Aerospace

Science

and Technology,

1$98,

no 1, 61-73

A Basic Experimental Investigation of Passive Control Applied to a Transonic Interaction

R. Bur, J. Dklery, B. Corbel, D. Soulevant, R. Soares
ONERA, 29, Avenue de la Division Leclerc, BP 72, F-92322 Chktillon Cedex.

Manuscript

received

July 15, 1996; accepted December

3, 1996.

Bur R., DClery J., Corbel B., Soulevant 61-73.

D., Soares R., Aerospace

Science

and

Technology,

1998, no 1,

Abstract

Passive control applied to a turbulent shock wave/boundary layer interaction has been investigated by considering a two-dimensional channel flow. The field resulting from application of passive control has been probed in great detail by using a two-component laser Doppler velocimetry system to execute mean velocity and turbulence measurements. Four different perforated plates have been considered, as also the solid wall reference case. The performed measurements have shown that passive control deeply modifies the inviscid flowfield structure, the unique strong shock being replaced by a lambda shock system. This fractionning of the compression induces a substantial reduction of the wave drag associated with the interaction. On the other hand, the combined injection-suction effect taking place in the control region provokes an important thickening of the boundary layer. There results an increase of the friction drag which nearly outbalances the gain in wave drag. A determination of the total drag in the control region was made. It was found that passive control induced a modest decrease of this drag compared to the solid wall case. Also, the rugosity of the holes is an important source of drag (excrescence drag) which contributes to compromise the potential benefit of the passive control technique. 0 Elsevier, Paris

Keywords: Boundary
reduction.

layers - Turbulence

- Shock waves - Transonic

flow - Passive control

- Drag

Une etude fondamentale du contr6le passif appliquk B une interaction transsonique. Une etude experimentale detaillee dune interaction onde de choclcouche limite turbulente avec effet de controle passif a CtC executee dans un Ccoulement de canal. La region oti sapplique le controle a CtC sondee a laide dun velocimtttre laser en configuration bidirectionnelle, afin de mesurer finement les champs moyen et turbulent dans cette region. Le cas de reference de la paroi pleine et quatre configurations avec controle passif, utilisant des plaques perforees differentes, ont Cte Ctudies. Les mesures ont montre que le controle passif modifie profondement la structure du champ exteme, le choc unique intense &ant remplace par un systeme de chocs en lambda. Ce fractionnement de la compression conduit a une reduction substantielle de la trainee donde. Dautre part, leffet combine de linjection et de laspiration a travers la grille perforce provoque un Cpaississement important de la couche limite et entraine une augmentation de la trainee de frottement. Un bilan de la trainee totale a CtC effectue dans la region de controle. 11 montre que le controle passif permet dobtenir une faible diminution de la trainee du cas de reference. En revanche, la trainee induite par les trous de grille (trainee dexcroissance) peut fortement compromettre leffet benClique du controle passif. 0 Elsevier, Paris Mets-cl&: Couche limite - Turbulence
Reduction trainee. - Onde choc - Ecoulement transsonique - Controle passif -

Aerospace

Science and Technology,

1270-9638,

98/01/O

Elsevier,

Paris

62 Subscripts c Symbols

R. Bur et al.

NOTATIONS

c.f
CF

designates the conditions at the local boundary layer edge designates the conditions at the origin of the investigated domain (X = 220 mm)

h
Hi

plate excrescence drag coefficient total drag coefficient cavity depth boundary layer incompressible shape parameter

I - INTRODUCTION
Shock waves and their interaction with the boundary layer play a major role in determining the performance of transonic transport aircraft. It has been suggested that one way to reduce the harmful effects of strong shock waves on the off-design performance of airfoils is to use a passive control device in the shock wave/boundary layer interaction region. The principle of passive control consists in establishing a natural circulation between the downstream high pressure face of a shock and its upstream low pressure face. This circulation is achieved through a closed cavity, placed underneath the shock foot region, the face in contact with the outer flow being made of a perforated plate. Experimental and theoretical studies [l-8] have shown that, in some circumstances, passive control may produce a reduction of an airfoil drag, while postponing to higher incidences the limit of buffet onset. In fact, a clear effect of passive control is to reduce the wave drag by inducing a smearing of the compression provoked by the shock. The same mechanism is at work in supersonic air intakes where flow retardation is accomplished through a succession of shocks to reduce entropy increase, and hence increase efficiency. Another advantage of passive control would be to fix the shock location, preventing large amplitude oscillations which may occur at high incidence (buffeting phenomenon). However. the boundary layer manipulation resulting from the combined injection-suction effect, with in addition the excrescence drag of the perforated plate, provoke an increase of the friction drag which may outbalance the wave drag reduction. Thus, as far as airfoil drag is concerned, the potential benefit of passive control results from a delicate balance between two opposite tendencies. One of the objectives of the European Unionsponsored EUROSHOCK I project was to contribute to the understanding and modelling of the physical phenomena involved in a shock wave/boundary layer interaction under control conditions [9, lo]. In this context, the experimental facet of the study performed at ONERA [ll] focuses on a local analysis of the interaction with a view to establishing a detailed picture of the flowfield. including both its mean and turbulent properties. This analysis is detailed in the present paper. These experiments have been executed in a transonic channel flow by considering the interaction between the shock crossing the channel and the boundary layer developing on one of the channel

H, = d;z= Jb (l-gw 2 6 .L16g (l-@Y

i k fringe distance in the LDV probe volume turbulent kinetic energy approximated by -k = ;[1L2 + 7P + $(u2 + PP)] streamwise extent of the control region (L = 70 mm) Mach number static pressure stagnation (reservoir) pressure stagnation temperature X-wise and Y-wise instantaneous velocity components mean values of 7~and 21 fluctuating parts of u and v turbulent normal stresses turbulent shear stress reference velocity (U, = 377 m/s) streamwise coordinate along the channel lower wall, origin at the test section entrance (passive control region starts at X = 235 mm) coordinate normal to the channel lower wall boundary layer thickness boundary layer displacement thickness s* = Jj(l 0 - gyY thickness

Y 6 s*

boundary layer momentum 0 = Jf %(I - $)dY

o*

boundary layer mean flow kinetic energy thickness o* = s, J&l - $)dY

light-wave length density local mean velocity vector direction

A Basic Experimental

Investigation

of Passive Control/he

etude fondamentale

du contr&le passif

63

walls. This arrangement allows to work with a thick boundary layer and hence to obtain accurate definition of its properties during the interaction process. The precise objective of this study was to define, mainly by means of laser Doppler velocimetry (LDV) explorations, the properties of the interaction domain, firstly in the absence of control (solid wall reference case), secondly under passive control conditions for four perforated plates, having different geometrical characteristics. A method to determine the total drag in the control region is exposed in order to define an optimum passive control configuration. Additional experiments have been executed to determine the excrescence drag of the perforated plates. For this purpose, a uniform flow was established over the plates whose holes were obturated, the drag being deduced from the growth of the boundary layer momentum thickness. II - EXPERIMENTAL CONDITIONS

II.1 - Test set-up arrangement These experiments have been executed in the S8 Basic Research Transonic-Supersonic Wind Tunnel of the ONERA Fluid Mechanics Laboratory at ChalaisMeudon, France. This facility is a continuous wind tunnel supplied with dessicated atmospheric air. The stagnation conditions, nearly constant throughout the tests, had the following average values: - Stagnation pressure: psto = 92 000 Pa; - Stagnation temperature: Tst,, = 300 K. The test set-up used for the present study is shown in Figure 1. It is constituted by a transonic channel having a height of 100 mm and a span of 120 mm in the test section itself. The lower wall is rectilinear, the upper wall being made of a contoured profile designed to produce a uniform supersonic flow of nominal Mach number equal to 1.4. The two side walls are equipped with high quality glass windows to allow visualizations and LDV measurements. A second throat, of adjustable cross section, is placed in the test section outlet, approximately

Contoured

wall

I Perforated plate

\ Cavity Dimensions set-up

in mm tunnel.

Fig. 1. - Passive
1998. no 1

control

experimental

in the S8 wind

450 mm downstream of the study region, making it possible: - To produce, by choking effect, a shock wave whose position, and hence intensity, can be adjusted in a continuous and precise manner; - To isolate the flowfield under study from pressure perturbations emanating from downstream ducts. Such a device notably reduces unwanted shock oscillations. In what follows, X designates the streamwise distance measured from the channel entrance section and Y the distance normal to the study wall (see Fig. 1). Passive control is applied on the rectilinear lower wall in a region where the outer inviscid flow Mach number is equal to 1.33. The upstream edge of the control cavity, which nearly spans the entire test section, is located at 235 mm from the transonic channel entrance. Its length L has been defined in such a way as to approximately reproduce the ratio L/6, existing on an airfoil (S, is the boundary layer thickness upstream the cavity origin); i.e., L/6, 21 18. Since, in the present experiments S, N 4 mm, the value L = 70 mm has been adopted for all the tests. The cavity, or control region, thus extends from X = 235 mm to X = 305 mm. Its depth h has a fixed value equal to 60 mm. A first study [12] having demonstrated that h has a negligible influence on the control mechanism, provided it is not too small (at least 10 mm for the present conditions), an unrealistically high value of h has been adopted to facilitate possible LDV measurements inside the cavity. The interchangeable perforated plate is fixed to the wall, upstream and downstream of the cavity, with Araldite glue, the plate being made flush with the wall. In spite of these precautions, a slight perturbation could not be avoided at the junction between the plate and the wall. Two lateral narrow beams, of width 2 mm, are placed on each side of the cavity to prevent the bending of the plate which could be caused by the pressure difference between the cavity and the outer transonic flow. Air-tightness is insured by a seal placed around the cavity and along the plate. A plastic tape is sticked on each side of the perforated plate to prevent unwanted lateral circulation. All the pressure orifices were located in the vertical median plane of the test set-up. The rectilinear study wall was equipped with 10 orifices (diameter: 0.4 mm) upstream of the control region and 8 downstream of it. In order to avoid possible perturbation of the laser Doppler velocimetry (LDV) measurements, the perforated plates were not equipped with pressure tappings. After completion of these tests, 7 orifices were installed on the plates to determine the pressure distribution in the control region itself. The pressure tubings were directly fixed to the plates with Araldite glue. It was checked that this equipment did not affect the flow. Moreover, five pressure taps were located on the bottom part of the cavity.

64

R. Bur et al. this case, it was possible to reduce the probe volume diameter down to 0.1 mm and the minimum distance to the wall to 0.1 mm. In order to obtain a correct signal-to-noise ratio, the LDV system was operated in the forward scattering mode. In this configuration, the collecting part of the system is opposite to the emitting part and placed on the other side of the wind tunnel. The collecting part is equipped with a Cassegrain telescope having a diameter of 200 mm. The light from the telescope passes through interference filters that separate the green and blue components which are then applied to two photomultipliers whose signals are processed by Dantec type 55 L counters connected to the acquisition system. The emitting and collecting parts are mounted on two separate tables allowing accurate displacements (step of 0.01 mm) along three orthogonal axes. The motions of the tables are computer controlled and synchronized in such a way that the collecting optics follows the probe volume during its displacement. The complete probing sequence can be entirely automated, the probe volume being displaced along a preprogrammed path. At each measurement point, a sample of N couples of the instantaneous values of the velocity components II and ?I is acquired for further processing. Proper seeding of the flow to obtain an acceptable acquisition rate is a delicate and major problem in LDV. In the present study, where maximum velocities are in the transonic range (around 400 m/s), the flow was seeded with submicronic (0.5 pm diameter) droplets of parafine oil injected in the wind tunnel settling chamber. 11.2.2 - Mean velocity and Reynolds tensor components determination The size of the sample was N = 2000 which gives an acceptable statistical uncertainty for the first and second order statistical moments. The mean velocity and Reynolds tensor components were computed by the classical formulae: - Mean velocity components:
U=AV 1 Ui : N c i=l ,v 1 6 = N i=l llf ; c

For the study of the interaction without control (reference case), a rectilinear solid wall extending from the entrance to the end of the channel was manufactured to avoid the perturbations caused by junctions between different parts. This wall was equipped with 24 pressure orifices whose locations are included between X = 135 mm and 385 mm. The contoured wall, facing the study wall, was equipped with 23 orifices whose indications were used to monitor the shock location. It was thus possible to place the shock at exactly the same location in the outer inviscid stream for the various tested configurations. II.2 - Techniques processing of investigation and data

11.2.1 - The laser Doppler velocimetry (LDV) system The flows under study were qualified by means of: - Schlieren visualizations; - Measurements of wall pressure distributions; - Probings with a multi-component LDV system, this action constituting the core of the experimental effort. For the present study, where the flows were nominally two-dimensional, the two-component version of the ONERA three-component LDV system was used [ 131. The light source is a 15W - Argon laser, used in the present tests at a power of 5W, whose beam is separated into two beams of wave length 0.488 pm (blue color) and 0.5 145 pm (green color) by means of a semi-transparent mirror. The two original beams are first split by classical beam splitters and then traverse Bragg cells to enable the system to detect the velocity direction by causing the fringes forming in the probe volume to move at a shifting frequency which can be varied from 0 to 8 MHz. The four beams are focused by the emission lens to constitute two fringe patterns inside the measuring volume whose diameter was equal to 0.2 mm. The two fringe patterns were rotated at f 45 with respect to the X-direction, allowing the simultaneaous measurement of two velocity components in a vertical plane. In the two-component version of the LDV system, the optical adjustments were such that the fringe spacing %had the following values: - Green color (X = 0.5145 ,um): i = 13.82 pm; - Blue color (X = 0.488 pm): i = 13.30 pm. The fringe patterns comprised about 15 fringes in the probe volume. The Bragg cell frequency was set to 15 MHz. The minimum distance to the wall achievable (safely) by the LDV system, in the two-component version, was equal to 0.3 mm. In the experiments aimed at the determination of the excrescence drag of the perforated plates, a onecomponent arrangement was employed, since only the X-wise velocity component li was of interest. Then, the green color was used, with fringes perpendicular to the wall and a fringe spacing of 13.048 IAm. In

- Normal stress components:

.li

~ X(7& u/z = (N _ 1) i=l

- q2 ;

- Shear stress component:

A Basic Experimental

Investigation

of Passive Control/he

etude fondamentale

du contr8le passif

65

The above determination of the mean velocity components allow to compute the local Mach number and define boundary layer global properties; i.e., displacement thickness 6*, momentum thickness 8, mean flow kinetic energy thickness O* and The local incompressible shape parameter Hi. Mach number is deduced from the mean velocity vector modulus by assuming a constant stagnation temperature throughout the flow, even in the boundary layer. Such an assumption is legitimate in a transonic adiabatic flow. The boundary layer integral thicknesses are obtained by assuming a linear interpolation between the wall and each first measured point of the velocity profiles. The two-component version of the LDV system did not allow the determination of the spanwise, or transverse, velocity component w whose mean value is in principle zero (the flow being nominally twodimensional) but whose fluctuating part is non-zero. Thus, measurement of the true turbulent kinetic energy Ir, was not possible. The values of k given here are approximations computed by assuming: z= Hence: i(IIIL+;;Iz).

the range 220 5 X (mm) <_ 365 (it is reminded that the perforated plate extends between X = 235 mm and X = 305 mm) and separated by a distance AX = 5 mm from X = 220 mm to X = 325 mm and AX = 10 mm from X = 325 mm to X = 365 mm. Some additional explorations have been realized here and there the median plane (at + 20 mm of it). The comparison between the different explorated planes leads to very similar velocity profiles and to an acceptable two-dimensionality of the mean flow out of the sidewall boundary layers. Determination of the peqorated plates excrescence drag. The excrescence drag of the perforated plates has been deduced from the boundary layer development. To obtain sufficient accuracy in the velocity profile determination, these measurements were performed with a finer mesh close to the wall (minimum spacing AY = 0.05 mm). The boundary layer has been probed along 17 vertical lines, 5 mm apart, whose X locations are within the range 230 5 X (mm) 5 310. II.3 - Tested configurations In addition of the solid wall reference case, four different plates, numbered 1 to 4 and whose nominal characteristics are given in Table 1, have been tested. These plates were provided by DASA-Airbus. The holes were drilled by an electron beam technique. The thickness of the plates was equal to 1 mm. For Plates 1 and 2, the holes are everywhere normal to the surface. For Plates 3 and 4, the holes are inclined with respect to the surface (in the downstream direction) in the upstream half part of the plate, and normal to it in its downstream half part. For the study of passive control, including the reference case (solid wall), the location of the shock wave in the outer part of the flow (i.e., outside the interaction region) was the same for all the tested configurations. This location, which was fixed at mid distance between the origin and the end of the control region, was monitored by considering the pressure distribution on the channel upper wall, the shock position being accurately defined by adjustment of the second throat section. Maintaining a fixed shock
Table 1. - Nominal characteristics of the tested plates.

k= gF+$+$F+p)].
One should be aware of this fact when making comparisons with calculations. The above expressions give the field quantities with an accuracy depending of uncertainties affecting: - The LDV system calibration (uncertainties on the fringe distance, on the Bragg frequency); - The determination by the counters of the frequency of the light scattered by the particles; - The statistical treatment of the sample. For the present experiments, the field properties have been determined with an accuracy of: - 1% of the maximum velocity modulus for the mean velocity components; - 5 to 8% for the normal stress components; - < to 10% for the turbulent shear stress component. 11.2.3 - Domain explored with the LDV system Passive control study. The llowfields produced by the various tested configurations, including the reference case (solid wall), have been explored along lines normal to the wall (Y-direction) extending from the surface (Y = 0) to an altitude Y = 22 mm and contained in the test section median plane. This extent was chosen to be sure to cover the entire dissipative layer and a part of the outer inviscid flow. Each exploration contained 42 measurement points unevenly distributed in order to refine the probing in the vicinity of the wall (minimum spacing AY = 0.1 mm). The interaction domain has been explored along 30 vertical lines whose streamwise locations X were in 1998, n I

66

R. Bur et al. oblique shocks, in the sense of the strong solution to the Rankine-Hugoniot equations. This structure, which has a lambda shape, is generally attached to shock-separated flows at transonic speed. In well separated flows, the upstream compression waves rapidly coalesce to constitute the separation or leading shock. In this case, these three shocks meet at a bifurcation, or triple point. The present flow pattern, for which the compression waves have no time to coalesce, corresponds to a situation nearly coincident with incipient shock-induced separation which occurs for an upstream Mach number i?n, of 1.30. The present value MO = 1.34 should lead to the formation of a tiny separation bubble. However, LDV measurements have not detected negative values for the streamwise velocity component. Thus, if it exists, the bubble must be excessively small.

location insured an identical pressure rise for all the configurations, which allows a meaningful evaluation of a possible positive effect of passive control. In the experiments aiming at the estimation of excrescence drag, the second throat was adjusted in such a way that the flow in the channel be entirely supersonic. Hence, the flow over the plates was uniform with a Mach number equal to 1.36. In these experiments, the holes were obturated by strips of Scotch tape sticked to the plate inner face (cavity side). III - PRESENTATION III.1 - Introductory OF THE RESULTS remarks

In the following sections, the flow resulting from passive control is described by considering firstly its global properties, revealed by schlieren photographs and surface pressure distributions; secondly, a more refined description is given by examining the field properties obtained from LDV probings. Emphasis is placed on mean flow characteristics, since they are of main interest for evaluation of passive control effectiveness. Turbulent quantities, whose knowledge is essential to validate theoretical models, are presented in a more synthetic manner. In order to qualify the state of the incoming flow, LDV measurements were performed at the X = 195 mm station; i.e., 40 mm upstream of the beginning of the control region. The onecomponent arrangement was employed to improve the determination of the boundary layer near the wall. The boundary layer upstream of the interaction has a thickness around 4 mm and displacement and momentum thicknesses of 0.44 mm and 0.24 mm, respectively. Its incompressible shape parameter is equal to 1.41, which is typical of a well behave (fully) turbulent boundary layer. III.2 - Flow visualizations

Fig. 2. - Schlieren (Solid wall).

photograph

of the

flowfield.

Reference

case

111.2.1 - Reference case (Solid wall) The schlieren photograph shown in Figure 2, reveals the flow structure for the reference case (solid wall). It corresponds to a classical transonic shock wave/boundary layer interaction of moderate strength. One notes the thickening of the boundary layer in the shock foot region and the associated wave pattern forming in the outer inviscid flow. It is constituted by compression waves emanating from the interaction upstream part. This front zone is followed by a small triangular region of still supersonic flow terminated by a nearly normal shock, sometimes called the trailing shock. The compression waves and the trailing shock meet in a region from which starts the unique shock crossing the channel. A precise analysis of such a flow pattern shows that these shocks are strong

111.2.2 - Flow with passive control The schlieren visualization of the flow with passive control is shown in Figure 3. It corresponds to Plate 1 (normal holes, diameter 0.15 mm). One immediately notices the major differences brought by the control action. Now, the boundary layer starts to thicken suddenly at the cavity origin. Then, it takes a wedgelike shape until it meets with the trailing shock; afterwards, its thickness remains nearly constant. Hence, one of the passive control effects is to induce an increase of the friction drag. The clear region visible close to the wall, downstream of the interaction origin, is not due to the existence of a large separated region, but to the special character of the density distribution across a turbulent boundary layer having undergone a strong destabilization. The rapid thickening of the boundary layer, which is provoked by the injection effect taking place in the upstream part of the control region, is felt by the contiguous supersonic flow as a ramp effect. There results the formation of the oblique leading shock wave (Cl), downstream of which the Mach number is still supersonic. This supersonic region is terminated by the trailing shock (C,) through which the velocity

A Basic Experimental

Investigation

of Passive Control/he

etude ,fondamentale du contrBle passif

67

0.7

P/PSI,
l

0.6 I .

0 x 0 a .

Reference Plate 1 Plate 2 Plate 3 Plate 4 .

case

Cavity

?J
l a . 7,

pressure Plate 3

Fig. 3. - Schlieren (Plate I).

photograph

of the flowfield.

With

passive

control

Fig.

4. - Comparison

of the surface

pressure

distributions

vector changes from an upward direction to become nearly parallel to the wall. The two shocks (Cl) and (C2) meet at the triple point I, from which starts the unique shock (C,), to constitute a well defined lambda shape. In the absence of control, such a pattern is associated to strong transonic shock wave/boundary layer interaction with separation. The triple point I is the origin of a slip line, barely visible on the photograph, which separates the flow having passed through the shock (C,) from the flow having traversed successively (Cl) and (C,). Indeed, the rise in entropy (or loss in stagnation pressure) is less important through the consecutive shocks (C,) and (C,) than across the unique shock (C,), the upstream Mach numbers and downstream pressure levels being identical in the two flows. Thus, one can conclude that passive control produces a decrease of the wave drag, a unique strong shock being replaced by two weaker shocks over a great part of the channel flow. The total entropy production by the shock waves will be less, compared to the reference case where the shock (C,) occupies nearly the entire channel. This mechanism is well known in the design of supersonic air intakes where supersonic-subsonic compression is achieved through a succession of shocks in order to reduce efficiency loss. A careful examination of the schlieren pictures shows the existence of wavelets revealing that the flow is still supersonic behind the trailing shock (C,), in a region in contact with the boundary layer. Going towards the channel central part, this supersonic region disappears, the shock (C,) becoming stronger on approaching the triple point 1. The schlieren pictures obtained with Plates 2, 3 and 4 have shown similar modifications of the flow structure in the control region. III.3 - Surface pressure distributions The surface pressure distributions on the channel lower wall for the reference case and the four control cases are plotted in Figure 4 (the local static pressure 199x. 110 1

is referenced to the freestream stagnation pressure

Pstd

Concerning the reference case, the decreasing part of the distribution corresponds to the expansion in the supersonic part of the channel. The minimum pressure is reached at X = 245 mm. If one assumes an isentropic relation, the local value of the Mach number in the inviscid flow, contiguous to the boundary layer, is there equal to 1.33. The interaction with the shock wave produces a rapid rise of the pressure to a downstream nearly constant level p/pstC, = 0.63, giving an outer flow Mach number equal to 0.84 (here, the isentropic assumption may be less accurate). The general shape of the curve is typical of a transonic interaction without (noticeable) separation. The surface pressure distribution obtained with Plate 1 shows that the start of the pressure rise practically coincides with the cavity forward edge which is located at X = 235 mm. The local outer flow Mach number just upstream of this abscissa is equal to 1.31; i.e., slightly lower than in the reference case. The curve exhibits the three inflection points typical of shock wave/boundary layer interaction with separation. In this situation, the pressure first undergoes a rapid rise associated with the leading oblique shock (C, ). Then, the slope decreases with the tendency (weakly marked in the present case) towards an intermediate constant level (the plateau of an interaction with extended separation). Thereafter, the pressure gradient increases again, remaining less intense than in the first part of the interaction. Finally, after a third inflection, the pressure tends towards the constant downstream level. The five pressure taps located in the cavity give the same value p/p,,, = 0.472 which is indicated in Figure 4. It practically corresponds to the intermediate inflection point. The surface pressure distributions measured for the three other plates lead to no obvious differences between the four distributions. It is to be noticed that the cavity pressure is slightly higher for Plate 3 (inclined holes, diameter 0.15 mm) than for the other plates (p/pStO = 0.480 instead of 0.472). No

68

R. Bur et al. value of 1.18. In the triangular supersonic region, comprised between (C,) and the trailing shock (C,), the Mach number varies from 1.18 to 1.08, the flow undergoing an isentropic compression. The associated pressure rises correspond, respectively, to the quasidiscontinuity observed in the first rising part of the distributions shown in Figure 4 and to the level of the second inflection point. The Mach number on the downstream face of (C,) varies between 1.02 in the vicinity of the boundary layer edge and 0.98 at the outer border of the explored domain. A region of supersonic flow, bounded by iso-Mach line iU = 1, exists downstream of (C,), the outer inviscid flow undergoing a nearly isentropic compression to a Mach number of 0.85 at the end of the explored region. Y (mm)

convincing explanation of this difference has been found. Hence, differences between the curves corresponding to the tested perforated plates are extremely small. By considering these results, as also the visualizations, it is not possible to decide about an influence of the hole diameter and/or inclination. III.4 - Mean flowfield properties

The mean and turbulent properties of the flowfield will be discussed from the measurements performed with the LDV system. This discussion will be based on a limited number of typical results in order to have a synthetic view of the investigated phenomena. Also, since the influence of the plate characteristics is weak, only the case of Plate 1 will be discussed thoroughly and compared with the reference case. The contour lines of the Mach number are traced in Figure 5 for the reference case, by adopting the same scale (a) for the X-wise and Y-wise distances in order to have an objective view of the field structure, and also with an Y-wise dilatation (b) to have a better resolution of the boundary layer region. This tracing shows the system of compression waves generated by the initial thickening of the boundary layer inner region. The apparent thickness of the shock is due to an insufficient refinement of the measurement mesh (X-wise spacing of 5 mm) which did not permit a correct capture of discontinuities oriented along a normal direction. Y (mm)

(4

220 04 Fig. 6. - Contour (Plate I ). lines

270

320

X (mm)

of the Mach

number.

With

passive

control

(4

220 (4 Fig. 5. - Contour wall). lines

270

320

X (mm)

of the Mach

number.

Reference

case (Solid

The contour lines of the Mach number of the flow under passive control are traced in Figure 6 with identical Y-and X-scales (a) and with an enlarged Y-scale (b). This figure shows that, in the outer inviscid flow, the leading oblique shock (Cl) provokes a first supersonic compression of the flow from an upstream Mach number of 1.30, to a downstream

Figures 7 and 8 show mean velocity vector plots with a greater dilatation of the Y-scale emphasizing respectively the near solid wall and perforated plate regions. Some streamlines are superimposed on these plottings to visualize the fluid motion. In the case with passive control, the picture (Fig. 8) reveals a region where the mean velocity component E is negative. Its maximum vertical size is approximately 0.8 mm, for a longitudinal extent of 35 mm (between X = 270 mm and X = 305 mm). Thus, the back flow region is extremely flat and is contained in a thin zone in contact with the wall. The streamlines reveal the rapid turning of the flow in the downstream part of the control region. On this tracing, the streamlines penetrate into the cavity in the part of the perforated plate beginning at X = 260 mm, which indicates a suction effect in a region where the local pressure is still inferior to the cavity pressure, according to the pressure distribution shown in Figure 4. This apparently too early beginning of the suction effect is also observed for the other plates. The longitudinal evolutions of the boundary layer characteristic thicknesses are plotted in Figure 9,
Armpace Science md Technology

A Basic Experimental

Investigation

of Passive Control/he

etude fondamentale

du contrGle passif

69

Y W-4

5 ,Y(mm)

0 -2 2 X (mm) Fig. 7. - Mean velocity vector plot with region. Reference case (Solid wall). dilatation of the near wall

0 -2 220

240

260

280 dilatation

300 fl mm)

320

Fig. 8. - Mean velocity vector plot with region. With passive control (Plate I).

of the near wall

respectively for the reference case (a) and the passive control case (b). For the flow under control, the displacement thickness 6* reaches a maximum of 4.7 mm at X N 31.5 mm, leading to a ratio 6$,,/S,* = 9.06, to be compared with the value 5.1 of the reference case. This large difference in the increase of s*, while the thickening of the boundary layer is comparable in the two cases, proves that passive control provokes a greater destabilization of the boundary layer; i.e., the velocity profiles are less filled. After going through its maximum, S* decreases slowly in both cases. For the passive control configuration, downstream values of S are affected by a scatter due to a bad localization of the boundary layer outer edge. Moreover, the rapid rise of S* taking place in the first part of the interaction is felt by the outer supersonic flow as a ramp effect of angle: cp = tan-l(&*/dX). In the passive control case, the value of p is 3.7, which is nearly equal to that of the reference case (3.5). The essential difference is that, in the presence of control, the amplitude of the S* rise is much more important. The evolution of the momentum thickness f3 is of special interest since it is a measure of momentum loss undergone by the flow because of dissipative effects. Thus, its rise gives an indication on the friction drag exerted on the surface over the extent of the interaction domain. For the flow under control, 19 reaches a maximum of 2.04 mm at the end of the explorated domain, to which corresponds a ratio 0,/o, = 9.27, to be compared with the value 6.14 of the reference
1998. no 1

Fig. 9. - Evolutions

of the boundary

layer

characteristic

thicknesses.

case. At this step, it can be concluded that the friction drag is greater when passive control is applied. The results relative to the other tested plates lead to no obvious effect of the plate characteristics on the mean flow properties. Table 2 summarizes typical values of the boundary layer global characteristics
Table 2. - Typical values of the boundary layer global characteristics.

70

R. Bur et al. Y (mm)

20 t

for all the tested configurations, by giving the maximum values reached by 6* and the incompressible shape parameter Hi and the final value of 0. Due to measurement uncertainty, these results do not show any significant effect of the hole diameter or inclination. III.5 - Turbulent flowfield properties

The turbulent field properties are strongly modified by the presence of passive control in the interaction region. The contour lines of the turbulent shear stress -uv (normalized by the square of the reference velocity U,) are plotted in Figures 10 and 11, with an enlarged Y-scale, respectively for the reference case and the control case. It is seen that the maximum in shear stress is almost twice higher under passive control conditions (0.0075 against 0.004 for the reference case) and located farther to the wall, both of them being reached near X = 340 mm. Moreover, the turbulence level downstream of the interaction region remains very high, particularly when passive control occurs (its decreasing after the maximum being rather slow). Y (mm)
20 r

220 Fig. 12. ~ Contour case (Solid wall). lines

270 of the turbulent

320 kinetic energy.

X (mm) Reference

Y (mm) 2o t
10

0 220 Fig. 13. - Contour control (Plate 1). 270 lines of the turbulent kinetic 320 energy.

X (mm) With passive

0 270 lines of the turbulent shear 320 stress.

220 Fig. 10. - Contour (Solid wall).

X (mm) Reference case

Y (mm)

10

220 Fig. 11. - Contour control (Plate 1). lines

270 of the turbulent

320 shear stress.

X (mm) With passive

= 310 mm, whereas they occur downstream for the turbulent shear stress (X 2 340 mm). This lag of the shear stress is a feature commonly observed in shock wave/boundary layer interactions [ 141. In order to allow a comparison between the reference case and all the tested plates, the X-wise variations of the local maxima of the turbulent kinetic energy and shear stress are shown in Figure 14. The delayed rise in these quantities for the reference case is due to the fact that the interaction starts at X 21 265 mm on the solid wall, whereas it begins at X = 235 mm on the perforated plates. Moreover, it is clear that passive control produces a higher rise in turbulence levels, which is mainly due to the larger distortion of the velocity profiles, but the plottings reveal a somewhat different behaviour between plates with normal holes (Plates 1 and 2) and inclined holes (Plates 3 and 4). The rise in turbulent kinetic energy is more important for the inclined holes, the results for Plates 1 and 2, on the one hand, and Plates 3 and 4, on the other hand, being in good agreement. The effect on the shear stress is less important. The plottings also exhibit the out-phasing between I%and -u%, the maximum maximorum in -uu occurring farther downstream than the one in k for each configuration. III.6 - Determination of the total drag coefficient in the control region One way to evaluate the efficiency of the passive control device is to compute the total drag force F (and its associated coefficient C,) acting on the surface where control is applied, for each tested plate and the

The contour lines of the (pseudo) turbulent kinetic energy Ic (normalized by U,) are plotted in Figures 12 and 13. The general rise in turbulence intensity in the passive control case is also observed with Ic. However, the maxima of k (0.016 for the reference case and 0.022 for the control case) are located around X

A Basic Experimental

Investigation

of Passive Control/he

etude fondamentale

du contr6le passif

71

0 x 0 A 0.008

Reference Plate 1 Plate 2 Plate 3 Plate 4

case

(- )max/UO

where the P and F vectors are respectively associated with pressure and viscous forces, ? is the velocity vector and n the unit normal vector relative to the considered element of surface. The projection of this relation on the longitudinal X axis allows to obtain the friction drag F. To evaluate this balance, it is assumed that the pressure in the entry and exit sections is constant along Y which may constitute an approximation. Also, having seen that the pressure in the cavity is practically constant and that the velocities are quite below those of the external flow, the possibility of the cavity contribution to the drag was not considered. The associated total drag coefficient is defined as:

0.006 1 0.004 B a.0 .

8 .
(mm)

0.002

I x

250 I Cavity

300

350

Fig. 14. - Streamwise shear stress maxima.

variations

of the turbulent

kinetic

energy

and

where F is the friction force per unit of surface and the reference values pO, U, (freestream) are the values at the coordinates X = 220 mm, Y = S,. Table 3 gives the values of the total drag coefficient CF obtained by this momentum balance calculation. This table also indicates the corresponding percentage of decrease in CF, by comparison with the solid wall, defined as: ACF =
CF(solid) CF(solid) C~(porons) *

solid wall. For this calculation, the measured flow properties were used. By comparison with the value obtained for the reference case, we will be able to know if passive control has a favourable effect on shock wave/turbulent boundary layer interaction. The drag computed here is the component of the force applied to an element of surface (chosen as a portion of the channel lower surface). As this surface is rectilinear and parallel to the Y = 0 axis, the force results only from the action of friction. In making this evaluation, the flow is assumed two-dimensional; i.e., constant along the spanwise direction. The force acting on the surface element is determined by calculating the momentum balance for the control volume (S) defined as follows: - The surface element at ordinate Y = 0 is comprised between the abscissas X = 220 mm and 365 mm (respectively the first and last stations of the LDV probing), and includes the region of control; - The upper boundary is rectilinear and parallel to the surface. It is placed at an ordinate higher than the maximum boundary layer thickness; - The entry and exit sections are perpendicular to the surface and located at the abscissas X = 220 mm and 365 mm, respectively. The momentum balance gives on this control surface w
(P + F+ .11 .s
1998, no 1

Table

3. - Total

drag

coefficient.

Solid (Ref)

Plate I normal holes 0.15 mm 0.00972 1.9

Plate 2 normal holes 0.30 mm 0.00926 6.6

Plate 3 inclined holes 0.15 mm 0.00946 4.5

Plate 4 inclined holes 0.30 mm 0.00952 3.9

ClACF(%)

0.0099 I

pp.

Z)QdS

= 6,

The above results show a slight decrease (around 4%) of the perforated plates average coefficient value compared to the solid wall value. This modest gain of drag is of the order of the measurement uncertainty; so that, no definitive conclusion can be drawn from these results. Moreover, they do not give clear indications about differences which could be due to the characteristics of the plates (diameter and inclination of the holes). As seen in a previous study [5], this shock location (i.e., the shock centered on the cavity) is not the best one to obtain the most important reduction of the drag in the control region; better results were found for farther downstream shock locations.

72

R. Bur et al. of the perforated plates

III.7 - Evaluation excrescence drag

In these experiments, the shock wave is located downstream of the control region, the expansion part of the channel being followed by a region of uniform flow of Mach number 1.36. The explorations were comprised between the control region origin (X = 235 mm) and its end (X = 30.5 mm) and executed with the one-component version of the LDV system, special precautions being taken to make measurements as close to the wall as possible. Also, before calculating the boundary layer integral thicknesses and deducing the excrescence drag, the measured velocity profiles were faired, when necessary, to reduce the scatter of the results. For a flat plate boundary layer (zero pressure gradient), the Von Karman integral equation reduces to the following relation between the X-derivative of the momentum thickness and the skin friction coefficient:
--L-.

twice the solid wall value. However, these results do not show clear tendencies due to differences in the characteristics of the plates. IV - CONCLUSION A detailed experimental investigation of transonic shock wave/boundary layer interaction under passive control conditions has been performed in a channel type how in order to work with a thick boundary layer allowing refined explorations. The solid wall reference case and four passive control configurations, with different perforated plates, have been studied by using a two-component LDV system to determine the mean and turbulent flowfield properties. For all the tested configurations, the shock location in the outer flow was maintained fixed (middle of the control region) which insured the same overall pressure rise. The exploitation of the results shows the following tendencies: - When passive control is applied, the wall transpiration occurring in the upstream part of the perforated plate induces a viscous ramp effect which provokes the formation of a leading oblique shock wave. The supersonic zone following this shock is terminated by a nearly normal trailing shock, the two shocks meeting at a triple point from which starts the main shock. Thus, the inviscid part of the flow adopts a lambda pattern, typical of large shock-induced separation in transonic flows. The net effect of this flow organization is to lead to a substantial reduction of the wave drag, the compression being achieved through a two-shock system instead of a unique strong shock; - The boundary layer destabilization is more important in the interactions with passive control than in the reference case. Thus, when control is applied, a reversed flow region forms and the rise in the boundary layer momentum thickness over the interaction region is SO% greater with passive control than in the reference case. This result tends to prove that the friction drag is significantly higher when passive control is applied. This negative effect could offset the gain in wave drag; - As far as the mean flow properties are concerned, there is no conclusive effect of the perforated plates characteristics (hole diameter and inclination), the observed differences being within the uncertainty margin of the results; - Application of passive control provokes a more important rise in the turbulence level of the flow, compared to the reference case. Thus, the maxima in turbulent shear stress and kinetic energy can be respectively 90% and 40% higher than for the interaction on the solid wall. This behaviour has to be correlated with the greater distortion of the mean velocity profiles occurring under passive control conditions. It is probable that the more intense turbulence production is not a direct consequence of

C<

d0

dX Thus, for a distance over which the variation of H can be considered as linear, it is a simple matter to determine the skin friction, which is then constant. By applying this method (a straight line is fitted to the data points by the least square method) to the reference case, one finds: c, = 0.00354. This value is in fair agreement with a classical Aat plate formula which gives [ 151:
c, = 0.00341. Table 4 gives the values of the skin friction coefficient deduced from the derivative dO/dX (assumed constant over the plates) for the all tested configurations. The table also indicates the corresponding percentage of increase in C,, by comparison with the solid wall, defined as:

AC,=
Table 4. - Perforated

Cf

(holes)

f(solid)

Cf (solid)

plates

excrescence

drag coefficient

Solid (ReO

Plate I normal holes 0.15 mm 0.00753 113

Plate 2 normal holes 0.30 mm 0.00832 135

Plate 3 inclined holes 0.15 mm

Plate 4 inclined holes 0.30 mm 0.00718 103

Cf

0.00354 -

AC,(%)

The evaluation of the skin friction for perforated plates leads to an average value which is more than

A Basic Experimental

Investigation

of Passive Control/he

&de fondamentale

du contrdle passif

passive control effect, but results from the greater mean flow destabilization produced by the injection taking place in the upstream part of the control region. Moreover, there is a clear effect of the plates characteristics on the turbulence properties, the
maximum turbulence levels being substantially higher for the inclined holes cases; - A determination of the total drag in the control region was made. It was found that passive control

induced a modest decrease (around 4%; i.e., of the order of the measurement uncertainty) of this drag,
compared to the solid wall case, Moreover, it was not

possible to deduce clear tendencies about the effect of the characteristics of the tested perforated plates
(diameter and inclination of the holes), and also to

propose an optimal passive control device; - Measurements performed with obturated holes, in a
uniform slightly supersonic flow (Mach number equal to I.36), show that the skin friction of the perforated

plates is more than twice the skin friction

of the
an

solid wall (0.008 instead of 0.0035, considering

average value for the four perforated plates). Thus, the excrescence drag of the wall in the control region may compromise the benefit of passive control, as far as a reduction in drag is looked for. The obtained results have allowed to establish
well documented data banks which will be useful to improve and validate methods used to predict the performance of transonic airfoils equipped with control devices. Considering the modest gain (if any) achieved with passive control, the future studies will be oriented towards the consideration of active control where suction is performed in the cavity. Also, a combination

of passive and active control can be envisaged to limit

boundary layer thickening, while keeping the gain in wave drag brought by passive control.

REFERENCES
[ 1] Thiede P., Krogmann P. - PassiveControl of Transonic Shock/Boundary Layer Interaction. IUTAM Symposium Transsonicum III, Gottingen, Germany, May 24-21, 1988, Springer-Verlag, Berlin, 1989. [2] Bohning R., Jungbluth H. - Turbulent Shock/Boundary Layer Interaction with Control. Theory and Experiment. IUTAM Symposium Transsonicum III, Giittingen, Germany, May 24-27, 1988, Springer-Verlag, Berlin, 1989. [3] Raghunathan S., McIlwain S.T. - Further Investigations of Transonic Shock Wave/Boundary Layer Interaction with Passive Control. Journal of Aircraft, Vol. 27, no 1, Jan. 1990, 60-65.

[4] Kim I., Chokani N. - Navier-Stokes Study of Supersonic Cavity Flowfield with Passive Control. Journal of Aircraft, Vol. 29, no 2, March-April 1992, 217-223. [5] Bur R. - Passive Control of a Shock Wave/Turbulent Boundary Layer Interaction in a Transonic Flow. La Recherche Aerospatiale, no 1992-6, Nov.-Dec. 1992, 1l-30. [6] Chokani N., Squire L.C. - Transonic Shock Wave/Turbulent Boundary Layer Interactions on a Porous Surface. Aeronautical Journal, Vol. 97, May 1993, 163-170. [7] Schnerr G.H., Dohrmann U., Sadi O., Zierep J. Numerical and Experimental Investigation of Passive Control of the Shock/Boundary Layer Interaction in a Transonic Compressor Cascade. ICFM-II, Beijing, China, July 7-10, 1993. [8] Kim I., Sung B. - Computation of Passively Controlled Transonic Wing. Journal of Aircraft, Vol. 32, no 2, March-April 1995, 349-354. [9] Bohning R., Doerffer P. - Wind Tunnel Tests of Shock/Boundary Layer Interaction with Passive Control. Porous Wall Flow Investigation. Numerical Simulation of the Wind Tunnel Tests. EUROSHOCK Report TR AER 2-92-49/1.3, Karlsruhe University, Germany, Dec. 1995. [IO] Yeung A.F.K., Squire L.C., Faucher X. - The Passive Control of the Interaction between Swept Shocks and Boundary Layers. EUROSHOCK Report TR AER 2-92-49/l .2, Cambridge University, England, 1995. Bur R., Delery J. - Study of PassiveControl Applied to a Transonic Shock Wave/Boundary Layer Interaction. EUROSHOCK Report TR AER 2-92-49/1.4, ONERA, Chatillon, France, Jan. 1996. Chanetz B., Pot T. - Etude fondamentale sur le controle passif applique a une interaction onde de choclcouche limite en transsonique. Premiers resultats dessais. ONERA Report RT 75/7078 AN, Chatillon, France, July 1988. Boutier A., dHumieres Ch., Soulevant D. - Three Dimensional Laser Velocimetry: a Review. 2nd International Symposium on Applications of Laser Anemometry to Fluid Mechanics, Lisbon, Portugal, July 1984. Delery J. - Experimental Investigation of Turbulence Properties in Transonic Shock/Boundary Layer Interactions. AIAA Journal, Vol. 21, no 2, Feb. 1983, 180-185. Cousteix J. - Turbulence et couche limite. CepaduesEditions, Aerodynamique, 1989.

Acknowledgments.

- The present research has benefited of the financial support of the European Union through the EUROSHOCK I project.

1998, no 1