Instructions for preparing the manuscript of the papers for Proceedings of INTER-NOISE 2022 GLASGOW Philip Morris 1 and Darryl Douglas 2 The Pennsylvania State University Department of Aerospace Engineering 229 Hammond Building University Park, PA 16802, USA

ABSTRACT Aircraft designs may include an engine that exhausts over the upper fuselage of the aircraft. This design has many benefits including reduced noise below the aircraft. However, the scrubbing of the high-speed turbulent jet flow over the fuselage has the potential to cause structural fatigue. This paper describes an experimental/computational study of the surface pressure fluctuations induced by the supersonic jet flow. The model configuration consists of a rectangular convergent-divergent nozzle that exhausts over a flat plate. The plate is a continuation of the lower lip of the nozzle. Pressure measurements are made at several locations on the plate surface. In addition, schlieren visualization is used to identify the jet’s shock cell structure. Numerical simulations are performed using hybrid RANS/Large Eddy Simulation (LES). Results are compared between the simulations and the experiments. The importance of shock turbulent boundary layer interaction to the surface loading is identified.

1. INTRODUCTION

Many recent designs for manned and unmanned aircraft feature an engine exhaust integrated into the upper surface of the fuselage. An example is the Northrop Grumman X-47B which is an unmanned combat aerial vehicle (UCAV). It is designed for carrier-based operations. The embedded exhaust results in low observability. However, the hot supersonic jet exhaust interacts with the fuselage aft-deck and may cause structural damage or fatigue. Due to the sharp throat on a typical combat aircraft vehicle, the jet exhaust will contain shocks, even when operating in a nominally on-design condition. It is hypothesized that the shock turbulent boundary layer interaction will generate unsteady surface pressure loading at frequencies with the potential to cause structural damage.

Previous studies of the effect of an aft deck (and side walls) on the flow and acoustics of a supersonic rectangular jet include that by Tinney et al. [1]. They investigated the properties of an approximately 4:1 aspect ratio multistream nozzle with a design pressure ratio of 5.0. Core nozzle pressure ratios between 3.0 and 5.0 were considered. Aft deck surface pressures, schlieren visualization and far field pressures were acquired. Reynolds-Averaged Navier Stokes (RANS) simulations were performed and the agreement with the measured mean pressure on the deck and side walls was good. In a series of papers (Berry et al.[2], Magstadt et al.[3], Ruscher et al.[4])

1 pjm@psu.edu 2 dad5668@psu.edu

a Shea mar ce 21-24 AUGUST SCOTTISH BENT caso

made near and far field acoustics measurements of an effectively two-stream rectangular nozzle flow exhausting over an aft deck. Far-field microphone and aft deck pressure measurements were made. Particle Image Velocimetry and high speed schlieren were also used. They observed a 34 kHz oscillation throughout the experimental domain which they attributed to Kelvin-Helmholtz instabilities inside the nozzle. The effects of heating one or both streams were also reported.

The present paper reports on flow and surface pressure measurements for a 4:1 aspect ratio rectangular jet exhausting over an aft deck. The focus is on surface pressure fluctuations in the hundreds of Hertz, as relevant for structural response. In the interest of length only selected results from an ongoing research project are presented. Limited results from a hybrid RANS/LES are also presented.

The next section describes the experimental facility, the instrumentation, and the numerical simulation approach. This is followed by averaged schlieren visualizations. Average and unsteady surface pressures are then presented. Following a brief description of the numerical approach, some preliminary unsteady flow simulations are given. Finally, conclusions based on the completed results are provided. 2. EXPERIMATAL FACILITIES, INSTRUMENTATION, AND NUMERICAL SIMULATIONS 2.1. Penn State Jet Aeroacoustics Facility The experiments were conducted in the Penn State Jet Aeroacoustics Facility. A schematic is shown in Figure 1. The facility is capable of simulating 1-inch equivalent diameter unheated or heat-simulated jets to a jet exit Mach number of 1.7. Heated jets are simulated using helium-air mixtures to match the heated jet acoustic Mach number. The jet temperature and pressure are controlled by means of a control panel and data are acquired and recorded using a LabVIEW workstation. Further details about the history of the facility, its design and operation can be found in Doty [5], Doty and McLaughlin [6], and Veltin [7].

Figure 1. Schematic of the Penn State Jet Aeroacoustics Facility.

2.2. Nozzle and aft deck model

The rectangular jet nozzle has a 4:1 aspect ratio with a height of 0.443 in. and a width of 1.76 in. Downstream of the throat, the top and bottom walls are parallel, and the sidewalls diverge. The design Mach number equals 1.26, based on quasi one-dimensional isentropic flow relations. A previous model design had a separate aft deck section (see Lurie [8]) that was aligned with the

lower lip of the nozzle. However, it was found that this generated either a forward or backward facing step at the lower jet exit which, in turn, generated spurious shocks and expansions. The new nozzle and aft deck design is shown in Figure 2. The deck is integrated into the lower wall of the nozzle, which eliminates any step at the jet exit. Also shown in Figure 2 are the locations of holes for the pressure transducers along axial lines at the jet center and quarter span and in line with the nozzle exit sidewall.

Figure 2. Sketch of nozzle design: (a) isometric view, (b) CAD vertical centerline cutaway and, (c) CAD spanwise nozzle centerline cutaway.

2.3 Pressure measurements

For the pressure measurements, Endevco models 8507C and 8510B pressure transducers were used. These transducers were powered by Endevco model 136 DC amplifiers. Signals from the amplifier were transmitted to an NI DAQ system. Raw data was recorded in the DAQ system without any filtering.

The total record length for each run condition was 409600 samples with a sample frequency of 300.3 𝑘𝑘𝑘𝑘𝑘𝑘 . Mean pressure was calculated using the unfiltered pressure data. For the RMS and PSD pressure calculations a band-pass IIR filter was applied. This was done to remove high frequency content due to the pinhole arrangement of the sensors, the sensor resonance, and to isolate frequencies typically associated with Shock Turbulent Boundary Layer Interaction (STBLI) shock foot motion and sonic fatigue. The band-pass IIR filter had half power frequencies of 50 𝐻𝐻𝐻𝐻 and 2 𝑘𝑘𝑘𝑘𝑘𝑘 . A Hamming window was applied to the data with a window length of 16384 samples and a 50% overlap.

2.4 Schlieren visualization

A Z-type schlieren system was used for flow visualization. The components included are the light source side optics, mirror 1, mirror 2 and the camera side optics. The subcomponents of the light source side include a white LED, adjustable collimation adapter with a 75mm achromatic doublet lens and an adjustable rectangular slit. Both mirrors 1 and 2 are 8-inch diameter parabolic mirrors with a focal length of 64 inches. On the camera side is a razor blade mounted to traverses and a Phantom High-Speed Miro 310 camera with either a 300mm or 500mm focal length achromatic doublet lens.

2.5 Numerical simulations Numerical simulations were performed using Wind-US. Wind-US was developed by the National Project for Applications-oriented Research in CFD (NPARC) Alliance as a CFD research tool for government, industry, and academia [9]. Within the Wind-US package the combined RANS/LES model developed by Bush [10, 11] has been used. This model uses a shear stress limiter to transition between the SST RANS model and LES. Only the results of the simulations are provided here. Additional details are given by Douglas and Morris [12].

3. RESULTS

Experiments and simulations were performed for three jet operating conditions. Nozzle Pressure Ratios of 2.30 (overexpanded), 2.62 (nominally on design), and 3.50 (underexpanded). All the results included here are for unheated jets. The focus will be on the overexpanded case. 3.1 Schlieren visualization

Figure 3 shows schlieren photographs for the three operating conditions. Evident for each case

(a)

(b)

(c)

Figure 3. Instantaneous aft deck schlieren. (a) NPR=2.30, (b) NPR=2.62, (c) NPR=3.50.

are shocks that are generated upstream of the nozzle exit. These are generated at the sharp nozzle throat. Munday [13] observed this behavior for a circular conical nozzle resulting in a double-shock pattern in a slightly underexpanded jet: one set of shocks from the throat and one from the jet exit.

As expected, the shock angles increase as the jet Mach number increases. From shock- expansion theory, the intersection of the first shock from the upper nozzle lip with the aft deck would occur at approximately / 0.64. x H = 3.2 Surface pressure measurements

Figure 4a) shows the mean pressure levels on the jet centerline and the quarter and half span locations. The first peak in the mean pressure occurs at the first transducer location, / 0.67. x H = Subsequent peaks occur at / 2.23 x H = and / 3.79. x H = From the schlieren images the first peak corresponds to the shock intersection from the upper nozzle lip, however the integrated nature of the schlieren makes it difficult to identify the causes of the other peaks. The peaks at the quarter span location follow a similar trend, but the mean pressure at the half span is almost uniform and equal to atmospheric pressure. It should be recalled that the half span transducers are located below the nozzle side wall shear layer.

(a) (b) Figure 4. NPR=2.30 aft deck pressure measurements on the deck centerline, quarter span and

half span. (a) Mean pressure, (b) RMS pressure.

As seen in Figure 4b) the highest RMS pressures are on the deck centerline. The locations of these peaks do not align with the peaks in the mean pressure. So far as the first shock intersection is concerned, the numerical simulations indicate the presence of a separation bubble due to STBLI. All the RMS peaks have values between 150 and 155 dB with respect to 9 2.9 10 − × psi ( 6 20 10 − × Pa).

Because of the relatively coarse spacing of the transducers a sweep of the NPR from 2.2 to 2.4 in steps of 0.05 was performed. It was expected that the feet of the shocks would move relative to the transducer locations, giving some indication of the range of RMS levels possible. Figure 5 shows the variation of the RMS levels for two transducer, one at / 0.67 x H = and a second at / 3.79. x H = The variation in the RMS levels is considerable, varying between 146 and 162 dB for the transducer at / 3.79. x H = Thus, the RMS levels shown in Figure 4b) should be taken as a general indication of the pressure fluctuation amplitudes and are likely to be different were the transducers positioned slightly differently.

Of particular interest in structural response is the low frequency content of the surface pressure fluctuations. Dupont et al. 14 performed STBLI experiments and noted four identifiable zones. Of interest here is the reflected shock zone as it contains surface pressure fluctuations in the hundreds

of Hertz range due to the motion of the shock foot and the separated flow just downstream. Though the configuration considered by Dupont et al. 14 was different from the present geometry, a similar behavior is expected.

Figure 5. Variation of RMS pressure levels on the deck centerline as a function of NPR for transducers at x/H =0.67 and x/H =3.79.

Figure 6 shows the power spectral density of the surface pressure at centerline pressure sensor locations corresponding to the measured peak RMS values. The frequency resolution is 18Hz and the pressure signal is filtered between 50 and 2kHz. A clear peak occurs at approximately 120Hz and associated harmonics. Time-resolved schlieren measurements, not included here, were digitally analyzed and show the same dominant frequency associated with the back-and-forth motion of the shock.

Figure 6. NPR=2.30 experimental PSD along the deck centerline. The sensor locations correspond to the locations of peak RMS values.

3.3 Numerical simulations

In a complementary study numerical simulations have been performed using the hybrid RANS/LES turbulence model in WindUS. These simulations are ongoing and only some preliminary results are given here for the NPR=2.30 case. Figure 7 shows instantaneous contours of Mach number for a vertical slice taken on the jet centerline. The shock structure upstream of the

nozzle exit is visible and a comparison with the schlieren image in Figure 3a) shows good qualitative agreement (given the integrated nature of the schlieren image). Just downstream of the nozzle exit the jets starts by contracting, as expected for an overexpanded jet. Munday noted that this didn’t occur for the conical nozzle in his experiments as the flow had a small radial component at the jet exit. However, for the present jet geometry the top and bottom walls are parallel to each other, and it is the side walls that expand.

Figure 7. Simulated centerline Mach number contours, NPR=2.30.

A comparison of the simulated mean pressure on the jet centerline for NPR=2.30 is shown in Figure 8. Measurements are also shown for a range of NPR from 2.25 to 2.35. The agreement between the predictions and measurements is generally good in terms of levels but the locations of the peak pressure are not captured exactly. Note that the simulation includes predictions upstream of nozzle exit.

Figure 8. Predicted and measured mean surface pressures on the jet centerline, NPR=2.30.

Figure 9 shows the corresponding predicted and measured RMS levels. The simulated values are generally higher than the measurements by as much as 6-8 dB, especially downstream of

/ 2.0. x H = A number of iterations with different grid configurations and turbulence model parameters made little difference to the predicted levels. Testing is continuing, as well as simulations for the on-design (NPR=2.62) and underexpanded (NPR=3.50) cases.

4. CONCLUSIONS

Experiments and simulations have been performed for a supersonic rectangular jet exhausting over an aft deck. Flow visualization using schlieren and surface pressure measurements have been made. Preliminary hybrid RANS/LES simulations have also been performed. The mean pressure

measurements show peaks corresponding to the intersection of shocks with the aft deck surface. Peaks also appear in the RMS levels, but at locations downstream of the mean peaks. PSD measurements in the peak RMS level locations show low frequency fluctuations corresponding to the motion of the shocks. The numerical simulations provide some agreement with the measured mean surface pressures, but the RMS levels are higher than the measurements by 6-8 dB in further downstream locations.

Calculations are underway for the other experimental operating conditions and will be reported when complete. 5. ACKNOWLEDGEMENTS

The authors are grateful to Pratt and Whitney, a Raytheon Technologies company for their support in the earlier stages of this work. 6. REFERENCES

1. Tinney, C. E., Panickar, P. and Vogel, P. Aeroacoustics of a planar multistream supersonic nozzle with aft deck and sidewalls. AIAA Journal , 56(10) , 3926-3937 (2018). 2. Berry, M. G., Magstadt, A. S., Glauser, M. N., Ruscher, C. J., Gogineni, S. and Kiel, B. V. An acoustic investigation of a supersonic, multi-stream jet with aft deck: characterization and acoustically-optimal operating conditions. AIAA Paper 2016-1883 (2016). 3. Magstadt, A. S., Berry, M. G., Coleman, T. J., Shea, P. R., Glauser, M. N., Ruscher, C. J., Gogineni, S. and Kiel, B. V. A near-field investigation of a supersonic, multi-stream jet: locating turbulence mechanisms through velocity and density measurements. AIAA Paper 2016- 1639 (2016). 4. Ruscher, C. J., Gogineni, S., Kiel, B. V., Giese, A. J. and Viswanath, K. Noisy flow structures in a heated and unheated jet produced by a three-stream rectangular nozzle with an aft deck. AIAA Paper 2016-1634 (2016). 5. Doty, M. J., “An experimental investigation of the aeroacoustic properties of high-speed, helium/air mixture axisymmetric jets,” Ph.D. thesis, The Pennsylvania State University (2002). 6. Doty, M. J. and McLaughlin, D. K. Acoustic and mean flow measurements of high-speed, helium-air mixture jets. International Journal of Aeroacoustics, 2(3) , 293-333 (2003). 7. Veltin, J., “On the characterization of noise sources in supersonic shock containing jets,” Ph.D. thesis, The Pennsylvania State University (2008). 8. Lurie, M. B., “The prediction of unsteady aerodynamic loading in high aspect ratio wall bounded jets,” Ph.D. thesis, The Pennsylvania State University (2015). 9. Bush, R. H., Power, G. D., and Towne, C. E., WIND: The production flow solver of the NPARC Alliance, AIAA Paper 98-0935 (1998). 10. Bush, R. H., “ A two-equation large eddy stress model for high sub-grid shear, ” AIAA Paper 2001-2561 (2001). 11. Bush, R. H., “ A three dimensional zonal Navier-Stokes code for subsonic through hypersonic

propulsion flowfields, ” AIAA Paper 88-2830 (1988). 12. Douglas, D., Morris, P. J., Behlman, N. and Low, K. An experimental and numerical investigation of rectangular embedded jet exhausts. AIAA/CEAS Aeroacoustics Conference, Southampton, UK, 14-17 June 2022. 13. Munday, D. Flow and Acoustics of Jets from Practical Nozzles for High-Performance Military Aircraft. Ph.D. thesis, University of Cincinnati (2010). 14. Dupont, P., Haddad, C., and Debieve, J. F. Space and time organization in a shock-induced separated boundary layer. Journal of Fluid Mechanics , 559 , 255–277 (2006).