STARTING OF GENERIC INLET WITH BLUNTED WEDGES

Bluntness e ̈ect of gas-compressing wedges on starting and §ow structure in an air inlet was investigated experimentally. The inlet was of internal compression type with §at walls and rectangular cross section. The experiments were carried out in the wind tunnel UT-1M at Mach numbers M = 5 and 8 and Reynolds numbers Re∞L from 2.8 ·106 to 23 ·106. The §ow characteristics were measured by panoramic optical methods. Data demonstrating in§uence of wedge bluntness radius on the inlet starting were obtained at di ̈erent Mach and Reynolds numbers as well as at di ̈erent contraction ratios. Ambiguity of the §ow regime in the inlet under certain conditions was found.


INTRODUCTION
Starting of supersonic inlet is an interesting problem both from theoretical and practical points of view and attracted many researches (see, for example, [116]).The problem arises when the inlet is shaped as a converging channel.In order to provide the inlet starting, one should limit the channel constriction that results in limiting of air compression.
It is known that leading edges of a hypersonic inlet should be blunt to exclude the surface overheating.Moreover, the sharp leading edges will be blunted during a hypersonic §ight due to ablation.In [4], it is shown that blunting of leading edges can result in signi¦cant decrease of pressure recovery and mass §ow rate as well as in unstart of the inlet.However, most other studies deal with sharp leading edges.Maybe, there was supposed that the leading edge bluntness plays a secondary role in the gas §ow.But in reality, as the present investigation shows, even small blunting of the leading edges at high contraction ratio can in §uence the inlet starting similar to the signi¦cant increase of Mach number or decrease of Reynolds number.
In this work, a schematized inlet of internal compression type with §at walls and rectangular cross section was studied.Such inlets are not used practically, but the observed phenomena can meet with real §ow con¦gurations.
The present paper is the completion of studies started in [15] and continued in [16].The subject of the study in [15] was a pair of wedges located on sharp or blunt plate.In [16], the pair of wedges was replaced by the inlet and the in §uence of plate-and cowl-bluntness on the §ow was investigated.In the present paper, the same model as in [16] was used, but the study is focused on the wedgebluntness e¨ect on the inlet starting.

TEST MODEL CONFIGURATION AND FLOW CONDITIONS
Figure 1 presents the experimental model.On the §at plate 1, the wedges 2 with cowl 3 are located.The wedges have changeable leading edges 4 for variation of the bluntness radius in the range from 0 to 4 mm.The cowl is transparent behind the leading edge to provide the visibility of the §ow inside the inlet.In some tests, the wedge manufactured of a transparent material was used.The shape of the model studied can be characterized by the following nondimensional values a¨ecting the §ow in the inlet: the relative distance of the inlet from the plate leading edge X 0 /W 0 = 1.29 and the relative inlet height H/W 0 = 0.8.The channel contraction ratio η = W 0 /W t was varied by replacement the wedges: when the wedge thickness was 37.5 mm, the throat width was 25 mm and  89.0 770 η = 4. Correspondingly, when the wedge thickness was 33.3 mm, the throat width was 33.3 mm and η = 3. Figure 1b illustrates the shock waves and rarefaction waves in the twodimensional (2D) §ow of nonviscous gas at M = 5.Behind the wedge shocks, the pressure is 4.8 times higher in respect with the undisturbed §ow.The shock re §ected from the symmetry line increases the pressure 3.4 times additionally.Thus, the pressure inside the rhombus located between the points A and B is 16.4 times higher than that in the undisturbed §ow.Between the points C and D, the pressure falls almost to the undisturbed level (P/P ∞ = 1.08).
The tests were carried out in the TsAGI wind tunnel UT-1M.It worked in the Ludwieg-type mode.The steady §ow duration was 40 ms.The averaged characteristics of undisturbed §ow are presented in Table 1.
Scatter of total pressure values in di¨erent runs was ±6% of the mean pressure at P t = 62.9 bar and ±10% at P t = 33.6 bar.The total temperature scatter was ±2%.The surface temperature was approximately 293 K; so, the temperature factor T w /T 0 at M ∞ = 5 was 0.56 and at M = 8, it was 0.39.

EXPERIMENTAL METHODS
Investigations of heat transfer and pressure distribution were performed by the thin luminescent paints sensitive to the temperature or to the pressure, correspondingly [17].The methods use the temperature quenching of luminescence of organic dye.At heat-transfer investigation, the measured increase of temperature during the determined time interval is used to calculate the heat §ux from gas to the surface in each point.The surface §ow is visualized by the method recently developed at TsAGI [18].It is based on measurements of small movements of oil marked by contrast §uorescent particles.In addition, the method presents visulization of shear stress distribution.
The above described methods are applied in the present work similar to [16] for investigation of external and internal §ow as well.To do that, some parts of the cowl and one wedge were made of a material transparent to the exciting ultraviolet (UV) light and to the luminescent visible light.The transparent parts were used in two ways: (i) to measure the heat §ux to the transparent surface; and (ii) to measure the heat §ux to the opposite nontransparent surface.Thermal characteristics of appropriate material were used at solving the equation of heat di¨usion.The data presented below demonstrate a satisfactory agreement between the experimental results and calculated heat §ux values on the cowl and on the plate under the cowl.
According to the estimations, the total related measurement error lies in the range from 10% to 30% depending on the absolute value of heat §ux and pressure.

STARTING OF A CONVERGENT CHANNEL
The channel can be referred as started if the channel ¤swallows¥ the supersonic jet coming to the front area of the channel W 0 (Fig. 2, left).In contrast, if some part of the coming jet ¤splash out¥ and does not get inside the channel, the channel is referred as unstarted (Fig. 2, right).
Considering the staring process in the one-dimensional (1D) approach at constant characteristics of undisturbed supersonic §ow, one can determine three ranges of contraction ratio η = W 0 /W t (see Fig. 2, W t is the throat width).
(1) if η < η 1 , the channel is always started and a supersonic §ow exists inside the channel.Autostarting is provided, and no starting device is necessary; (2) if η 1 < η < η 2 , two §ow regimes are possible: with supersonic speeds inside the channel (started channel) or with subsonic speeds inside the channel (nonstarted or blocked channel).In the second case, the bow shock wave forms in front of the channel and some part of the coming jet, corresponding

Figure 2 Ranges of channel contraction ratio
to the entrance area W 0 , §ow outside.The implementation of the regime depends on the prehistory of the §ow.At nonstart, the total pressure losses are higher than at start; and (3) if η > η 2 , starting of the channel is impossible and the §ow with detached bow wave realizes.
The value of η 1 can be determined from the mass §ow rate equation Here, ρ ∞ and U ∞ are the gas density and velocity in the undisturbed §ow; and ρ * and U * are similar values in the channel throat.It assumed that normal shock forms ahead the channel entrance.Behind the normal shock, between the channel entrance and throat, gas accelerates isentropic to the sound speed U * and its density decreases to the corresponding value ρ * .Equation ( 1) implies the Kantrovitz£ relation [1]: where γ = C p /C v is the ratio of speci¦c heats at constant pressure and volume.
At η > η 1 , the maximal contraction ratio η 2 , permitting the supersonic §ow inside the channel, is also limited by the channel throat, and the value of η 2 can be also obtained from Eq. (1).But in this case, the gas density in the throat should be de¦ned using the value ρ ∞ and taking into account the total-pressure losses in the channel due to the gas viscosity and presence of shock waves.In the special case of isentropic §ow (similar to the contrary Laval nozzle), η 2 = 25 at í ∞ = 5 and η 2 = 190 at í ∞ = 8.But the total-pressure losses in the shock waves decrease signi¦cantly the η 2 -values: for the 2D inviscid §ow in the channel, shown in Fig. 1b, η 2 = 14.7 at í ∞ = 5 and η 2 = 51 at í ∞ = 8.Gas viscosity and blunting of leading edges diminish additionally the η 2 -values, as will be shown below.
In §uence of gas viscosity and leading-edges bluntness can be assessed approximately substituting the real throat width by an e¨ective width W t − 2(δ * + - * ) where δ * and - * are the thicknesses of boundary and high-entropy layers.Then, let us obtain for the e¨ective channel contraction ratio: The e¨ective coe©cient of channel contraction η ef does not consider the losses in the internal shocks.It also does not consider the three-dimensional (3D) e¨ects of interference between the wedge shocks and the boundary layers generated on the plate and cowl surfaces.Therefore, the η ef -value is not a stringent criterion of inlet blockage.It re §ects only The thickness of high-entropy layer can be estimated by assuming the following [19] (Fig. 3): the bow shock wave is equidistant to the leading edge surface; the thickness of the jet de §ected download (on the declined surface of the wedge) is equal r cos θ; gas expands isentropic from the total pressure behind the normal shock P st to the pressure on the inclined surface of the sharp wedge P e ; and the highentropy layer is uniform.The equation of mass §ow rate gives Here, -is the thickness of high-entropy layer; and ρ -, T -, and U -are the gas density, temperature, and speed in the high-entropy layer at big distance from the wedge leading edge.The last values can be estimated in dependence on the ratio P e /P st .The following relation can be obtained for the displacement thickness of high-entropy layer - * (subscript e refers to the nonviscid §ow on the sharp wedge): - The relation (3) shows that high-entropy layer gets thinner when wedge angle θ increases due to the increase of pressure P e and thinning the jet de §ected on the inclined surface of the wedge.For the angle θ = 15 • , the following estimates have been obtained: - * /r = 0.46 at M = 5 and - * /r = 0.62 at M = 8.
Starting of the inlet with contraction ratio η 1 < η < η 2 in a wind tunnel depends on the character of §ow start in the wind tunnel [7,8].In the impulse wind tunnel, usually, the ¤wave¥ start takes place.It means that the nozzle and the test chamber are vacuumed before the start.Then, the shock wave passes through the nozzle of the wind tunnel and through the channel of the inlet.At that, the supersonic §ow realizes usually in the inlet.But in the present study, it is revealed that in the inlets with contraction ratios η 1 < η < η 2 at some §ow characteristics, both subsonic or supersonic §ow regime can realize (see below).
In contrast, in long-duration wind tunnel, the §ow start is usually slow.At that, in the inlet with contraction ratio η 1 < η < η 2 , the subsonic §ow with bow shock wave is always set.

FLOW PATTERNS
The presented below experimental data demonstrate that one of three §ow patterns shown in Fig. 4 can realize in the inlets under investigation.
Regular §ow with internal shocks forms at low in §uence of gas viscosity and small bluntness of the leading edges (Fig. 4a).The shocks generate narrow local separation zones, in which gas §ows spirally in normal direction.The §ow near each wedge is the same as near the single wedge up to the point of intersection between the boundaries of wedge-in §uence regions.Behind this point, a more complicated §ow structure forms.It is investigated in detail in [20].
Symmetrical global separation zone (Fig. 4b) forms at blockage of a channel with moderate contraction ratio.If the inlet is located near the plate leading edge, the detached bow wave would be formed in front of the inlet.Due to the big distance from the plate leading edge to the inlet, the interference between the bow shock wave and plate boundary layer leads to formation of global separation zone and separation shock (see Fig. 4b).Location of the separation line depends on the related inlet height H/X 0 (see Fig. 1).At large values H/X 0 , the separation line is located near the plate leading edge.The separated boundary layer reattaches inside the inlet, near the wedges and the throat.
Formation of nonsymmetrical global separation zone with reverse §ow is possible in the inlet with big contraction ratio.In this case, the boundary of the separation zone ahead the inlet is conical as shown in Fig. 4c.But at some tests, symmetrical separation zone shown in Fig. 4b was observed at large contraction ratio, as well.

EXPERIMENTAL DATA. FLOW STRUCTURE AND HEAT TRANSFER
Flow structure and heat transfer were studied mainly at Mach number M = 5. Figure 5 demonstrates impact of wedge-leading-edge bluntness on heat transfer over the plate and cowl surfaces.At the cowl absence (Fig. 5Á), regular §ow with local separation zones, generated by the wedges, forms even at large wedge bluntness radius r = 2 mm and at high contraction ratio η = 4.It happens because some part of the gas, compressed by the shocks, §ow out along the wedges.
At the cowl presence (Fig. 5b and 5c), the regular §ow persists when the wedges are sharp or blunted weakly.But signi¦cant blunting, for example, by the radius r = 2 mm, results in high total-pressure losses and leads to the channel blockage and formation of global separation zone.The §ow can be symmetric or nonsymmetric.At the test, demonstrated in Fig. 5b (right), the separation §ow was nonsymmetric and the separation zone was conical in front of the channel.Near the top boundary of the separation zone, where the §ow reattaches to the plate, heat transfer was enhanced and near the opposite boundary, it was weakened.Similar nonsymmetry of heat transfer distribution was inside the channel.
Figure 6 gives additional information concerning the §ow pattern inside the inlet at large wedge bluntness when nonsymmetrical separation zone forms.The picture presents stream lines and visualizes shear stress distribution as well.On the cowl, a converging line can be seen in front of the throat.The line is removed from the channel symmetry line.Near the cowl, only some part of §ow penetrates into the throat and signi¦cant part of gas §ows down to the plate.The convergence line is similar to the line of sinks.Line of sources is formed on the plate (see Fig. 6b), under the line of sinks.From the source line, gas §ows to the plate leading edge and to the throat as well.The shear stress distribution corresponds to the §ow pattern: on the cowl (see Fig. 6a), shear stress decreases with the distance from the leading edge due to the thickening of the boundary layer and reaches a minimum at the convergence line; and on the plate (see Fig. 6b), the shear stress at ¦rst increases with the distance from the source line due to the §ow acceleration and then decreases with thickening of the boundary layer.
Distribution of pressure coe©cient C P is presented in Fig. 7 for the same conditions as St-number distribution.At the sharp wedges, the C P -distribution is similar to the St-distribution: local increase of pressure occurs behind the separation zones generated by the wedges and in the intersection of the separation zones.Much higher pressure increase is visible behind the wedge shocks and behind the shocks intersection point.But at big bluntness of the wedges, the pressure distribution is symmetrical in contrast to the St-number distribution.At that, almost isobarick region forms on the plate in front of the inlet.The reasons of formation of symmetric or nonsymmetric separation zones are not established.Apparently, the separation §ow is not stable and sensitive to small di¨erence in the wedge angle relative the §ow direction.
In the range from 0 to 1 mm, i. e., at regular §ow in the inlet, the variation of bluntness radius in §uences weakly the St-number distribution.Figure 8   turbulent transition occurs at the distance about 80 mm from the plate leading edge so that the wedge shocks interact with the turbulent boundary layer.On the cowl, the distance from the leading edge to the shocks intersection point is shorter, and the wedge shocks interact with the laminar boundary layer.But inside the separation zones, the transition occurs.This is testi¦ed by the comparison of maximum values of St-number on the cowl and on the plate: they are close to each other.Blunting the wedges with r ≥ 2 mm leads to the blockage the inlet and to signi¦cant increase of St-number both on the plate and on the cowl.The maximal St-number on the cowl is much lower than on the plate (see Fig. 8) because the separation shock does not reach the cowl.
In the undisturbed region on the sharp plate (at X < 190 mm), the experimental data are close to the predictions for laminar and turbulent boundary layers (see Fig. 8a).This concerns the measurements ahead the inlet (X < 129 mm) and inside it (165 < X < 190 mm) as well and demonstrates acceptable accuracy of the heat-transfer measurements on the plate through the transparent cowl.Consistency of the measurements with calculations on the cowl itself (X = 150176 mm, see Fig. 8b) is not so good.This can be connected both with measurements errors and the forward shock in §uence along the laminar §ow.At regular §ow, on the symmetry line of the plate and the cowl, two maxima are seen (see Fig. 8).The ¦rst one corresponds approximately to the intersection point of the separation shocks (here, St-number is 4 times bigger than in front of the interference region).The second maximum corresponds to the intersection point of the shocks generated by the wedges.At this point, St-number increases 1015 times depending on the §ow regime.
Figure 9 demonstrates heat transfer distribution on the inner surface of the sharp wedge at the presence of sharp cowl and slightly blunted plate.The panoramic investigation showed weakly variation of the heat transfer coe©cient along the height of the wedge; therefore, the St-distribution is presented for the middle section (Y = 40 mm) only.Comparison of the experimental data with the prediction shows that at í = 5 and Re ∞L = 23 • 10 6 , the transi-Figure 9 Stanton distribution along the symmetry line of the wedge at í = 5 and Re∞L = 23 • 10 6 : 1 ¡ experiment; 2 ¡ prediction for the turbulent boundary layer; 3 ¡ prediction for the laminar boundary layer; X = 0 corresponds to the wedge leading edge tion occurs on the wedge surface near its leading edge.At the same conditions, as shown above, laminar §ow is maintained on the cowl up to the region of interference between the wedge shocks.The di¨erence is partly due to the pressure increase and, correspondingly, to the increase of the local Reynolds number (approximately 1.5 times).Maybe, more important was the in §uence of shock vibrations.

INFLUENCE OF WEDGE BLUNTNESS ON THE INLET BLOCKAGE
The St-number value on the plate at the point with coordinates X = 235 mm and Z = 0 (see Fig. 8a) is used as the indicator of inlet blockage.Near this point, the ¦rst pick of heat transfer coe©cient forms.The St-number at this point varies slightly with wedge bluntness but only until the regular mode of §ow maintains.
When the inlet blockage happens, the St-number increases sharply.Figure 10 presents St-number at the characteristic point (X = 235 mm and Z = 0) in dependence on the related bluntness radius r/W t at M = 5.From Fig. 10, it follows that blockage of the inlet with η = 4 occurs at related wedge bluntness r/W t = 0.06 (r = 1.5 mm), and the inlet with smaller contraction ratio η = 3 is started even at maximal value of related bluntness radius studied r/W t = 0.12 (r = 4 mm) (in Fig. 10 and following ¦gures, the started inlets are marked by open symbols and the unstarted ones by ¦lled symbols).The boundary layer on the wedges near the throat is turbulent as can be concluded from Fig. 9. Thus, for the blockage conditions of the inlet with η = 4, the following estimation can be obtained: displacement thickness of the high-entropy layer generated by the wedge leading edges is - * = 0.9 mm at r = 1.5 mm, displacement thickness of turbulent boundary layer on the wedge surface near the throat is δ * = 0.79 mm (at that, the Reynolds number, calculated using the local §ow characteristics and the length S = 145 mm, is Re e,S = 16 • 10 6 ); the e¨ective contraction ratio is η ef = 4.6.For the inlet with η = 3, the maximal tested bluntness radius is r 3 = 4 mm, the corresponding displacement thicknesses are - * = 2.4 mm and δ * = 0.72 mm, and corresponding value of the e¨ective contracting ratio is η ef = 3.7.This value is much lower than in the previous case that correlates with the fact that at η = 3, the channel is started.Assuming that the critical value of contraction ratio is the same in both cases (η ef = 4.6), the blockage of the channel with η = 3 should occur at wedge bluntness radius r 3 = 8.5 mm.
Reynolds number has a great in §uence on the inlet blockage (Fig. 11).As mentioned above, blunting the wedges up to r = 4 mm (r/W t = 0.12 and η ef = 3.7) of the inlet with contraction ratio η = 3 does not result in channel blockage when the Reynolds number is high enough (Re ∞L = 22 • 10 6 ).But when the Reynolds number is signi¦cantly lower (Re ∞L = 8.5 • 10 6 ), the same inlet is blocked at much smaller wedge bluntness r = 2 mm (r/W t = 0.06, Fig. 11a).At that, the corresponding value of e¨ective contraction ratio for turbulent boundary layer is η ef = 3.4.
Figure 11Á re §ects an unexpected fact: at lower Reynolds number Re ∞L = 8.5 • 10 6 and í = 5 and η = 3, there is a range of related bluntness radii r/W t = 0.060.11(2 < r < 4 mm), wherein the inlet can be started and nonstarted as well.This fact will be discussed below.At bigger contraction ratio η = 4, Reynolds number impact on the §ow in the inlet gets stronger (Fig. 11b): at Re ∞L = 22 • 10 6 , the bluntness radius r = 1.5 mm (r/W t = 0.06 and η ef = 4.6) is necessary to block the inlet and at Re ∞L = 13 • 10 6 , even very small bluntness radius r = 0.3 mm (r/W t = 0.012 and η ef = 4.3) is enough for blockage.The reason of so strong Reynolds in §uence is not de¦ned.
At higher Mach number M = 8 and Re ∞L = 5.6 • 10 6 , the inlet with η = 4 is blocked, when the wedges are even sharp.At that, the corresponding value of e¨ective contraction ratio is η ef = 4.15, taking into account characteristics of laminar boundary layer on the wedge surfaces.The inlet with smaller contraction rate η = 3 can be started with signi¦cantly blunted wedges (Fig. 12): at Re ∞L = 5.6 • 10 6 , the blockage occurs when the related bluntness radius is r/W t = 0.06 (r = 2 mm and η ef = 3.4, Fig. 12a).Reducing the Reynolds number leads to lower value of critical bluntness radius: at Re ∞L = 2.8 • 10 6 , r/W t = 0.052 (r = 1.75 mm and η ef = 3.4, Fig. 12b).
Figure 12 shows that at Mach number M = 8 as well as at M = 5, there is a range of wedge bluntness, wherein the §ow regime is ambiguous.At M = 8 and Re ∞L = 2.8 • 10 6 , it happens in the ranges r/W t ≈ 0.050.11,1.75 ≤ r ≤ 3 mm, and η ef = 3.63.9.This phenomenon is probably was not observed up to now in short duration wind tunnels.Figure 12b demonstrates that in the case of regular §ow, the Stanton number, measured at the characteristic point (X = 225 mm and Z = 0), continued to decrease monotonically behind the branching point.It means that the observed phenomenon is not connected with the laminarturbulent transition.Apparently, the causes of the §ow ambiguous lies in uncontrolled di¨erences in forming rate of the §ow in the wind tunnel.The di¨erences can, for example, be in the rate of pressure rise or in the degree of vacuum before the test run.The other possible reason is the di¨erence in concentration of water vapor in the gas.

CONCLUDING REMARKS
Starting of generic inlet with sharp or blunted wedges was investigated experimentally at Mach numbers M = 5 and 8 and Reynolds numbers Re ∞L from 2.8 • 10 6 to 22 • 10 6 .The §ow structure and heat transfer were studied as well.The inlet was of internal-compression type with §at walls and rectangular cross section.
Increasing the blunting of wedge leading edges results in decreasing of heat transfer until the regular §ow structure with internal shocks and local separation zones maintains in the inlet.Exceeding a critical value, depending on the §ow characteristics and channel contraction ratio, leads to the inlet blockage and sharp increase of heat §ux.
At maximal Reynolds numbers studied, regular §ow in the inlet of internalcompression type is possible at contraction ratio up to η = 3, exceeding signi¦cantly the value η ≈ 1.6 for the autostarting inlet.At that, considerable wedge blunting is permissible (up to r = 24 mm at inlet width 100 mm).
Sensitivity to the wedge blunting enhances with increase of Mach number or channel contraction ratio and with decrease of Reynolds number as well.At high Mach number and low Reynolds number, even small wedge blunting can lead to the inlet blockage.
Ambiguity of §ow regime in the inlet is revealed at experiments in the wind tunnel of impulse type: at some §ow conditions, there is a range of wedge bluntness wherein subsonic or supersonic §ow regime can realize randomly.
At big contraction ratio, the inlet blockage can be accompanied by formation of nonsymmetric separation zone, in which gas §ows spirally and backward relative to the undisturbed §ow.
The performed experiments con¦rm applicability of the developed panoramic methods at investigations of the outer and the inner §ows, as well.

Figure 1
Figure 1 Photograph of the model (a) and cross section of the wedges (b): 1 ¡ plate; 2 ¡ wedges; 3 ¡ cowl; and 4 ¡ changeable leading edges of the wedges.Dimensions are in millimeters

Figure 3
Figure 3 Formation of a high-entropy layer the trend approaches the blockage or away from it at variations of bluntness radius and Reynolds number.The presented below data show that the inlet under study approaches the blockage when η ef closes to the value η ef = 4.5 at M = 5 and to η ef = 4 at M = 8.The thickness of high-entropy layer can be estimated by assuming the following[19] (Fig.3): the bow shock wave is equidistant to the leading edge surface; the thickness of the jet de §ected download (on the declined surface of the wedge) is equal r cos θ; gas expands isentropic from the total pressure behind the normal shock P st to the pressure on the inclined surface of the sharp wedge P e ; and the highentropy layer is uniform.The equation of mass §ow rate gives

Figure 5
Figure 5 Distribution of Stanton number on the plate and inner surface of the cowl at í = 5, Re∞L = 23 • 10 6 , and η = 4: (a) the plate at the cowl absence; (b) the plate at the cowl presence; (c) the cowl; 1 ¡ the plate; 2 ¡ the wedges; 3 ¡ shadows of the wedges; 4 ¡ nontransparent part of the cowl; 5 ¡ shocks; 6 ¡ separation zones; 7 ¡ reattachment zones; and 8 ¡ boundaries of the separation zone on the plate.Left column refers to r = 0 mm and right column to r = 2 mm

Figure 6
Figure 6 Surface stream lines and shear stress visualization in the inlet with blunted wedges (r = 4 mm) at M = 5, Re∞L = 23 • 10 6 , and η = 4: (a) cowl; (b) plate; 1 ¡ convergence line; and 2 ¡ divergence line demonstrates St-number distribution along the symmetry line of the plate and the cowl at two values of wedge bluntness: r = 0 and 2 mm.At r = 0 mm, laminar

Figure 7
Figure 7 Distribution of pressure coe©cient CP in the inlet with cowl at í = 5, Re∞L = 23 • 10 6 , and η = 4: (a) plate; and (b) cowl.The notations are the same as in Fig. 5. Left column refers to r = 0 mm and right column to r = 2 mm

Table 1
Flow characteristics