Ok 1989 Main Ref P 55 Entrainment MechanismA693012

DYNAMICS AND CONTROL
(V)
of
TURBULENT SHEAR FLOWS

)T
.vC
Report of Research Activities
SL ECTE' __
UN 2 6 1989
During the Period

October 1986 to May 1989
Dhct i
Supported by the
OFFICE OF NAVAL RESEARCH

under the
UNIVERSITY RESEARCH INITIATIVE PROGRAM

Contract No. N00014-86-K-0679
A Research Program
of the
DEPARTMENT OF AEROSPACE ENGINEERING

UNIVERSITY OF SOUTHERN CALIFORNIA
Los Angeles, California 90089-1191
, .
-I-CJI q
Project Director: L. G. Redekopp
age
Assistant Director: R. F. Blackwelder
Contract Monitor: M. Reischman
89
2"
41
DYNAMICS AND CONTROL
5TURBULENT
NTIS
DiC
OF
SHEAR FLOWS
i ,3
Di ,t
EXECUTIVE SUMMARY
SCIENTIFIC OBJECTIVES
The objectives of this research program are summarized in three statements.
Ia)
Provide a DYNAMICAL DESCRIPTION of various turbulent shear flows having technological significance
b) Define the CONTROLLABILITY of key dynamical events ancl processes which impact
global,
engineering performance measures.
c) Formulate effective and efficient CONTROL STRATEGIES for specific types of flows
5and
performance measures.
Research progress toward these goals is summarized in eight separate reports following the
Executive Summary. These reports set forth the goals accomplishments, and future plans of
different sub-projects led by individual, or teams of, co-investigators. There is a considerable

amount of interaction and cooperative effort among different co-investigators on several
5
I
topics (free shear layer dynamics and G6rtler instability, in particular), but no attempt has
been made to merge combined efforts on an individual topic into a single report.
I2
RESEARCH PERSONNEL
The team of co-investigators consists of seven full-time faculty and one adjunct professor
from the Aerospace Engineering department and one faculty member with a joint appoint-
3
3
ment in both the Aerospace Engineering and the Mechanical Engineering departments. As
presently comprised, this research team represents several changes in personnel since the
beginning of the project. First, a change in directorship of the project was made during
the first fiscal year. Professor R. E. Kaplan, the project director and chairman of the De-
partment of Aerospace Engineering, left the project in March 1987 to assume the position
of Vice-Provost for Computing within the university administration. The department ad-
ministration has been reorganized under two co-chairmen, Professors E. P. Munts and L.
G. Redekopp. Professor Redekopp, an original co-investigator on the the project, is now
serving as project director with Professor R. F. Blackwelder serving as assistant director.
A second change involved the addition of Professor J. A. Domaradzkd as a co-investigator.
Professor Domaradzki joined the department as an assistant professor in January 1987. His
expertise is in the area of computational fluid dynamics with considerable experience in the
computation of turbulent flows. Professor Domaradzki was, for the first two and one-half
years, supported by institutional funds as promised from the USC School of Engineering
through a Powell Foundation Grant in support of the ONR-URI program. In October 1989
3
I
he will begin receiving partial support from ONR-URI funds. A third change was instituted
in October 1988 when Dr. R. H. Liebeck, an adjunct professor in Aerospace Engineering was
added to the research team. Professor Liebeck is renowned for his expertise in wing theory
and design practice and is, together with Professor Blackwelder, supervising a project on
the control of laminar separation bubbles on low Reynolds number airfoils. A report of the
SI
activity on this topic is included in the Research Reports section.

Project funds have provided training support for a substantial group of research person-
nel. This includes half-time support for a post-doctoral research fellow (Dr. J. Hertzberg),
support for six Ph.D. students (H. Asano, E. Ikeda, P. LeBlanc, D. Park, J. Poon and J.
Wiggert), support for two M.S. students who have already completed their studies (F. Pray
12
I
Prost-Domasky), plus support for several undergraduate laboratory assistants. Instiand S.

tutional funds derived from the Powell Foundation have provided support for an additional
Ph.D. student (Liu Wei).
BUDGETARY ISSUES
Funding uncertainties and variations over the last two years have had a measureable
impact on the overall program and its management. During the second fiscal year (Octo.
ber 1,1987 to September 30, 1988), the budget was reduced by 39.4% from that originally
proposed. During the current fiscal year (i.e., the third year of the project) the budget has
been reduced 23.1% compared to the originally proposal amount. These reductions forced
several revisions of the research plan, impacted graduate student recruitment, prevented ap-
pointment of two post-doctoral researchers, slowed the development of several experimental

facilities, postponed some equipment purchases, and caused us to significantly reduce the
Iparticipation
5
5
of the co-investigators during the summer months. Some of the impact of these
cuts was softened by judicious use of Powell Foundation funds (S 70K/yr.). However, those
funds were only committed for project use during the first-three years of the program and,
hence, careful planning is required so as to sustain the vitality of the program in the face of
reduced discretionary resources.
4
INSTRUMENTATION AND FACILITIES

In spite of the budgetary variations, the ONR-URI project has had a major impact
on laboratory instrumentation and facilities. Two major facilities have come on line since
the beginning of the project: a large water channel and an isothermal dump combustor
facility. The large water channel is currently being prepared and instrumented for studies in
turbulent boundary layer control with application to drag reduction. The dump combustor
Sfacility has already been used extensively for studies of passive control of mixing enhancement
1
*3
in confined geometries. A third facility is currently being tested for use in studies of the
G6rtler instability. The latter facility was delayed considerably as a consequence of the
Saforementioned
budget adjustments.
ONR-URI funds were used to significantly enhance the instrumentation capability in our
laboratory. Most importantly, we purchased two Laser-Doppler Velocimeter systems which
are being used in conjunction with experiments in the large water channel and the dump
Icombustor
facility. Due to budget limitations, the LDV systems that were purchased were
downgraded from the two, 2-component systems which were part of our original equipment
proposal. Instead, we purchased limited optics and only one signal processing unit so the
two systems are presently only capable of measuring one velocity component. Enhancements
required to make both systems operate as independent, 2-component LDV systems will
3depend
5
on future budgets. In addition to these major instrument purchases, project funds
have been used to purchase six hot-film channels, fourteen hot-wire channels, together with
a variety of other laboratory instruments like oscilloscopes, transducers, amplifiers, etc.
Another area where this project has had a major impact is in the area of computational and data analysis capabilities. Together with institutional support through the Powell Foundation, ONR-URI funds have been used to dramatically increase our computational
environment through hardware purchases (six SUN 3/50 workstations, one SUN 3/60 color
workstation, one SUN 3/280 fieserver, four Maxum 286 Turbo PCs, two 60 mb tape drives,
one Apple Laser Printer, one HP Laser-jet Printer, one HP Graphics plotter, and five 16channel A/D converters) plus several major software purchases and a high-speed link to the
San Diego Supercomputer Center.
5
I
i
LABORATORY INTERACTION
We have had extensive and profitable interaction with several scientists from the Naval
Weapons Center at China Lake, CA over the duration of the project. A goodly number
of exchange visits between NWC personnel (Drs. Schadow, Koshigoe and Gutmark) and
3USC
1
co-investigators (Professors Ho and Huerre) have occurred thus far and we anticipate a
continuation of this interaction. The subject of this interaction is jet mixing for supersonic
combustion and scram-jet technology.
Quite recently we have begun an interaction with the Naval Underwater System Center
1
...........
at Newport, RI regarding the effect of particulates on boundary layer transition. This interaction involves Dr. Duncan Brown and his associates at NUSC and Professors Blackwelder
and Browand at USC. The present plans include some experiments in the large water channel
at USC which was brought on line and instrumented by URI funds.
Another area of interaction is with the LAURA (low altitude unmanned research aircraft)
project at NRL and the HALE (high altitude long endurance) project at NADC. Professors
Liebeck and Blackwelder are guiding research efforts along these lines and Professor Liebeck
has been involved with the evaluation team at NRL for the LAURA project.
Professor Huerre visited NRL and, in addition to presenting a seminar, had discussions
3with
several researchers concerning the dynamics of mixing layer flows. He also participated
in and gave an invited lecture at the "Joint ONR/AFOSR Colloquium on Active Control
3and
5
3
Turbomachinery Hydroacoustics" in May 1988.

We have also had extensive interaction with the turbulence research group at the NASA
Ames Research Center. Professor Blackwelder was a visiting scientist at the Center for
Turbulence Research summer program in 1987, and Professors Domaradzki, Ho, and Huerre
participated as visiting scientists in the 1988 summer program. Professors Ho and Huerre
each presented tutorial lectures on topics related to turbulent mixing. Joint research efforts
with scientists at NASA Ames and USC co-investigators are on-going.
I
I
!
I
I
I5
U
BOUNDED SHEAR FLOWS
EXPERIMENTAL APPROACHES
Principal Investigator: R. F. Blackwelder
Research Summary
Description of Scientific Research Goals
The principal objective of the research is to understand the dynamics of the bursting process well enough that a significant reduction in drag can be achieved. If the low speed streaks
associated with the bursting phenomenon in a turbulent boundary layer can be prevented
from lifting away from the surface, then the bursting process would be disrupted. Hence
turbulent production would be reduced, and a lower drag should follow. The first problem is
that the low speed streaks occur randomly in space and time. Research into aligning these
low speed streaks using wall roughness elements has been performed. Selective suction has
been used to prevent the streaks from lifting off the wall. By using our knowledge of the
eddy structure and selectively interacting with it, significantly less suction should be needed
than for standard boundary layer control.
Significant Results
3
I
Cylindrical rods of 4v/u, in diameter aligned in the streamwise direction were placed on
the wall of the fiat plate model. These longitudinal roughness elements (denoted by LREs)
acted as nucleation sites for the formation of the low speed regions. For y+ < 10, these LREs
reduced the meandering of the low speed streaks such that the streaks were always close to
one of the elements. Thus the probability of finding a streak at a particular location was
greatly improved. This control method is quite important if one wants to modify the low
speed streaks which usually appear randomly in space and time.
The LREs were used to anchor the streaks into known locations and then suction was
applied. The suction was generated over a 3.5cm x 3.5cm area with the fluid withdrawn
through a porous plate under the roughness elements. The amount of suction was controlled
by a variable speed pump and was applied selectively under the streaks by masking portions
of the porous plate between the LREs. Figure 1 shows that, above each of the LREs, the
mean velocity was reduced by 1.5 - 2.0u, at y+ = 12 depending upon the amount of suction.
This results from the elevated no slip condition on the LREs and supports the results that
the low speed streaks were less random. The number of streaks, N, were counted using an
algorithm that determined the number of times the velocity crossed below a fixed threshold.
The threshold was one rms value below the mean velocity measured on the clean flat plate.
As expected, the number of streaks decreased considerably as the suction was applied (cf.,
Figure 2). Even when the threshold was adjusted for the new local mean and rms values,
the detected number of low speed streaks was reduced as the suction increased.
The suction increases the net drag as manifested by the increased mean velocity at a
fixed elevation above the wall (cf., Figure 1). However, the intent of the suction was to
prevent the low speed streaks from lifting and forming the ejection phase of the motion
which should reduce the turbulence production downstream. A measure of this activity
is shown in Figure 3 in which the shear layer detection frequency obtained by the VITA
technique was measured. The detection parameters were not altered from those used on
the dean fiat plate. The detection frequency decreased slightly when the LREs were added
and decreased further with the suction. The total decrease shown in Figure 3 is 20-30%,
suggesting that this method may offer some means of control of the near wall eddies.
To provide for a zero mass flow system, the fluid removed from beneath the low speed
streaks must be added back into the wall region, possibly beneath the higher speed regions
(although this has not been attempted). This selective injection would reduce the gradient
at the wall in those regions resulting in further drag reduction at the expense of a more
complex system. Ideally, the suction needs to be only applied when a lift up or ejection is
occurring from a low speed streak. To adapt this method to the wall layer would require
additional geometrical constraints or some intelligence to determine when the suction should
be applied. Thus, a more complex system would be necessary, but would require a reduced
expenditure of energy.
Plans for Next Year's Research
The main plans for the coming year are to conclude the present phase of the research
and present it at the IUTAM Symposium on Structure of Turbulence and Drag Reduction to
be held at the Federal Institute of Technology in Zurich. The course of the research will be
altered slightly in order to study the global effects of suction upon the turbulent boundary
layer. This is considered important because so little is known about the effects of uniform
suciton upon the structure of the near wall region. Only one recent work (Antonia, et al.,
J. Fluid Mech., 190, 217, 1989) has studied the effects of the suction upon the boundary
layer structure and it included little detailed data. Although the present work indicated that
the suc lon provided the expected trends in the bursting frequency and other statistics, the
suction area was too small to examine an asymptotic state.
IC
oi
-C
'
\ j
%)S cc
-Ip
040
-W
222 K
22
C4
A~c
0-
lu-
,q!,
4ii
a034
List of Publications /Reports/Presentations

1. Papers Published in Refereed Journals
None
2. Technical Reports
5
3Upon
Ron F. Blackwelder, Some Ideas on the Control of Near-Wall Eddies, AIAA-89-1009.

R. F. Blackwelder and J. B. Roon, The Effects of Longitudinal Roughness Elements
the Turbulent Boundary Layer, AIAA 88-0134.
3.
Presentations
Ron F. Blackwelder, "Some Ideas on the Control

of Near-Wall Eddies", Invited paper
at the AIAA 2nd Shear Flow Conference, Tempe, AZ, March 13-16, 1989.
J. B. Roon and R. F. Blackwelder, "Modification of Turbulent Boundary Layers
by Longitudinal Roughness Elements", American Physical Society, Division of Fluid
Dynamics, 40th Anniversary Meeting, November 1987.
J. B. Roon and R. F. Blackwelder, "The Effects of Longitudinal Roughness Elements
Upon the Turbulent Boundary Layer", AIAA 26th Aerospace Sciences Meeting, January 1988.
List of Honors/Awards
R. F. Blackwelder, Elected Fellow of the American Physical Society
R. F. Blackwelder, Elected Associate Fellow of the American Institute of Aeronautics
and Astronautics
R. F. Blackwelder, Keynote Address, International Symposium on Transport Phenomena, Tokyo, Japan
List of Participants
R. F. Blackwelder, Professor
J. B. Roon, Ph.D. student
SF.
Pray, M.S. student
CONTROL OF LAMINAR SEPARATED SHEAR LAYERS ON AIRFOILS

Principal Lvestigators: R. F. Blackwelder, R. H. Liebeck
Research Summary
The primary goal of this research is a better understanding of the transitional instability
mechanism which has a controlling effect on the extent of laminar separation bubbles occurring on airfoils operating at low Reynolds numbers (R, < 10'). With applications which
include, for example, low speed unmanned aircraft, these airfoils experience an incremental
drag penalty which becomes more severe as the extent of laminar separation region increases
at lower R,. Earlier transition has been shown to correlate with smaller bubbles and reduced
drag. Results include a detailed experimental investigation of the low Reynolds number airfoil flowfield over a range of incidence, R, and flow disturbance configurations such as free
stream turbulence, acoustic disturbances and surface modifications. Ultimately, a greater
knowledge of the separation bubble and instability mechanism would encourage the following:
1.
Development of active control of the separated shear layer, such as local acoustic
forcing, to accelerate transition within the separation bubble in cases where the drag
penalty severely inhibits performance.
2. Increased confidence in low Reynolds number airfoil flowfield prediction methods
which incorporate stability calculations for determining transition characterstics.
3. Design methods incorporating calculations based on more reliable prediction methods for this separated flow phenomenon resulting in more efficient (lower drag) airfoils.
Significant Results
Experimental work since this research was incorporated into the URI program last year
represents a continuation of work previously funded by ONR under contract N00014-84K-0500. Baseline boundary layer and drag data were obtained for a new 11.75" chord
LA2573A airfoil model over a Reynolds number range from 235,000 to 500,000. These results
successfully reproduced drag estimates from a 6" chord model where previous measurements
were made at R, = 235,000. Mean velocity profiles normalized with the boundary layer
displacement thickness also collapsed data from both models at particular chord positions,
as well as for different Reynolds numbers at positions upstream of transition. This result
also supports transition as a controlling factor in the growth of the separation bubble and
drag. Spectral analysis was made of the velocity signal inside the boundary layer to
ooverall
document the evolution of boundary layer instability and fluctuations through separation
and transition for different airfoil incidence and Reynolds numbers. A spectral peak which
develops in the separation region during transition has been found to correspond to the most
10
amplified waves from a stability theory calculation. This calculation incorporated FalknerSkan reverse flow profiles to represent the experimental flow. The measured spectral peak
frequency follows the stability theory predictions as the airfoil chord length and Reynolds
number were varied. Preliminary estimates of the amplification rates also agree with those
predicted by stability calculations.
Tests have been conducted in which Wheeler vortex generators were applied to the airfoil
model surface upstream of the laminar separation point. Results indicate that a decrease in
drag as high as 30% from the clean airfoil is possible at Reynolds numbers below 400,000
with proper vortex generator sizing and spanwise spacing. The generators also showed a
drag improvement cver the case in which a standard grit roughness transition strip was
used. Optimization of vortex generator size and spacing, and an analysis of drag sensitivity
to airfuil incidence, remains to be investigated.
Experiments will investigate the spanwise coherence of the separated shear layer instability to determine the extent to which a local forcing disturbance might be used to accelerate
transition over the entire airfoil surface. The effect of a local acoustic forcing and the addition of airfoil sweep will also be considered in examining the evolution and amplification of
the separation bubble instability.
List of Publications/Reports/Presentations
None
2. Technical Renorts
None
3. Presentations
None
R. H. Liebeck, McDonnell Douglas Fellow
I
I
U
|1
R. F. Blackwelder, Professor
R. H. Liebeck, Adjunct Professor
P. J. LeBlanc, Ph.D. student
Mark Emanuelli, Undergraduate Assistant
David Herriot, Undergraduate Assistant
Scot Hutcherson, Undergraduate Assistant
Michael Kerho, Undergraduate Assistant
I
IU
I
I
12
CONTROL OF THE LARGE-SCALE FEATURES IN TURBULENT MIXING

LAYERS
Principal Investigator: F. K. Browand
Research Summary

The broad goal is the explore the requirements for the control of turbulent shear flows.
Control is effected by introducing perturbations which exert influence upon the underlying
(turbulent) vortex structure. The turbulent mixing layer has been chosen as a candidate
flow, and the response to a variety of imposed excitations will be investigated. Experimental results will be compared with numerical solutions of various evolution-equation models.
These equations are simpler than the Navier-Stokes equations, yet hopefully contain the
essential features needed to describe the large- scale flow evolution. The connection between
experiment and theory will provide a strong predictive capability for the exploration of control strategies.
Significant Results
Perturbations can be applied at the origin of the mixing layer by means of acoustic
forcing. One excitation of interest is periodic in time and periodic across the span of the
flow. Wave lengths and periods are comparable to the streamwise spacing of developing
vortices and their average passage period, respectively. Other excitations are designed to
produce discrete space-time vortex structure; for example, vortex pattern imperfections c t
dislocations. The motion and development of pattern imperfections is a central comparisor
with evolution equation models.
We have constructed and tested an array of speakers designed to provide acoustical
excitation. Sixteen small dynamic speakers are arranged in a linear array across the wind
tunnel ceiling directly above the splitter plate trailing edge (origin of mixing layer). All
speakers are operated at constant amplitude, but the phase of any speaker can be varied.
A variety of spanwise modes can be generated by simply switching the phase of various
speakers between the two states 0 degrees and 180 degrees. Using a three microphone array,
and an automated data acquisition system, the acoustic field in the cross section of the plate
trailing edge has been surveyed. The results show that low wave number modes are easily
excited at the plate trailing edge because these waves are near resonance at frequencies of 1
kHz or less. The higher standing wave modes (corresponding to more than four wave lengths
across the span) are more difficult to excite. The forcing signal decays exponentially away
from the wind tunnel ceiling and, at the central plane, the acoustic field is again dominated
by lower modes. The presence of unwanted lower modes in this case is probably due to slight
differences in amplitude of the different speakers.
13
1
m
Vortex pattern imperfections, which appear similar to dislocations in a regular crystal

lattice, arise naturally in the nonlinear evolution of the mixing layer flow. Such dislocations
are the major pattern defect and are important in understanding the transition process. We
have documented the distribution of dislocations across the span, and have also counted the
number of dislocations in a suitable space-time area as the flow develops downstream. These
numbers are being compared with the predictions from our nonlinear evolution-equation
model.
Dislocations which arise naturally cannot be studied in detail because they appear randomly in space and time. We have therefore perfected a more sophisticated forcing procedure
to produce artificial imperfections in the vortex patterns. Pulse trains containing two frequencies differing by about 10% are produced by computer. The pulse trains are output
to various speakers across the span via shaping networks and amplifiers. These differing
frequency acoustic pulse trains produce velocity perturbations at the origin of the flow and
impose differing separations upon the growing vortex structure. The dislocations which form
from each application of the pulse train are highly reproducible. (Only small amounts of
input forcing are required.) The flow field surrounding the dislocation can be constructed
by utilizing many identical pulses. In this way, very high spatial resolution can be achieved
with a small number of measuring probes. We are presently using three hot wire probes and
twenty pulses for a total of sixty span measurement stations surrounding the defect.
The principal preliminary result is that a dislocation-once-formed is a virulent structure.
It steepens dramatically while travelling downstream and nucleates additional defects in the
immediate vicinity. The evolution equation model displays a similar behavior, and we are
quite encouraged by this qualitative agreement. The results are discussed in publication (b),
and the companion publication by Yang, Huerre, and Coullet.
Our plan for the coming year is to study the evolution of a variety of defect structures
in more detail. We would like to obtain a global picture of the defect evolution. This will
include quantitative measures of defect motion within the pattern, and the number and
placement of additional defects which are nucleated. These results will be compared with
the evolution equation model. We will also begin to measure various components of the
evolving vorticity field as well as the velocity fluctuations.
I
I
I
1. Papers Published in Refereed Journals/Conference Proceedings
(a) F. K. Browand and C.-M. Ho. "Forced, Unbounded Shear Flows," Nuclear Physics B,
(Proc. Suppl.) 2 1987 InternationalConference on the Physics of Chaos and Systems
Farfrom Equilibrium, M. Duong-Van, editor, North Holland.
(b) F. K. Browand and S. Prost-Domasky. "Experiment on Pattern Evolution in the 2-D
Mixing Layer," New Trends in Nonlinear Dynamics and PatterningPhenomena: The
Geometry of Non Equilibrium,NATO ASI Series B, P. Coullet and P. Huerre, editors,
New York/London Plenum, 1989.
(c) F. K. Browand and S. Prost-Domasky, "A Technique for Acoustic Excitation of SepShear Flows: Preliminary Results," Nonlinear Interaction Effects and Chaotic
Motions - Volume 7, M. Reischman, M. Pardoussis, R. J. Hansen, editors, ASME
Meeting, Chicago (Book no. G00447).
Iarated
3Winter
INone
3. Presentations
F. K. Browand, "Forced, Unbounded Shear Flows," The Physics of Chaos and Sys.
tems Farfrom Equilibrium, Monterey, California, January 11-14, 1987.
F. K. Browand, "Experiments on Pattern Evolution in Mixing Layers," New Trends in
Nonlinear Dynamics and PatterningPhenomena: The Geometry of Non Equilibrium,
Cargese, France, August 2-12, 1988.
IF.
3
F. K. Browand, "Pattern Evolution in Mixing Layers," International Congress of

Theoretical and Applied Mechanics, Grenoble, France, August 21-27, 1988.
K. Browand, "A Technique for Acoustic Excitation of Separated Shear Flows: Preliminary Results," ASME Winter meeting, Chicago, Illinois, November 27-December
2, 1988.
F. K. Browand, Elected Fellow of the American Physical Society, November 1988.
I
I
15
U
I
List of Partidpants
F. K. Browand, Professor
S. Prost-Domasky, Research Assistant
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
16
DIRECT NUMERICAL SIMULATIONS OF FLOW CONTROL

Principal Investigator: J. A. Domaradzki
Research Summary
3
I

The principal goal is to simulate numerically specific fluid flows of interest to the Navy
in order to better understand their physics, in particular those physical features which can
be effectively controlled by either external disturbances or boundary conditions.
Two types of flows were chosen for investigation: free convection and boundary layer flow
over a concave surface. The understanding of convective flows is important in the task of
designing heat pipes, heat exchangers, and cooling of electronic equipment. The fluid flow
over a concave surface develops a regular pattern of Gbrtler vortices which can serve as a
general model of counter- rotating vortices encountered in other transitional and turbulent
boundary layers. The behavior of these vortices is important in many practical situations;
for example, determining efficiency of turbine blades, drag on objects moving in fluids, etc.
The goals of this research are to develop numerical codes capable of simulating these
phenomena, investigate the possibility of influencing heat transfer through modification of
boundary conditions, describe energy transfer in the transitional and turbulent boundary
layers over concave walls, and prepare new methods of delaying transition in such flows.
An additional research topic, partially supported by the ONR-URI during the last year,
involved an investigation of energy transfer in numerically simulated isotropic turbulence.
This work was done in collaboration with Dr. R. S. Rogallo of NASA-Ames and was also
a part of the Summer Program 1989 of the Stanford/NASA-Ames Center for Turbulence
Research.
The goals of the energy transfer analysis are to better understand dynamics of interactions
between large and small scales of turbulence. The clear understanding of these interactions
is necessary in order to assess how small scales of turbulence can be influenced by modifying/controlling only large scales which are more amenable to experimental manipulation
than the small scales.
Significant Results
3
*
In the first year of this effort we were able to obtain new physical results in the area
of convection. Our rapid progress in this area was possible because we were able to use a
numerical code which was developed previously under contract from DOE to simulate convection with an imposed mean shear. We have performed a number of numerical simulations
of the convection between two rigid, horizontal plates with nonuniform-temperature boundary conditions at the lower plate. The boundary conditions were chosen to excite n pairs
17
of counter-rotating convective rolls. For given horizontal size of the computational domain,
runs with different values of the parameter n were performed. It was shown that different
numbers of convective rolls could be accommodated in the same physical domain. In particular, it was possible to generate through the nonuniform-temperature boundary conditions
rolls with sizes significantly different from the size of rolls encountered in natural convection
at the same Rayleigh number. The control over the size of the convective elements allowed
about 15-20% heat transfer increase as compared with the case of natural convection.
In the area of numerical simulations of G~rtler vortices, the first year effort was devoted
to the modification of the flat plate boundary layer code to include curvature effects, and
the modification of the pressure boundary conditions in the full time-splitting scheme to
eliminate the velocity divergence errors.
In the last year the numerical code has been implemented on the Cray X-MP at the
San Diego Supercomputer Center. Through a number of separate proposals to the SDSC, a
total of eighty-seven hours of CPU time were obtained to perform the numerical simulations.
About thirty hours have been used so far.
The following results have been obtained in our studies to date. The growth rates of
unstable Gbrtler vortices obtained from the linear stability theory were compared with the
growth rates predicted by the Navier-Stokes solver. The results agree to four digits after
the decimal point. This level of accuracy is less than that attained by divergence-free codes
and comparable to the accuracy obtained by using the full time splitting schems. Thus
it seems that various modifications of the pressure boundary conditions in the full timesplitting scheme designed to reduce the velocity divergence errors do not improve the overall
accuracy of the code.
Simulation of the G~rtler flow were performed with different initial conditions for the
velocity field (a random initial condition and an initial condition composed of the most
unstable mode of the linear theory). The results were compared with the experiments of
Swearingen and Blackwelder (JFM 182, 255 (1987)) and a good agreement was obtained in
the laminar regime. At the present time we are probing the beginning of the transitional
regime. The numberical simulations properly reproduce all qualitative features of this regime
as observed in the experiments.
The most difficult, technical problem encountered in these simulations was related to
a need to adequately resolve two boundary layers with different thicknesses induced by
the counter-rotating vortices: the thick boundary layer in the upwash region and the thin
boundary layer in the downwash region. The mesh points in the direction normal to the
plate have to be carefully distributed and the minimum resolution requirements for accurate
simulations are 128 mesh points in the vertical.
Regarding the topic of the energy transfer in isotropic turbulence studied during the
Stanford/Ames Summer Program 1989, detailed measurements were made of energy transfer
among the scales of motion in incompressible turbulent fields at low Reynolds numbers generated by direct numerical simulations. It was observed that the transfer resulted from triad
interactions that were nonlocal in k (wavenumber) space, but the energy always transferred
locally. The ONR-URI support was used to investigate the energy transfer at high Reynolds
numbers in the framework of the Eddy-Damped Quasi-Normal Markovian (EDQNM) apS18
I
I
proximation. The EDQNM theory predicts the same transfer mechanism in the inertial
range that is observed at low Reynolds numbers (i.e., predominantly local transfer caused by
nonlocal triads). Coupling between very large and very small scales of turbulence revealed
by our analysis suggests that it may be possible to influence dynamics of the small scales by
active control/modification of the large scales.
Next year's effort will be devoted principally to applying a modified boundary layer
code to investigate transition to turbulence in the G6rtler flow. In particular we plan to
use the Reynolds stress and energy equations to assess the importance of different physical
effects (inflectional velocity profiles in spanwise and normal direction, three-dimensional
disturbances, low speed streaks, etc.) in the transition to turbulence.
I
I
I
I
I
I
I
I
I
19
Domaradzki, J. A., "Heat Transfer Enhancement in Rayleigh-Binard Convection"
accepted for publication Int. J. Heat Mass Transfer, April 1989.
Domaradzki, J. A. and Rogallo, R. S., "Local Energy Transfer and Nonlocal Interactions in Homogeneous, Isotropic Turbulence" to be submitted to Phys. Fluids.
None
3. Presentations (invited seminars)
"Local Energy Transfer and Nonlocal Interactions in Isotropic Turbulence"
* University of Houston, Houston, Texas, April 27, 1989
. National Center for Atmospheric Research, Boulder, Colorado, May 19, 1989
None
J. A. Domaradzki, Assistant Professor
Liu Wei, Ph.D. student (supported by Powell Foundation Grant as part of institutional support in conjuction with the ONR-URI contract).
I
I
I
I
I
I
20
CONTRuL OF ENTRAINMENT IN CONFINED SHEAR LAYERS

Principal Investigator: C.-M. Ho
Research Summary
The control of mixing processes between two fluids is crucial in improving the efficiency
of many engineering applications such as combustors, rocket engines, chemical lasers and
heat exchangers. Since mixing and entrainment are dominated by the unsteady evolution
of coherent structures in the shear layer between the streams, active and passive forcing
techniques have been developed to control this evolution. A large increase in entrainment
has been achieved in some cases involving active forcing in shear flows and passive and
active forcing in asymmetric jets. However, these studies were made in unconfined flows
while almost all engineering applications are confined. Confinement can significantly alter
the development of coherent structures in shear layers, particularly in cases where the shear
layer reattaches to a wall such as in sudden duct expansions and dump combustors. Thus,
the goals of the present experimental research are to understand the effect of confinement
on forced shear layers.
Significant Results
(i) Facility and Instrumentation
A new multi-configuration water channel facility has been built to study the effects of
active and passive control on confined three dimensional reattaching shear layers. After
testing several flow systems, this versatile channel is in full operating condition. Although
the configuration can be varied from a single-sided backward facing step to a fully asymmetric
rectangular sudden expansion, experiments to date have concentrated on a 2:1 aspect ratio
rectangular jet which undergoes a sudden expansion with uniform step height (h). The use
of a rectangular jet of small aspect ratio rather than an axisymmetric or plane confined
jet comprises the passive forcing of the shear layer. Figure 1 shows a schematic of this
configuration.
Two diagnostic systems have been developed to study this flow. Hydrogen bubble flow
visualization is used to qualitatively examine the dynamics of large vortex rings in this
configuration and to study the structure of the reattachment region. Two-component laser
doppler velocimetry is used to quantitatively study the steady state flow. Data acquisition
is entirely computer controlled.
*
I
I
21
(ii) Flow-Field Characteristics

Preliminary visualization studies reveal that a large asymmetric vortex ring, created
by impulsively starting the rectangular jet, experiences rapid breakdown as a result of the
confining effects of the walls. Where a free vortex ring of this shape would undergo several
axis switching. as a result of self induction before decaying, the confined ring is slowed and
distorted by the walls, as well as the attendant image vortices, and breaks down soon after
a single axis switch. Since the development of the shear layer surrounding a rectangular
jet can be dominated by coherent vortex structures resembling such a starting vortex, these
results suggest that shorter reattachment lengths and high mixing levels can be expected in
the minor axis plane under steady state conditions.
LDV measurements confirm this picture of the flow structure and, in fact, reveal that
no time-averaged recirculation zone is found in the minor axis plane. Instead, a region of
slightly positive mean flow with moderate fluctuations is found downstream of the jet. This
is a new and significant finding. In order to compensate for the streamwise mass flux, there
must exist circumferential mass transfer. In other words, in the region immediately after the
step, which is usually a fuel rich region in a ram jet, mixing can be greatly enhanced. These
results demonstrate that passive forcing in the form of jet asymmetry can have a significant
effect on the flow field of a confined shear layer.
There is strong recirculation, however, in the major axis plane. This is shown in Figure
2 where the time-averaged streamwise velocity is plotted versus downstream position at a
distance of 0.125 inches from the end of the minor axis, and 0.2 inches from the end of the
major axis. The recirculation zone length is 10 step heights, which is slightly longer than
expected for a step Reynolds number of 12,000.
(i) The lack of a recirculation zone in the minor axis plane is an extremly important
finding in the control of confined shear layers. We will concentrate our effort in examining
this phenomenon from the deterministic structure point of view.
(ii) A rotating valve has been installed recently in order to study the effect of active
global forcing on the shear layer recirculation regions. Preliminary results indicate futher
significant effects on the flow field structure.
I
I
IEn
I
ow
U),
L
On0U)0I
L
LO
II
oas/ouT
-ueow
cJ
CD
23
3List
i
of Publications /Reports /Presentations

C.M. Ho, T. Austin and J.R. Hertzberg "Entrainment of 3-D Shear Layers". Paper
by the Fourth Asian Congress of Fluid Mechanics, Hong Kong, 1989.
3accepted
INone
3. Presentations
J.R. Hertzberg and C.-M. Ho "Vortex Dynamics in an Isothermal Dump Combustor".

Poster Session at the Twenty-Second Symposium (International) on Combustion,
University of Washington, Aug. 14-19, 1988. Abstract published in Symposium
Abstract book, Combustion Institute 1988.
J.R. Hertzberg and C.-M. Ho "Vortex Dynamics in a Rectangular Sudden Expansion".
Paper presented at the Forty-First Annual Meeting of the Division of Fluid Dynamics,
American Physical Society. Abstract published in Bull. A.P.S. Vol. 33, no. 10 (1988),
p 2233.
3None
Ho, Chih-Ming, Professor
IJean
3Mike
Hertzberg, Research Associate

Bailey, Undergraduate Assistant
Chris Niestepski, Undergraduate Assistant

Scott Reader, Undergraduate Assistant
1
1
1
1
24
I.
U
I
I
THEORETICAL UNDERPINNINGS: AMPLITUDE AND PHASE DESCRIPTIONS OF COHERENT STRUCTURES

Principal Investigators: P. Huerre and L. G. Redekopp.
Research Summary
The dynamics and control of coherent structures in free shear flows such as mixing layers
and wakes is to be analyzed in terms of a suitable superposition of interacting instability
waves subjected to external forcing. Rational analytical models of the global dynamic. are
to be obtained in terms of nonlinear evolution equations governing either the amplitude or
the phase of the wave. The spatio-temporal dynamics of the structures under both natural
and controlled conditions will be studied numerically on both types of models. Efficient flow
control strategies are to be developed to alter the long-range interactions between structures.
Streamwise vortices are known to play a crucial role in the transition from laminar to
turbulent flow in boundary layers. Analytical and numerical studies of the development of
Gbrtler vortices on a concave plate are to be conducted to arrive at a fundamental understanding of the primary instability evolution and of the development of secondary instabilities.
Significant results
325
Progress has been made in 4 distinct areas as summarized below.

(1) A phase dynamics description of the evolution of Kelvin- Helmholtz vortices in stratified mixing layers has been rationally derived from the basic governing equations in the
viscous critical layer limit [Huerre (1987)]. The combined translational and Galilean invariance of the initial problem has been shown to give rise to strong coupling between the
structures and a large-scale, horizontal, velocity field. Two types of phase instabilities have
been identified. A modified Eckhaus instability can occur as local "compressions" and "expansions" of the vortex array are induced along the stream. This mechanism, which is
primarily one-dimensional, could account for phase decorrelations between structures. The
second type of phase instability is three-dimensional in character, the most amplified disturbance being obtained at a wavenumber inclined to the flow direction. Finally, in the region
of stable configurations, anisotropic phase waves may propagate on the vortex structures.
(2) Concurrently, Mr. Ruixin Yang, a Ph.D. student supported by the ONR core program,
has carried out a numerical study of a two-dimensional model of pattern evolution in *ixng
layers (Yang, Huerre, and Coullet (1988)]. This investigation is a joint collaboration with
F. K. Browand who is performing companion experiments. In contrast with the previous
case, the field equation governing the amplitude evolution of the waves has been postulated.
*26
The invariance properties of the problem, together with the presence of shear, place strict
constraints on the nature of the linear operator, as observed in the context of convection by
Swift and Hohenberg at Bell Laboratories. Interactions with the large scale field have been
inferred from the asymptotic model described above under item (1).
This non-potential pattern evolution model has been shown to reproduce many of the
features observed in mixing layers. Experimentally one observes a downstream evolution
toward an ordered quasi-two-dimensional pattern. Imperfections in the pattern lead to the
formation of dislocations. If the initial state is chosen to be turbulent, the model results
reveal a similar behavior: random input gives rise to organized spanwise vortices in the
long-time evolution, with the presence of dislocations. When a mixing layer is acoustically
perturbed at slightly different frequencies along the span, controlled dislocations can be
produced experimentally which generate additional defects further downstream. Dislocation
pairs are also produced in the model by suitably choosing the initial state. The results of
numerical simulations indicate that new defects are then nucleated [see Fig. 1]. The last type
of input which has so far been analysed for the model involves noisy, random, infinitesimal
perturbations superimposed on a perfect two-dimensional array of vortices. Remarkably,
dislocations are found to be spontaneously generated in the flow as a result of intrinsic phase
instabilities [see Fig. 2]. This result may provide an important clue to the mechanism of
dislocation nucleation in shear flows.
(3) A rational model for the amplitude evolution of coherent structures in confined mixing
layers has been derived for both two-dimensional and three-dimensional motion. The model
represents a significant reduction of the full equations of motion in that the cross-stream
structure is separated from the streamwise and transverse variations. The resulting model
is an integro-partial differential equation in one- or two-space coordinates plus time. The
model accounts for interactions between all unstable wavenumbers. Thus, transfer of energy
to larger scales and vortex pairing are captured by the model. This is in contrast to a phase
dynamics description which focuses on the large scale interactions between structures whose
amplitude is entirely slaved to phase variations. The nature of the finite-amplitude dynamics
described by the present model is found to depend on both the structure of the critical layer
encompassing the inflection point of the basic velocity profile and the super-criticality of the
profile. The space-time scalings implicit in the two-space dimensional model are consistent
with Squire's transformation of the plane wave dispersion relation. A numerical code is
currently being developed to study the response of the model system to controlled inputs.
(4) The linear evolution of wave packets in boundary layers on a concave wall with suction
has been studied analytically and numerically. The application of suction removes technical difficulties associated with the growth of the boundary layer. The basic velocity profile
remains unchanged in the streamwise direction. It has been shown numerically [Park and
Huerre (1988)] that the impulse response of the flow takes the form of a wavepacket which
is convected downstram. Thus, the GOrtler instability is convective in character and it is
expected to be extremely sensitive to external perturbations. This result has been confirmed
by a study of the dispersion relation in the complex frequency plane: there are no saddle
points at finite distance. A nonlinear evolution model of the Ginzburg-Landau variety has
been derived in the vicinity of the critical G~rtler number. The cubic-nonlinearity was found
to be destabilizing. G6rtler vortices, therefore, appear via a subcritical stationary bifurcation. A full numerical code has been developed to characterize the finite-amplitude branch
away from threshold when the motion is assumed to be uniform in the streamwise direction.
It has been confirmed that the bifurcation is subcritical. As perturbations grow away from
the basic state to reach the finite-amplitude solution, the mean velocity profile exhibits extremely strong inflection points. In the presence of streamwise fluctuations these would be
the site of Kelvin-Helmholtz instabilities.

Research in the above four areas will be actively pursued.
3
I
Preliminary analytical investigations of the motion of dislocations have been undertaken.

In the absence of a large-scale vertical vorticity field, one can appeal to available results in
anisotropic nematic liquid crystals. The velocity of dislocations is then linearly proportional
to Sq. Here Sq is the departure of the background wavenumber from the value at which
dislocations are stationary. We would like to verify this scaling law numerically and determine
the effects of the vertical vorticity field. Perturbation analyses will be used to obtain the
dominant contribution from non-potential effects. Coefficients appearing in the pattern
evolution model will be calibrated by fitting the properties of the model to known linear
and nonlinear stability characteristics. This will supply valuable information to compare
the results of the model with the experiments. Concurrently, we shall characterize the main
statistical properties of the turbulent regime in situations where defects are spontaneously
nucleated (defect-mediated turbulence).
The amplitude evolution model will be used to study the response of the mixing layer
to one- and two-dimensional disturbances. Vortex pairing and collective interactions will be
studied using the one-space dimensional model. The two-space dimensional model will be
used to study spanwise-varying inputs and comparisons with experimental observations by
Browand will be made. the evolution model will then be used to explore forcing scenarios
which either enhance or delay mixing layer growth and the level of three-dimensionality. A
control theory for the one-space dimensional model will be initiated.
In the G6rtler problem, the formulation needs to be generalized to include the secondary
instability of Gbrtler vortices to streamwise dependent perturbations. It is important to
determine if the finite-amplitude G~rtler branch is always unstable to Kelvin-Helmholtz
disturbances. Fortunately, the time scales pertaining to the development of G6rtler vortices
and to the Kelvin-Helmholtz mechanism are well-separated. We intend to take advantage
of this feature to set up a multiple scale analysis of the development of Kelvin-Helmholtz
waves on a slowly-varying Gortler field. This study will provide valuable information on the
dynamics of spatially-periodic, inflectional profiles.
I.
I
(b)
(a)
Figure 1. Nucleation of dislocations
by disloc
ations.
I
3
(a)
dislcatins.(d)
x(b
*
I
U
(c)
*
(d)
Figure 2. Nucleation of dislocations

by phase perturbations.
28
Yang, R., Huerre, P., and Coullet, P. (1988). "A Two-Dimensional Model of Pattern
Evolution in Mixing Layers." To appear in the proceedings of "New Trends in Nonlinear Dynamics and Pattern-Forming Phenomena: The Geometry of Non Equilibrium.:
NATO ASI Series B, P. Coullet and P. Huerre, editors, New York/London: Plenum.
Huerre, P. (1987). "Evolution of Coherent Structures in Shear Flows: A Phase Dynamics Approach." Nuclear Physics B (Proc. Suppl.), 2:159.
Djordjevic, V. D., and Redekopp, L. G. (1989). "Nonlinear Stability Properties of
Subsonic Mixing Layers with Symmetric Temperature Variations." accepted for publication in Phil. Trans. of Roy. Soc. London.
2. Tecaiical Reports
Djordjevic, V. D., Pavithran, S., and Redekopp, L. G. (1989). "Stability Properties
of Subsonic Mixing Layers." AIAA Paper No. 89-1020.
3. Presentations
Redekopp, L. G. (1989). "Stability Properties of Subsonic Mixing Layers." AIAA
Second Shear Flow Conference, Tempe, AZ, March 12-16, 1989.
Park, D., and Huerre, P. (1988). "On the Convective Nature of the G6rtler Instability." Bulletin of the American Physical Society, vol. 33, p. 2252.
Yang, R., Huerre, P. and Coullet, P., (1988). "Models of Pattern Formation and
Dislocations in Free Shear Flows." Bulletin of the American Physical Society, vol.
33, p. 2256.
Huerre, P. (Dec. 21, 1988). "Structures coherentes et ondes d'instabilite dans le
ecoulements cisailles." Seminar, Department of Mechanics, Ecole Polytechnique,
Palaiseau, France.
IFluid
Huerre, P. (Jan 9-11, 1989). "Hydrodynamic Instabilities in Free Shear Flows: spatiotemporal descriptions." Invited series of lectures, DARPA/URI Winter School in
Dynamics, Institue for Nonlinear Science, University of California, San Diego.
Huerre, P. (May 3, 1988). "Nonlinear Dynamical Systems", Joint ONR/AFOSR Colloquium on Active Control and Turbomachinery Hydroacoustics, Washington, D.C.,
May 3-4, 1988.
Huerre, P. (May 2, 1988). "Order and Disorder of the Large Scales in Free Shear
Flows", Seminar, Naval Research Laboratory, Washington, D.C.
Huerre, P. (March 9, 1988). "Global Instability and Chaos in Free Shear Flows",
Department of Mechanical and Materials Engineering Seminar, Washington State
University, Pullman.
Huerre, P. (Jan. 29, 1988). "Local and Global Evolution of Instability Waves in Free
Shear Flows", Fluid Mechanics Seminar, California Institute of Technology, Pasadena.
SHuerre, P. (Jan. 22, 1988). "Large-Scale Dynamics in Open Flows", Nonlinear Science Seminar, University of California, Santa Barbara.
Huerre, P. (Dec. 16-22, 1987). "Spatio-temporal Evolution of Coherent Structures."
Invited lecture, Second International Workshop on Instabilities and Nonequilibrium
Structures, Universidad Federico Santa Maria, Valparaiso, Chile.
Huerre, P. (Nov. 24, 1987). "Spatio-Temporal Dynamics in Free Shear Flows",
Invited Lecture, 40th Annual Meeting of the Division of Fluid Dynamics, Eugene,
Oregon, Nov. 22-24, 1987.
Asano, H., Huerre, P. and Redekopp, L. G. (1987). "Finite Amplitude, Three- Dimensional, Kelvin-Helmholtz Instability Waves in Bounded Shear Layers." 40th meeting
of the Division of Fluid Dynamics of the American Physical Society, November 1987,
Eugene, Oregon.
U
I
Huerre, P. (Sept. 11, 1987). "Dynamique des Structures Coherentes et Ondes

d'Instabilite dans lee Ecoulements Cisailles", Special Seminar, Universite d'Orsay,
Paris, France.
Huerre, P. (Aug. 20, 1987). "Heuristic Introduction to Chaos", Miniworkshop, China
Lake Naval Weapons Center, California.
Huerre, P. (July 23, 1987). "Amplitude and Phase Evolution of Coherent Structures
in Free Shear Flows", invited talk, American Mathematical Society Conference on
"The Connection between Infinite Dimensional and Finite Dimensional Dynamical
Systems", University of Colorado at Boulder.
Huerre, P. (Mar. 18, 1987). "Hydrodynamic Instabilities in Open Flows: A Bird's
Eye View", Workshop on "Propagation in Far From Equilibrium Structures", Les
Houches, France, March 11-18, 1987.
30
Browand, F. K., and Huerre, P. (Jan. 13, 1987). "The Structure of the Turbulent
Mixing Layer", with F.K. Browand, International Conference on the Physics of Chaos
and Systems Far from Equilibrium, Monterey, California, January 11-14, 1987.
Huerre, P. (Oct. 24, 1986). "Phase Dynamics of Large Structures in Benard Convection and Stratified Mixing Layers", Applied Mathematics Colloquium, University of
Arizona, Tucson.
*
P. Huerre, May 1989, TRW Award for Teaching Excellence, School of Engineering,
USC.
L. G. Redekopp, Elected Fellow of the American Physical Society, November 1987.
P. Huerre, Professor
L. G. Redekopp, Professor
H. Asano, Ph.D. student
D. Park, Ph.D. student
R. Yang, Ph.D. student (not supported by ONR-URI contract)
It
I
I
I
I
I
I
i
31
PHASE CONTROL OF VORTEX STRUCTURES IN SHEAR FLOWS

Principal Investigator: T. Maxworthy
Research Summary

Persistent patterns of vortex structures emerge as a result of instability processes in many
shear flows. These patterns exhibit a high degree of spatial coherence and determine the
characteristic scales which dominate the turbulent energy specturm. One approach toward
control or influence of the turbulent scales is through external control of the phase relationships between vortex structures on length scales which encompass many individual features.
The goal of this task is to produce laboratory flows containing an extensive pattern of vortex
structures driven by a common instability mechanism, and then to explore the response of
these patterns to control inputs which principally influence the phase relationship within the
pattern. In particular, we seek to clarify the response of vortex patterns to phase disturbances which are incommensurate with the underlying instability scale.
SiAnificant Results
1
3
*32
The first of two pieces of apparatus to study problems of vortex dynamics and control
has been built and placed in operation. It has been designed to complement the numerical
work of Professor Domaradzki and the theoretical work of Professor Huerre on unsteady
G~rtler instability. Here we study the controlled spin-down of a rotating cylinder initially
containing a fluid in solid-body rotation. As the boundary layer on the concave cylinder
surface grows it becomes unstable to a temporally growing GBrtler instability. The flow can
be observed in a number of ways: initially by reflection of external lighting from aluminum
flakes suspended in the flow and by observing a cross-section of the evolving flow using a
sheet of laser light projected normally to the cylinder surface. In the later approach, velocity
measurements can be made using the "laser speckle" technique. Visualisation is enhanced
by placing the cylinder in a square box filled with refraction-index-matching fluid to remove
distortion created by the cylinder. A number of experiments have been run in which the
cylinder was brought rapidly to rest and the results correlated using a simple extension of the
original Grtler theory. Following the generation of the initial instability, at a well defined
wavelength, the vortex pattern undergoes a rapid and, as yet, ill-understood transition to
a state with approximately double the original vortex spacing. This process includes the
apprearance of dislocations in the vortex rings and a sudden amalgamation of vortices of the
same sign. This wavelength-doubling process continues until the cylinder is filled with a very
disturbed turbulent flow containing discernable azimuthal vortices. Preliminary experiments
on the slow spin-down of the cylinder show, as found in the work of Weidman, that stability
is enhanced and that the vortex "doubling" process is less violent. This is presumably due
to the stabilizing effect of system rotation on the generation of the vortices, a phenomenon
discussed recently by Anderson and a group of the Royal Institute of Stockholm for the case
of rotating, Poiseuille flow in a curved channel (to appear in JFM).
A second laboratory experiment is currently under construction. It consists of two concentric, rotating cylinders with working fluid in the annular gap between them. The inner
cylinder is machined with a wavy surface of known wavelength and amplitude. This provides
a perturbation to the unstable Taylor flow that is incommensurate with that which would
occur in its absence. The adjustment of the flow to this boundary condition has been the
subject of theoretical work by Professor Huerre and his predictions will form the preliminary
basis for our interpretation of the results.
5
3
With ONR sponsorship we have developed methods to automatically track small, neutrallybouyant particles in complex flow fields. These techniques and those of laser speckle velocimetry are ideal for the study of the flow we are considering here. The coming year will
concentrate on velocity measurements within the unsteady Gbrtler flow experiment and the
attempt to understand the complex vortex interactions and reorganizations during the transition to larger vortex scales. The spatial-modulated, circular Couette flow apparatus will
be used to explore the characteristics of the system and to observe, if possible, the states
predicted by the Huerre analysis. Here also, speckle velocimetry will be used to characterize
the velocity field of these flows and their temporal evolution.
None
None
3. Presentations
None
T. Maxworthy, Smith International Professor of Mechanical and Aerospace Engineering
*
I-
3
mm
inmnmnanna
i nnnani
T. Maxworthy, Professor
E. Ikeda, Ph.D. student
J. Wiggert, Ph.D. student
I
I
I
I
I
I
I
U
I
U
I
I
I
I
34
MATHEMATICAL ANALYSIS OF THE NAVIER STOKES EQUATIONS

Principal Investigator: S. S. Sritharan
Research Summary
This research involves three inter-related topics in turbulence theory: (i) investigation
of the structure of attractors and invariant manifolds; (ii) global unique solvability of viscous flow past unbounded geometries; and (iii) optimal control theory. Research is being
performed to understand the functional attractors characterizing three dimensional viscous
motion. We study both the conventional as well as the regularized Navier Stokes equations in
this respect. The attractor of a turbulent flow may be of fractional dimension. It is therefore
of interest to find finite dimensional manifolds ("surfaces") which contain the attractor. If
such manifolds are invariant with respect to the time evolution of the hydrodynamic solution
operator, then the coordinates of these manifolds would provide an efficient computational
tool. The fundamental element in the mathematical analysis of shear flows is the global
unique solvability theorem. This result would be basic to numerical analysis, attractor theory, and also to optimal control theory. There are global unique solvability theorems for
viscous flow in bounded domains (at least for sufficiently low initial enstrophy.) Much less
is known regarding viscous flow in channel-type configurations. One of our major goals is to
investigate this problem for unbounded channel geometries.
The central topic of our research is of course, the control of fluid flows. We have made a
major breakthrough in developing optimal control theory for a certain class of hydrodynamic
situations.
Significant Results
A. Mathematical theory of channel flows. This is a joint research with Professor J. G.
Heywood of the University of British Columbia. We have proved the existence and uniqueness
of two-dimensional, time-dependent, viscous flow past unbounded configurations with outlets
which open at infinity. The two dimensional problem is particularly difficult since the finite
energy solutions cannot carry flux through the outlets. The task was to find a flux carrying
infinite energy vector field.
B. Optimal control theory of viscous flows. We have devloped a theory for driving a given
flowfield to a desired flow field by means of control forcings. Existence of an optimal solution
minimizing a suitable cost quadratic functional is established. Necessary and sufficient conditions for minimum has been worked out using the adjoint problem. When a linearization
of the Navier Stokes equations was used in the above analysis, the task of computing the
feed back map reduces to that of finding the solution of a Riccati type partial differential
equations. (These results will be presented in the SIAM meeting at San Diego in July, 1989.)
35

Stimulated by the breakthrough made in the optimal control of hydrodynamics the past
year, we have decided to focus on two control problems. The first one is the task of finding
the optimal way to accelerate an obstacle from rest to a given speed in a given time. The
reverse of this problem is also important as it models the aerobraking maneuver. The second
problem deals with the task of finding optimal velocity distributions (suction and blowing)
on an obstacle to minimize the drag. A mathematically related problem is shape optimization in viscous flow. The cost functional for each of these problems can be easily written
down and is quadratic in nature.
1
5
1.Papers Published in Refereed Journals

"Analysis of Regularized Navier Stokes Equations - I," submitted to Pacific Journal
of Mathematics, March 1988, joint author Y. R. Ou.
"Invariant Manifold Theory for Hydrodynamic Transition," submitted to Acta Mathematica, July 1988.
Analysis of Regularized Navier Stokes Equations - II," submitted to Pacific Journal
of Mathematics, Y. R. Ou, February, 1989.
"Analysis of Regularized Navier Stokes Equations - II," ICASE report 89-14, joint
author Y. R. Ou.
"Invariant Manifold Theorems for the Navier Stokes Equations," Proceedings of the
Oberwolfach workshop on Theory and Numerical Methods for Navier Stokes Equations, to appear on Springer Lecture Notes.
3. Presentations (invited seminars)
"Invariant Manifold Theorems for the Navier Stokes Equations" SIAM Annual meeting, July 1988, Minnesota.
"Invariant Manifold Theorems for Conventional and Regularized Navier Stokes Equations" Invited Seminar, Mathematisches Forschunginstitut Oberwolfach, West Germany, September, 18-24, 1988
36
"On the Existence Theory of Viscous Time Dependent Flow Past Noncompact Boundaries," Invited Seminar in Mathematical Sciences Department, San Diego State University, February, 23, 1989.
*
None
S. S. Sritharan, Assistant Professor
Y. R. Ou, Graduate Student, presently staff scientist ICASE (not supported by URI
contract).
J. G. Heywood, Professor, University of British Columbia, consultant.
I
I
I
I
I
I
I
I
I
I
I
37
Ii
I
l
I
I
AIAA-88-0134
The Effects of Longitudinal
Roughness Elements upon the
Turbulent Boundary Layer
R. F.Blackwelder and J. B. Roon,
University of Southern California, Los
Angeles, CA
Velocity profiles and bursting statistics
were measured in the resence of cylindrical
longitudinal rougbness elements (IREs) and
compared to similar work by Johaysen and
Smith. 1
An algorithm was devised to detect
low speed streaks (ISS) from the hot-wire
rake data, allowing an estimation of the
UIs ability to reduce spatial
The velocity
of the sublayer structure.
profiles agree with Johansen and Smith's
result that the effects of the taqs are felt
only for y+ j 15.
The analysis of the.
brting statistics and LSS algorithm mt~x
shows tht the flow relaxes bark to a fa

plate boundary layer between the
REs.
crude first approximation is that the UM

Simply imos the no-slip boundary =3dcition
atdiameter.
an elevation
to thei
While the equivalent
LP~s migt serve
asa
nucleation site for the LSS, spatial
rand mness of the structure is effected only
very close to the REs.
AIAA 26th Aerospace Sciences Meeting

January 11-14, 1988/Reno, Nevada
For permission to copy or republish, contact tw American Inftut O Aermmuics and Astronautics
37M LEUaM Promenafe, S.W., Wuahinqan, D.C. 2904
I.
0
|
AIAA-89-1009
SOME IDEAS ON THE CONTROL OF
NEAR-WALL EDDIES
Ron F. Blackwelder
University of Southern California
Abstract
The near-wall region of bounded turbulent flows

consists of streamwise vortices, low speed streaks, intense shear
layers, inflectional velocity profiles, oscillations and ejections
of low speed fluid out into the logarithmic layer. A mass of
data has accumulated over the past 25 years concerning this
important chain of events, however the general picture is still
rather suggestive rather than conclusive. Some proposals are
offered as means of interacting with and/or interrupting the
sequence of these events in the wall region.
AIAA 2nd Shear Flow Conference
March 13-16, 1989 / Tempe, AZ
Astronaubcs
For permssn to cop or repubrmh, contact the American Insdbat of Aeronaucs and
39
370 L'Effant Promenade, S.W., Wastington, D.C. 20024
I
~Reonnted
From
Nonlinear Interaction Effects and Chaotic Motions - Volume 7
Editors: M. M. Reischman. M. P. Paidoussis. R. J. Hansen
The American Society of
(Book No. G00447)
Mechanical Engineers
I
3
A TECHNIQUE FOR ACOUSTIC EXCITATION OF SEPARATED

SHEAR FLOWS: PRELIMINARY RESULTS
F. X. Browand and S. A. Pro.rt-Domasky
Department of Aerospace Ergineering
Los Angeles, California
I
I
The
turbulent
region
contained
between two parallel streams of unequal
speed is the protorypic separated flow. The
controlling influence is partially understood

3asfollows. Shear flows which are tree from
boundaries, such as lets, wakes, and mixing
layers, usually artse as a result of fow
separation. When laminar, these flows are
application of small acoustic perturbations

(or excitations by a trailing edge flap) can
cause the flow to grow more rapidly or less
extremely unstable, and boundary laver

vorticity is collected into discrete clumps.
forming the largest scale features of the
rapidly
depending
upon
the
forcing
frequency. The effect of forcing is either to
promote or discourage interactions among
developing turbulent flow.

When the
boundary layer at separation is turbulent, a
similar process seems operative, and the
the large, nearly spanwisa vortices. Here we

study a
means for producing complex
unbounded
result is again rapid formation of lumps of

vorticity. These vortical regions are-to first
of
rearrangements
approxJmadon-simply
the preexisting boundary layer vorticity.
Thus in flows which are approximately twodimensional. the vortices are onented with
vorticity perpendicular to the plane of
mvotion, while for axlsywnmetric flows, the
vortices arm rings or donuts lying in planes
perpendicular to the axis of symmetry.
The
downstream
growth
of
the
turbulent region is the result of interactions
turbulent
sheer
flows
are
particularly
sensitive to the application of a small,
external disturbance.
This conclusion is
supported by a variety of expenments
performed over the past 15-20 years (c.f. Ho
and Huerre (11 Browand and Ho (2r. and by
a lesser but increasing number of numerical
(computational) experiments. Ho and Huerre
contains an extensive bibliographV
The
between
neighboring
vortices--often
involving mergings to form larger structures
with greater spacings
The application of a
small, soatiallv coherent perturbation-- for
example. !)v use of an oscillating flea at the
plate trailing edge or let nOzze-creates a
stronger, more conerent vortex structure. In
the subsequent downstream development.
these vortices undergo interactions and
ABSTRACT
acoustic forcing. The forcing is applied with

a linear arrangement of small speakers
placed across the span of the flow in the
wind tunnel ceiling.
Flow response is
observed with a rake of hot wires placed

across the span of the flow at selected
downstream positions,
1. INTROOUCT1ON
It
is
ohylsical
well
established
mechanism
that
resoonsible
for
trhis
rearrangements which can be quite different
171
40
I
I
To appear in the proceedings of "New Trends in Nonlinear Dynamics and

Patterning Phenomena: The Geometry of Non Equilibrium". NATO ASI Series B,
P. Coullet & P. Huerre, editors, New York/London: Plenum.
Experiment on Pattern Evolution in the 2-D Mixing Layer
F.K. Browand and S. Prost-Domasky

Department of Aerospace Engineering
Los Angeles, CA 90089-1191
INTRODUCTORY REMARKS
The first detailed, quantitative study of the 2-D mixing layer was completed by Liepmann
and Laufer in 1947. They described this technologically important flow in terms of the variation
of the mean velocity and various mean fluctuation intensities. The measurements were of high
quality, and have scarcely been improved upon in the intervening forty years. Yet they give very
little fundamental understanding of the structure of the flow. What has changed within the past
twenty years is the increased concern with process in turbulent flows. Today turbulent flows are
perceived to contain identifiable structure. The interaction of structureis the turbulent process.
Thus process attempts to provide a dynamical description of the flow, and is a more ambitious
undertaking than a simple description of the state of the flow.

The identification of structure in highly turbulent flows first came about by relatively
simple visualizations using dye or hydrogen bubbles. These techniques allowed the viewing of
selected volumes or areas within the flow in a manner not possible with single point velocity
measurements. Spatial patterns and pattern evolution are emphasized in these qualitative
visualizations. We are now beginning to ask quantitative questions about pattern evolutiog , This
has only become possible within the past ten years or so, as the result of inexpensive comlputers
having vastly increased storage capacity.
The present experiment is an attempt to describe a high Reynolds number, highly turbulent
shear flow from a geometrical point of view, and is an outgrowth of earlier work by Browand
and Troutt, (1980, 1985). Here we extend these concepts to include an evolution equation model
as has been shown to be effective in the study of the Rayleigh-Benard and Taylor-Couette flows.
The companion analytical/numerical effort by Yang, Huerre, and Coullet is presented in the
paper immediately following. The long-term goal is to provide a cornbined mathematical-experimental model which will quantitatively deicribe the evolution of the large scale
structure in the highly turbulent regions of the flow. To date, we concentrate on describing the
initial instability region where a small range of scales is present. The comparisons with theory
are largely qualitative at this stage.
DESCRIPTION OF THE EXPERIMENTAL APPARATUS
The experiment is performed in a large wind tunnel. A side view of the test section is
shown schematically in figure 1. The mixing layer is produced by the merging of two parallel
streams at the termination of a splitter plate. Flow speeds are on the order of 5-20 m/sec. The
inital thickness of the laminar shear layer, 6f, is about 2.3 mm. The width of the wind tunnel -
expressed in multiples of this thickness -
~41
is about 400. The Reynolds number, based upon
I.
I
Heat Transfer Enhancement
in
Rayleigh-Benard Convection
5J.
Andrzej Domaradzki

I
3with
3
Direct numerical simulations of 3-D Rayleigh-Benard convection were performed

non-uniform temperature boundary conditions at the lower plate.
The boundary
conditions excite convective states in a form of counterrotating convective rolls that
Iremain
stable for at least ten eddy-turnover times even if their size is significantly
different from the size of rolls encountered in the natural convection at the same
Rayleigh number.
For longer times applied temperature forcing becomes ineffective and
forced convective patterns lose their stability.
Changing the size of the convective
elements increases heat transfer by about 15-20% as compared with case of the natural
convection.
Accepted for publication in Int. J. Heat & Mass Transfer.
II
I
I
i4
I
Local Energy Transfer and Nonlocal Interactions
in Homogeneous, Isotropic Turbulence
J. Andrzej Domaradzki
Department of Aerospace Engineering, University of Southern California,
Robert S. Rogallo
NASA Ames Research Center, Moffet Field, CA 94035
IApril,1989
Abstract
Detailed measurements were made of energy transfer among the scales of motion in
incompressible turbulent fields at low Reynolds numbers generated by direct numerical
simulations.It was observed that although the transfer resulted from triad interactions
that were nonlocal in k space, the energy always transferred locally. The energy transfer calculated from the EDQNM theory of turbulence at low Reynolds numbers is in
excellent agreement with the results of the numerical simulations. At high Reynolds
numbers the EDQNM theory predicts in the inertial range the same transfer mechanism
which is observed at low Reynolds numbers i.e. predominantly local transfer caused
by nonlocal triads.The weaker, nonlocal energy transfer is from large to small scales at
high Reynolds numbers and from small to large scales at low Reynolds numbers.
To be submitted to Physics of Fluids.
43
I
!North-Hollird,
Nuclear Physics B (Proc. Suppl.) 2 (1987) 159-178
159
Aerdam
EVOLUTION OF COHERENT STRUCTURES IN SHEAR FLOWS.

A PHASE DYNAMICS APPROACH
If
Patrtck HUERRE
Department of Aerospace Engineering, University of Southern California.
Los Angeles. California 90089-1191
U.S.A.
A general amplitude evolution model is proposed to describe the development of instability waves in parallel free shear flows which are invariant under both space translations and Galilean transformations. It is argued that temporal and spatial modulations
of finite-amplitude states are strongly coupled with a marginally unstable large-scale
horizontal velocity field. As a result, an array of spatially periodic coherent structures
is shown to exhibit two distinct types of phase instabilities: an Eckhaus-like modulational instability dominated by two-dimensional disturbances and a fully threedimensional secondary instability generated by coupling between pattern deformation
and advection by the large scale field. Anisotropic phase waves can also travel on
the basic array, the global velocity field acting as a restoring force. Applications
of these concepts to the phase dynamics of Kelvin-Helmholtz vortices are briefly
considered.
31.
INTRODUCTION.
The evolution of free shear flows such as mixing layers, two-dimensional jets and
wakes is dominated, at high Reynolds numbers, by large scale vortical structures which
develop as a result of the instability of the basic velocity profile.
In particular, recent
experimental evidence1 4 has indicated that the instability of free shear layers leads to the
formalism of quasi-two dimensional spatial patterns exhibiting localized dislocations in the
spanwise direction. Some of the experimental results obtained by Browand are illustrated
in Figure 1.
The goal of the present work is to outline a phase dynamical description of
the long-range evolution of such features over many characteristic length scales and time
scales of the flow. The phase dynamics formlfism has been used with great success in the
past to study the slow variations of similarl spatia
conservative waves-, chemical
Couette flow
12
reactionss-,
patterns arising in weakly nonlinear
Rayleigh-B6nard
convection 9 - 1
and Taylor-
It is felt that a similar approach could be fruitful in describing the slow
meanderings of Kelvin-Helmholtz vortices as observed in Figure 1.
gSponsored
by the Office of Naval Research Under Contract No. N00014-86-K-0679.
0920-5632/871$03.500 Elsevier Science Pubishers B.V.

(North-Holland Physi Publishing Diion)
44
To appear in the proceedings of "Ne, Trends in konl inear Dynamics and Patterning
NATO ASI Series B, P. Couilec &
The Geometry of Non Equilibrium''.
Phenomena:
P. Huerre, eiTors, New YEFk7oE-ndon: Plenum.
A TWO DIMENSIONAL MODEL OF

PATTERN EVOLUTION IN MIXING LAYERS
R. Yang, P. Huerre
USA
P. Coullet
Laboratoire de Physique Thiorique
Parc Varose
Universiti de Nice
06036 Nice
FRANCE
INTRODUCTION
The evolution of coherent structures in shear layers and wakes provides a particularly
simple example of pattern dynamics in open spatially-developing non-equilibrium systems. As in dosed flows such as Rayleigh-Binard convection, one observes a wealth
of possible flow configurations involving dislocations, periodic arrays of vortices with
distinct orientation, quasi-two dimensional vortical arrangements, etc. However, in
contrast with Rayleigh-Bhnard convection, the presence of a basic shear lifts the orientational degeneracy: vortices tend to remain more or less perpendicular to the flow
direction.
The geometry of three-dimensional patterns in mixing layers has been extensively
studied from an experimental point of view by Browand and his colleagues [see for
instance Browand (1986), Browand & Ho (1987) and in this volume, Browand & ProstDomasky (1989)]. These experiments have clearly and carefully documented the generation of defects in nearly periodic arrangements of spanwise vortices. Recent experiments by Stuber & Gharib (1988) have demonstrated that complex spatio-temporal
regimes result from external forcing of dislocations in wakes. Lateral boundaries can
also considerably affect the direction of shedding and lead to oblique patterns as shown
by Williamson (1989). The present investigation should be viewed as an attempt at a
theoretical description of some of these phenomena in the specific context of Browand
45
I
i
!
i
IInvariant manifold theory for hydrodynamic
Itransition
SS.S.Srtharan'
3Department of Aerospace Engineering

g
Los Angeles,CA90089-1191
I
I
I
Abstract
The hyperbolicity property of the time periodic solutions to the Navier Stokes equations
is studied.Existence,uniqueness and analyticity of the invariant manifolds are established.

Submitted for publication in Acta Mathematica.
1 Supported
by the Office of Naval Research through the URI Research contract No:N0014-86-
5K-06T9
:46
To appear in the Proceedings of the Oberwolfach Meeting on Navier Stokes
Equations.
I
3
Invariant manifold theorems for the Navier Stokes equations
S .S.Sritharan

Los Angeles CA 90089-1191
We will present certain results on the structure of attractors for the Navier Stokes
equations. Ladyzhenskaya [4] proved that the time dependent viscous flow in two dimensional bounded domains can be characterized by a compact global attractor. This
attractor contains in particular the steady, periodic, quasi-periodic and almost periodic
3
g
solutions. Such a result for three dimensional flow would be of great interest in the theory
of turbulence. Another open problem is the task of proving the existence of global finite
dimensional invariant manifolds containing the attractor. Coordinates of such manifolds
5
p
would provide us with an effective means of computing the properties of the attractor
(which may be of fractional dimension.). In this report we will present certain preliminary results towards this subject. Let us first describe our results [8] on the hyperbolicity
of periodic solutions of the Navier Stokes equations in bounded domains. Analyticity

of the invariant manifolds is the central result. For earlier studies on hyperbolicity see
(5],(21.
Let 11 C R, n = 2 or 3 be a smooth bounded domain. We will consider the problem of
finding (u,
p) : fl x (0, oo)
-*
R" x R such that,
t,+(u.V)u=-Vp+Au
in a x (0,oo)
47
I.
*
Analysis of Regularized Navier-Stokes Equations1
I
Yuh-Roung Ou and S. S. Sritharan 2

Abstract
I
3
A practically important form of regularization of the Navier-Stokes equations is analyzed. Global existence and uniqueness is established for two classes of generalized solu-
tions. It is shown that the solution of this regularized system converges to the solution
5of
the conventional Navier-Stokes equations for low Reynolds numbers. Other results ob-
tained are the existence theory of periodic solution and the squeezing property of solutions.
Submitted for Publication in Pacific J. of Math.
I
i
1AMS subject classification number 35, 46, 76.
2This research has been supported by Faculty Research Innovation Fund of the University of
Southern California and by the Office of Naval Research through the URI Research Contract No.
N00014-86-K-0679.
48
ANALYSIS OF REGULARIZED NAVIER-STOKES
EQUATIONS -II
Yuh-Roung Ou'
Institute for Computer Applications in Science and Engineering
NASA Langley Research Center
Hampton, VA 23665.
and
S. S. Sritharan
Los Angeles, CA 9"089-1191
5
3
ABSTRACT
A practically important regularization of the Navier-Stokes equations have been analyzed. As a continuation of the previous work, we study in this paper the structure of the
attractors characterizing the solutions. Local as well as global invariant manifolds have
been found. Regularity properties of these manifolds are analyzed.
Submitted for publication in Pacific J. of Math.
NASA Contractor Report 181801
ICASE REPORT NO.
89-14
I
'This research was supported in part by the Office of Naval Research under the URI Research Contract
No. N00014-86-K-0679. Additional support was provided by National Aeronautics and Space Adminitration
under NASA Contract No. NAS1-18605 and the Air Force Office of Scientific Research under AFOSR Grant
No. 89-0079 while the first author was in residence at the Institute for Computer Application in Science and
Engineering (ICASE), NASA Langley Research Center, Hampton, VA 23665.
49
THE THEORY OF NONSTATIONARY VISCOUS FLOW

PAST PLANE DOMAINS WITH
NIONCOMPACT BOUNDARIES
gJ.
G. Heywood
Department of Mathematics
University of British Columbia
I
3
S.
S.
Szithara"
Department of Aerospace Engieering
I
I
I
ABSTRACT
In this paper we establish the existence and uniqueness of generalized solutions for time
dependent viscous flow past plane domains which have-outlets which may or may not expand
at infinity. These solutions carry the prescribed time dependent flux through the outlets.
I
*Supported by the Office of Naval Research under the
URI contract no: N00014-86-K-0679.
50
Nonlinear Stability of Subsonic Mixing Layers
With Symmetric Temperature Variations
1V.D.
by
D~de~
Department of Mechanical Engineering
University of Belgrade
Belgrade, Yugoslavia
and
t.G. Redekopp
5
3
3
ABSTRACT
The nonlinear evolution of stability waves in mixing layers of a
perfect gas with a symmetric mean temperature profile is studied
for subsonic Mach numbers in the high Reynolds number limit where
viscous and thermal diffusion effects enter first and dominate in
the critical
layer.
The
linear, neutral elgensolution of the
invlscid theory for temperature profiles having either an excess
or deficit of mean temperature in. the shear layer is used as a

basis for the weakly nonlinear, slightly viscous analysis.
The
coefficients of viscosity and thermal conductivity are assumed to

have a power law dependence on the temperature and the effect of
5
5
3
3
viscous dissipation is included.
An analytical expression for the
Landau constant, and other constants appearing in the nonlinear

evolution equation for the amplitude of the elgenmode, have been
obtained.
It is
found that the temperature excess/deficit at the
critical level and the Mach number have a strong nonlinear effect,
even to the extent of changing the sign of the Landau constant.
Accepted for publication in Phil. Trans. Roy. Soc. London.
AIAA-89-1020
STABILITY PROPERTIES OF SUBSONIC
MIXING LAYERS
V. Djordjevic, S. Pavithran and

L.G. Redekopp
Abst act
Lineaz and nonlineaz stability propertie of subsonic =iing Jay-.

ers with vziable mean tenmpezature profls are discued. Spatial
stability charactertics we pr nted as a function of the Mach mumbet and the velocity rado for muing layers with two diferent &silies of tgeMperaure proiles . TMe citical velocity rio fcr which
the absolute-con-ective character of the iwstbiity chaga at ,usIons -Mach u=bers and t=perature rat= is also conmpute
.&
analysis of the noninear. seld.intezactiou of mazinal dstbances in

layers is also sunmwaised. Th~e two mildag Iaye= &MM
two asizi
in the symmetry properies of the mea low proiles. It is fiun that
the Mach, znmber and temnperature ratio, together With the proile
synimertry, can signigcaatly change the nture of the biwrmtim froms
the inal
state- Me nature of the bt
a can chanue fArm
superriticalto subcAitcal at high subsonic Mach ausubas and the
p ofth biurcation canges from smonary to op as te poa
symmety7 is bok
de.
AIA 2nd Shear Flow Conference

March 13-16, 1989 / Tempe, AZ
E 370
For permission to copy or republish, contact the American Institute of Aeronautics and Astronautics
L'Enfant Promenade, S.W., Washington, D.C. 20024
52
Abstract for 22nd Combustion Symposium Poster Presentation

VORTEX DYNAMICS IN AN ISOTHERMAL
RECTANGULAR DUMP COMBUSTOR
J. Hertzberg and C.M. Ho

Aerospace Engineering Dept.
University of Southern California, L.A. CA 90089-1191
Due to the enhanced mixing properties of small aspect
ratio elliptic jets interest has developed recently in
asymmetric combustion chambers for ramjets. Little is currently
known about mixing and shear layer reattachment in confined
asymmetric jets. Therefore, an experimental study of isothermal
flow in a rectangular duct which undergoes a sudden expansion
is being conducted in order to understand the basic dynamics
of this type of configuration. The long range goal of this
study is to be able to control reattachment and mixing and
thus, combustion efficiency, through active or passive means.
The first phase of the work involves flow visualization
of the roll up vortices in the free shear layer and the
structure of the region where the shear layer reattaches to
the walls. To this end a rectangular ducted dump water channel
has been constructed of plexiglas. Two jets are being
investigated: a 3.8 cm square and a 3.8 x 7.6 cm rectangular
crossection. The dump section can be varied from a single sided
backward facing step to a full centered rectangular 6:1 area
increase. The jet Reynolds number can range from
2000 to
10,000.
Preliminary results have been obtained for an impulsively
started rectangular jet with a 6:1 area increase, Re = 2000.
Using hydrogen bubble flow visualization the free shear layer
at the jet exit has been observed to roll up into rectangular
vortices which quickly become distorted due to the self
induction effect of asymmetric vortex rings. Since the radius
of curvature of the ring is smallest in the corners, these
regions are convected ahead of the rest of the ring until the
ensuing distortion increases the curvature in the rest of the
ring, which catches up. The same effect causes the ring as a
whole to switch major and minor axes. By the time this occurrs
the ring has spread so that the ends of the new major axis are
now close to the wall, which causes the section of vortex ring
there to stretch and intensify. At this point the effects of
confinement dominate the dynamics of the vortex corners as the
corner sections of the ring are bent through 180 degrees by the
perpendicular sidewalls, resulting in an apparent reversal of
vorticity.
U
"This work is supported by a U.R.I. contract from the
of Naval Research.
goffice
I
53
THE t-UUii 1 "I PAQIM %#AJNM t

August 19-23. 1989
ta'a
Ur-ILUL
Hong Kong
ENTRAINMENT OF 3-D SHEAR LAYERS

C. M. Ho, T. Austin and J. Hertzberg
ABSTRACT
in engineering devices involving chemical reaction,
it is important to be able to control the mass transfer between
two streams of fluids. Ho and Gutmark (1] used a small aspect
ratio elliptic jet which
can significantly
enhance
the
entrainment. This efficient passive control technique is due to
the unsteady deformation of 3-D vortical structures caused by
self-induction.
11.Passive
and Active Control of Free Shear Layers

The possibility of controlling the mass transfer in the
transverse direction of a shear layer is important in improving
the efficiency of devices with chemical reaction.
In a
two-dimensional
flow,
the
coalescense
of
the vortical
structures in the shear layer was identified to
be the main
entrainment mechanism
(2].
If the
vortex merging is
prohibited, the growth of the mixing layer is stopped or even
decreased (3].
On the other hand, if the shear layer is
actively perturbed by the
subharmonics of
the most amplified
frequency, multiple vortices will merge simultaneously (4].
The growth of the shear layer is increased.
The spatial development of the shear layer is extremely

sensitive to the
initial
perturbations or
the boundary
conditions.
The alternative approach of controlling the
enhancement is to change the upstream boundary condition of the
shear layer. Ho and Gutmark [1,
5] found that a jet with an
elliptic nozzle can entrain much more fluid than that of a
circular or a plane jet (Fig. 1].
This passive control method
is even more advantageous for engineering applications, because
no delicate forcing arrangement is required. More importantly,
they identified a new entrainment mechanism; the self-induction
of an elliptic vortex ring pakes the structure switch its axis
orientation, the vortex el~ment near the minor axis moves
outward and makes a large amount of ambient fluid move into the
jet near the minor axis region.
This
concept can be
generalized and used in other flow configurations, such as
combustion chamber.
2.Applications in Combustion
The ramjet is a device which can be benifited by the
entrianment contol technique.
There is a short distance from
the flame holder to the exhaust nozzle. Combustion needs to be
accomplished
during
short
residence
time
and combustion
54
Iinstability
advantages of
has been a troublesome problem. The
using an elliptic jet
in the ramjet has been shown to be
phenomenal. Schadow et al. [61 used a 3:1 elliptical nozzle
in a jet with combustion.
They found that the centerline
temperature
increased
sharply a short distance from
the
At the end of the jet potential core, the
elliptic nozzle.
temperature
was much higher than that of a circular jet [Fig. 21.
instability problem was alivated by using
The combustion
nozzle.
They used a triangular
another type of asymmetric
nozzle and injected fuel near the tips of the triangle [7].
In
this way, the fuel is mixed by the small eddies near the tips
do not trigger the combustion
and the
large structures
instability.
3.Supersonic Asymmetric Jet
The spreading rate of a supersonic shear layer is much
slower than that of a subsonic flow and the combustion
efficiency of supersonic flame is hindered by the low
mixing rate.
Actually, this is the most pressing problem in
developing a hypersonic aircraft. In a preliminary experiment,
we found that the small aspect ratio rectangular nozzle could
enhance the entrainment as it did in the subsonic flow [8].
In general, the supersonic flows spread slower than that
of subsonic flows. However, the entrainment improvement of the
rectangular nozzle over the circular nozzle is about the same
for both supersonic and subsonic jets in the far downstream
locations. Near the Nozzle, the variations of the
the
shear layer
indicating
ratios,
area
cross-section
spreading, are
similar for supersonic and subsonic cases.
Hence, we expect that the supersonic rectangular jet will
entrain more mass than that of a supersonic circular jet in the
region near the nozzle.
4.Asymmetric Jet in Confinement
Most of the combustion is taking place inside a confined
space. The entrainment process is very different
from a shear
layer in the free space.
As
has been pointed out in our
previous findings [1]
the unsteady evolution of the vortex
the mixing region. We
fluids into
structures dan engulf the
used a 2:2 aspect
ratio
rectangular jet
in
a confined
environment.
The deformation of the vortical structures was
found to be much more convoluted than
those in a free jet.
These deformations are produced by the local 3-D shear layer
In other words, a
image vortex.
and the induction of the
properly designed confinement should be able to facilitate the
mixing process.
3References
3[1]
Ho, C.M. & Gutmark, E. : J. Fluid Mech. 179, 383 (1987).
55
I.
3
[2] Winant, C.D. & Browand, F.K.

J. Fluid Mech. 6,
237
(1974).
(31 Oster, D. & Wygnanski, I. : J. Fluid Mech. 123, 91 (1982).
[4] Ho, C.M. & Huang, L.S. : J. Fluid Mech. 119, 443 (1982).
Jou, W.H., Knoke, G.S. & Ho, C.M. : DOE Ejector Conf. (1988).
[5] Ho,
(1982).
C.M. &
Gutmark, E.
Bull.
[6] Schadow, K.C., Wilson, K.J.,

paper No. 84-1260, (1984).
Am. Phys.
Soc. 27,
1184
Lee, M.J. & Gutmark, E. : AIAA
[7]
Schadow, K.C.,
Gutmark, E.,
Parr, T.P.,
Wilson, K.J. : JANNAF Combustion Conf. (1986).
Parr, D.M., &
[8] Jou, W.H., Knoke, G.S.
Ejector Conf.
&
Ho,
C.M.
DOE
(1988).
I
UASSFL0W a.'P C JET v..
/-
5/,.
CENTMUNE
0
r/R
TEDPRAI"URE
OSTRISUnON
" /
I44
2400
"\
//
2-0 Jr
2200
UASFOW ELUPTC JEr v.& ORCULAR .JET

N
2000
1800-
__r_
Fig. I Mass entrainment rate
Fig. 2 Centerline temperature
II
I
56
-"
| II I
I I
Abstract of a paper presented at the International Congress

of Theoretical and Applied Mechanics, Grenoble, France,
August 22-28, 1988.
Vortex Patterns in
High Reynolds Number Shear Flow
by F. K. Browand and S. ?rost-Domasky

The turbulent region contained between tvo

unequal
speed
parallel
is the prototypal separated flow.
streams
of
Such a mixing layer
occurs in external flows separating from wing or body surfaces, and in
streams.
3turbulence
5high
flows
internal
many
or
steps
facing
backward
having
co-flowing
In the laboratory, this flow is created by merging

streams
at
the
termination
of
two
low
a splittcr plate.
The
laminar boundary layers from the splitter plate
form
the
region
of
shear; in the lateral (spanwise direction), the flow is sensihly

uniform.
As a result of the initial
Sspan.
Ideally, these vortices are perfectly

flow
tvo-dimensional
influence the flow).

flow
the
vorticity
in the
layer is gathered into vortex tubes which are aligned along the
shear
instability,
do
break
(the
which
do
in the
irregularities
in the lateral direction, and the naturally

and
evolving flow contains both skew vortices

defects
produce
walls are far removed and do not
lateral
But small, unrvoidable
symmetry
to
oriented
not conserve phase
measure of defect density is found
other.
more
A ouantitative
r '- lfral.
to
increase
singular
in the
downstream
direction until an asymptotic value is reached.

The development and maintenance of
can
structure
this
two-dimensional
be conveniently studied by application of a variety of
space/time acoustic forcings utilizihg a linear arrangement
3speakers
3with
3
located
in the
wind tunnel ceiling.
driven by computer generated pulse trains.

a
rake
of
hot
wires
selected dovnstream positions.

span
3
*
and
structure.
vortex
placed
of
small
This phased array is
Flow response is observed
across
the span of the flow at
Plotting velocity
as
function
of
time gives a useful visualization of the large scale vortex

The formation and subsequent evolution of defects will
be
discussed.
57
Ii
I
Abstract Submittal Form for the 41 st Annual Meeting
of the Division of Fluid Dyanmics of the
American Physical Society

'Vortex Pat=erns in Hich Revnolds Number Shear F'low; BROWAIND, F.K. & PROST-.DOMASKY,
Uni-r. So.
Calif. --bulent structtlre ;n

ix-q
laver is studied in terms
of the aeomezr'z of the vort.-ices wnhcn make un
the larzest scale features.
Defects i.n ortex
structure can me auantified by making simultaneous -7eloci- measurements at many =oin4s alona
the stan (the -r-nsiationallv invarianz directZon)
Vor-ex zat-erns are never defec- :ree,
S.
the develoolina
and the number c- defects is shown 7o increase

in the downstream direction until an asvmptotic
value is reached. Acoustic forcina -sapplied
utilizing a linear arrangement of small speakers 6ocated in -ice wind tunnel ceiling.
This
phased array is driven by computer-generated

pulse trains, and allows the evolution of de-
fects to be studied in
detail.
work was supported by the ONR/URI Contract No. N00014-86-K-0679.

*This
II
I
U
!
I
I
1
I
58
I
I
Abstract. Submittal Form for the 41st Annual Meeting
of the Division of Fluid Dyanmics of the

American Physical Society
On the Convective Mature of the G6rtler Instability*, PARK, 0. & HUERRE, P., Univ. So. Calif.--The
Gdrtler instability arising in the Doundary layer on a
concave plate is studied from the point of view of its
convective/absolute nature.
To avoid difficulties due to
the nonseparable character of the problem, the boundary
layer velocity field is chosen to be the asymptotic suc-
tion velocity profile of constant momentum thickness along the stream. This basic flow field is characterized
by a unique neutral curve for the Gortler instability and
the complex dispersion relation is obtained numerically
for arbitrary wavenumbers kx and kz along the streamwise
Direct calculaand spanwise directions respectively.
tions indicate that the impulse response (or Green's
function) takes the form of a wavepacket propagating
downstream. The packet remains localized in the streamwise direction but its "wing tips" diffuse rapidly in the
spanwise direction. One can conclude that the GFrtler
G~rtler vortices observed in
instability is convective:
practice are likely to be very sensitive to upstream
perturbations: they are the result of the amplification
of external fluctuations.
*Supported by ONR under QNR/UR! N0O0i4-86-K-0679.
Models of Pattern Formation and Dislocations in

Free Shear Flown, YANG, R. & HUERRE, P., Univ. So. Calif.
& COULLET, P., Observatoire de Nice.--The three-dimensional evolution of Kelvin-Helmholtz vortices in free
shear flows is modelled a-la Swift-Hohenberg by a twodimensional equation for the amplitude of the vortices.
As a result of Galilean invariance, deformations of the
structures are coupled to a large scale vertical vorticity field.
Numerical and analytical studies of the
model indicate that collective oscillatory notions may
arise when an initially two-dimensional array of vortices
is perturbed in the skew-varicose mode of instability.
In such collective oscillations, the restoring force is
provided by the vertical vorticity field.
Spatial inhomogeneities in the local wavenuber can also induce
pattern dislocations which are qualitatively similar to
those observed in free shear layers by F.K. Browand.
Lines of constant phase meet at the center of a defect
and the amplitude of the perturbations decays to zero.
Interactions between several of these phase singularities
are studied numerically.
~ONR/URI
I
U
U.S.NOOO14-B6-K-0679.
Office of Naval Research under
by the No.
Supported
Contract
59
Abstract for Nov. 1987

A.P.S. Meeting Presentation
Modification of Turbulent Boundary Layers by
Longitudinal Roughness Elements.* JASON B.ROON,
RON F. BLACKWELDER, Univ. So. Calif. and
MOHAMED GAD-EL-HAK, Univ. Notre Dame-Velocity
profiles and bursting statistics were measured
in the presence of Longitudinal Roughness Elements (LRE's) and compared to similar work by
Johansen and Smith (1986).
An algorithm was
devised to count low speed streaks (LSS) and
measure their length from the hot-wire rake
data, enabling determination of the LRE's ability to minimize the LSS's spatial randomness.
The velocity and turbulence intensity profiles
agree with Johansen and Smith's result that the
effect of the LRE's are felt
only out toy+ 1Q.
However the analysis of the bursting statistics
and LSS algorithm output suggest that while the
LRE's might serve as a nucleation site for LSS
formation, their ability to lock the fully
formed LSS's onto the LRE's is minimal and
hence spatial randomness was not eliminated.
* Supported by NASA under Contract
NASI-18292
and ONR-URIP N00014-86-K-0679.
Abstract for Nov. 1988 A.P.S. meeting presentation

VORTEX
DYNAMICS
IN
RECTANGULAR
SUDDEN
EXPANSION J. Hertzberg and C.M. Ho, University

An experimental study of a 2:1
of So. Cal.-aspect ratio rectangular jet which undergoes a
6:1 sudden expansion is being conducted in order
to understand entrainment and mixing in this type
of configuration. Preliminary results have been
flow using
obtained for impulsively started
hydrogen bubble flow visualization. The effect of
self induction and confinement on the rectangular
starting vortex ring greatly enhance the unsteady
distortion of the ring. Hence, large entrainment
Further hydrogen bubble
ratios are expected.
visualization in
continuous
quasi-instantaneous velocity
3
3
flow have revealed
profiles
in the
recirculation and reattachment regions. Laser

doppler anemometry studies are also in progress
to provide quantitative information on the flow

structure.
This work issupported by an O.N.R.-
O.R.Z.
60
I.
SSOCIETY
I
SUITE 1400
for INDUSTRIAL and APPLIED MATHEMATICS
117 SOUTH 17TH STREET
PHILAOELPHIA. PA 190
(215) 5"1292n
Regularity of Inertial Manifolds for Semi linear

Evolution Equations
Yuh-Roung Ou
ICASE
NASA Langley Research Center
23665
Hampton VA
and
S.S. Sritharan
90089-ii91
Los Angeles CA
I
J.G. Heywood
Theory of Nonstationary Viscous Flow Past Plane

Domains With Noncompact Boundaries
Department of Mathematics
aUncrsity of British Columbia
BC
Existence theory of nonstationary viscous flow

past plane domains has been an open problem for
ICanada
and
Soutlets
S.S. Sritharan
SUniversity of Southern California
Los Angeles CA
9OO89-9l
*years
5Optimal
EngTneering
Unersty of SouernaceiEnia,
Uos n
les
C
aliforia.
Los Angeles
CA
the past decade.

In this talk we present the
existence and uniqueness of generalized solutions
for time dependent flow past domains which have
which may or may not expand at -infinity.
These solutions carry the prescribed time dependent flux through the outlets. The corresponding
three dimensional problem is relatively simpler
and has already been resolved due to the works of
Heywood and of Ladyzhenskaya and Solonnikov a few
ago.
Control Theory of Navier Stokes Equations
Sritharan
SS.S.
Department of-Aerospace
It has been realized recently by several authors

that the solutions of certain class of evolution
equations can be characterized by exponentially
attracting Lipschitz global manifolds. These
manifolds contain the global attractor. Coordinatization of these manifolds will provide us with
an exact Galerkin method for the solution of the
evolution system. These manifolds are modelled
on the spans of the eigenspaces of certain self
In this talk we will discuss
adjoint operator.
the regularity properties/of these manifolds. We
will show in particular that the higher, dimensional manifolds are automatically C i .
90089-191.
In this talk we will present certain results on

an application of distributed systems theory to
a hydrodynamic problem. Existence of optimal
solution that minimizes a.given cost functional
is established. As a special case we then conIn
sider the linearized Navier Stokes equation.
this case the original system, the adjoint system
and the feedback equation are combined to give
the Riccati partial differential equation for
the computation of the feedback map.
6
61
Iae
3"Navier
Stokes Equations"
Mathematisches Forschunstitut OBERWOLFACH
Sept. 18-24, 1988
S. SRITHARAN:
Invariant manifold theor-ms for :he 4avier-Stokes
equations
Invariant manifold theory lays a bridge between the onset dynamics of
"urbulence and the theory of finite dimensional dynamical systems. In
this talk we will describe the existenceuniqueness and analyticity of
local invariant manifolds along with the characterizat4on of the nonlinear hydrodynamic semigroup and the monodromy operator. These mnifolds are constructed in the neighborhood of each periodic solution to
the Navier Stokes equations establishing the hyperbolicity of these
solutions. For a particular regularization of the Navier Stokes equations
(for which we have global existence and uniqueness theorem) we are able
to prove the existence of global invariant (also knom as the inertial)
manifolds. This result is significant since the (generalized) solution
of the regularized system has a strong limit (as the regularization
parameter approaches zero) to the weak solution of the Xavier Stokes
equations.
U
I
I
SIAM Annual Meeting
July 11-15, 1988

University of Minnesota
*#.56/4:45PM
Invariant Manifold Theorems For the Xavier Stokes

S.qua tions
We have established the hyperbolicity of
stationary and time periodic solutions of the
Xavier Stokes equations with analytic stable and
unstable manifolds. Moreover, the spectral
properties of the monodromy operator (the frechet
derivative of the nonlinear semigroup) have also
been studied. For a regularized version of the
Xavier Stokes equations, both local and global
invariant manifold theorems have been established.
~University
S.S. Sritharan
Department of
Engineering
of Aerospace
Southern California
Los Anigeles, CA
90089-1191
62

Ok 1989 Main Ref P 55 Entrainment MechanismA693012

Uploaded by

Copyright:

Available Formats

Ok 1989 Main Ref P 55 Entrainment MechanismA693012

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Ok 1989 Main Ref P 55 Entrainment MechanismA693012

Uploaded by

Copyright:

Available Formats

DYNAMICS AND CONTROL

TURBULENT SHEAR FLOWS

Report of Research Activities

During the Period

OFFICE OF NAVAL RESEARCH

UNIVERSITY RESEARCH INITIATIVE PROGRAM

DEPARTMENT OF AEROSPACE ENGINEERING

Project Director: L. G. Redekopp

Assistant Director: R. F. Blackwelder

Contract Monitor: M. Reischman

DYNAMICS AND CONTROL

engineering performance measures.

different sub-projects led by individual, or teams of, co-investigators. There is a considerable

activity on this topic is included in the Research Reports section.

Prost-Domasky), plus support for several undergraduate laboratory assistants. Instiand S.

Ph.D. student (Liu Wei).

pointment of two post-doctoral researchers, slowed the development of several experimental

INSTRUMENTATION AND FACILITIES

on future budgets. In addition to these major instrument purchases, project funds

Turbomachinery Hydroacoustics" in May 1988.

Principal Investigator: R. F. Blackwelder

List of Publications /Reports/Presentations

Ron F. Blackwelder, Some Ideas on the Control of Near-Wall Eddies, AIAA-89-1009.

Ron F. Blackwelder, "Some Ideas on the Control

J. B. Roon, Ph.D. student

Pray, M.S. student

CONTROL OF LAMINAR SEPARATED SHEAR LAYERS ON AIRFOILS

CONTROL OF THE LARGE-SCALE FEATURES IN TURBULENT MIXING

Principal Investigator: F. K. Browand

Description of Scientific Research Goals

Vortex pattern imperfections, which appear similar to dislocations in a regular crystal

F. K. Browand, "Pattern Evolution in Mixing Layers," International Congress of

DIRECT NUMERICAL SIMULATIONS OF FLOW CONTROL

Description of Scientific Research Goals

CONTRuL OF ENTRAINMENT IN CONFINED SHEAR LAYERS

(ii) Flow-Field Characteristics

of Publications /Reports /Presentations

J.R. Hertzberg and C.-M. Ho "Vortex Dynamics in an Isothermal Dump Combustor".

Hertzberg, Research Associate

Chris Niestepski, Undergraduate Assistant

THEORETICAL UNDERPINNINGS: AMPLITUDE AND PHASE DESCRIPTIONS OF COHERENT STRUCTURES

Progress has been made in 4 distinct areas as summarized below.

Plans for Next Year's Research

Preliminary analytical investigations of the motion of dislocations have been undertaken.

Figure 1. Nucleation of dislocations

Figure 2. Nucleation of dislocations

Huerre, P. (Sept. 11, 1987). "Dynamique des Structures Coherentes et Ondes

D. Park, Ph.D. student

R. Yang, Ph.D. student (not supported by ONR-URI contract)

PHASE CONTROL OF VORTEX STRUCTURES IN SHEAR FLOWS

Description of Scientific Research Goals

J. Wiggert, Ph.D. student

MATHEMATICAL ANALYSIS OF THE NAVIER STOKES EQUATIONS

Plans for Next Year's Research

1.Papers Published in Refereed Journals

shows tht the flow relaxes bark to a fa

crude first approximation is that the UM

very close to the REs.