 0 = 300 K and P 0 = 351.6 KPa at the compressor tank. The cross-sectional area is 5.08 cm x 15.24 cm at the exit of the facility nozzle. The isolator test section (5.08 cm x 15.24 cm ), which is directly connected to the exit of the facility nozzle, is 0.911 m long, as depicted in Fig.1. The numerical simulation was extended from the facility nozzle plenum to the entrance of the isolator test section to provide the inflow conditions for the isolator section. The exit of the isolator section was pressurized to ( Pb = ) 166 KPa to create a chok ing condition in the isolator. A converged solution of the rectangular isolator with P b = 166 KPa was used as the initial condition for all the bleed slot simulations, as shown in Figure 2. A large boundary-layer separation can be observed in the upper corner in both the streamlines and skin-friction coefficient, C f, which represent an asymm etric flow feature in the rectangular isolator. This flow mechanism is caused by the facility nozzle. Since the nozzle is not symmetric (Fig. 1), the flow is accelerated more rapidly along the bottom wall than the upper wall of the nozzle throat, which provides more momentum flow out of the nozzle. This increase in momentum plays a significant role further downstream of the nozzle, particularly in the corner regions. The displacement and momentum thicknesses increase faster in the upper corner than in the bottom corner (Fig. 3). Eventually, the upper corner of the flow is mor e likely to separate than the bottom corner, as occurs with a high back pressure (Pb = 166 KPa ). Three different exit pressures for the bleed slots (Pbleed ) were tested to study their impact on scramjet isolator performance: P bleed = 10.1 (1/10 atm. pressure), 101 (1 atm. pressure), and 37.9 KPa . Note that additional computational domains were added to the existing isolator solution to simulate the bleed slots, with a finite wall thickness of 1.016 cm and a nominal area of 7.742 cm 2 . Thus, the bleed slot simulations are not considered to be boundary condition problems, but rather three-dim ensional computations. The flow conditions inside the slot com putational domains were set to zero velocity and the pressure corresponding to the P bleed test condition. This initial condition is to represent an instantaneous bleed to the specified Pbleed , or a sudden opening of the bleed slot matching the P bleed condition. Comp utational G rid The computational domain for the isolator test section consisted of 31 blocks w ith each block dim ensioned to 5 X 41

X 41 grid points in the x -, y - and z- directions, respectively. Note that because the isolator is symmetric along the centerline plane, only half of the plane was computed in this study. Additional computational blocks for the bleed slots were added to the existing isolator domains to simulate the effects of bleeding through a finite wall thickness. Ten different bleed slot locations and orientations were tested in studies of their effectiveness in controlling the shock-wave/ turbulent-boun dary-layer interaction in a scram jet isolator, as illustrated in Fig. 4. M ost of the configurations are aligned in the streamwise direction (A-H), with two in the spanwise direction (I-J).The num ber of grid points for the bleed slots varies depending on their locations and orientations, which are denoted in Figs. 4 (A-J). T he leading edge of the shock train is initially in the vicinity of the bleed slots, except for Cases E, F, I and J. All the leading edges of the bleed slots start at x = 0.247 m , except for Cases E and F. The Case E slot is approximately 0.0762 m upstream of the shock system, while the Case F slot is located 0.0762 m downstream. For Cases I and J, the slots are placed upstream of the shock-train system to bleed out the low momentum flow in the spanwise direction on the top and side wall, respectively. Similarly, Case H is designed to remove the low momentum flows at the corner regions. In this case the total area of the bleed slots is twice that of the other configurations. The bleed slots for Cases G, H an d J are on the side wall, while the others are on the top w all. III. RESULTS AND DISCUSSION The main goal of this research is to understand the effects and performance of boundary-layer bleed in a scramjet isolator, especially in preparation for an experimental research program to be performed subsequently. The results include simulations of ten different bleed slot locations and orientations for a rectangular isolator (5.08 cm x 15.24 cm.). These configurations w ere tested with a high back-pressure con dition, P b = 166, to study their effectiveness in controlling the shock-wave/turbulent boundary layer interaction in the isolator. In addition, three different exit bleed pressures (Pbleed ) were tested to study their effects on the scramjet isolator performance. For P bleed = 10.1 KPa , the exit-pressure boundary condition is lower than the mass-averaged static pressure in the isolator, particularly at the x = 0.24722-0.33527 m slot location (Fig. 5). With the specified low-pressure boundary condition, the bleed slots represent an overall mass expulsion effect. However, in the case of Pbleed = 101 KPa (1 atm ), the exit pressure at the first half of the bleed slot is higher than the pressure in the isolator, while the pressure in the second half of the slot location is lower than the internal pressure. This exit boundary condition could cause simultaneous air mass injection into and expulsion from the isolator. This exit condition simulates the expulsion of isolator air to atmospheric pressure inside the Test Cell 19 facility. The results showed, however, that having a high P bleed reduces the effectiveness of the bleed slots. This will be discussed in the following section. Finally, three slot configurations were chosen to be tested in the AFRL/PRA Test Cell 19 wind-tunnel facility based on the current findings, and experimental resources. For these cases, simulations were performed with Pb = 166 KPa and P bleed = 37.9 KPa to provide information for the experimental design process. P bleed was based on the approximate pressure condition outside the isolator, with the air exhausted to a vacuum chamber (20.7 KPa) , instead of the Test Cell facility (1 atm ). Unfortunately, no experimental data is available for comparison at this time. T he num erical results are presented in terms of streamlines, mass-averaged quantities, such as static pressure, percentage of air mass flow loss, and total pressure recovery ratio, and contour plots of static pressure, and skin-friction coefficients. Table 1 shows the performance of different slot configurations with various P bleed , in terms of shock location relative to the initial shock position, X 0 , without bleeding then normalized by isolator duct height (H=5.08 cm ), and the percentage of air mass loss. The shock position, X, is defined as the location where a 5% increase in the normalized massaveraged static pressure. The static pressure is normalized by the pressure from the no back-pressure solution. A positive value of (X S - X 0 ) / H indicates the shock train is pushed downstream relative to the original shock location due to bleeding. On the other hand, a negative value shows that the shock system has traveled upstream, which is not a desirable bleed slot design for a scramjet isolator. Solutions for Back Pressure Pb = 166 KPa with Pbleed = 10.1 and 101 KPa For the low P bleed ( = 10.1 KPa ) condition, the bleed slots overall performed better than for the high P bleed case in controlling the shock train in the isolator test section (Table 1). Among all the test configurations, the bleed slots, which potentially bled out the low momentum flow in the boundary layer around the corner regions and pushed the shock system further downstream, as in Cases H and J were more effective based on the mass-averaged static pressure data, as show n in Fig. 6a. Although C ase H pushed the shock train approximately 3.6493 duct heights dow nstream as shown in Table 1, it lost 7.88% of the incoming air mass, much more than the other cases, because the total bleed slot area is twice as large as the other configurations. Sim ilarly, this case has the largest loss (7.68%) in term of total pressure recovery ratio as compared to the isolator without bleeding (Fig. 5). On the other hand, Case J showed better performance than any other

configuration in terms of lower loss in air mass (2.18% ) and total pressure recovery (1.99% ). In addition, the shock train is pushed 2.2736 duct heights downstream of the isolator (Table 1.) The bleed slots in the plane of symmetry (Cases D, E and F) do not exhibit any significant contribution in controlling the shock-wave/turbulent-boundary-layer interaction for the rectangular isolator. On the contrary, the shock train is pushed as much as 1.2657 duct heights upstream of the original location w ithout bleeding (Table 1). T his could be caused by the removal of air mass not only in the streamwise direction, but also in the spanwise direction, i.e., the flow in front of the separation bubb le at the upper corner to the centerplan e. Consequently, some of the high-momentum boundary-layer flow has been bled out and the separation region becomes larger because of the increase in the adverse pressure gradient. This flow feature can be observed from skinfriction coefficient con tours, as shown in Fig. 7b. In Case A, the low momentum flow or separated region is bled out, but a separation bubble is created on the bottom corner, as illustrated in Fig. 7a. A similar ph enomenon occurs in Case G (Fig. 7c). As discussed earlier, Case H appears to perform better in controlling the shock train because the low-momentum boundary-layer flows on the corner regions could be bled out (Fig. 7d), and the shock system is pushed further downstream. Penalties are paid in the form of higher losses in air mass and total pressure recovery ratio. In the cases of the spanwise bleed slots, Case J is more effective than Case I in terms of reducing the size of the separation bubble and its location is downstream of the bleed slot, as shown in Figs. 7e and f. The separation regions are indicated by the dashed line where C f = 0 in Figs. 7e and f. In addition, C ase I has a small separation bubble on the lower corner. Again, this is due to the removal of low-momentum boundary-layer flows in both upper and bottom corners, as in Case J. With a high P bleed = 101 KPa , none of the slot locations improved the isolator performance. On the contrary, all the test cases including Cases H and pushed the shock train upstream of the bleed slots (T able 1). In addition, the overall mass-averaged static pressure in the vicinity of the slot locations (shown in Fig. 8a) is higher than the static-pressure data without bleeding (Fig. 5). In some cases, such as Cases D, E, and F, the pressure is much higher than the specified Pbleed , which indicates that some external air is injected into to the isolator through the slot. The results showed that none of the bleed configurations lose as much air mass as in the lower P bleed condition. This is especially true for Cases A and H, as listed in Table 1. The air mass losses in these two cases seem ed to be minimal compared to the previous case (Table 1), which strongly suggested that air was not only removed from the isolator but also added through the bleed slots. As illustrated in Fig. 8b for Cases A and H, the front portion of the slot showed mass injection into the isolator, because the overall static pressure is lower than Pbleed . On the other hand, a portion of the air is removed from the isolator, because the internal pressure is higher than Pbleed . The air removal from the isolator seemed in general to be more effective in controlling the shock-wave/turbulent-boun dary-layer interaction in the P bleed = 10.1 KPa case than in the P bleed = 101 KPa condition. Solutions for Back Pressure, P b = 166 KPa and Pbleed = 37.9 KPa Based on the current numerical findings and experimental resources, three bleed slots were identified to be tested in the AFRL/PRA wind-tunnel facility. For these cases, simulations were performed with Pb = 166 KPa and P bleed = 37.9 KPa to provide information for the design of the experimental model and set up. The P bleed was based on the approximate pressure condition outside the isolator, with the air exhausted to a vacuum chamber at 20.7 KPa , instead to the Test Cell facility (at 1 atm ). No experimental data are available for comparison at this time. All three designs involve bleed slots located on the side wall, as in Cases G, H and J. In these simulations, the total bleed area was set to be the same for all the cases, approximately 7.742 cm 2 with a wall thickness of 1.016 cm . Thus, the area of each corner side slot in Case H is half the size of the slot in Case G. With the specified P bleed = 37.9 KPa , the external pressure is lower than the mass-averaged static pressure inside the isolator, so that these three configurations would produce the bleeding effect. The numerical results showed that the 2 corner side slots perform better than the other two designs in terms of pushing the shock system approximately 2.52 duct heights downstream, as shown in Table 1, with a loss of nearly 2.4% in incoming air mass as compared to 5.3% for the side slot (Case G). Overall, the vertical side slot (Case J) has the better performance in terms of less air mass (Table 1) and total pressure recovery ratio losses (Fig. 9b). Again the results show that bleeding low -momentum flows on the corner of the rectangular isolator could potentially improve the isolator perform ance. IV. SUMMARY Numerical simulations w ere performed to study the effects of boundary-layer bleeding in a rectangular scramjet isolator. Ten different bleed slot locations and orientations were tested to improve the isolator performance. A back pressure condition (P b = 166 KPa ) was imposed on the exit of the isolator to create a shock train in the isolator. The pressure at the bleed slots was set to 10.1 and 101 KPa , i.e ., 1/10 and 1 atmosp heric pressure condition, respectively. Overall, placing the bleed slots on the side wall improves the isolator performance in moving the shock train downstream,

particularly slots in the corner regions and vertical slot designs. The results show that bleeding out the low-momentum flows near the corners improves the isolator performance, which indicates that the bleed slots are more effective when the exit pressure is set lower than the internal pressure in the isolator. In addition, numerical results w ere used to provide information for the design of an experimental program involving 3 slot designs to be run in the AFRL/PRA T est Cell 19 wind-tunnel facility. Eventually the numerical solutions will be compared to the experimental data. V. ACKNOWLEDGM ENTS This study was sup ported by an d performed at W right-Patterson Air Force Base, under Contract F33615-03-M-2373. The support for this project is sincerely appreciated. Additional support took the form of a grant of computer resources from ASC and ERDC MSR C Department of Defence High Performance Computing centers. REFERENCES 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. Morris, M.J., Sajben, M., and K routil, J.C., Experimental Investigation of Normal Shock/Turbulent Boundary Layer Interactions with and without Mass Removal, AIAA Journal, Vol. 30, No. 2, 1992, pp. 359-366. Delery, J.M., Shock Wave/Turbulent Boundary Layer Interaction and its Control, Progress in Aerospace Sciences, Vol. 22, 1985, pp. 209-280. Harloff, G. J. and Smith, G.E., Supersonic-Inlet Boundary-Layer Bleed Flow, NASA-CR-195426, 1995. Stanew sky, E., Delery, J., Fulker, J., and de Matteis, P., Drag Reduction by Shock and Boundary Layer Control, Results of the Project EUROSHOCK II, Notes on Numerical Fluid Mechanics, Vol. 80, 2002. McCormick, D. C., Shock/Boundary-Layer Interaction Control with V ortex Generators and Passive Cavity, AIAA Journal, Vol.31, No.1, 1993, pp.91-96. Smith, A. N., Babinsky, H., Fulker, J. L., and Ashill, P. A., Con trol of Normal Shock Wave/Turbulent Boundary-Layer Interaction Using Streamwise Grooves, AIAA Paper 2002-0978, 2002. Hafenrichter, E. S., Lee, Y., McIlw ain, S. T., Dutton, J. C., and Loth, E., Experiments on Normal Shock/Boundary Layer Interaction Control Using Aeroelastic Mesoflaps, AIAA Paper 2001-0156, 2001. Tam, C.-J., Baurle, R.A., and Streby, G.D., Numerical Analysis of Stream line-Traced Hypersonic Inlets, AIAA Paper 2003-0013, 2003. Smart, M. K ., and T rexler, C. A., Mach 4 Performance of a Fixed-Geometry Hypersonic Inlet with Rectangular-to-Elliptical Shape Transition, AIAA Paper 2003-0012, 2003. White, J. and Morrison, J.H., A Pseudo-Temporal Multi-Grid Relaxation Scheme for Solving the Parabolized Navier-Stokes Equations, AIAA Paper 1999-3360, 1999. Roe, P.L., Some Contributions to the Modeling of Discontinuous Flows, Proceedings of the 1983 AMS-SIAM Summer Seminar on Large Scale Computing in Fluid Mechanics, Lectures in Applied Mathematics , Vol. 22, 1985, pp.163-193. Edwards, J.R., A L ow-Diffusion Flux-Splitting Scheme for Navier-Stokes Calculations, Computers & Fluids, Vol. 26, No. 6, 1997, pp. 635-659. Menter, F.R., Zonal Two Equation 6-T Models for Aerodynamic Flows, AIAA Paper 1993-2906, 1993.

12. 13.

P b = 166 KPa , P bleed = 10.1 KPa CASE A B C D E F G H I J ( XS - X0 ) / H 1.3811 0.7825 0.1925 -0.5041 -1.2657 -0.2136 1.2921 3.6493 0.8773 2.2736 % air mass loss 5.615 5.467 6.423 4.510 3.838 4.897 5.866 7.883 3.414 2.180

P b = 166 KPa , P bleed = 101 KPa ( XS - X0 ) / H -1.8023 -2.5391 -3.0333 -3.5742 -3.7562 -3.4880 -2.3660 -1.5884 -3.1233 -2.8571 % air mass loss 0.303 1.490 2.594 2.651 0.450 3.257 0.672 0.219 0.560 0.583

P b = 166 KPa , P bleed = 37.9 KPa CASE 2 corner side slots side slot vertical side slot ( XS - X0 ) / H 2.5155 -0.2154 0.8390 % air mass loss 2.374 5.331 0.834

Table 1. Normalized shock train locations and percentage air mass losses for different cases.

Figure 1. Schematic of AFRL/PRA Test Cell 19 wind-tunnel facility.

(a) Pressure contours and streamlines along the symmetric plane and a plane 0.000267 m from the side wall.

(b) Skin-friction coefficient contours

Figure 2. Contour plots with streamlines for a rectangular isolator test section with Pb = 166 KPa .

Figure 3. Displacement and momentum thicknesses near the side wall of a rectangular isolator without back pressure.

P b = 166 KPa Figure 4. Mass-averaged properties along the facility nozzle and isolator test sections without bleeding.

(A) 15 x 11 x 21

(B) 15 x 5 x 21

(C) 15 X 5 X 21

(D) 15 x 3 x 21

(E) 15 x 3 x 21

(F) 15 x 3 x 21

(G) 15 x 21 x 6

(H) 15 x 21 x 14 x 2

(I) 3 x 41 x 21

(J) 3 x 21 x 41

Figure 5. Bleed slot configurations and grid dimensions.

(a) Static pressure plots

(b) Total pressure recovery ratio plots

Figure 6. Mass-averaged properties for Pb = 166 KPa and Pbleed = 10.1 KPa (1/10 atm ).

(a) Case A

(b) Case D

(c) Case G

(d) Case H

(e) Case I

(f) Case J

Figure 7. Skin-friction coefficient contours for various bleed slot configurations. Dashed contour lines in (e) and (f) represen t C f = 0.0.

(a) Static pressure plots

Case A

Case H

(b) Pressure contours and streamlines at the centerline of bleed slots for Case A and H. Figure 8. Mass-averaged properties and contour plots for Pb = 166 KPa and Pbleed = 101 KPa (1 atm ).

(a) Static pressure plots

(b) Total pressure recovery ratio plots

Figure 9. Mass-averaged properties for Pb = 166 KPa and Pbleed = 37.9 KPa .

Вам также может понравиться