Effects of trips on the oscillatory flow of an axisymmetric hypersonic inlet with downstream throttle

Wenzhi GAO,Zhufei LI,Jiming YANG,Yishn ZENG,*

aSchool of Mechanical Engineering,Hefei University of Technology,Hefei 230009,China

bDepartment of Modern Mechanics,University of Science and Technology of China,Hefei 230027,China

1.Introduction

Hypersonic inlets must operate in a started mode for efficient operations as the air capture and compression portion of hypersonic airbreathing vehicles.1However,a hypersonic inlet can be unstarted through various factors,such as improper design,low flight Mach number,and high backpressure,among others.Among these factors,a high backpressure is induced by fuel injection and combustion in the combustors,which is encountered at the acceleration or cruise stages of flights.

Unstart problems caused by combustor backpressures have been largely investigated in wind tunnels2–5or through numerical simulations.6,7High combustor backpressures are mimicked primarily with mechanical throttling of flow,2–4,8–10mass injection,11,12or heat release.5,13According to prior studies,separated flows emerge and propagate upstream under the influences of backpressures.These separated flows can remain at a certain location with a quasi-steady motion of shock trains under a relatively ‘low” backpressure.Once the backpressure is sufficiently high,separations would propagate upstream and eventually unstart the inlet flow.Then,oscillatory flow can appear as a violently unsteady mode of unstart flow.2–4,8Wagner et al.2observed a high-amplitude oscillatory flow in an inlet/isolator model as the inlet unstarted.Tan et al.3,8investigated oscillatory flows and the unstarting process of a rectangular inlet.Substantial separations and pressure fluctuations were observed and specified as ‘big”or ‘small”buzzes according to oscillatory amplitudes.Li et al.4studied the unsteady behaviors of a two-dimensional inlet in a shock tunnel;the results indicated that the unstart flows were accompanied with a shock wave oscillation.

Oscillatory flows are highly detrimental to inlet performances and flight operations.Flow separations and spillages reduce the captured oxidizer,leading to combustor flameout and thrust loss.Pressure fluctuations and shock oscillations generate unsteady aerodynamic loads that may harm structuralsafety.Therefore,preventing oscillatory flows is necessary.Based on the findings of previous works,some studies have been conducted to prevent or delay oscillatory/unstart flows,14–16using mainly active control methods.Valdivia et al.14investigated the effects of energizing the sidewall boundary layer of an isolator on the unstart process.Vortex generators,vortex generator jets,and their combinations were tested,and the combined method improved the inlet’s sustainment to backpressure and reduced pressure fluctuations.Im et al.16used a plasma actuator to manipulate unstart flow.The plasma actuation could arrestboundary layer separations and delay the unstart process.

Additional instruments of manipulation would increase flight loads and complicate vehicle operations,although the above control devices can provide effective mitigations on unstart flows.Passive flow control devices,such as vortex generators,may be suitable and are worth studying in terms of robustness.Furthermore,discrete roughness arrays(also called trips)mounted on the forebodies of flight vehicles17,18can be a more suitable choice than vortex generators,because these trips are part of the vehicles.Trips are designed to enhance boundary layer transitions through streamwise vortexes generated in wake flows.Previous studies showed that trips could suppress the separations of started flows19or mitigate unstarts caused by large bluntness,20which are similar to the roles that vortex generators exert on separated flows.21However,to the best of the authors’knowledge,few studies have focused on the effects of trips on oscillatory flows.

Consequently,the effects of trips on the oscillatory flows of a hypersonic inlet are investigated experimentally in this paper.An axisymmetric inlet model is selected to avoid corner vortexes encountered in configurations with sidewalls.Schlieren imaging and transient pressure measurements are adopted to record unsteady flow features during experiments.The effects of trips on oscillatory flows are analyzed,and instructions for practical applications are provided.

2.Experimental facilities and methods

2.1.Facilities and experimental conditions

Experimental studies are conducted in a conventional wind tunnel running in a blowdown mode at the Nanjing University of Aeronautics and Astronautics22and a shock tunnel at the University of Science and Technology of China.4Detailed descriptions of the two facilities can be found in the previous literatures.The shock tunnel is a facility of the authors’institution and can be operated conveniently.Corresponding studies are also conducted in the conventional wind tunnel,because studies on oscillatory flows of a hypersonic inlet have rarely been conducted in impulse facilities.

In Table 1,the freestream Mach numbers of the conventional wind tunnel and the shock tunnel are 6.0(nominal)and 5.9,respectively.A corresponding total temperature and a total pressure of both wind tunnels are regulated to supply the freestream with unit Reynolds numbers of 5.41×106m-1and 5.03×106m-1,respectively.The experimental time of the conventional wind tunnel is much longer than that of the shock tunnel.Based on the time(t)history of the total pressure(pt)in Fig.1,the conventional wind tunnel can supply 4 s of stable flows in a blowdown mode,while the experimental time of the shock tunnel is approximately 20 ms for equilibrium interface operations.Each case is fired twice in the shock tunnel to ensure repeatability of experiments in the impulse facility.

2.2.Test model and measurements

The test model is an axisymmetric hypersonic inlet designed with a shock-on-lip Mach number of 6.5.23The model shown in Fig.2 has an inlet capture diameter of 128 mm,and the forebody nose is blunted with a radius of 0.8 mm.The captured flows are compressed with an initial conical surface inclined at 10°and the following curved surfaces with a total turning angle of 9.7°upstream from the inlet entrance.The flowturning angle at the entrance is 9°,while the internal compression section smoothly transits to a 50 mm horizontal isolator with a height of 4.72 mm.The total and internal contract ratios of the inlet are 6.41 and 1.58,respectively.A 90 mm long section with a divergent angle of 1.3°acts as a simplified combustor chamber,which leads to an exit height of 6.72 mm.

Oscillatory flows are generated through a block(Fig.2(a))fixed at the model exit with a Throttle Ratio(TR)of 66%.The TR is defined as follows:

Table 1 Test conditions of conventional wind tunnel and shock tunnel.

whereAt,bis the geometrical throat area of the flow passage in the block region,andAeis the exit area of the inlet model without blockage.

1 mm thick Teflon trips are carved in a diamond shape with side lengths of 3 mm and an internal angle of 60°.In Fig.3,12 trips are circumferentially adhered to the end of the removable nose part,located at 75 mm downstream from the nose in the axial direction.

Schlieren imaging and surface pressure measurements are adopted to record the inlet flows.In Fig.2(a),a round region of 200 mm in diameter provides limited schlieren observation in the shock tunnel,while the 300 mm diametric observation region of the conventional wind tunnel can cover the entire external flowfield.To obtain the pressures on the centerbody surfaces,13pressure transducers(NS-2,ShanghaiTM Automation,Inc.)are mounted and labeled asS3toS15(see Fig.2(a))in the shock tunnel.The corresponding 18 pressure transducers labeled asC1toC18(CYG1001,Kunshan Shuangqiao Sensor,Co.Ltd.)are connected to the centerbody surfaces in the conventional tunnel.

Schlieren imaging is equipped with high-speed cameras for the wind tunnels,and the typical frame rates are 15000 Hz for the shock tunnel(Photron FASTCAM SA5 1000 K-M2)and 4000 Hz for the conventional wind tunnel(Phantom V710).The acquisition rates of pressure are 1 MHz for the shock tunnel and 1000 Hz for the conventional wind tunnel.Pressure measurements and schlieren imaging are synchronized with a digital delay generator(Stanford Research Systems Model DG645)in the shock tunnel,but are recorded independently in the conventional wind tunnel.

Fig.1 Chamber pressure time histories of two wind tunnels during a typical run.

Fig.2 Experimental model.

Fig.3 Drawing and photograph of trips.

3.Experimental results

3.1.Started flow,TR=0%

The effects of the trips are evaluated primarily under started flows as the TR equals 0%before the experiments of oscillatory flows.The schlieren photos obtained from both facilities demonstrate well-started flows in Figs.4 and 5,while the smaller schlieren mirrors of the shock tunnel provide a limited view of the external flow field in Fig.5.For the untripped case of the conventional wind tunnel in Fig.4(a),the external shock system consists of leading-edge shocks,compression waves generated from curved surfaces,and separated shocks near the entrance.The separated shocks and separations in Fig.4(a)exhibit a typical flow pattern of the boundary layer separation.24Static pressures(p,normalized with the freestream pressurep0)along the centerbody surfaces in Fig.6 further indicate the scale of separation regions,as a pressure plateau starts fromX=0.23 m(the separation point)and ends with a pressure rise atX=0.28 m(the reattachment point).

Trips exhibit prominent effects on the started flow in the conventional wind tunnel.Based on the schlieren image,additional shock waves are induced near the trips,and no perceptible separation could be noticed near the entrance in Fig.4(b).In Fig.6,the surface pressures of the tripped case are evidently lower than those of the untripped case asX＞0.23 m,mainly because of the suppression of the separations.The surface pressures of the tripped case overlap with those of the untripped case asX＜0.23 m,suggesting the negligible effects of trips on the airflow compression.

The effects of trips on started flows are different in the shock tunnel.According to the schlieren images in Fig.5,no perceptible separation could be observed in Fig.5(a)for the untripped case,while negligible variations of the external flowfield can be noticed despite the trip-induced shocks in Fig.5(b).The surface pressure data in Fig.7 indicate a good coincidence between tripped and untripped cases,although theS11signal of the tripped case is blanked because of the transducer damage.

The discrepancies between the started flows in the two wind tunnels may be caused by the differences in the natural transition locations of the centerbody boundary layers.The transition would start earlier in the shock tunnel than in the conventional tunnel under a similar freestream unit Reynolds number considering the smaller nozzle size and the higher level of disturbances caused by the impulse operation mode.25Given an earlier transition process,no perceptible separation is observed in the shock tunnel,because the boundary layers can sustain the pressure gradient induced by the incident cowl shock wave.However, flow separations are generated by the cowl shock wave/boundary layer interactions in the conventional tunnel,where the centerbody boundary layers may still be laminar.

The discrepancies of the untripped flows further influence the effects of the trips.For the separated,started flow in the conventional wind tunnel,the trips show prominent suppressions of the separation regions and influences on the centerbody pressures.However,minor modifications of the flow structures or surface pressures are exerted on the unseparated,started flow in the shock tunnel.The mechanisms of trips are related to streamwise vortexes in the wake flow.On one hand,the streamwise vortexes induce additional disturbances in the boundary layers thereby enhancing the transition process,leading to a good sustainment to the pressure gradient without separations.On the other hand,they could energize the flow in the separation regions and decrease the separation scales.These effects probably cause the variations of the started flow in the conventional tunnel.However,the streamwise vortexes have a limited influence on the main flow structures or the air compression of unseparated flows,resulting in minor effects on the started flow in the shock tunnel.

Fig.4 Schlieren images of conventional wind tunnel as TR=0%(horizontal knife).

Fig.5 Schlieren images of shock tunnel as TR=0%(vertical knife).

Fig.6 Static pressures of the centerbody surfaces as the TR=0%in the conventional wind tunnel.

Fig.7 Pressure distribution of the centerbody surfaces as the TR=0%in the shock tunnel.

3.2.Oscillatory flow,TR=66%

3.2.1.Schlieren images

(1)Conventional wind tunnel

An oscillatory flow appears under a TR of 66%.Schlieren images and surface pressure signals exhibit typical features of oscillatory flows.Schlieren images of a typical oscillatory cycle are illustrated in Figs.8 and 9,where shock waves are labeled with numbered arrows.The untripped results of the conventional wind tunnel are analyzed firstly.In Fig.8(a),the moment when the inlet unstarts is labeled as the start of an oscillatory cycle,i.e.,time zero.The inlet is unstarted by exit throttling,of which the detailed process has been illustrated in the previous literature.2–4Briefly,the equilibrium between the capture and the exhaustion of airflow is broken by the exit blockage,leading to an accumulation of high-pressure gas in the inlet duct.Under a high backpressure,flow separations formulate and propagate upstream,which eventually unstart the inlet.

The separation regions enlarge and the separated shocks further move upstream,driven by the high backpressure in the inlet duct as depicted in Fig.8(a)–(c).The external separations can reach the nose regions and the leading-edge shocks are detached in Fig.8(c).The high-pressure gas in the inlet duct is exhausted through the exit and external flow spillage.The external shock waves move downstream,and the inlet undergoes a short phase of restarting as the backpressure decreases,as presented in Fig.8(d)and(e).The inlet restarts at the moment of 2.25 ms,and the external flowfield is similar to that of the started flow,comparing Fig.8(e)with Fig.4(a).The inlet-air-capture capability restores in the restarting process that leads to accumulating high-pressure gas,and the inlet unstarts again at 3.75 ms as illustrated in Fig.8(f).Then a new oscillatory cycle starts.

According to the schlieren images in Fig.8,the oscillatory flow undergoes a process of‘inlet unstarts→ upstream movement of separated shocks→downstream movement of separated shocks→ inlet restarts→ inlet unstarts.” The oscillatory period is 3.75 ms for the untripped case of the conventional wind tunnel.The shock wave fronts in Fig.8 are blurring,probably due to the limited 90 μs exposure time of schlieren imaging.

Trips generate more remarkable effects on oscillatory flows than those on started flows in the conventional tunnel.Compared to Fig.8,Fig.9 also exhibits a similar flow process,which is ‘inlet unstarts → upstream movement of separated shocks→downstream movement of separated shocks→inlet restarts→ inlet unstarts.” However,the separation regions are prominently suppressed.The farthest locations of the sep-arated shock feet are nearly axially 109 mm upstream from the inlet entrance,as shown in Fig.9(c).This location is also approximately 142 mm downstream from the nose region.The separated shock fronts of the tripped case are steeper and closer to the model surfaces than those of the untripped case when Fig.9(b)–(d)are compared to Fig.8(b)–(d).

Fig.8 Schlieren images during a typical buzz cycle of the untripped case in the conventional wind tunnel(horizontal knife).

The discrepancies that can be observed between Figs.8 and 9 demonstrate the effects of trips on the oscillatory flow structures,which are probably related to the streamwise vortexes induced from trips.In the previous literature,19the streamwise vortexes can enhance the transition process and momentum exchange from the main flow to the boundary layers.Both effects would improve the sustainability of the boundary layers to the adverse pressure gradient.The streamwise vortexes bring additional interactions with the separations regions,especially for opponent interactions as external separations moving upstream,leading to steep shock fronts.The steep shock fronts indicate strong shock strength,suggesting a good resistance to the separations caused by backpressures.Therefore,the movement scales of external separations are reduced.The oscillatory period of the tripped case decreases from 3.75 ms to 3.25 ms because of the effects of trips.

(2)Shock tunnel

The oscillatory flows in the shock tunnel are also divided,similar to those in the conventional wind tunnel,into ‘inlet unstarts→upstream movement of separation region→downstream movement of separation region→inlet restarts→inlet unstarts”,as illustrated in Figs.10 and 11.The separation scales of the untripped case exceed the observing scope in Fig.10(c).However,15000 Hz schlieren imaging(with an exposure time of 7 μs)provides a good time-resolution of oscillatory flows,and the shock waves in Figs.10 and 11 are clearer than those in Figs.8 and 9.According to the schlieren images in Fig.10,the oscillatory period is 4.0 ms for the untripped case in the shock tunnel.

Trips also generate remarkable influences on the oscillatory flows in the shock tunnel.Firstly,the scale of external separations is limited to within the scope of the schlieren image in Fig.11(c).The farthest upstream locations of the separated shock feet are approximately axially 131 mm downstream from the nose region as compared to 142 mm of the conventional tunnel in Fig.9(c).A similar distance of separated shocks indicates comparable oscillatory flows between the two wind tunnels.Secondly,the shock fronts and separated regions are more disordered in Fig.11(b)than those in Fig.10(b).These results suggest phenomena of interactions between the streamwise vortexes and the separation regions,especially for opponent interactions as the separated shocks move upstream.Similar but weakly disordered flow structures can be observed in Fig.11(d),indicating weak influences of the streamwise vortexes as the separations move downstream.Under the effects of the trips,the oscillatory period decreases to 3.13 ms for the tripped case in the shock tunnel.

Fig.9 Schlieren images during a typical buzz cycle of the tripped case in the conventional wind tunnel(horizontal knife).

Fig.10 Schlieren images during a typical buzz cycle of the untripped case in the shock tunnel(vertical knife).

Fig.11 Schlieren images during a typical buzz cycle of the tripped case in the shock tunnel(vertical knife).

3.2.2.Pressure signals

The dynamic pressure signals measured in the shock tunnel also exhibit typical characteristics of oscillatory flows.The static pressure time histories of three survey points(S8,S12,andS14)are shown in Fig.12,from which prominent fluctuations of the pressure signals can be noticed during the experimental time.Furthermore,the magnitude and amplitude of the pressure fluctuations increase as the flow moves downstream,indicating a higher backpressure in the inlet duct.However,the pressure signals measured in the conventional wind tunnel fail to capture transient characteristics because of the limited acquisition rate and low resonance of the transducer system.

Evident discrepancies of pressure time curves exist between the untripped and tripped cases in Fig.12.Pressure signals of the tripped case exhibit higher magnitudes than those of the untripped case.For pressure signals ofS8in Fig.12(a),the fluctuation amplitude of the tripped case is nearly four times that of the freestream(4p0),compared to 2p0of the untripped case.The corresponding fluctuation amplitudes ofS12andS14signals are approximately 32p0and 90p0,respectively,for both the untripped and tripped cases.

Fast Fourier Transform(FFT)is performed to analyze the effects of trips on the time frequency(f)of the pressure signals.Fig.13 shows that the most prominent peaks of pressure signals are 260 Hz and 304 Hz for the untripped and tripped cases,respectively,in which the vertical coordinates are the unsteady pressure amplitudes(Ap,normalized with freestream pressurep0).These results indicate that trips would increase the dynamic frequency of the surface pressure signals,which is consistent with the analysis of the schlieren images.In Figs.10 and 11,the oscillatory periods are 4.0 ms and 3.13 ms for the untripped and tripped cases,respectively,in which the frequencies are nearly 250 and 319.5 Hz.The timefrequency values of the schlieren images coincide well with those of the pressure signals in the shock tunnel.

The static time-average pressure signals of both facilities are also analyzed to evaluate the time-averaged backpressure encountered in the oscillatory flows.The pressure magnitudes of the conventional wind tunnel are averaged with an integral time of 2.0 s as compared to all the buzz cycles for the shock tunnel.Fig.14 exhibits the time-averaged pressure（¯p）distributions of the centerbody when compared to the corresponding values of the started flows as the TR equals 0%.For the pressure curves of oscillatory flows in Fig.14,surface pressures increase monotonically as the flow moves downstream.The maximum values of the shock tunnel are 113p0and 92p0for the tripped and untripped cases,respectively,asX=0.358 m in Fig.14(a).For the conventional wind tunnels in Fig.14(b),the corresponding values are 139p0and 144p0,respectively,asX=0.36 m.

In Fig.14(a),the time-averaged pressures of the tripped case are evidently higher than those of the untripped cases in the shock tunnel.The discrepancies between pressure values generally increase as the flow moves downstream,and the maximum value is 21p0for the farthest downstreamS15transducer.The high values indicate high backpressures of the tripped case,which may cause a reduction in the internal movement time in an oscillatory cycle.

Fig.12 Static pressure time histories of typical survey points in the shock tunnel.

Fig.13 FFT results of pressure signals of typical survey points in the shock tunnel as the TR=66%.

The effects of trips on the time-averaged pressures of the conventional wind tunnel differ from those of the shock tunnel.In Fig.14(b),the pressure values of the tripped case are higher than those of the untripped case upstream from the throat section(X=0.32 m),but lower in the isolator.Besides,the maximum pressure value of the conventional tunnel is obviously higher than that of the shock tunnel,considering a similar freestream Mach number and measurement locations.The discrepancies may be due to two reasons.Firstly,the pressure transducers are connected to the model surfaces with conduits of 1.5 m long and 1.2 mm in diameter in the conventional tunnel.The propagation of sonic waves in the conduits costs a time order of the oscillatory period and thus they are insufficient for measuring dynamic pressure signals.High-pressure gas accumulated in the transducer cavities at the moment of‘pressure peak” fails to be expelled during the oscillatory period,especially for inlet sections with a high pressure magnitude and amplitude.This phenomenon would cause the ‘retention”of high-pressure gas,thereby enlarging the time-averaged pressure magnitude.Secondly,the flow separations of the untripped,started flowfield would further affect the pressure distributions of the oscillatory flows in the conventional tunnel.The separated shocks in Fig.4(a)can induce a high pressure magnitude in the inlet duct,which may cause a high isolator backpressure for the untripped case in Fig.14(b).

Fig.14 Time-averaged surface pressures of the oscillatory flows.

4.Discussions

Specifically,trips prominently reduce the upstream range of the oscillatory flows in both wind tunnels.The separated shock fronts of the tripped cases are steeper and more disordered than those of the untripped cases.Variations of the flow structures lead an increased oscillatory frequency and pressure magnitude.However,the mechanisms dominating the variations of the oscillatory flows remain unclear.The effects of the trips on the prevention of unstart or oscillatory flows must also be evaluated.

Figs.15 and 16 depict the dynamic pressure signals of a typical oscillatory cycle in the shock tunnel.Combined with the schlieren images in Figs.10 and 11,an oscillatory cycle is divided into five phases.Taking the untripped case as an example,Phases 1 and 2 are distinguished through the movement of external separations.According to the schlieren images in Fig.10(a)–(c),the time interval of the upstream movement of the external separations is defined as Phase 1.Subsequently,the inlet restarting process depicted by Fig.10(c)–(e)is defined as Phase 2.Phases 3–5 correspond to the internal movement of the oscillatory flows,which are divided based on the pressure signals.Phase 3 ends at the ‘wave trough” of theS14signal in Fig.15(b).Phase 4 ends at the instant when theS9signal starts to climb in Fig.15(a),indicating the duration as the backpressure transmits fromS14toS9.Finally,Phase 5 ends as the inlet unstarts,and a new cycle begins.

Similar to those in the untripped case,the pressure signals of the tripped case are also divided into Phases 1–5 in Fig.16.The phase durations of the tripped and untripped cases are listed in Table 2.The durations of Phases 1,3,and 5 are substantially shortened for the tripped case,and the time variations of Phases 2 and 4 are negligible,as shown in Table 2.Pressure distributions at the beginning and end moments of each phase are extracted and plotted in Fig.17.The time-averaged pressure in an interval of±0.02 ms is chosen for the instantaneous pressure at timet.Fig.17 is mainly used to analyze the variations of centerbody pressure magnitudes during the whole phase,and not to characterize the unsteady process of pressure changes.

As shown in Fig.17,the pressure curves generally increase along the streamwise direction,exhibiting a typical flow pattern propelled by a downstream backpressure.However,variations of the pressure magnitude during each phase are different for external and internal inlet sections.The effects of trips on the oscillatory flow process are discussed by combing the pressure signals in Figs.15–17 and the schlieren images in Figs.10 and 11.

Fig.15 Dynamic pressure signals of a typical buzz cycle of the untripped case in the shock tunnel.

Phase 1 corresponds to the upstream movement process of external separations as the inlet begins to unstart.In Fig.17(a),the centerbody pressure magnitudes of the internal sections decrease,whereas the pressure values of the external surfaces increase during Phase 1.Both the tripped and untripped cases display similar tendencies,although the pressure magnitudes of the tripped case are higher than those of the untripped case.Moreover,in the inlet duct section,the pressure decrements of the untripped case are evidently higher than those of the tripped case.From the schlieren images in Figs.10(b)and 11(b),the high-pressure gas in the inlet duct is exhausted through the external spillages and the inlet exit,leading to a pressure decrement in the internal sections.Under the effects of the trips,the upstream ranges of the oscillatory flows are substantially reduced,which shortens the duration of Phase 1 and narrows the scale of external spillages.These effects further reduce the exhaustion of the high-pressure gas in the inlet duct and the pressure decrement of the internal sections.The increments of external pressures are caused by the upstream movements of the separated shocks and separated flow.The tripped case in Fig.11(b)displays steeper shock fronts and stronger compression than those of the untripped case in Fig.10(b),which leads to higher pressure values of the external surfaces in Fig.17(a).

Phase 2 corresponds to the downstream movement process of external separations,i.e.,the restarting process.During the process,the airflow capture ability of the inlet is gradually restored,while the internal high-pressure gas is exhausted through the external spillages and the inlet exit.The centerbody pressures generally decrease during Phase 2 for both the untripped and tripped cases,as shown in Fig.17(b).The pressure decrements of the tripped case are comparable to those of the untripped case,and the pressure values are significantly higher than the latter.The similar pressure decrement may be related to the similar time internals of Phase 2.However,the external pressure values of the tripped case are higher than those of the untripped case at the end of Phase 2.This finding contradicts the results in Fig.7,which occurs when the inlet starts.The stream wise vortexes generated from the trips may have caused the contradiction.Such vortexes enhance the momentum exchanges of the main flows,which reduces the responses of the transducers by hindering the exhaustion of highpressure gas in the cavities, although these have negligible effects on air compression.The pressure data in Phases3–5 support the assumption,as the external pressures of the tripped case gradually decrease due to this phenomenon in Fig.17(c)–(e).

Meanwhile,the air capture ability is restored as the inlet restarts at the end of Phase 2.The captured high-speed air flow would gradually spread downstream and accumulate in the duct sections during Phase 3.The high-speed air flow modifies the internal flow structures and decreases the surface pressures.However,the accumulated air flow tends to generate a high backpressure from the blockage sections and propagates upstream.In Fig.17(c),theS15pressure signal increases during Phase 3,indicating the locations reached by the backpressure.Despite theS15survey point,the centerbody pressures decrease in Fig.17(c)for both the tripped and untripped cases.Compared to the untripped case,the time required for the formation and propagation of a backpressure is shorter for the tripped case due to a higher initial pressure.Therefore,the tripped case exhibits a shorter time interval in Phase 3 than that in the untripped case.

Fig.16 Dynamic pressure signals of a typical buzz cycle of the tripped case in the shock tunnel.

In Phases 4 and 5,the backpressures further propagate upstream and eventually unstart the inlet.Driven by the backpressures,the internal pressures increase during Phases 4 and 5 for both the tripped and untripped cases,as shown in Fig.17(d)and(e),respectively.Moreover,the external pressures of the untripped case remain constant under the started flows,while those of the tripped case decrease due to the low responses of transducers caused by the effects of streamwise vortexes.Phase 4 characterizes the time at which the backpressure propagates from theS14survey point to theS9survey point.As shown in Table 2,the duration of Phase 4 is minimally influenced by the trips.However,the mechanisms dominating the process are difficult to reveal without visualized flow data.Subsequently,Phase 5 begins with an initial rise ofS9pressure signals and ends with an unstart of the inlet.The trips prominently shorten the time interval from 0.50 ms to 0.14 ms.The shorter duration of the tripped case may be caused by the higher initial pressure values compared to those of the untripped case,which requires a smaller pressure increment and a shorter time to reach the unstart threshold.

According to the above analysis,the streamwise vortexes generated from the trips substantially modify the separated flow structures and reduce the upstream movement ranges.These effects shorten the time consumed by the upstream movement and reduce flow spillages caused by external separations.The reductions of both the flow time and flow spillage would limit the exhaustion of high-pressure gas in the inlet duct sections,resulting in a lower pressure decrement and a higher pressure magnitude of the tripped case than those of the untripped case.The higher pressure magnitude can also affect the subse-quent flow processes of the oscillatory flows.During the restarting process,the effect of trips on the time intervals and pressure decrement of the internal sections are minimal,and the internal pressure values are higher for the tripped case than those for the untripped case.The higher pressure values of the internal sections further reduce the time,during which the backpressures formulate downstream the isolator.Subsequently,the effects of trips on the upstream propagation in the isolator and internal contract section are mainly reflected on the higher pressure magnitude;the influences on the propagation time are minimal.However,the trips significantly shorten the duration of the process from the arrival of the backpressure near the entrance to the unstarting of the inlet.The mechanisms may be related to the higher internal pressure caused by the trips,although it is deeply difficult to analyze these without a visualization of the internal flows.

The external pressure magnitudes of the tripped case are prominently higher than those of the untripped case.This may be related to two reasons.On one side,the steeper separated shocks tend to increase the strength of the airflow compressions.On the other hand,the streamwise vortexes generated from the trips enhance the momentum exchanges between the main flows and the boundary layers,which in turn,decrease the responses of the transducers as the vortexes hinder the exhaustion of high-pressure gas in the cavities.

Generally,trips substantially increase the oscillatory frequency and time-averaged pressure values.These are unfavorable to oscillatory flows.Specifically,the reduced duration of Phase 5 in Figs.15 and 16 increases difficulties in the detection and prevention of unstart/oscillatory flows.However,trips substantially reduce the external separation ranges,which is favorable to the control of oscillatory flows.The present experimental results are not very pertinent to the prevention of inlet unstart as the present flows are in a highly oscillatory mode.The effects of trips on unstart preventions need to be studied elaborately in the future.

Table 2 Time intervals of various phases in an oscillatory cycle of the shock tunnel.

5.Conclusions

The effects of 1 mm thick trips on oscillatory flows of an axisymmetric hypersonic inlet are investigated experimentally in a conventional wind tunnel and a shock tunnel.Oscillatory flows are generated through a fixed block at the model exit with a TR of 66%,while the corresponding started flows are also studied as the TR equals 0%.The main results are as follows.

(1)For the separated,started flow in the conventional wind tunnel,the 1 mm thick trips exhibit prominent suppressions on the separation regions and evident variations on the centerbody pressures.However,minor modifications of the flow structures or surface pressures are exerted by the trips on the unseparated,started flow in the shock tunnel.

Fig.17 Centerbody pressure distributions of the start and end moments of various phases.

(2)Trips can prominently suppress external separations,shorten oscillatory cycles,and modify pressure signals of the oscillatory flows in both wind tunnels.Trips can reduce the upstream movement ranges of separated shocks from the nose regions to locations axially 142 mm downstream for a 251 mm long forebody in the conventional wind tunnel.The upstream oscillatory ranges of the untripped case exceed the schlieren scope in the shock tunnel,which are suppressed to locations approximately 131 mm downstream from the forebody nose by the trips.The oscillatory periods are decreased from 3.75 ms to 3.25 ms in the conventional tunnel and from 4.0 ms to 3.13 ms in the shock tunnel,respectively.According to results in the shock tunnel,the tripped cases generally exhibit higher pressure magnitudes and higher oscillatory frequencies than those of the untripped cases.The increment of time-averaged pressure is up to 21 times the freestream static pressure for the farthest downstream transducer.However,the pressure signals in the conventional tunnel are insufficient for analysis of oscillatory flows due to measurement limitations.

(3)Effects of trips on oscillatory flows are probably related to the streamwise vortexes in wake flows.Trips improve the momentum exchanges between main flows and the boundary layers and enhance the transition process,which would improve the sustainment of the boundary layers to the adverse pressure gradient.Besides,the streamwise vortexes bring additional momentum exchanges to separation regions and lead to steeper shock fronts,especially for opponent interactions as separations move upstream.The steeper shock fronts indicate stronger shock strength,suggesting better limitations to the separations caused by downstream backpressure.Therefore,the movement scales of external separations are reduced,which further leads to increments of the oscillatory frequency and pressure magnitudes.

(4)The present results indicate that trips can exert substantial effects on oscillatory flows of hypersonic inlets.However,the flow mechanisms dominating the effects are still not clarified.Further investigations are still needed,such as studies on different types of inlet configurations,parametric research on various backpressure levels,and the flow mechanisms of trips.The effects of trips should be concerned in the design and studies of hypersonic vehicles,especially for investigations of unstart flows.

Acknowledgements

This study was co-supported by the China Postdoctoral Science Foundation(No.2017M612059),the Fundamental Research Funds for the Central Universities of China(JZ2015HGBZ0471),and the National Natural Science Foundation of China(Nos.11402263 and 11132010).

1.Van Wie DM.Scramjet inlets.In:Murthy SNB,Curran ET,editors.Scramjet propulsion.Reston,VA:AIAA;2000.p.447–511.

2.Wagner JL,Valdivia A,Clemens NT,Dolling DS.Experimental investigation of unstart in an inlet/isolator model in Mach 5 flow.AIAA J2009;47(6):1528–42.

3.Tan HJ,Li LG,Wen YF,Zhang QF.Experimental investigation of the unstart process of a generic hypersonic inlet.AIAA J2011;49(2):279–88.

4.Li ZF,Gao WZ,Jiang HL,Yang JM.Unsteady behaviors of a hypersonic inlet caused by throttling in shock tunnel.AIAA J2013;51(10):2485–92.

5.Im S,Baccarella D,McGann B,Liu Q,Wermer L,Do H.Unstart phenomena induced by mass addition and heat release in a model scramjet.J Fluid Mech2016;797:604–29.

6.Trapier S,Deck S,Duveau P.Delayed detached-eddy simulation and analysis of supersonic inlet buzz.AIAA J2008;46(1):118–31.

7.Wang WX,Guo RW.Numerical study of unsteady starting characteristics of a hypersonic inlet.Chinese J Aeronaut2013;26(3):563–71.

8.Tan HJ,Sun S,Yin ZL.Oscillatory flows of rectangular hypersonic inlet unstart caused by downstream mass- flow choking.J Propul Power2009;25(1):138–47.

9.Wang CP,Xue LS,Tian X.Experimental characteristics of oblique shock train upstream propagation.Chinese J Aeronaut2017;30(2):663–76.

10.He XZ,Zhou Z,Qin S,Wei F,Le JL.Design and experimental study of a practical osculating inward cone waverider inlet.Chinese J Aeronaut2016;29(6):1582–90.

11.Do H,Im S,Mungal MG,Cappelli MA.The influence of boundary layers on supersonic inlet flow unstart induced by mass injection.Exp Fluids2011;51(3):679–91.

12.Shimura T,Mitani T,Sakuranaka N,Izumikawa M.Load oscillations caused by unstart of hypersonic wind tunnels and engines.J Propul Power1998;14(3):348–53.

13.Liu QL,Passaro A,Baccarella D,Do H.Ethylene flame dynamics and inlet unstart in a model scramjet.J Propul Power2014;30(6):1577–85.

14.Valdivia A,Yuceil KB,Wagner JL,Clemens NT,Dolling DS.Control of supersonic inlet-isolator unstart using active and passive vortex generators.AIAA J2014;52(6):1207–18.

15.Kodera M,Tomioka S,Kanda T.Mach 6 test of a scramjet engine with boundary-layer bleeding and two-staged fuel injection.Reston,VA:AIAA;2003.Report No.:AIAA-2003-7049.

16.Im S,Do H,Cappelli MA.The manipulation of an unstarting supersonic flow by plasma actuator.J Phys D:Appl Phys2012;45(48):485–502.

17.Berry S,Daryabeigi K,Wurster K.Boundary-layer transition on X-43A.J Spacecraft Rockets2010;47(6):922–34.

18.Matthew PB,Steven PS,Thomas JJ.Effect of freestream noise on roughness-induced transition for the X-51A forebody.Reston,VA:AIAA;2008.Report No.:AIAA-2008-0592.

19.Gao WZ,Li ZF,Yang JM.Flow characteristics experiments of a hypersonic axisymmetric inlet with nose bluntness.Acta Aeronaut Astronaut Sin2015;36(1):302–10[Chinese].

20.Zhang HJ,Shen Q.Experimental studies of leading edge bluntness effects on hypersonic inlet.J Propul Technol2013;34(10):1316–20[Chinese].

21.Lu FK,Li Q,Liu CQ.Microvortex generators in high-speed flow.Progress Aerosp Sci2012;53:30–45.

22.Xu X,Cheng KM,Wang ZJ,Xu Y.Flow field calibration of NUAA Φ0.5m hypersonic wind tunnel.J Exp Fluid Mech2009;23(4):77–81[Chinese].

23.Gao WZ,Li ZF,Yang JM.A hybrid CFD/characteristics method for fast characterization of hypersonic blunt forebody/inlet flow.Sci China:Phys,Mech Astr2015;58(10):38–45.

24.De´lery J.Physical introduction.In:Babinsky H,Harvey JK,editors.Shock wave-boundary-layer interactions.New York:Cambridge University Press;2011.p.39–43.

25.Schneider SP.Hypersonic laminar–turbulent transition on circular cones and scramjet forebodies.Progress Aerosp Sci2004;40(1):1–50.