Escolar Documentos
Profissional Documentos
Cultura Documentos
University of Waterloo, Department of Mechanical Engineering, Waterloo, Ontario, Canada, N2L 3G1
An experimental investigation of the interaction of an annular air jet with a methanol spray flame was
conducted using phase Doppler interferometry (PDI) to measure fuel droplet size and velocity as well as
gas-phase velocity, and planar laser-induced fluorescence (PLIF) to determine instantaneous and timeaveraged reaction zone location, as well as quantitative OH concentration. Temperature measurements were
made using Pt/Pt-10% Rh thermocouples. PLIF imaging of the OH radical showed that the spray flame had a
dual reaction zone structure with annular air off but a single reaction zone with annular air on. Peak OH
concentrations were measured to be 5400 PPM, regardless of the annular air flow rate. The OH concentration
peaks always occurred on the fuel-lean side of the reaction zone, while temperature peaks occurred on the
fuel-rich side. Gas-phase velocity vector fields show that annular air tends to channel the flow towards the
centreline, and large droplet velocity vector fields show that the large droplets follow this trend as well,
suggesting that the annular air jet assists in confining large spray droplets. Overall visible flame length is
reduced by over 50% with annular air, providing a flame well-suited to compact combustion chambers. 1999
by The Combustion Institute
NOMENCLATURE
Roman
A 5 Area viewed by collection
optics, (cm2)
A11, A10 , A01, A00 5 Einstein coefficients for
spontaneous
emission/
absorption (1/s)
Ap 5 Area of laser sheet imaged by a pixel (cm2)
B10 5 Einstein coefficient for
stimulated
absorption
(cm3/J-s2)
Bv 5 Rotational
constant
(cm21)
c 5 Speed of light (3 3 1010
cm/s)
D 5 Droplet diameter (mm)
D10 5 Arithmetic mean diameter (mm)
D32 5 Sauter mean diameter
(mm)
E 5 Laser pulse energy (J)
E(n) 5 Laser energy lineshape
function (J/Hz)
Ei 5 Incident laser energy (J)
*Corresponding author. E-mail: metin@sunwise.uwaterloo.ca
COMBUSTION AND FLAME 117:661 684 (1999)
1999 by The Combustion Institute
Published by Elsevier Science Inc.
662
663
TABLE 1
PDI System Optical Parameters Used in the Present
Work
Laser
Argon ion laser
Transmitting optics
Beam expanding telescope ratio
Beam separation
Transmitting lens focal length
Probe volume waist (1/e2)
Receiving optics
Receiving lens focal length
Lens diameter
Spatial filter slit width
Collimating lens focal length
Magnification
Separation between detectors
A, B, and C
Orientation
supplied to the atomizer nozzle from a pressurized holding tank at approximately 6 bar, and
metered through a calibrated rotameter. The
entire burner assembly is mounted in the vertically up-firing position on a manual three-axis
traversing system. The burner was operated at a
fixed fuel flow rate of 0.42 g/s, with annular air
flow rates of 0, 4.77, and 9.52 l/s, corresponding
to Reynolds numbers (based on the bluff body
diameter of 50 mm) of 0, 13,000, and 26,000,
respectively.
Phase-Doppler Interferometry System
PDI allows simultaneous determination of
droplet size and velocity by measuring the phase
difference and frequency of light scattered when
a droplet passes through the intersection of two
laser beams (the Doppler burst). In the basic
PDI arrangement, the Doppler bursts are ob-
500750 mW
2:1
40 mm
250 mm
117 mm
300 mm
72 mm (f/4.2)
150 mm
250 mm
0.833
16.32 mm (A-B)
49.96 mm (A-C)
30 off forward axis
served by two photodetectors arranged at different points in space. The two detectors observe the same Doppler burst but with a phase
difference due to their spatial separation. This
phase difference is directly proportional to the
droplet diameter, and the frequency of the
Doppler burst is directly proportional to the
droplet velocity. The theory of operation of the
phase-Doppler system is well established (e.g.,
[6 8]) and therefore will not be repeated here.
The PDI system used in this work is an
Aerometrics DSA fibre-optics based system
with two velocity component capability. The
system is operated in first order refraction
mode, with the receiver located at a 30 off-axis
angle. A Coherent Innova 90 water-cooled argon ion laser is used, operating at approximately
600 mW output. System optical parameters are
summarized in Table 1.
Planar Laser-Induced Fluorescence System
PLIF allows non-intrusive instantaneous 2-D
imaging of species concentration. According to
the principles of quantum mechanics, a molecule can be raised to an excited state by interaction with a photon of a frequency corresponding to the energy difference between the initial
lower energy state and the desired excited state.
The molecule can then relax either by interaction with neighbouring species (quenching) or
by the emission of a photon corresponding to
664
S DF E
B 10
c
z ~ At!.
1
h Lt
E~ n ! g~ n ! d n ~ f n f Jn 0!
(1)
DF
n hc v n
f n~T! 5 exp 2
kT
f J~T! 5
DG
hc v n
1 2 exp 2
kT
665
TABLE 2
Spectroscopic Constants for OH
OH (A2 2 X2P (1,0))
Constant
A 00
A 10
A 11
Q 0, Q 1
V
B10
Bv
vv
(2)
hcB n
B nhcJ~ J 1 1!
~2J 1 1! exp 2
.
kT
kT
(3)
Fy
A 01
A 00
1 T 00
Q0
Q0
T 11A 11 1 T 10 A 10 1 V T 01
5
Q1 1 V
(4)
S DF E
SD
B 10
c
1
hL
z ~ A p!~F y!
E~ n ! g~ n ! d n ~ fn f Jn 0!
V
.
4p
(5)
666
S D
E Ray 5 n
Ei
~N As RayV!~ AL!.
A
(6)
S D
n pp 5 n
Ei
~N AsRayV! L.
hn
(7)
Temperature Measurements
Temperature measurements were made using a
75 mm type S (Pt/Pt 2 10%Rh) thermocouple
mounted on a stainless steel and ceramic probe.
No corrections for radiative losses, catalytic
effects or droplet impacts were made to the
mean temperature measurements obtained.
RESULTS AND DISCUSSION
Photographic images, made using laser sheet
illumination through the centreline of the flame
to highlight spray particles, are presented in
Figs. 35. The imaged height of the flame in
these figures was 150 mm, with laser sheet
lighting through the lower 70 mm. Figure 3,
obtained using 1/8 sec exposure time, shows the
spray flame with annular air off. The flame
stabilized approximately 12 mm above the spray
nozzle, and the stabilization point was steady
and symmetric about the spray axis. An inner
reacting core approximately 13 mm in diameter
was visible inside the flame, with little variation
667
668
Fig. 6. One hundred shot average composite OH fluorescence image, annular air off.
669
Fig. 8. One hundred shot average composite OH fluorescence image, annular air 5 4.77 l/s.
670
Fig. 10. One hundred shot average composite OH fluorescence image, annular air 5 9.52 l/s.
671
consistent with other measurements and computations for laminar and turbulent diffusion
flames reported in the literature [15, 16]. Noise
appearing as isolated intensity spikes in the
centre region for the annular air-off case is due
to elastically-scattered light from droplets
present in high concentrations in this region.
Figures 1315 present measurements of mean
672
true for the annular air 5 9.52 l/s case, Fig. 15,
which shows distinct mean temperature and OH
concentration peaks only in the plane 25 mm
downstream of the nozzle. Again, the temperature peak occurs slightly inside (fuel rich side)
of the mean OH concentration peak. It is
apparent that there is little difference in centreline temperature for the two annular air-on
cases tested, although both these cases result in
considerably hotter centerline temperatures
than the air-off case.
Measurements of droplet Sauter mean diameter obtained in planes 25, 60, and 100 mm downstream of the nozzle are presented in Figs. 1618.
Included in these figures are best fit RosinRammler parameters at locations corresponding to the peak volume flux. At these locations,
the Rosin-Rammler distribution model provides reasonable agreement with data, generally
with the arithmetic mean diameter D 10 within
10% and the Sauter mean diameter D 32 within
5% of the measured values. The RosinRammler size distribution model is given by
S S DD
D
m 5 1 2 exp 2
X
(8)
673
674
higher velocities associated with increased annular air flow rate would mean that droplets
require less time to travel downstream, which
means they would have less time to evaporate
before travelling a given distance. Other factors
which affect the size and volume flux distribution are the changes in flame structure that arise
with varying annular air flow rates, which affect
local evaporation rates, and hence droplet size
and trajectory histories.
Gas-phase mean velocity was obtained from
raw data sets by filtering the data to generate a
file consisting of only small droplet data. The
cut-off diameter was typically 5 mm, resulting in
a turbulent Stokes number of approximately 40
(defined as the turbulent time scale divided by
the droplet drag response timescale) in a region
with an integral time scale of 2 ms and relative
velocity of 10 m/s. Figure 23 presents a diameter-velocity correlation plot, obtained in the z 5
25 mm plane, approximately 18 mm off centreline in a region containing a wide spectrum of
droplet sizes under highly turbulent conditions.
As can be seen, there appears to be a clear
cut-off in measured velocity for diameters near
10 mm, below which there is essentially no
change in droplet velocity, suggesting that droplets in this size range are accurately following
the gas-phase flow.
Figures 24 to 26 present the gas-phase velocity vector fields for all annular air flow rates
675
Fig. 24. Gas-phase velocity vector plot with annular air off.
676
677
678
and its uncertainty is estimated to be approximately 1.5%.
The other term in the phase/diameter relationship is the relative index of refraction m
between the fuel and air. Although this is known
to a high degree of precision, it is a function of
droplet temperature. Since the fuel droplet can
be at any temperature from below ambient to
near its boiling point, the index of refraction is
variable between these limits. For methanol, the
index of refraction ranges from 1.3108 to 1.3329
over the temperature range from 10C to near
boiling at 64.5C [19], resulting in a sizing
uncertainty of approximately 4%. Selection of
index of refraction at the mid-point of the range
limits this error to 62%. Recent work by
Schneider and Hirleman [20] examined the effect of index of refraction gradients on particle
sizing, as larger particles would not be isothermal in a combusting environment and radial
gradients in temperature and hence refractive
index would be present. It was concluded that,
for atmospheric pressure flames, there is insufficient temperature and refractive index range
to cause a significant sizing error. However, in
high-pressure combustion systems with fuels
able to reach several hundred degrees, this
effect should be addressed.
The diameter-phase relationship assumes
that the particles are spherical, and at least of
the same order as the incident light wavelength.
As the particle size decreases, oscillations in the
phase/diameter relationship can occur, and care
must be taken in selecting the appropriate collection angle in order to size small particles
accurately. Naqwi and Durst [21, 22] performed
a detailed theoretical analysis of the scattering
process, and concluded that a PDI system could
be configured to accurately size particles as
small as 0.5 mm, at the expense of overall size
range. Rather than examining this problem
from a theoretical point of view, Ceman et al.
[23] and OHern et al. [24] performed a set of
experiments using a vibrating orifice aerosol
generator capable of accurately producing droplets as small as 3.7 mm, and assessing the
Aerometrics Phase Doppler Particle Analyzers
sizing ability directly. Oscillations in response
were found to occur at the two commonly
employed collection angles tested. Sizing errors
were found to increase dramatically below 10
679
680
be noted that there is a temperature dependence of the absorption lineshape due to Doppler and pressure broadening. In the range of
1000 to 2000K, the Doppler linewidth ranges
from 0.20 to 0.27 cm21, and the pressure broadened linewidth ranges from 0.11 to .08 cm21 and
the resulting overlap integrals range from 0.84
to 0.68 E/Dn respectively. Hence, any uncertainty in temperature can lead to some uncertainty in the overlap integral.
Another uncertainty in the laser fluence
arises from the laser itself in terms of shot to
shot variation of total and point laser power.
Since laser power measurements were made
with a time-averaging instrument, and LIF images were made using a single laser pulse, it was
necessary to assess the variability of the laser
power on a shot to shot basis. This was accomplished by imaging the light scattered by the
laser sheet when directed onto a smooth surface, and recording the total intensity count of
all pixels in the sheet image. The variation of
this total count on a shot-to-shot basis represents the variation of total sheet energy, again
on a shot-to-shot basis, which was found to be
approximately 10%. A similar quantification,
looking at a single pixel in the sheet image,
showed a 15% variation, suggesting some variability in laser light distribution across the sheet
on a shot-to-shot basis.
Some uncertainties are associated with the
energy distribution in the laser sheet across its
width and along its height. As long as the
intensity distribution across the width of the
sheet does not result in localized partial saturation, the resulting fluorescence is linear with
intensity and the total fluorescence arising from
an excited volume in space is related to the total
laser energy exciting that volume, regardless of
energy distribution. The absence of significant
localized saturation was verified by measuring
the fluorescence intensity in a steady flame
while increasing laser power, and ensuring that
fluorescence intensity increased linearly. Variations in intensity along the height of the sheet
were approximately 610%, and were compensated for in the data processing stage.
Perhaps the most significant source of uncertainty in the estimation of OH concentration is
the value used for the quenching rate, Q. The
quench rate is dependent on knowledge of the
681
682
temperatures and relative importance. In some
cases, such as lean combustion of non-sooting
fuels, radiation can be considered as radiant
exchange between the thermocouple junction
and the far surroundings, neglecting exchange
between hot gases. However, in other flames
where particulate loadings are high, or where
hot gases may be participating, these losses are
extremely difficult to quantify. Attya and
Whitelaw [28] estimated that the radiation correction is less than 5% for an 80 mm thermocouple wire where the burning mode of the
spray flame was primarily a diffusion flame.
Uncertainties and errors associated with fuel
droplet strikes on the thermocouple junction
itself, or on the lead wires near the junction, do
not appear to have been quantified. Conduction
losses can be minimized by using long leadwires,
and orienting the thermocouple in the flow to
minimize temperature gradients in the vicinity
of the junction.
Although the effect is not well quantified,
catalytic effects involving reactions at the thermocouple surface can produce large errors.
Bare platinum thermocouples are particularly
susceptible, since platinum is a particularly effective catalyst for a large number of reactions.
Pita and Nina [29] investigated the catalytic
effect in a premixed propane/air flame, as well
as in a 20C hydrogen/air non-reacting flow, and
concluded that substantial errors associated
with catalytic effects are present in radical-rich
flame zones. The magnitude of the error was
not quantified, depending strongly on local
composition. It was also found that, within the
ranges tested, the effect was independent of
thermocouple size and local flow velocity.
In this work, severe catalytic action was initiated upon shut-down after operating with an
annular air flow rate of 4.77 l/s, a condition
which caused heating of the bluff body due to
flame impingement. On shut-down, methanol
left in the nozzle vaporized and emerged from
the nozzle as a vapour jet which caused sufficient catalytic action on contact with the thermocouple that it was seen to glow. Turning on
the fuel spray and directing it at the thermocouple enhanced this reaction, and caused a temperature reading over 1600K, Hence, in this
case, the error induced by the catalytic effect is
on the order of 1300K when the fuel spray is at
683
REFERENCES
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
684
Twentieth Symposium (International) on Combustion,
The Combustion Institute, Pittsburgh, 1984, p. 327.
16. Smooke, M. D., Xu, Y., Zurn, R. M., Lin, P., Frank,
J. H., and Long, M. B., Twenty-fourth Symposium (Intl)
on Combustion, The Combustion Institute, Pittsburgh,
1992, p. 813.
17. Bilger, R. W., Joint Intl Conference Australia/New
Zealand and Japanese Sections, The Combustion Institute, 1989.
18. Keyes, D. E., and Smooke, M. D., Comb. Sci. Technol.
67:85 (1989).
19. Dean, J. A., Langes Handbook of Chemistry 14th
Edition, McGraw Hill, New York, p. 5.130, 1992.
20. Schneider, M. and Hirleman, E. D., Applied Optics
33:2379 (1994).
21. Naqwi, A. A., and Durst, F., Part. Part. Syst. Charact.
8:245 (1991).
22. Naqwi, A. A., and Durst, F., Part. Part. Syst. Charact.
9:66 (1992).
23. Ceman, D. L., OHern, T. J., and Rader, D. J., ASME
Fluids Engineering Division FED V.95, ASME, 1990, p.
61.
24. OHern, T. J., Rader, D. J., and Ceman, D. L.,
Proceedings of the ASME Fluid Measurements and
Instrumentation Forum, San Diego, 1989, p. 49.
25. McDonell, V. G., and Samuelsen, S., Liquid Particle
26.
27.
28.
29.
30.
31.
32.
33.
34.