On streamwise vortices in high Reynolds number supersonic axisymmetric jets

S. A. Arnette, M. Samimy, and G. S. Elliott

Citation: Physics of Fluids A: Fluid Dynamics (1989-1993) 5, 187 (1993); doi: 10.1063/1.858803
View online: http://dx.doi.org/10.1063/1.858803
View Table of Contents: http://scitation.aip.org/content/aip/journal/pofa/5/1?ver=pdfcov
Published by the AIP Publishing

Articles you may be interested in

An experimental investigation of the flow dynamics of streamwise vortices of various strengths
interacting with a supersonic jet
Phys. Fluids 26, 086102 (2014); 10.1063/1.4892008

Spatial evolution of supersonic streamwise vortices

Phys. Fluids 26, 074102 (2014); 10.1063/1.4886097

Vorticity spectra in high Reynolds number anisotropic turbulence

Phys. Fluids 17, 088102 (2005); 10.1063/1.1989387

Numerical study of cavitation inception in the near field of an axisymmetric jet at high Reynolds number
Phys. Fluids 12, 2444 (2000); 10.1063/1.1289394

Streamwise vortices in an underexpanded axisymmetric jet

Phys. Fluids A 3, 1848 (1991); 10.1063/1.857914

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitationnew.aip.org/termsconditions. Downloaded
to IP: 147.188.128.74 On: Sat, 30 May 2015 11:33:52
 On streamwise vortices in high Reynolds number supersonic
axisymmetric jets
S. A. Arnette, M. Samimy, and G. S. Elliott
Department of Mechanical Engineering, The Ohio State Unive?sity, Columbus, Ohio 43210
(Received 25 February 1992; accepted 27 August 1992)
Pitot pressure measurements and flow visualizations were used to investigate streamwise
vortices previously observed in underexpanded jets. A simple model was developed, which gives
reasonable agreement with the pressure measurements. A converging nozzle and
converging-diverging nozzle of design Mach number 1.5 were used to generate jet flows of
equivalent Mach numbers up to 2.5 (stagnation to ambient pressure ratios up to
17.1). By operating the nozzles fully expanded, overexpanded, and underexpanded, insight
was gained into both the occurrence and cause for formation of the vortices. Spatially
stationary streamwise vortices were found to exist in the near-field region around the
circumference of underexpanded jets in the vicinity of the jet boundary. Short exposure
visualizations show the vortices persist much farther downstream with a loss of spatial
organization. Visualizations suggest adjacent vortices have streamwise vorticity of
opposite sign, so the action of adjacent vortices is to either pump jet fluid radially outward or
entrain ambient fluid radially inward toward the jet. The downstream extent, strength,
and number of vortices around the jet circumference increase with degree of underexpansion.
A large number of vortices is found near the nozzle exit. Fewer vortices of larger scale
are found farther downstream, indicative of a merging process. The absence of the vortices in
fully expanded and overexpanded jets suggests the vortices are a consequence of a
Taylorkoertler-type instability.

1. INTRODUCTION Zapryagaev and Solotchin’ used schlieren photogra-

phy and pressure measurements to demonstrate the pres-
There are many reasons for investigating the structure ence of stationary streamwise vortices in an underex-
of supersonic jets. Among these is the need to better con- panded jet issuing from a conical Mach 1.5 nozzle at a
trol the mixing of jet and ambient lluid. Given the coupling pressure ratio of 10. The vortices were proposed to be of
between the flow and emitted acoustic energy, deeper in- the Taylor-Goertler type with adjacent vortices having
sight into jet noise will also require greater understanding streamwise vorticity of opposite sign. A vortex merging
of the structure of these flows. process is suggested by schlieren images which show the
The shock cell structure of jets operated at underex- number of vortices to decrease with increasing downstream
panded conditions is well known. Adamson and Nicholls’ distance. Novopashin and Perepelkin’ used Rayleigh scat-
found the main characteristics to be those presented in Fig. tering to generate average cross-sectional density maps of
1. Upon encountering the lower ambient pressure at the highly underexpanded axisymmetric jets (stagnation-to-
nozzle exit, the gas passes through an expansion fan as the ambient pressure ratio of 100) issuing from a sonic orifice
first step in the pressure equilibration process. The expan- and found a drastic departure from axisymmetry, which
sion waves reflect from the jet boundary as compression they referred to as a “petal structure.” A connection be-
waves, which coalesce to form the intercepting shock. At tween the petal structure and the nozzle was established
low pressure ratios, the intercepting shocks meet at the jet when an angular rotation of the nozzle resulted in an
axis. For higher-pressure ratios, however, these shocks are equivalent rotation of the lobes. The lobed structure of the
connected by a normal shock, or Mach disk, as shown in density maps is likely a consequence of stationary stream-
Fig. 1. In both cases, reflected shocks are formed that in- wise vortices. Krothapalli et al. lo demonstrated the exist-
tersect the jet boundary and reflect as expansion waves. ence of stationary streamwise vortices in a moderately un-
The cell structure repeats itself downstream until the flow derexpanded axisymmetric jet issuing from a converging
becomes subsonic everywhere in the jet. nozzle at a pressure ratio of 5.1. Pitot pressure measure-
Many investigators have performed studies of the axi- ments show sinuous type variation of the total pressure
symmetric or helical large-scale structures that occur in around the jet circumference, suggesting streamwise vorti-
the jet shear layer as a result of instabilities. It has been ces exist and possess significant strength. Laser sheet light-
shown that these structures are important to the noise pro- ing visualizations of the jet cross section were obtained by
duction process (McLaughlin et a1.,2 Seiner and Kresja,3 collecting the scattering from condensed water particles
Morrison and McLaughlin,4 Moore,’ Lepicovsky et al. ,6 formed when the water originally present in the ambient
and Gutmark et aL7). Although possibly important to air as vapor was entrained and mixed with cold supersonic
noise generation and mixing processes, the existence of sta- fluid. Strong indentations into the region marked by the
tionary streamwise vortices in underexpanded supersonic condensed water particles support the presence of stream-
jets has been the subject of much less study. wise vortices.

187 Phys. Fluids A 5 (l), January 1993 0899-8213/93/010187-l 6$06.00 @ 1993 American Institute of Physics 187
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitationnew.aip.org/termsconditions. Downloaded
to IP: 147.188.128.74 On: Sat, 30 May 2015 11:33:52
 Jet Boundary TABLE I. Range of flow conditions studied for the sonic nozzle.
Expansion FansA--
c /,- Intercepting Shock
Mr h&&nb RedlO

1.0 1.89 0.375

1.25 2.59 0.515
1.5 3.67 0.727
1.75 5.33 1.05
2.0 7.82 1.55
2.25 11.6 2.29
2.5 17.1 4.26

Mach Disk / i

FIG. 1. Shock cell structure in the plume of an underexpanded jet. mm. The pressure measuring system was capable of mea-
surements to within *3.5 kPa (0.5 psi), utilizing a Bour-
don tube-type pressure gauge. As the investigation focused
The main objective of this study was to further inves- on spatially stationary streamwise vortices, the frequency
tigate the existence and detailed characteristics of stream- response of the system was not a concern.
wise vortices in supersonic jets, with methods similar to Pitot pressure measurements were taken through 180”
those employed by Krothapalli et al. lo Jets covering a wide of the jet circumference. The number of measurement
range of pressure ratios were investigated, issuing from points was determined by dividing half of the circumfer-
both a converging nozzle and a converging-diverging noz- ence for the given measurement radius by the sensing di-
zle of design Mach number 1.5. Unlike the conical Mach ameter of the probe. Adjacent measurement points are sep-
1.5 nozzle used by Zapryagaev and Solotchin,’ the Mach arated by an arclength equal to the sensing diameter of the
1.5 nozzle in this study was designed with the method of probe. Because of the alternating diverging and converging
characteristics to generate uniform flow at the nozzle exit. tlow around the shock cells, the Pitot pressure measure-
The cited studies indicate streamwise vortices are an im- ments in these experiments are not of the true total pres-
portant feature of underexpanded supersonic jets. In this sure that would exist behind a normal shock. This would
work, fully expanded, overexpanded, and underexpanded be measured only if the probe face were exactly perpendio
supersonic jets were studied to draw more general conclu- ular to the flow direction. Only the component of the total
sions concerning the relevance of the observed streamwise pressure corresponding to flow velocity perpendicular to
vortices to supersonic jets. the Pitot probe face, which is aligned parallel to the nozzle
exit cross section, is measured. However, since the interest
II. EXPERIMENTAL PROCEDURE here is in azimuthal variation due to vortices, the acquired
measurements are considered useful.
The experiments were conducted at The Ohio State Schlieren images with various knife edge settings were
University Aeronautical and Astronautical Research Lab- acquired with a charge coupled device (CCD) camera and
oratory (AARL). An air storage capacity of 41 m3 at recorded onto super VHS videotape. Illumination was pro-
pressures up to 16.5 MPa allows the jet facility to be run vided by a 5 W, continuous wave Spectra Physics model
continuously. The system allowed the stagnation pressure 2020 argon-ion laser. In a second visualization system,
of the jet to be maintained to within f 1% of the set point. light scattered from condensed moisture in the jet mixing
This, along with the daily variation of the ambient pres- region was used to visualize the jet flow. The optical layout
sure, led to a small run-to-run variation in the equivalent is shown in Fig. 2. Vertical sheets (approximately 0.2 mm
Mach numbers about the reported set points. thick) parallel to the face of the nozzle were used to obtain
Pitot pressure measurements and flow visualizations jet cross sections at various downstream locations. This
were performed on a sonic nozzle with an exit diameter of sheet orientation is referred to as a cross-sectional view. As
12.7 mm and a Mach 1.5 nozzle with an exit diameter of shown in Fig. 2, some of the images were recorded with the
19.05 mm. Experiments are referred to in terms of the aid of a molecular iodine filter. Without the filter, the tech-
equivalent Mach number instead of the stagnation-to- nique is the commonly employed laser sheet lighting. The
ambient pressure ratio. The equivalent Mach number is continuous-wave argon-ion laser was used in conjunction
defined as the Mach number which would result in an
isentropic expansion for a given stagnation to ambient
pressure ratio. The equivalent Mach numbers investigated, TABLE II. Range of flow conditions investigated for the Mach 1.5 noz-
along with the Reynolds numbers based on throat condi- zle.
tions, are presented in Tables I and II for the sonic and
M/ Pe,g/&,b RedlO’
Mach 1.5 nozzles, respectively.
Pitot pressure data were acquired with a standard 1.5 3.67 1.03
2.0 7.82 2.19
probe mounted on a three-axis traversing system. The tra-
2.5 17.1 4.79
versing system was equipped with a digital position read- 1.3 2.70 0.756
out on each axis, allowing positional accuracy to bO.013 1.18 2.36 0.661
mm along each axis. The probe sensing diameter was 0.76

188 Phys. Fluids A, Vol. 5, No. 1, January i993 Arnette, Samimy, and Elliott 188

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitationnew.aip.org/termsconditions. Downloaded
to IP: 147.188.128.74 On: Sat, 30 May 2015 11:33:52
 Frequency Doubler Filter Absorotion Profile

, Spectral Content of Unfiltered Image

V v+Av

I
L to super VHS

V v+Av
FIG. 2. Schematic of the filtered MielRayleigh scattering optical setup.

FIG. 3. Schematic of the background absorption process in the filtered

Mie/Rayleigh scattering technique.
with a CCD camera set to a “long” exposure time to col-
lect average jet cross sections. For these images, the re-
corded intensity for a given optical system (fixed laser
power, camera position, polarization, etc.) is determined shape of the absorption profile is that of a notch or band-
by the size and number density of the scattering particles reject filter. The iodine vapor filter was operated in the
and the amount of background light present. This tech- thermally broadened regime by evacuating the vessel to
nique is beset by unwanted reflections from the outer noz- very low pressures (approximately 0.2 Torr) . This resulted
zle surfaces and scattering from stray particles in the am- in relatively steep cutoff slopes and allowed the absorption
bient fluid outside the jet. For these reasons, the molecular of unwanted “background” light and the passage of Dop-
filter was employed in a filtered Mie/Rayleigh scattering pler shifted light. A schematic of this situation is presented
technique for the collection of all instantaneous cross sec- in Fig. 3. The top graph shows a schematic of the filter
tions. The illuminating light was provided by a Spectra absorption profile. The middle graph, on the same horizon-
Physics Nd:YAG laser, injection seeded to provide a nar- tal frequency scale as the top graph, shows the spectral
row linewidth and approximately 50 GHz tuning capabil- content of any given image. Intense unshifted background
ity. The beam was passed through a frequency doubler to light is present at the illuminating laser frequency. Scatter-
obtain the second harmonic at a wavelength of 532 nm. ing from the condensed water particles in the flow is Dop-
The pulse duration of the Nd:YAG laser is 9 nsec, effec- pler shifted to a higher frequency. Although molecular
tively freezing the flow in the collected images. A less-than- Rayleigh scattering possesses a significant linewidth be-
maximum pulse energy of approximately 300 mJ was used. cause of molecular thermal and acoustic motions, the scat-
Images were collected with a CCD camera and stored on tering from particles is of relatively narrow linewidth as the
super VHS tape. Presented images were subsequently dig- greater mass particles are insensitive to such motions. In
itized for output. A more complete description of the fa- the standard laser sheet lighting technique, all this light is
cility and instrumentation is provided by Arnette.” accumulated in the recorded image. With the tilter, the
Miles et al. I2 demonstrated that a molecular filter con- situation is as pictured in the bottom graph of Fig. 3. With
taining iodine vapor can be used in conjunction with 532 the laser frequency tuned to the right edge of the absorbing
nm illumination from an Nd:YAG laser to discriminate region, unshifted background light is absorbed by the filter
detected light based on frequency content and referred to and Doppler shifted light is passed, resulting in an image
the technique as filtered Rayleigh scattering. In principle, free of background light.
the ability to discriminate recorded intensities based on The situation presented in Fig. 3 is somewhat idealized
frequency, coupled with the well-known Doppler shift ef- in that the Doppler shifted light from the imaged region
fect, makes nonintrusive planar velocimetry possible. Ko- covers a fmite linewidth dependent on the range of veloc-
mine et al. l3 developed a similar system capable of instan- ities present in the region and the illumination/collection
taneous velocimetry for an illuminated plane in a seeded optics orientation. As a result, the absorption of Doppler
flow field. For a fixed filter volume and fixed number of shifted light is possible. In terms of jet cross sections, the
moles of absorbing medium contained therein, the absorp- actual situation progressively deviates from the ideal case
tion spectrum is determined by the temperature and pres- with increasing downstream distance, as the velocity pro-
sure of the absorbing medium. This issue is discussed at file of the jet decays from a top-hat shape to a bell shape.
greater length by Miles et al. l2 The two extremes are re- As a result, use of the filtered Mie/Rayleigh scattering
ferred to as the thermally broadened regime and the technique was confined to the first tive nozzle exit diame-
collision-broadened regime. In both cases, the general ters downstream of the nozzle exit. Further. the laser was

189 Phys. Fluids A, Vol. 5, No. 1, Januaty 1993 Arnette, Samimy, and Elliott 189
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitationnew.aip.org/termsconditions. Downloaded
to IP: 147.188.128.74 On: Sat, 30 May 2015 11:33:52
 SIR = I .20, pa.avs= 376 kPa
r/~ = 1.30, P,,,avs= 308 kPa
. r/R = 1.40, po.avs= 266 kPa
r/R = 1 JO, Po,avg= 210 kPa
* r/R = 1.60,po.avg= 168 kPa
r/R = 1.70, Po.=“s= 135 kPa
r/R = 1.80, po.avg= 117 kPa
r/R = 1.90, po,avn= 106 kPa

laser)

k-k (To detector)

0.8
FIG. 4. Scattering schematic for the calculation of Doppler shift.
0.7

90 110 130 150 170 190 210 230 250 270

tuned to the highest frequency (the rightmost position in Theta (Degrees)
Fig. 3) that resulted in sufficient background suppression.
The realized Doppler shift is a function of the orienta-
FIG. 5. Azimuthal variation of the Pitot pressure for Mj=2.0 issuing
tion of the illumination and receiving optics and the mag- from the sonic nozzle at X/D=2.00.
nitude and direction of the flow velocity. This situation is
depicted in Fig. 4, where the scatterer moves with velocity
components V, and V, in the plane of the page, k, is the where the two fluids have mixed sufficiently on a molecular
wave vector of the collected scattering, and ki is the wave level for vapor in the low-speed fluid to condense.
vector of the illuminating light. The Doppler shift in the Mention should be given to the size of the scattering
detected scattered light, Av, is given by particles. Dibble et al. l4 estimated the condensed particle
(2VJ 0 size as 100 nm. Taking this value as a characteristic diam-
Av=Tsin 5, eter (assuming spherical particles) gives an r//z ratio of
0 0.094 (where Y is the particle radius and /z is the laser
where A is the wavelength of the illuminating light and 8 is wavelength) for the 532 nm Nd:YAG wavelength, which
the angle between the incident wave vector and the col- falls in the transitional region between Mie and classical
lected scattering wave vector. For the cross-sectional Rayleigh scattering theories. This estimate seems reason-
views, the camera and filter were positioned approximately able for the current study. A distinct polarization direction
30” away from the jet axis in the same horizontal plane as effect on the scattered intensity was present, but the scat-
the jet axis. A tradeoff situation existed between maximiz- tering could not be attenuated to near extinction by vary-
ing the realized Doppler shift and minimizing the distor- ing the polarization direction. This suggests the scattering
tion of the jet cross section. The chosen camera position was near the transition between Rayleigh and Mie scatter-
caused a small distortion of the cross sections in the in- ing, so the term filtered Mie/Rayleigh scattering is used.
cluded images, but represented a good compromise be- Particle size is important to the particles’ ability to accu-
tween the two competing effects. rately track the flow. Fourguette et al. t6 studied a Mach
The cross-sectional visualizations are of the scattering 1.5 air jet with the same method of condensed water par-
from condensed water particles. The jet air is dried to low ticle formation used here. Using the results of Samimy and
moisture levels in the supply system. The jet exhausted to Lele,” an estimate of 350 nm was found for the maximum
ambient air with typical relative humidities from 55%- allowable particle size.” Assuming a particle size on the
75%. Condensed water particles formed around the pe- order of 100 nm as suggested previously, the condensed
riphery of the jet when ambient air was entrained and the particIes in the current experiments should accurately fol-
ambient water vapor encountered the low-temperature, low the flow.
high-speed jet fluid. The jet temperatures encountered by
the ambient water vapor were well below the saturation
Ill. RESULTS AND DISCUSSION
temperature for all encountered ambient temperatures and
relative humidities. Accordingly, the visualization tech- As discussed earlier, previous investigators have shown
nique was not sensitive to ambient conditions. This method that spatially stationary streamwise vortices located
of visualizing condensed particles in the mixing region has around the circumference of underexpanded jets cause
been used in the investigation of supersonic jets by Dibble variations in azimuthal Pitot pressure measurements at a
et aZ.,14 Samimy et aL,15 Krothapalli et al,” and Four- constant radius.**” Figure 5 presents Pitot pressure mea-
guette et al., I6 and in the study of planar shear layers by surements taken at X/0=2.00 around the circumference
Clemens and Mungal,17 Messersmith et al. ,” and Elliott of an lWj’2.0 jet issuing from the sonic nozzle. Measure-
et al. l9 Clemens and Munga117’20 called this technique ments through 180” at various measurement radii, r (non-
“product formation” to indicate the flow is highlighted dimensionalized with the radius of the nozzle exit, R = D/

190 Phys. Fluids A, Vol. 5, No. 1, January 1993 Arnette, Samimy, and Elliott 190

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitationnew.aip.org/termsconditions. Downloaded
to IP: 147.188.128.74 On: Sat, 30 May 2015 11:33:52
 2), are presented. In Figs. 5-15, measured Pitot pressures notes total pressure, s denotes entropy, and R denotes the
are nondimensionalized with the average Pitot pressure for gas constant for air. The entropy increase is given by
that measurement radius. Nondimensionalizing with a sin-
gle value (such as the stagnation pressure) masked the s,-s,=gln( [ 1+(~)(~~-1)1(2:::~~~~)]
indentations at the outer measurement radii, where the
flow consists largely of entrained ambient air and the Pitot
pressure is low. The vertical scale of the azimuthal Pitot
profile plots is varied to minimize crowding of the profiles
-Rln[l,(~)(M:-l)], (3)

at different measurement radii. where cp denotes the constant-pressure specific heat, y de-
In the MjE2.0 case issuing from the sonic nozzle, the notes the ratio of specific heats, and M, denotes the Mach
density gradient across the Mach disk was strong enough number ahead of the normal shock. The measured pressure
to be seen by the unaided eye, and the streamwise location outside the jet boundary would obviously be the static pres-
was found directly by manually positioning the probe at sure of the ambient fluid, &,b. If stationary streamwise
the Mach disk and recording the position from the travers- vortices were present at the jet boundary, such that adja-
ing stages. The axial location of the measurements in Fig. cent vortices had opposite senses of rotation (streamwise
5 (1Mj=2.0) is approximately 5 mm downstream of the vorticity of opposite sign), the hypothetical vortex sheet
Mach disk. Six well-defined local minima are seen in the and the jet boundary of the actual jet would be distorted.
figure, and the maximum variation level is approximately At locations where adjacent vortices rotated such that fluid
230 kPa at r/R = 1.20 (6 1% of the average Pitot pressure is entrained into the jet, the jet boundary would be dis-
at r/R= 1.20). As proposed by Zapryagaev and Solotchin,8 placed closer to the jet centerline than if the vortices were
it is believed that each of the indentations in the pressure not present, and vice versa. If such a distortion of the jet
measurements around the jet circumference, resulting in a boundary existed, a Pitot pressure profile around the vor-
departure from axisymmetry, indicates the presence of two tex sheet jet at a radius equal to that of the unperturbed jet
counter-rotating, spatially stationary, streamwise vortices. boundary would see alternating regions of jet and ambient
Adjacent counter-rotating vortices gives rise to a pumping fluid. In regions of jet fluid, the measured Pitot pressure
action. Depending on the actual senses of rotation of the would be significantly higher than in regions of ambient
adjacent vortices, either jet fluid is pumped radially out- fluid, azimuthal profiles nominally like those of Fig. 5
ward or ambient fluid is entrained radially inward toward would be expected. The maximum measurable variation in
the jet centerline. the simplified model is
Whether the demonstrated azimuthal variations in Pi-
tot pressure can be considered evidence of stationary, k’o,max =Po,2 -Pamb . (4)
counter-rotating, streamwise vortices must be addressed. The term maximum variation is used to denote the largest
Consider a simplified model of the underexpanded jet issu- difference in Pitot pressure values between a local maxi-
ing into ambient fluid, where (as in the current work) the mum and an adjacent local minimum. No attempts are
jet and ambient fluids are the same. In the model, the finite made to incorporate the pressure variation across the vor-
thickness of the shear layer between the jet and ambient tices into this simple model. The model gives values for the
fluids is neglected, so the jet boundary separating jet and extreme pressures that would exist on either side of an
ambient fluid becomes an infinitesimally thin interface of individual vortex present in a distribution of adjacent vor-
infinite vorticity. Rayleigh= considered this “vortex sheet” tices having alternating senses of rotation.
model in his investigations of jet stability. The hypothetical Significant differences exist between the vortex sheet
“vortex sheet” jet boundary is a pressure boundary, outside model and the actual flow that suggest the measured max-
which is quiescent ambient fluid and inside which is the imum Pitot pressure variation will be less than the value
underexpanded jet. If it is assumed the jet fluid expands to calculated with the model. The finite thickness of the ac-
ambient pressure at the jet boundary, the Mach number of tual shear layer makes the calculation of the variation mag-
the supersonic jet at the jet boundary can be calculated nitude based on quiescent ambient fluid and perfectly ex-
from the known stagnation pressure, po,l. The Pitot pres- panded jet fluid unrealistic. Because of this gradual
sure measured inside the jet boundary, po,2, would be the shearing action, the Mach numbers and velocities of the
total pressure behind the normal portion of the bow shock local maximum and minimum at a single measurement
set up ahead of the Pitot probe (in the limit of ideal probe radius are closer than those assumed in the model. This
alignment with the flow direction). Utilizing one- would result in a smaller theoretical maximum variation in
dimensional gasdynamic theory, with the assumption of a model that accounts for finite shear layer thickness. It is
ideal gases having constant ratio of specific heats, the mea- also significant that the flow between the intercepting
sured Pitot pressure is connected to the entropy increase shock and the jet boundary, referred to as the compression
across the shock, and is given by region, passes through a series of compression waves as it
travels downstream. In doing so, the total pressure of the
po2=po
9 I
1 et-(~z-WRl
, flow in the compression region continuously decreases with
increasing downstream distance. Further, the oblique ori-
where subscripts 1 and 2 represent quantities before and entation of the compression waves in this region results in
after the normal shock, respectively. The symbol p. de- decreasing Mach number with increasing radial distance

191 Phys. Fluids A, Vol. 5, No. 1, January 1993 Amette, Samimy, and Elliott 191

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitationnew.aip.org/termsconditions. Downloaded to
IP: 147.188.128.74 On: Sat, 30 May 2015 11:33:52
 TABLE III. Comparison of simplified model and experimental variations present (consistent with the idea that the vortices result
in Pitot pressure azimuthal profiles. from Taylor-Goertler instability) in both the calculated
and measured variations, the magnitudes are significantly
different.
Experimental variation at X/D= 1.00 Indentations in the azimuthal Pitot profiles exist at the
Finite shear thickness same angular position for all of the presented measurement
Theoretical Vortex radii in Fig. 5. Grouping the indentations in the azimuthal
variation sheet Mean Low High profiles that occur at a single angular location for multiple
1.50 244 loo 217 194 241 measurement radii allows a better comparison between the
2.00 470 255 422 381 444 model and the measurements to be made by neglecting
2.50 763 310 633 623 657 intermediate pressures in the shear layer and compression
region. For a single indentation, the maximum pressure is
taken as the local maximum at the measurement radius
from the jet centerline, as shown by Zapryagaev and displaying the greatest local maximum. Likewise, the min-
Solotchins These issues suggest the assumed model is clos- imum is taken as the smallest local minimum for the iden-
est to actuality near the nozzle exit, before the shear layer tified indentation. When locating the extreme pressures
has grown significantly and before significant recompres- this way, the minimum aZways occurred at a larger mea-
sion of the expanded flow occurs. The most upstream mea- surement radius than the maximum. Not all indentations
surements of this study, however, were collected at X/D in the azimuthal profiles spanned several measurement ra-
= 1.00. Limited spatial resolution prevented measurements dii, suggesting the streamwise vortices around the jet cir-
closer to the nozzle. cumference are of varying size and strength. Only those
The limitations of the current experiments will also indentations that could be identified over several measure-
cause the measured variation to be less than the theoretical ment radii were chosen for comparison. The angular loca-
variation. The maximum possible variation could be mea- tion of the local maxima and minima for an indentation
sured only if the Pitot probe had a perfect spatial resolu- varies a small amount between different measurement ra-
tion and if the face of the probe were aligned exactly per- dii, which is expected, since the probe had limited spatial
pendicular to the flow velocity. The probe resolution was resolution and the measurements reflect an average over
relatively poor, especially close to the nozzle exit where the the probe face.
shear layer had not acquired significant thickness and the The resulting mean, low, and high variations are re-
jet had not expanded to its full size. The face of the probe ported in columns 3, 4, and 5 of Table III, respectively. In
in these experiments was aligned parallel to the nozzle exit order to indicate that an allowance has been made for the
plane. Because of the flow curvature in the shock cell, the finite shear layer thickness in calculating the variations, the
flow velocity was not perpendicular to the face of the columns are labeled “finite shear thickness.” These values
probe. The reported measurements are, therefore, only an exhibit much better agreement with the model, indicating
average of a component of the actual variation. Another the simple model is conceptually correct. The model rep-
problem in comparing the measured variations to those resents the greatest possible azimuthal variation in Pitot
calculated with the simplified model is measurements were pressure. As a result, the highest variations for each equiv-
collected only at discrete radii. As a result, it is probable alent Mach number agree better with the model than do
that the maximum variations that could have been mea- the mean variations. A total of 5, 6, and 5 indentations
sured were not. were identified in the Iwj= 1.5 (Fig. 6), 2.0, and 2.5 (Fig.
The theoretical maximum variations calculated with 9) cases, respectively. The smaller number of indentations
the simplified vortex sheet model for the Mj= 1.5, 2.0, and at Mj=2.5 than at Mj=2.0 stems from difficulty in ob-
2.5 cases are presented in column 2 of Table III. The mea- taining profiles at X/D= 1.00 in the Mj=2.5 jet, which is
sured Pitot pressure azimuthal variations compared to the discussed below in conjunction with Fig. 9. In general, the
model are taken from data at X/D= 1.00 since the model is number of vortices increases with equivalent Mach num-
best suited for comparison to measurements close to the ber. The discrepancy between the calculated and measured
nozzle. The simplest variation that can be taken from the variations increases with increasing equivalent Mach num-
measured Pitot pressures is the difference between a local ber. This is expected, since the data for the three cases are
maxima and the adjacent local minima in the azimuthal collected at the same downstream location. At X/D= 1.00,
profile at a single measurement radius. If the shear layer in streamline curvature of the underexpanded jet increases
the experiments approached a vortex sheet, this variation with increasing equivalent Mach number, causing the
should approach the model value. The maximum local probe to see a smaller portion of the actual Pitot pressure
variations contained in the azimuthal profiles at a single (and resulting variation). The collection of data only at
measurement radius for the three equivalent Mach num- select radii is also a problem.
bers are presented in column 3 of Table III. The column is Figures 6-8 present the azimuthal Pitot pressure pro-
labeled “vortex sheet” to indicate the presented variations files for the Mj= 1.5 jet issuing from the sonic nozzle at
are contained in the azimuthal profile of a single measure- X/D= 1.00, 2.00, and 3.00, respectively. At X/D= 1.00
ment radius. Although the trend of increasing variation (Fig. 6)) a large number of indentations are found in the
magnitude with increasing equivalent Mach number is profiles, indicating the presence of a large number of

192 Phys. Fluids A, Vol. 5, No. 1, January 1993 Arnette, Samimy, and Elliott 192

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitationnew.aip.org/termsconditions. Downloaded to
IP: 147.188.128.74 On: Sat, 30 May 2015 11:33:52
 r/R
r/R
‘r/R
i/R
r/R
=
=
=
=
=
0.90, pqavp = 334
1 .OO, po,avg= 285
1 .lO, po,avg’x 242
1.20, pO,aVp
1.30, pO,avg
= 130
= 109
kPa
FPa
kPa
kPa
kPa
1.2
:*
t
. . ...p
I
i/R = 0.90, &,svg = 254. kPa I
r/R=1.00,p4.~C=206’kPa..:
i~=1.10,po.av~~168:LPa,
t
1.1 c ; :

.90 110 130 150 170 190 210 230 250 270 90 110 130 150 170 190 210 230 250 270
Theta (Degrees) Theta (Degrees)

FIG. 6. Azimuthal variation of the Pitot pressure for Mj= 1.5 issuing FIG. 8. Azimuthal variation of the Pitot pressure for Mj= 1.5 issuing
from the sonic nozzle at X/D= 1.00. from the sonic nozzle at X/D=3.00.

streamwise vortices. The maximum measured variation of

shock cell. Similar tb the Mi= 1.5 case at X/D= 1.00, the
approximately 100 kPa at r/R = 1.OOrepresents35% of the
Mj=2.5 jet at X/D= 1.00 displays 10-12 indentations in
average Pitot pressure gt r/R= 1.00. At X/D=2.00 (Fig.
the azimuthal Pitot pressure profiles, again indicating the
7), the indentations in the azimuthal profiles have evolved
presence of a large number of vortices. Also similar to the
into only three or four well-defined maxima and minima of
Mj= 1.5 case, strong indentations are located in the
much decreased strength relative to the indentations at
8=90”-1 lo” region. Novopashin and Perepelkin’suggested
X/D= 1.00. This indicates vortex merging has occurred
the departure from axisymmetry results from the amplifi-
between X/D= 1.00 and 2.00, in accord with the findings
cation of perturbations supplied by the nozzle. No notice-
of previous investigators.s,‘O At X/D=3,00 (Fig. 8), the
able imperfections were found in the nozzle at this, or any
profiles have evolved into an almost axisymmetric shape.
other, angular location. Despite the appearance of the in-
Figures 9-12 present similar azimuthal Pitot pressure
dentations in the 8=90”-110” region, the strongest varia-
profiles tracking the downstream development of the
tion in pressure units occurs at r/R = 1.90 at 8~ 170” (Fig.
Mi=2.5 jet issuing from the sonic nozzle. The axial loca-
9) and corresponds to approximately 310 kPa (45% of the
tion for Fig. 9 (X/D-1.00) was approximately 19.0 mm
average Pitot pressure at r/R = 1.90).
upstream of the Mach disk in the diverging portion of the

1.3 . 2.25 -
. r/R = 1 SO, po.avg= 352 kPa
Q r/R = 0.90, po,avg7 326 FPa A r/R = 1.90, po.avg= 682 kPa
r/R = 1 .OO, po we = 259 kPa 2.05 - + r/R = 2.20, po+“s = 163 kPa
.Z. r/R=l.lO,p0’*vg=194kPa. i
a r/R = 1.20, p&e 4 147 kPa
1.85 -

1.65

31.45 -
P
a
e1.25 -
&
1.05 -
1

0.85 -

0.8 0.65
90 110 130 150 170 190 210 230 250 270 90 110 130 150 170 190 210 230 250 270
Theta (Degrees) Theta (Degrees)

FIG. 7. Azimuthal variation of the Pitot pressure for Mj= 1.5 issuing FIG. 9. Azimuthal variation of the Pitot pressure for Mj=2.5 issuing
from the sonic nozzle at X/D= 2.00. from the sonic nozzle at X/D= 1.00.

193 Phys. Fluids A, Vol. 5, No. 1, January 1993 Arnette, Samimy, and Elliott 193
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitationnew.aip.org/termsconditions. Downloaded
to IP: 147.188.128.74 On: Sat, 30 May 2015 11:33:52
 1.75 .,
A r/R = 1.80, pD,avg= 611 kPa 0 r/R = 1.30, pO,avg= 525 kPa
1.65 . r/R = 2.00, po,avs= 641 kPa
r, r/R = 2.20, po,avg= 443 kPa f, r/R = 1.70, po’avg:=
r/R=lSO,poayg=518kPa-
436 kPrt
1.55 h . r/R = 2.40, po,avg= 2.57 kPa l r/R = 1.90, p&g = 332 kPa

1.25

0.85
0.75
0.65
055’-““““““““’ .

90 110 130 150 170 190 210 230 250 270 90 110 130 150 170 190 210 230 250 270
Theta (Degrees) Theta (Degrees)

FIG. 10. Azimuthal variation of the Pitot pressure for Mj=2.5 issuing FIG. 12. Azimuthal variation of the Pitot pressure for Mj=2.5 issuing
from the sonic nozzle at X/D=2.00. from the sonic nozzle at X/D=S.OO.

The data of Fig. 10 (X/D=2.00) were collected 6.5 The radii displaying the maximum variation for the
mm upstream of the Mach disk. The maximum variation Mj’2.5 case are r/R= 1.90, 2.20, 1.70, and 1.50 at X/D
level of approximately 360 kPa occurs at r/R=2.20 (81% = 1.00, 2.00, 3.00, and 5.00, respectively. This suggests the
of the average Pitot pressure at r/R = 2.20), and the profile vortex tubes experience curvature as a result of the flow
exhibits approximately 11 local minima. The data of Fig. curvature in the vicinity of the jet boundary. Similar trends
11 (X/D=3.00) are collected approximately 6 mm down- were found in the downstream evolution of the Mjz2.0
stream of the Mach disk. The azimuthal profiles display case issuing from the sonic nozzle. The number of local
approximately six local minima, and the maximum varia- minima encountered in the Mj=2.5 case are approxi-
tion of approximately 400 kPa occurs at r/R = 1.70 (63% mately 11, 11, 6, and 8 at X/D= 1.00, 2.00, 3.00, and 5.00,
of the average Pitot pressure at r/R= 1.70). The shock respectively, suggesting that merging occurs between X/D
cells are no longer detected at the axial location of Fig. 12 =2.00 and X/D=3.00. The occurrence of the greatest
(X/D=5.00). At X/D=5.00, the maximum variation of variation at X/D=3.00 suggests the vortices gain strength
approximately 210 kPa occurs at r/R= 1.50 (41% of the as they merge, although the opposite was found to be true
average Pitot pressure at r/R = 1.50). for the Mj= 1.5 case (Figs. 6 and 7). The increase in the
number of local minima from X/‘D=3.00 to X/D=5.00
for the Mjz2.5 case suggests a vortex tearing process
might be involved in the decay of the vortices.
1.8 = 1.56; po+& = 763 kPa Measurements were attempted at r/R= 1.70 and r/R
7 1.70, po,avg.= 638 kP+
1.7 = 1.86, po,avg= 497 kPa
= 1.80 for the case shown in Fig. 9, but there appeared to
1.6
4 2.pO; poavp = 377 kPa be present a low-frequency, small-amplitude “spatial jitter”
I 2.20; pO,avp= 266 kPa
= 2.40; pqavg = 186 kP+ of the vortices (within the frequency response of the pres-
1.5
sure gauge). Schlieren images acquired at 30 frames/set
1.4 confirmed this to be the case. The encountered fluctuations
1.3 were not a result of probe vibration. This jitter, combined
with the relatively small size of the vortices in this early
stage of their development (Zarge spatial pressure gradi-
ents), made the assignment of accurate average values very
difficult. This phenomenon was significant only for the
0.9 Mi=2.5 case at X/D= 1.00, and affected noticeably only
0.8 those points between well-defined maxima and minima (re-
gions of high positive or negative slope in the figures).
0.7
\ When these fluctuations occurred, the mean of the low and
0.6’ * ’ ’ ’ * ’ ’ * ’ ’ ’ ’ ’ ’ 1 ’ high values was recorded. If the range of the fluctuations
90 110 130 150 170 190 210 230 250 270
Theta (Degrees) was observed to be more than approximately 25% of the
difference in the local maxima and local minima, measure-
FIG. 11. Azimuthal variation of the Pitot pressure for Mj=2.5 issuing ments were not taken.
from the sonic nozzle at X/D=3.00. The azimuthal Pitot pressure profiles for all of the

194 Phys. Fluids A, Vol. 5, No. 1, January 1993 Arnette, Samimy, and Elliott 194

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitationnew.aip.org/termsconditions. Downloaded
to IP: 147.188.128.74 On: Sat, 30 May 2015 11:33:52
 n r/R = 1 SO, po,avg= 352 kPa- . r/R = 1 X0, po,avs= 702 kPa
A r/R = 1.90, pO,avg= 682 kh F r/R = 2.00, po.avg= 527 !@a
c r/~ 7 2.2~4pq,avr = 163 kPa t . r/R = 2.20, po,avg= 250 kPa
,-

1 .l

0.9

o.& ” ” ” ” ” / t ” ”
90 110 130 1.50 170 190 210 230 250 270 90 110 130 150 170 190 210 230 250 270
Theta (Degrees) Theta (Degrees)

FIG. 13. Azimuthal variation of the Pitot pressure for Mj= 1.5 issuing FIG. 14. Azimuthal variation of the Pitot pressure for Mj=2.5 issuing
from the Mach 1.5 nozzle at X/D=2.00. from the Mach 1.5 nozzle at X/D=2.00.

equivalent Mach number cases listed in Table I display data of Figs. 14 and 15 is for the Mach 1.5 nozzle operated
clear trends (data not included is presented by Arnette”) . at MjE2.5 at X/D=2.00 and 5.00, respectively. The azi-
For the jets issuing from the sonic nozzle, the equivalent muthal profiles at X/D=2.00 display a maximum varia-
Mach number is a convenient indicator of the degree of tion of approximately 80 kPa at r/R=2.00 (15% of the
underexpansion (i.e., the degree of underexpansion in- average Pitot pressure), and 10-12 indentations are evi-
creases with equivalent Mach number). At a given down- dent. The variations in the azimuthal pressure profiles are
stream location, both the number and dimensional small compared to the analogous variations of Fig. 10 for
“strength” of the indentations increase with degree of un- the sonic nozzle operated at the same equivalent Mach
derexpansion. Each of the indentations is believed to result number, also at X/D=2.00, where a maximum variation
from the presence of two counter-rotating vortices. The of 360 kPa was encountered. The profiles of Fig. 15 at
vortices persist in an average sense for a greater down- X/D=5.00 display axisymmetry, indicating that the
stream distance with increasing equivalent Mach number. streamwise vortices have ceased to exist in an average sense
Although the data follow very strictly the trend of at this downstream location. The azimuthal Pitot pressure
increasing strength of the streamwise vortices with increas- profiles of the sonic nozzle operated at the same equivalent
ing degree of underexpansion, the equivalent Mach num-
ber is not the only varying parameter for the cases listed in
Table I. Crist et aLz3 showed the axial length of the shock
cells is strongly dependent on the degree of underexpan- 1.3
n i/R = 1.40, p,,+“* = 585 kPa
sion. The streamwise vortices are believed to be a manifes- . r/R = 1.60, pO,avg= 432 kPa
tation of instability arising from the streamline curvature + i/R = 1.80, pO,avg= 314 kPa
7 r/R = 2.00, po,avg.y.217 kPa
and radial velocity gradients in the underexpanded jet. 1.2
Since the flow curvature is dependent on the shock cell
geometry, comparisons that more effectively isolate the de-
gree of underexpansion as the lone independent variable
could be made by collecting data at similar axial locations
after normalizing with a characteristic dimension of the
shock cells (i.e., distance of the Mach disk from the nozzle
exit plane), and not the nozzle exit diameter. However,
X/D=2.00 falls in the vicinity of the Mach disk for the
Mj=2.0, 2.25, and 2.50 jets, and the data agree very well :

with the cited trends concerning the number and strength

of the vortices.
Pitot pressure azimuthal profiles were also generated O.S1 * ’ ( ’ ’ f ’ * ’ ’ * ’ p ’ * ’
90 110 130 150 170 190 210 230 250 270
for the Mach 1.5 nozzle at various equivalent Mach num- Theta (Degrees)
bers. The data of Fig. 13 is for the Mach 1.5 nozzle oper-
ated fully expanded at X/D=2.00. The fully expanded jet FIG. 15. Azimuthal variation of the Pitot pressure for Mj=2.5 issuing
displays the expected axisymmetric azimuthal profiles. The from the Mach 1.5 nozzle at X/D=5.00.

195 Phys. Fluids A, Vol. 5, No. 1, January 1993 Amette, Samimy, and Elliott 195

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitationnew.aip.org/termsconditions. Downloaded to
IP: 147.188.128.74 On: Sat, 30 May 2015 11:33:52
 Intercepting Shock ambient pressure is 4.7, in comparison to a ratio of 9.0 for
Mj=2.5 issuing from the sonic nozzle. These comparisons
Jet Boundary indicate the degree of underexpansion is important in de-
--A
termining the strength and downstream persistence of the
stationary streamwise vortices, specifically the streamwise
vortices are stronger and persist farther downstream with
an increasing degree of undercxpansion. It is important to
keep in mind the variation of the axial length of the shock
cells between nozzles for a given equivalent Mach number
when making the cited comparisons between the sonic and
Mach 1.5 nozzles. In addition to the perfectly expanded
jets mentioned earlier, the azimuthal Pitot pressure profiles
displayed axisymmetry in the two overexpanded jets issu-
Nozzle \ \ / / ing from the Mach 1.5 nozzle (Mj= 1.18 and 1.3, having
\--, nozzle exit to ambient pressure ratios of 0.64 and 0.74,
-/
respectively), indicating an absence of streamwise vortices.
FIG. 16. Conceptual Taylor-Goertler system in the underexpanded jet.
The above trends can be examined relative to the idea
that these streamwise vortices result from a Taylor-
Mach number at X/D=5.00 (Fig. 12) display significant Goertler-type instability, as proposed by Zapryagaev and
variation. Solotchin.* Referring to Fig. 16, if it is imagined that a
It should be noted that at M,-=2.5 for the Mach 1.5 cylinder of radius RI (with its surface in the region be-
nozzle, the ratio of the static pressure at the nozzle exit to tween the intercepting shock and the jet boundary) is ro-

FIG. 17. Long exposure laser sheet lighting images of Mj= 1.5 jet issuing from the sonic nozzle. Images A, B, C, and D were collected at X/D= 1.00,
2.00, 3.00, and 5.00, respectively.

196 Phys. Fluids A, Vol. 5, No. 1, January 1993 Arnette, Samimy, and Elliott 196

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitationnew.aip.org/termsconditions. Downloaded to
IP: 147.188.128.74 On: Sat, 30 May 2015 11:33:52
 FIG. 18. Filtered Mie/Rayleigh scattering images of yj= 1.5 jet issuing from the sonic nozzle. Images A, B, C, and D were collected at X/D= 1.00,
2.00, 3.00, and 5.00, respectively. No specific time relatlonship exists between the individual images.

tating with an angular velocity, fit, such that R,fII, = U, streamwise vortices and the absence of the vortices in fully
( Ut =flow velocity at the surface of cylinder 1 in the actual expanded jets issuing from both nozzles and overexpanded
flow), and a second larger nearly concentric cylinder of jets issuing from the Mach 1.5 nozzle support the proposal
radius R2 (with its surface just outside the jet boundary) is that the vortices originate from the Taylor-Goertler insta-
rotating with SJ,, such that R&l,= lJ,, where U, is the flow bility.
velocity just outside the jet boundary, the resulting system Since the flow curvature and radial velocity gradient in
is very much like the classical cylindrical Couette flow the underexpanded jet increase with Mj for a given nozzle,
stability problem. The Taylor-Goertler stability problem the strength of the vortices would increase with Mj for a
was first addressed by Rayleigh in the late 18OOs, who given nozzle if the vortices result from the Taylor-Goertler
established that the inviscid cylindrical Couette flow is un- instability. This was shown to be the case in the Pitot
stable if the stratification of angular momentum is such pressure azimuthal profiles. Likewise, since the degree of
that it decreases radially outward. The viscous problem underexpansion is much less in a Mach 1.5 nozzle than in
was f?rst solved by Taylor in 1923. A fundamental treat- a sonic nozzle operated at the same equivalent Mach num-
ment of their analyses is presented by Kundu.24 Taylor ber, the strength of the streamwise vortices would be much
showed viscous effects tend to stabilize the system. Since greater in the plume of the underexpanded sonic nozzle if
the flow between the intercepting shock and the jet bound- the Taylor-Goertler instability is the actual mechanism for
ary is supersonic and the velocity outside the jet boundary the vortices’ formation. Again, this agrees with the Pitot
is low, f2z,)fi2 in the supposed system. Since the large pressure measurements.
difference in velocity occurs over a small radial distance To further investigate the flow features discussed
[especially near the jet exit), RI/R, approaches unity, but above, various flow visualizations were performed on
is always less than unity by definition. Both of these facts cross-sectional jet views. Long exposure laser sheet lighting
suggest that such an instability mode is possible. The dem- images are presented in Fig. 17 for the Mj= 1.5 jet issuing
onstrated curvature of the vortex tubes containing the from the sonic nozzle. All presented cross sections are

197 Phys. Fluids A, Vol. 5, No. 1, January 1993 Arnette, Samimy, and Elliott 197
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitationnew.aip.org/termsconditions. Downloaded
to IP: 147.188.128.74 On: Sat, 30 May 2015 11:33:52
 FIG. 19. Long exposure Iaser sheet lighting images of Mj=2.0 jet issuing from the sonic nozzle. Images A, B, C, and D were collected at X/D= 1.00,
2.00, 3.00, and 5.00, respectively.

slightly distorted due to the off-axis position of the camera. sure data shows that 19= 170”corresponds to a maximum in
In all cases, the presented Pitot pressure measurements the measured pressure azimuthal profiles in all cases and
were taken from the top center of the jet (0=90”) to the vortex pairs tend to form on either side. Again, this sug-
bottom center (8= 270”)) traversing around the circumfer- gests that the geometry of the nozzle has important impli-
ence in a counterclockwise fashion referenced to the image cations for the development of the vortices, as originally
observer. The different sizes of the condensation rings in suggested by Novopashin and Perepelkin.g It was verified
the images are a result of camera positioning only. In ac- that a rotation of the nozzle resulted in an equivalent ro-
tuality, the diameter of the condensation ring grows with tation of the indentations in the condensation ring. Images
increasing downstream distance, as expected. Image A of C and D of Fig. 17 were collected at X/D=3.00 and 5.00,
Fig. 17 was taken at X/D= 1.00 and corresponds directly respectively. At these downstream locations (X/D= 3.00
to the measurements of Fig. 6. Several well-defined inden- and 5.00) the indentations in the condensation ring are less
tations are evident around the condensation ring. It is be- noticeable, indicating that the streamwise vortices cease to
lieved that these indentations can be identified with exist in an average sense.
counter-rotating vortices. Image B of Fig. 17 is at the same In Fig. 18, instantaneous (9 nsec laser pulse) filtered
downstream location as the pressure measurements of Fig. Mie/Rayleigh scattering images are presented for the same
7. The center of the condensed ring is located at approxi- Mj= 1.5 flow at X/D= 1.00, 2.00, 3.00, and 5.00 in images
mately r/R= 1.05, which corresponds very well to the Pi- A, B, C, and D, respectively. Again, the center of the
tot pressure azimuthal profiles of Fig. 7. A noticeable fea- condensation ring in image B occurs at approximately r/R
ture is the presence of two relatively large lobes at 8= 170 = 1.05. The instantaneous images show much more detail
and 290”. These appear in all of the acquired long exposure of the indentations in the condensation rings. Similar to the
images and are believed to be the result of disturbances long exposure image A of Fig. 17, image A of Fig. 18
supplied by the nozzle, although no noticeable defects were displays a large number of indentations around the circum-
found upon inspection. Inspection of the presented pres- ference of the condensation ring. Each of the dark inden-

198 Phys. Fluids A, Vol. 5, No. 1, January 1993 Arnette, Samimy, and Elliott 198

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitationnew.aip.org/termsconditions. Downloaded
to IP: 147.188.128.74 On: Sat, 30 May 2015 11:33:52
 FIG. 20. Filtered Mie/Rayleigh scattering images of Mj=2.0 jet issuing from the sonic nozzle. Images A, B, C, and D were collected at X/D= 1.00,
2.00, 3.00, and 5.00, respectively. No specific time relationship exists between the individual images.

tations into and bright indentations out of the condensa- Figure 19 presents long exposure laser sheet lighting
tion ring is thought to indicate the presence of a pair of images for the sonic nozzle operated at an equivalent Mach
counter-rotating vortices. The dark regions are thought to number of 2.0. Indentations in the condensation rings are
occur where the sense of rotation of the adjacent vortices is more noticeable than in the Mi= 1.5 case of Fig. 17, which
such that ambient fluid is entrained toward the center of is expected given the stronger variations in the Pitot pres-
the jet and sufficient mixing between moist ambient air and sure azimuthal profiles of the Mj=2.0 jet relative to the
supersonic jet air for condensation has not yet occurred. Mj= 1.5 jet. Image A at X/D= 1.00 displays many small
Conversely, the bright regions are thought to result where indentations. The left half of image B (X/D=2.00) dis-
mixed ambient air (possibly entrained on the opposite side plays six well-defined indentations, which corresponds ex-
of the vortices) and cold jet air containing condensed wa- actly to the Pitot pressure azimuthal profiles of Fig. 5 for
ter is pumped away from the jet toward the ambient. The the same angular region. The center of the condensation
alternating bright and dark regions can occur only if the ring in image B is at approximately r/R = 1.50. Again, the
condensation ring is surrounded by streamwise vortices of indentations in the condensation rings of images C and D
alternating sign (adjacent vortices have opposite senses of (X/D=3.00 and 5.00, respectively) become less notice-
rotation). The occurrence of these streamwise vortices in able, although indentations are still clearly present at X/D
counter-rotating pairs lends further credence to the idea E3.00.
that the vortices result from a Taylor-Goertler instability. In Fig. 5, the maximum variation in the measured axial
Images C and D display clear evidence of the presence of Pitot pressure was encountered at r/R = 1.20, which seems
counter-rotating vortices, even though the long exposure to be at odds with the apparent location of the indentations
images and the Pitot pressure azimuthal profiles show no in image B of Fig. 19. A possible explanation for the dis-
well-defined indentations. This indicates the streamwise crepancy can be formed with the knowledge that the Mach
vortices exist downstream of where they cease to exist in an number between the intercepting shock and the jet bound-
average sense, implying loss of spatial organization. ary decreases with increasing radial distance from the cen-

199 Phys. Fluids A, Vol. 5, No. 1, January 1993 Arnette, Samimy, and Elliott 199
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitationnew.aip.org/termsconditions. Downloaded
to IP: 147.188.128.74 On: Sat, 30 May 2015 11:33:52
 FIG. 21. Filtered Mie/Rayleigh scattering images of Mj=2.5 jet issuing from the sonic nozzle. Images A, B, C, and D were collected at X/D= 1.00,
2.00, 3.00, and 5.00, respectively. No specific time relatronship exists between the individual images.

ter of the jet. The ratio of the total pressure behind a the azimuthal Pitot profiles and long exposure images, the
normal shock (which is the quantity measured by a Pitot instantaneous images suggest significant deviation from the
probe in a supersonic flow) to the total pressure ahead of average distribution is present at any given instant. As with
the shock decreases sharply with increasing Mach number. long exposure images and Pitot pressure azimuthal pro-
As a result, an equivalent variation in total pressure in files, indentations in the instantaneous images for Mjc2.0
supersonic flows of low and high Mach number would be are more evident than those in the Mj=1.5 case. Again,
much more evident behind a normal shock in the high strong counter-rotating streamwise vortices appear to be
Mach number flow when detected by Pitot pressure mea- present instantaneously at X/D=5.00, although they have
surements. This is believed to be the reason for the asym- ceased to exist in an average sense at this downstream
metric distributions of the level of Pitot pressure variation location.
about the radius at which maximum variation is encoun- The downstream development of the streamwise vor-
tered. For example, in Fig. 5 the vortices are confined al- tices in the Mj= 2.5 jet are presented in Fig. 2 1 as captured
most entirely between r/R= 1.10 and r/R= 1.90, but the by filtered Mie/Rayleigh scattering. Image A is taken at
maximum variation is found at r/R = 1.20 or 1.30 (data at X/D= 1.OOand corresponds to the pressure measurements
r/R= 1.10, not shown in Fig. 5 to minimize crowding, is of Fig. 9. The center of the condensed ring occurs at r/R
presented by Arnette”). As a result, it is believed the vi- 22.00. The sheet enters the jet from the right, and the
sualizations are better indicators of the radial location of gradient in illuminating intensity across the jet is probably
the vortices than are the Pitot pressure measurements. a result of scattering by the high-density underexpanded
Figure 20 presents instantaneous filtered Mie/Rayleigh fluid in the potential core. The circumference of the jet is
scattering images of the Mjz2.0 jet issuing from the sonic seen to be dominated by a large number of small, counter-
nozzle. Images A, B, C, and D were collected at X/D rotating vortices. Image B of Fig. 21 is taken at X/D=2.00
= 1.00, 2.00, 3.00, and 5.00, respectively. While vortices and corresponds to the pressure measurements of Fig. 10.
are clearly present and stationary in an average sense from The center of the condensed ring occurs at r/R=2.25.

200 Phys. Fluids A, Vol. 5, No. 1, January 1993 Arnette, Samimy, and Elliott 200
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitationnew.aip.org/termsconditions. Downloaded
to IP: 147.188.128.74 On: Sat, 30 May 2015 11:33:52
 FIG. 22. Filtered Mie/Rayleigh scattering images of Mi=2.5 jet issuing from the Mach 1.5 nozzle. Images A, B, C, and D were collected at X/D= 1.00,
2.00, 3.00, and 5.00, respectively. No specific time relationship exists between the individual images.

Comparing images A and B, significant merging has oc- vortex merging has occurred. The bright ring in the center
curred between X/D= 1.00 and 2.00, as was concluded of the image indicates unusually moist supply air for that
from the corresponding pressure measurements of Figs. 9 particular experiment. This resulted from a failure to prop-
and 10. Image C is taken at X/D= 3.00, and can be com- erly cycle the parallel air dryers and is not an uncontrol-
pared to the azimuthal Pitot profiles of Fig. 11. The lable problem. Although accidental, the visualizations’ in-
smaller size of the ring is again due to camera placement. sensitivity to the jet air’s moisture content (as long as it is
The center of the condensed ring occurs at r/R =2.10. significantly less than the moisture content of the ambient
Image D was collected at X/D= 5.00, and is directly com- air) is demonstrated. The inner signal is a result of scat-
parable to the pressure measurements of Fig. 12. The cen- tering from condensation in the core flow within the inter-
ter of the condensed ring occurs at r/R = 1.80, demonstrat- cepting shock. For the core flow, the Mach number in-
ing the curvature of the shear layer surrounding the creases with increasing radial distance from the jet axis at
underexpanded jet. Instantaneous images like image D a given downstream location. The flow just inside the in-
show the presence of vortices larger in scale than those tercepting shock has expanded enough for condensation to
found at X/D=3.00. However, long exposure laser sheet occur. The condensation does not exist outside the inter-
lighting images and the pressure measurements of Fig. 12 cepting shock due to the higher temperature. In comparing
indicate the average presence of the vortices is much de- image B of Fig. 22 to the azimuthal Pitot profiles of Fig.
creased relative to the upstream locations. 14, although the indentations in the condensation ring of
Figure 22 shows the downstream evolution of the the instantaneous image are well defined, the Pitot profiles
Mj=2.5 jet issuing from the Mach 1.5 nozzle. Images A, indicate the streamwise vortices are of relatively small
B, C, and D were collected at X/D= 1.00, 2.00, 3.00, and strength. Evidence of streamwise vortices is present in im-
5.00, respectively. As expected, image A at X/D= 1.00 ages C at X/D=3.00 and D at X/D=5.00, although the
displays many small indentations. Image B at X/D=2.00 azimuthal Pitot profiles of Fig. 15 indicate that the vortices
displays fewer more well-defined indentations, indicating do not exist in an average sense at X/D=5.00.

201 Phys. Fluids A, Vol. 5, No. 1, January 1993 Arnette, Samimy, and Elliott 201
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitationnew.aip.org/termsconditions. Downloaded
to IP: 147.188.128.74 On: Sat, 30 May 2015 11:33:52
 IV. CONCLUSION and fabrication of the nozzles. Thanks are also expressed to
Professor Y. Krothapalli for furnishing Ref. 8 with an ac-
The existence and nature of streamwise vortices in
companying translation.
high Reynolds number, axisymmetric jets operating in
The support of the NASA Lewis Research Center for
overexpanded, fully expanded, and underexpanded regimes
this research under Contract No. NAG3-764 with Dr.
have been studied. The measurement of the azimuthal vari-
K. B. M. Q. Zaman is appreciated. Fellowships from the
ation of the Pitot pressure was used as an indirect measure
National Science Foundation to Arnette and the Ohio
of the strength of these vortices. This allowed comparisons
Aerospace Institute to Elliott are gratefully acknowledged.
over different Mach numbers and locations in the flow.
Both long exposure laser sheet lighting and instantaneous
filtered Mie/Rayleigh scattering were used as visualization ‘T. C. Adamson, Jr. and J. A. Nicholls, “On the structure of jets from
highly underexpanded nozzles into still air,” J. Aerosol Sci. 26, 16
techniques. The presence of spatially stationary, stream-
(1959).
wise vortices has been demonstrated indirectly in underex- ‘D. K. McLaughlin, G. L. Morrison, and T. R. Trout& “Experiments on
panded jets issuing from sonic and Mach 1.5 nozzles. The the instability waves in a supersonic jet and their acoustic radiation,” J.
maximum possible variations in the azimuthal Pitot pres- Fluid Mech. 69, 73 ( 1975).
3J. M. Seiner, and E. A. Kresja, “Supersonic jet noise and the high speed
sure distributions based on a simple model are calculated civil transport,” AIAA Paper No. ALAA-89-2358.
and compare favorably to the measurements. The action of ‘G. L. Morrison and D. K. McLaughlin, “Instability process in low
the counter-rotating vortices forms indentations in the re- Reynolds number supersonic jets,” AIAA J. 18, 793 ( 1980).
gion between the intercepting shock and the outer edge of ‘C. J. Moore, “The role of shear-layer instability waves in jet exhaust
noise,” J. Fluid Mech. 80, 321 ( 1977).
the shear layer surrounding the jet. The downstream de- “J. Lepicovsky, K. K. Ahuja, W. H. Brown, and R. H. Burrin, “Coher-
velopment of these vortices for equivalent Mach numbers ent large-scale structures in high Reynolds number supersonic jets,”
from 1.0 to 2.5 was investigated. As reported previously by AIAA J. 25, 1419 (1987).
Zapryagaev and Solotchins and Krothapalli et aZ.,‘O the ‘E. Gutmark, K. C. Schadow, and C. J. Bicker, “Mode switching in
supersonic circular jets,” Phys Fluids A 1, 868 ( 1989).
vortices were found to be spatially stationary for the first “V. I. Zapryagaev and A. V. Solotchin, “Spatial structure of flow in the
few jet exit diameter lengths downstream and exist in the initial section of a supersonic underexpanded jet,” Academy of Sciences
vicinity of the jet boundary. Good agreement was found USSR, Siberian section, Institute of Theoretical and Applied Mechanics
between the visualizations and Pitot measurements con- preprint No. 23-88, UDK 533.6.011, 1988 (in Russian).
‘S. A. Novopashin and A. L. Perepelkin, “Axial symmetry loss of a
cerning the number and angular location of the vortices supersonic preturbulent jet,” Phys. Lett. A 135, 290 (1989).
about the jet circumference. Visualizations and Pitot mea- ‘sA. Krothapalli, G. Buzyna, and L. Lourenco, “Streamwise vortices in
surements indicate significant merging of adjacent vortices an underexpanded axisymmetric jet,” Phys Fluids A 3, 1848 ( 1991) .
occurs, leading to an increase in the scale of the vortices ‘IS. A. Arnette, “An experimental investigation of the structure of high
Reynolds number supersonic jets,” MS thesis, The Ohio State Univer-
with increasing downstream location. sity, 1992.
All indications suggest the vortices result from the “R. B. Miles, W. R. Lempert, and J. Forkey, “Instantaneous velocity
Taylor-Goertler instability, as originally suggested by Za- fields and background suppression by filtered Rayleigh scattering,”
AIAA Paper No. AIAA-91-0357.
pryagaev and Solotchin.* This is strongly suggested by the “H. Komine, S. J. Brosnan, and A. B. Litton, “Real-time, Doppler global
absence of the vortices in perfectly expanded jets issuing velocimetry,” AIAA Paper No. AIAA-91-0337, 1991.
from both converging and converging-diverging nozzles j4R. W. Dibble, R. S. Barlow, M. G. Mungal, K. Lyons, B. Yip, and M.
and overexpanded jets issuing from a converging-diverging B. Long, “Use of Rayleigh scattering from condensed water vapor as a
means of imaging an underexpanded supersonic jet,” STAR Mtg.,
nozzle. Supporting arguments are ( 1) the vortices occur in Nashville, Tennessee, 1989.
counter-rotating pairs; (2) the vortices do not exist at the “M. Samimy, K. B. M. Q. Zaman, and M. F. Reeder, “Supersonic jet
same radius for all downstream locations, but instead ex- mixing enhancement by vortex generators,” AIAA Paper No. AIAA-
perience curvature with the flow, as is typical of Taylor- 91-2263, 1991.
16D. C. Fourguette, M. G. Mungal, and R. W. Dibble, “Time evolution of
Goertler vortices; and (3) the vortices display the expected the shear layer of a supersonic axisymmetric jet,” AIAA J. 29, 1123
trend of increasing strength with increasing curvature and (1991).
radial velocity gradient. “N. T. Clemens and M. G. Mungal, “Scalar mixing in the supersonic
The effect of increasing equivalent Mach number can shear layer,” AIAA Paper No. AIAA-9 l-1720, 199 1.
18N. L . Messersmith, J. C. Dutton, and H. Krier, “Experimental inves-
be summarized as ( 1) the vortices exist in an average sense tigation of large scale structures in compressible mixing layers,” AIAA
farther downstream, (2) the vortices at a given down- Paper No. AIAA-91-0244, 1991.
stream location are increased in strength, and (3) there is 19G. S. Elliott, M. Samimy, and S. A. Arnette, “A study of compressible
a larger number of vortices. Streamwise vortices exist in- mixing layers using filtered Rayleigh scattering,” AIAA Paper No.
AIAA-92-0175.
stantaneously downstream of where they cease to be spa- *“N . T . Clemens and M. G. Mungal, “A Planar Mie scattering technique
tially stationary. for visualizing supersonic mixing flows,” Exp. Fluids 11, 175 (1991).
2’M. Samimy and S. K. Lele “Motion of particles with inertia in a com-
ACKNOWLEDGMENTS pressible free shear layer,” Phys. Fluids A 3, 1915 (1991).
22Lord Ravleigh, “On the instability of iets,” Proc. London Math. Sot.
The authors express their gratitude to fellow graduate 10,4 (1879).
23S C&t, P. M. Sherman, and D. R. Glass, ‘Study of the highly under-
students V. Belovich, M. Reeder, D. Glawe, and J. Daw- expanded sonic jet,” AIAA J. 4, 68 (1966).
son for help in the experiments. Thanks also go to the staff 24P. K. Kundu, Fluid Mechanics (Academic. New York. 1990). _. DD. 368-
of the AARL at Ohio State and J. Dawson for the design 373.

202 Phys. Fluids A, Vol. 5, No. 1, January 1993 Arnette, Samimy, and Elliott 202

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitationnew.aip.org/termsconditions. Downloaded
to IP: 147.188.128.74 On: Sat, 30 May 2015 11:33:52