11th LDA Symposium
Analysis of oscillating shear layer
C.E.C. Cala, E.C. Fernandes and M.V. Heitor
Centro de Estudos em Inovação, Tecnologia e Políticas de Desenvolvimento, IN+
Dept. Engenharia Mecânica
Instituto Superior Técnico
Av. Rovisco Pais, 1049-001 Lisboa Codex
PORTUGAL
Abstract
This work reports the experimental study conducted on a curved oscillatory boundary
layer excited by a single frequency artificial signals. Physical analysis of turbulent
shear layer morphology was based on instantaneous sequences of shadowgraph images
and turbulent velocity characteristics by Laser Doppler Velocimetry measurements. The
data is being interpreted on a phase locked base to allow the characterization of the
unsteady inner and outer shear layers, that features to strong temporal and spatial
deformations imposed by external induced oscillations. The analysis of vortex pairing
dynamics allowed to conclude that the pairing process in this flow corresponds to the jet
column mode, as suggested for circular jet flow. The consequences of this flow
behavior to the combustion efficiency and pollutant reduction are discussed.
1. Introduction
Combustion under sufficiently fuel- lean conditions can have desirable attributes of high
efficiency and low emissions. However, as lean limit is approached acoustic or nonacoustic instabilities may arise giving rise to a premature blow-out. It is known that this
process is driven by fuel-air mixing characteristics, momentum oscillations of the
injected flow or shear layer hydrodynamic instabilities, particularly due to high
sensitivity of lean flames to these boundary conditions and also to the low flame speed.
All of these combine to narrower the stable operating windows.
The consequences of combustion instabilities in combustion chambers are often
undesirable due to high-pressure fluctuations (e.g. Sivasegaram et al., 1989) and
increased heat transfer to the wall (Azevedo et al., 1994), which may cause performance
degradation and structural damage (e.g. Bahr, 1996). In fact, the increased intens ity of
combustion at any given point, together with short and spatially localised flames, may
raise the likelihood of coupling with acoustic waves, in the context of the Rayleigh
criterium, thus causing combustor dynamic problems. It is clear that the amount of
potential damage is dependent on the amplitude and duration of the oscillation and on
the coupling process between the flame/flow and cavity transfers function. This has led
too much of the recent work on the control of combustion instabilities towards their
attenuation (e.g. Kemal and Bowman, 1996). In such cases an active control strategy is
most promising compared to traditional methods of achieving a desired improvement,
but is limited by the sensor/actuators technology available today (see for example the
report of Schadow et al., 1997) and by the sparse information covering the behavior of
bluff body stabilized flames, under resonance conditions, to different amplitude and
frequency of oscillations. In general it is known that in bluff body stabilized flames,
there are present two type of oscillations; weak oscillations associated with outer vortex
shedding process that may carry reaction if flame crosses the peck of maximum velocity
11th LDA Symposium
and strong oscillations associated with large scale deformation of the recirculation zone,
as implied by the presence of strong acoustic pressure gradient (see Fernandes 1998).
The reminding question is how flame is affected in the inner shear layer and contributes
to the unsteadiness support.
Due to the complexity associated with the unsteady reactive flows, and in order to better
understand the contribution of the shear layer and vortex unsteadiness to the flame
stabilization process, the study is carried out on an non-confined and non-reacting
turbulent lean premixed flame stabilized by central recirculation zone, induced by a
bluff-body, burning a mixture of C3 H8 and air and artificially actuated by acoustic
signals.
In this context, this work addresses the impact of upstream momentum oscillations, with
different amplitude and frequencies, on flow field downstream of the buff-body. The
experimental analysis conducted here is based on qualitative measurements undertaken
by shadowgraph techniques and quantitative measurements of local phase- locked
velocity obtained by combining an LDV with a microphone. The data is being
interpreted on a phase locked base to allow the characterization of the unsteady reacting
shear layer features to strong temporal and spatial deformations imposed by external
induced oscillations.
2. Experimental Procedures
2.1. Experimental Setup and Experimental Techniques
The set-up used throughout this work is that shown in Figure 1. The burner consists of a
nozzle with 80mm exit diameter. A 56mm diameter bluff-body was centrally placed in
order to stabilize the flame.
Figure 1. Experimental set-up
The burner was placed on a milling machine table in such a way that the profiles were
obtained by burner displacement. The arrangement allows three orthogonal movements,
(x, y and z direction) with an error less than 0.5mm.
A blower- motor assembly through a plenum with 860x740x500mm, in order to reduce
the pressure fluctuation, feed the airflow. A plate orifice connected to a “U” manometer
made the control. The propane was supplied through two rotameters allowing a
maximum flow rate of about 4.2g/s at PTN, with an error of 0.05g/s. The gases are
11th LDA Symposium
passed through screens and honeycomb to prevent swirling and breakdown the large
vortical structures in order to improve the mixing process.
The velocity field was measured with a multiline Laser Doppler Velocimetry (LDV)
system operating in the forward scatter mode. The coherent light from an Argon-Ion
laser with 2W maximum power at λ=514.5nm (green laser beam) was focused through
an optical system. The beam is splited in two with an approximately equal power; one
of them could be shifted continuously up to 16MHz by a rotating grate. The optical
system (splitter and rotating grate) is that used by Ferrão, (1993). This system was
configured for a receiving angle of 8.272º. The dimension of the measuring volume,
resulting of beams interception, is 44 x 44 x 606µm with 3.57µm fringes spacing. A
lens of 310mm focal length collects this control volume and focuses it in the
photomultiplier, which transforms it into a sinusoidal electric signal.
Three loudspeakers, Uni-Power model s-50, with 16, introduced the controlled
excitationΩ and 50W, attached near the exit of the jet, shifted geometrically 120º. The
driven input was sinusoidal signals from an audio-amplifier Rotel KB 890 feeded by a
Philips PM5131 generator with a working range of 0.1Hz up to 2MHz.
The flow motion was recorded with a high-speed camera, (Kodak Motion Corder
Analyzer, Model SR-Ultra) at 5000 frames per second with a freezing time of 1/10000s.
2.2. Data Processing
The understanding of the turbulent shear layer behavior is very important for the
development of many technological processes and practical devices. An accurate
description of the shear layer at a given location is for that very important.
This problem can be approached, in an experimental manner, by introducing controlled
disturbance within the fluid at some point, and then to study the behavior of this
disturbance downstream (Hussain and Reynolds, 1970). In the presence of such kind of
flow, with coherent structures, the instantaneous values, according Hussain and
Reynolds (1970) are defined to be composed by three terms:
~
U i ( t ) = U i + U i ( t ) + u ′i ( t )
(3.1)
~
Were U i (t ) , is the instantaneous value, U i is the long-time average mean, U i (t ) is the
periodic compone nt (the contribution of the organized waves, the coherent structures)
and u ′i (t ) is the turbulent fluctuations.
11th LDA Symposium
U(t)
u′
~
U
U
Figure 2 Time and phase averages of a random signal
Figure 2. Illustrates a typical acquisition scheme implemented for conditional sampling,
in order to extract the organized wave from the random turbulence (upper signal). The
lower is the reference signal from external source, used as clock to start the acquisition
process.
Conditional samplings were performed on the Data Translation-Fulcrum board; model
DT3009, witch combines a TMS320C40 DSP (Digital Signal Processing) from Texas
Instruments (for details see Fernandes and Heitor, 1998). Signals, velocity fluctuations
and pressure (reference signal) were recovered at 600Hz, the believed fundamental
response of the flow. Each record for phase averaging was composed by a set of 20 time
histories. With the same board an average of 60 Fast Fourier Transform for velocity was
performed.
3. Results and Discution
3.1. Mean Field
These flows were excited by single frequency artificial acoustic signals. Physical
analysis of turbulent oscillating shear layer, as shown in the shadowgraph images
presented in Figure 3, was obtained by providing the characterization of local
morphology of the flame front and shear layer characteristics for different amplitude of
oscillations in order to assess the non- linear response of the reacting shear layer.
11th LDA Symposium
z/R
Unforced flow
Forced flow
Vortex
2R
L
r/R
U0
Flame front
Figure 3. Experimental Conditions: St L=0.52, L=7mm, φ=0.6, BR=64%, Reacting
Flow
Under excitation the flame front is wrinkled, due its interaction with the induced
vortices, as can de seen in Figure 3. This reduces the local flow velocity and increases
the heat exchange between the hot products and cold mixture, which prevents a
premature extinction, widerrig the stable operating window. This interaction between
flame front (inner shear layer) and unburned mixture (outer shear layer) increases the
mixing between the hot products and the cold reactants. The vortex rolling up of the
outer shear layer engulfs the cold reactants witch are heated and burned during the shear
layers interaction. It is expected that this interaction can improve the combustion
efficiency and reduce the pollutant emissions, reducing the amount of unburned gases.
As suggested by Haile et al (1998), the visible length of the flame is shortened witch
can reduce the life time of burned gases in regions of high temperature, reducing then
the NO thermal formation. This shortness of the flow, under excitation, should be seen
in the Figure 4, where it is shown the mean axial velocity evolution along the central
line. The applied excitation corresponds to a Strouhal number of 0.52, based on the flow
thickness (L=7mm).
2
1.6
z/R
1.2
StL = 0
StL = 0.52
0.8
0.4
0
-4
-2
0
U, [m/s]
2
4
Figure 4. Mean axial velocity along the central line with and without oscillation;
U=8m/s, BR=64%
11th LDA Symposium
As suggested by Fernandes (1998) the combustion process can trigger flow instabilities
forcing the shear layer to roll- up. In order to understand the behavior of the shear layer
under external excitation and to ga in some control over the dynamics of the shear layer
development, isothermal flow is analyzed. Only the reactants were present in the same
amount needed to perform the reacting process, this way the instabilities from
combustion are isolated. Of course, the reactive flow (combustion flow) is much more
complicated than the non-reactive flow (isothermal flow) witch is presented in this
work, however, an understanding of simpler flows is a prerequisite to dealing with
complex flows.
Figure 5 a) and b) show high-speed motion pictures of non-forced and forced flow
respectively. From Figure 5 b) can be seen the enhancement of the vortex rolling- up
and the pairing process. Zaman and Hussain (1980) suggested that the occurrence of this
large-scale structures in turbulent shear flow, characterized by coherent vortical fluid,
with length scales of the order of the shear flow width play a dominant role in the
entrainment and mixing. An important feature of the dynamics of these structures is the
pairing. A better understanding of the behavior of these structures should be given by
quantitative measurements. The region marked with a circle, in Figure 3, will be
analyzed in detail.
0.4
z/R
a)
0.3
0.2
0.1
b)
z/R
0.4
0.3
0.2
0.1
1
1.2 1.4
r/R
1
1.2 1.4
r/R
1
1.2
r/R
1.4
1
1.2 1.4
r/R
1
1.2
r/R
1.4
1
1.2 1.4
r/R
Figure 5. Flow sequence, outer shear layer, acquired at 5000Fps with an exposition time
of 1/10000s. Isothermal flow φ=0.6, U=8m/s
a)Unforced flow
b)Forced flow at 600Hz
In Figure 6 it is shown the mean velocity, mean vorticity and spectra for inner and outer
shear layers. A gradual process of energy transfer from the forcing frequency
(fundamental) to the first subharmonic is shown. This frequency–halving according to
Zaman and Hussain (1980) revels that the pairing process is going on. The analysis of
the unsteady field (next) will show that this process starts approximately at z/R=0.25.
The spectra analysis doesn’t show any future pairing down stream, as suggested by
Zaman and Hussain (1980) this should be due the limited length before the end of
potential core. This suggests that the activity of pairing in inner shear layer corresponds
to the jet column mode and, in this particular issue; this flow behaves as a jet. Up stream
of the plane z/R=0.5, another energy distribution at higher frequencies is identified at
3f/2. This distribution should indicate the motion of small scales structures or the nonlinear response of the flow to the forcing signal. According to Husain and Hussain
(1982), Zaman and Hussain (1980) also found peaks of energy at 3f/2 for axisymmetric
jet flow under controlled excitation, in previous work. The energy spectra reveal a hump
11th LDA Symposium
at 30Hz, which should revel the bubble oscillation, as suggested by spectra of nonforced flow.
10
100
Frequency, [Hz]
1000
10
1
0
100
Freq uency, [Hz]
10
1000
100
frequency, [Hz]
1000
0.6
10
100
f requ ency, [Hz]
1000
z/R
0.5
10
100
Frequency, [Hz]
0.4
0.3
1000
0.2
10
0.8
1
1.2
100
frequency, [Hz]
1.4
10m/s
r/R
1000
Energy (arbitrary sca le)
100
Frequen cy, [Hz]
1.0E+010
1.0E+009
1.0E+008
10
100
Frequency, [Hz]
1000
Energy, [arbitrary sa cale]
Energy, [arbit rary scale]
10
1.0E+007
1.0E+005
1.0E+003
10
100
frequenc y, [Hz]
1000
1000
1.0E+007
1.0E+005
1.0E+003
10
100
frequency, [Hz]
1000
Vort.
3.1
2.5
1.8
1.2
0.5
-0.2
-0.8
-1.5
-2.2
-2.8
Figure 6. Mean velocity, vorticity and energy spectra for inner and outer shear layer;
U=8m/s, BR=64%
Since there are spatial and temporal variations in formation, shape, size, orientation,
convection velocity, interaction and breakdown of large scales structures, as suggested
by Zaman and Hussain (1980), the mean flow alone doesn’t reveal clearly their
occurrence and dynamical role. There, for a deeper understanding, conditional
measurements of velocity were taken without combustion.
3.2. Unsteady Velocity field
In Figure 7 time history velocity and vorticiy field, for the reactants alone, with no
flame present, is shown. It is seen that the vectors near the bottom and along the shear
layers are unsteady which can improve the mixing process. There is a reasonable
velocity gradient in radial direction, which is favorable for vortices growing.
11th LDA Symposium
1
0.6
2
3
z/R
0.5
0.4
0.3
0.2
5
4
0.6
6
10m/s
z/R
0.5
0.4
Vort.
3.1
2.5
1.8
1.2
0.5
-0.2
-0.8
-1.5
-2.2
-2.8
0.3
0.2
0.6
7
8
9
10
11
12
z/R
0.5
0.4
0.3
0.2
0.6
z/R
0.5
0.4
0.3
0.2
0.8
1
r/R
1.2
1.4
0.8
1
r/R
1.2
1.4
0.8
1
r/R
1.2
1.4
Figure 7. Two-dimensional representation of phase averaged velocity vectors and phase averaged
vorticity, over half- cycle of oscillation; U=8m/s, BR=64%.
The contour map reveals two regions of high vorticity. The inner region (internal shear
layer) with positive vorticity witch indicates counterclockwise rotating vortices and the
outer region (external shear layer) with negative vorticity, the clockwise rotating
vortices. The high vorticity at the bottom (Time 1) indicates the vortices shedding
points, from where they are convected down stream. For the inner shear layer; with the
center of the vortex at about z/R=0.25 (Time 4) the vorticity starts to decrease witch
indicates the occurrence of pairing. From this picture should be expeculated that the
pairing process at Time 6 is completed and the merged vortexes are convected down
stream as one new identity and, as suggested by Figure 6, no more pairing occurs. At
sometime a new vortex is starting to grow up in this shear layer. On other hand in the
outer shear layer the vortex is stronger and has longer lifetime. Taking again Time 1 as
starting point it is seen that the vortex grow from this time until more less Time 5 where
the pairing process starts and go on until Time 8. In fact, from the spectra, Figure 6, it is
evident that the merging process in the outer shear layer occurs while the vortices are
convected downstream but in the inner shear layer the merged vortex are only
convected after the end of the pairing process. Video images of vorticity field confirm
this statement. But, in reality, the inner vortex has other dynamical behavior. What is
seen from Figure 6 should be due the recovering process selected. In fact all signal were
recovered at 600Hz, but since there is a pairing process the frequency is halved at the
merging region and, the occurrences starts to happen two times slowly. So the
recovering process doesn’t have enough ability to seen what is going on with a halffrequency. To overtake this malfunction of the implemented process of recovering the
signal it is suggested a two folded recovering, at the fundamental frequency and at its
subharmonic. With combustion, as suggested by Driscoll et al (1994), it is expected that
11th LDA Symposium
the flame doesn’t create additional velocity gradient inducing vorticity to the reactant
side, the outer shear layer. So this pattern should be maintained.
In Figure 8 it is shown the effect of forcing on the turbulence intensity down stream. It
is seen that forcing, in the inner shear layer, decreases turbulence intensity. As
suggested by Petersen and Long (1992) it should be due the ability of the large scales to
organize the smallest scales. The same authors suggested that this scale organizing
process has direct implication for active control of mixing process.
Inner Shear Layer
III
0.8
II
0.6
0.4
I
0.2
0
Potencial Flow
1
0.1 0.2 0.3 0.4
((u'+Uosc)2)0.5/U
Outer Shear layer
StL=0.52
StL=0
1
II
0.8
0.6
0.4
I
0.2
0
0
1.2
II
0.8
z/R
1
z/R
1.2
z/R
1.2
0.6
0.4
I
0.2
0
0
0.1 0.2 0.3 0.4
((u'+Uosc)2)0.5/U
0
0.1 0.2 0.3 0.4
((u'+U osc)2)0.5/U
Figure 8. Forced flow versus unforced flow
Working Conditions: φ=0.6, U=8m/s, BR64%
Now combining the information contained in Figures 6, 7 and 8 it is possible to define
three distinct regions; region I where the vortex is formed, region II the pairing region
and the region III a full turbulent region with the merging process completed. It is
expected that in region II, during the merging process the mixing will be increased.
4. Conclusions
In this work velocity measurements in the wake of a bluff-body, simulating a flame with
a mean velocity of 8m/s and an equivalence ratio of 0.6, are reported for non-reactive
flow forced at 600Hz, corresponding to a Strouhal number based on flow thickness of
0.52. This excitation lies in the range where, according to Cho et al. (1998), the vortex
pairing can be controlled by means of the fundamental and its subharmonic forcing. The
pairing corresponds to the jet column mode, observed by Zamam and Hussain (1980) in
a circular jet flow. The data was interpreted on a phase locked base to allow the
characterization of the unsteady inner and outer shear layers features to strong temporal
and spatial deformations imposed by this external induced oscillations. A Fast Fourier
Transform procedure was implemented to characterize the pairing process along the
inner and the outer shear layers. With forcing was observed; that the height of the
recirculation zone decreases, reduction in the turbulence intensity along the axial
direction, reorganizing the small scales.
References
Azevedo, L.F.A.; Webb, B.W. and Queiroz, M. (1994). “Pulsed Air Jet Impingement Heat
Transfer”. Experimental Thermal and Fluid Science, 8, pp. 206-213.
Kemal, A. and Bowman, C.T. (1996). "Real-time Adaptive Feedback Control of Combustion
Instability". 26th Symp. (Int’l.) on Combustion, The Combustion Institute, pp. 1-14
Schadow, K., Yang, V., Culick, F., Rosfjord, T., Sturgess, G. and Zinn, B. (1997). “Active
Combustion Control for Propulsion Systems”, AGARD-R-820, pp.1- 38.
11th LDA Symposium
Sivasegaram, S.; Thompson, B.E. and Whitelaw, J.H. (1989). "Acoustic Characterization
Relevant to Gas Turbine Augmentors". J. Propulsion, 5(1), Jan.-Fev., pp. 109-115.
Hussain, A.K.M.F.; and Reynolds, W.C. (1970). "The Mechanics of an organized Wave in
Turbulent Shear Flow". J. Fluid Mechanics, 41(1), part 2., pp. 241-258.
Reynolds, W.C. and Hussain, A.K.M.F.; (1972). "The Mechanics of an organized Wave in
Turbulent Shear Flow. Part 3. Theoretical models and comparisons with experiments". J.
Fluid Mechanics, 54, part 2., pp. 263-288.
Ferrão, P.M.C.; (1993). "Análise Experimental de Chamas Turbulentas com recirculação".
PhD Tesis, in Portuguese.
E.C. Fernandes; (1998). "The Onset of Combustion-Driven Acoustic Oscillations ". PhD
Tesis.
E. Haile, O. Delabroy, D. Durox, F. Lacas and S. Candel; (1998). "Combustion
Enhancement by Active Control ". Springer, in fluid control.
J.F. Driscoll, D.J. Sutkus, W.L. Roberts, M.E. Post and L.P. Goss; (1994). "The
Strain Exerced by a Vortex on a Flame – Determined from Velocity Field Images ".
Combust. Sci and Thech, vol 96, pp. 213-229.
R.A. Petersen and T. A. Long; (1992). "Controled interations in a forced axisymmetric jet.
Part 2. The modulation of broadband turbulence ". J. Fluid Mech. (1992) vol 235, pp. 57-72.
Zaman, K.B.M. and Hussain, A.K.M.F. (1980). "Vortex pairing in a circular jet under
controlled excitation. Part 1. General jet response". J. Fluid Mechanics, vol. 101, part 3., pp.
449-491.
Hussain, A.K.M.F. and Zaman, K.B.M. (1980). "Vortex pairing in a circular jet under
controlled excitation. Part 2. Coherent strutures". J. Fluid Mechanics, vol. 101, part 3., pp. 493544.
Pan, J.C., Vangsness, M.D., Heneghan, S,P. and Ballal, D.R. (1991). "Scalar
Measurements in bluff-bodies Stabilized Flames using Cars Diagnostics". The American
Society of Mechanical Engineers (1991), 91-GT-302.
Cho, S.K., Yoo, J.Y. and Choi, H. (1998). "Vortex pairing in an axisymmetric jet using twofrequency acoustic forcing at low to moderate strouhal numbers". Experiments in Fluids 25, pp.
305-315.
Fernandes, E.C. and Heitor M.V. (1998). "On the Extension of a Laser-Doppler Velocimeter
to the Analysis of Oscillating Flames". Laser Techiques Applie d to Fluid Mechanics, 9th
International Symposium, pp. 385-399.