18th International Symposium on the Application of Laser and Imaging Techniques to Fluid Mechanics・LISBON | PORTUGAL ・JULY 4 – 7, 2016 Experimental study of laser ignition probability, kernel propagation and air and fuel droplet properties in a confined swirled jet-spray burner Javier Marrero Santiago 1,* , Antoine Verdier 1 , Gilles Godard 1 , Alexis Vandel 1 , Gilles Cabot 1 , Mourad Boukhalfa 1 and Bruno Renou 1 1: CORIA-UMR6614, Normandie Université, CNRS, INSA et Université de Rouen, 76800 Saint Etienne du Rouvray, France * Correspondent author: [email protected]Keywords: Swirl jet-spray flame, PDA, Spray ignition, Two-phase combustion, Flame kernel propagation, Laser ignition ABSTRACT Spray combustion is largely used in industrial applications such as aeronautical engines. New engine designs must reduce emissions and increase efficiency through lean combustion. This can carry flame stability problems and re- ignition in altitude has to be well addressed. Two-phase flow ignition and combustion have to be well understood in order to enable technological evolution in this direction. This investigation proposes a laser induced ignition study in a real confined swirled jet-spray burner. Phase Doppler anemometry (PDA) is used to characterise airflow velocity and fuel droplet size distribution and velocity in first place. A probabilistic approach to ignition is presented and propitious regions for ignition are identified. High-speed visualisation is used to track the flame kernel movement and development inside the burner in order to analyse the possible paths followed by the kernel towards a stabilised flame or towards extinction. Results show how local properties vary along the chamber. Airflow velocity and turbulent kinetic energy are very intense in the central region of the burner, over the annular co-flow. Fuel droplets of big diameters show high slip velocities, which are greater outside the air jet. These parameters control droplet evaporation and vapour repartition, and have a great impact on ignition probability. The probabilistic ignition study coupled to the kernel high-speed visualisation reveals that a flame kernel is more likely to survive if trapped by the outer recirculation zone (ORZ), showing great correlation with the airflow velocity field. Ignition probabilities grow towards the chamber lateral walls. The different steps of the development of a kernel towards a stabilised flame are compared to the pressure variation in the chamber. This investigation is useful for numerical simulation validations and contributes to scientific knowledge on two-phase ignition. 1. Introduction New aeronautical burner designs must move onto configurations with lower pollutant emissions and higher efficiencies. This demands a better comprehension of the processes involved in two- phase combustion and ignition. Aeronautical engines present very complex geometries. Here, combustion carries coupled multi-physical and chemical constraints. Flame temperature and burning velocity decrease for low fuel-to-air ratios, implying an increase in flame instabilities. This issue can be translated into more frequent flame extinctions and more difficult flame
19
Embed
Experimental study of laser ignition probability, kernel ...ltces.dem.ist.utl.pt/lxlaser/lxlaser2016/finalworks2016/papers/03... · on laser ignition probability will be presented,
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
18th International Symposium on the Application of Laser and Imaging Techniques to Fluid Mechanics・LISBON | PORTUGAL ・JULY 4 – 7, 2016
Experimental study of laser ignition probability, kernel propagation and air and fuel droplet properties in a confined swirled jet-spray burner
Javier Marrero Santiago1,*, Antoine Verdier1, Gilles Godard1, Alexis Vandel1, Gilles Cabot1, Mourad Boukhalfa1 and Bruno Renou1
1: CORIA-UMR6614, Normandie Université, CNRS, INSA et Université de Rouen, 76800 Saint Etienne du Rouvray, France * Correspondent author: [email protected]
The magnitude of the resulting vectors is displayed in Fig. 7 for the same three size-classes as
before. Mean slip velocities are more important close to the nozzle exit and for big droplets. The
[0-10] group experiences lower values than the others but these are not negligible. Above z=15
mm, peaks remain under 20 ms-1. For the [40-50] group, mean slip velocities grow up to 38 ms-1
and values above 20 ms-1 can be found below z=25 mm. The maximum slip velocities are found
18th International Symposium on the Application of Laser and Imaging Techniques to Fluid Mechanics・LISBON | PORTUGAL ・JULY 4 – 7, 2016
on the borders of the spray, at the same radial positions as droplet peak radial velocities. Here,
droplets still have high axial velocities and they have exited the high air velocity region. In the
centre, al low z, strong slip velocities are also observed. They are caused by the descending
recirculating air flowing against the rising droplets. Vapour produced at the centre will be
recirculated downwards along the centre. There is a minimum mean slip velocity for all size-
classes above z=10 mm placed over regions of medium turbulence intensity, where velocity
fluctuations are 40% of mean velocities and where k is maximal. Provided that large droplets
present far greater slip velocities than small droplets at many points along the profiles (>20 ms-1),
they will experience high evaporation rates. As they carry most of the fuel mass, they will
contribute significantly to fuel vapour production and repartition along the burner. The vapour
produced at the borders finds a region with low values of air velocity and may be entrained into
the ORZ.
Fig. 7 Mean slip velocity magnitude in non-reactive conditions for three size-classes. Purple triangles represent the
[0-10] µm group, green squares the [20-30] µm group and red circles the [40-50] µm group.
3.3. Ignition:
Ignition probability is strongly influenced by the aerodynamics of the flow inside the chamber
and by fuel droplet and vapour distribution. Air and droplet velocity fields have been discussed
on the previous sections. Fuel vapour is produced at a higher rate in high slip velocity regions,
and it will be transported by the airflow. Furthermore, there is a greater liquid presence in areas
covered by big droplet pathways, so the injection pressure, the injector design and the spray
angle play a very important role too. The coupling of these parameters plus the temperature
18th International Symposium on the Application of Laser and Imaging Techniques to Fluid Mechanics・LISBON | PORTUGAL ・JULY 4 – 7, 2016
distribution along the burner will directly impact on the ignition probability. Results also show
that the flame kernel follows different ways and different propagation directions in function of
local mixture properties and instantaneous airflow velocities. This may influence even more the
ignition probability than other parameters. Figure 8 presents the ignition probability results in
form of contour lines on the right half, superposed to a map of the airflow velocity magnitude
taking into account the three discussed velocity components. Airflow turbulent kinetic energy (k)
is recalled on the left half. Purple lines show how ignition probability increases towards the
burner walls and towards a vertical distance from the nozzle of z=20 mm. The minimum ignition
probability region is located along the centre of the burner, from r=0 to r=10 mm, where a flame
kernel rarely or never evolves into a stabilised flame. From r=10 to 30 mm, probability increases
strongly, from 0% to 70% in the better case (z=20 mm). The area under z=40 mm, over z=10 mm
and for r>30 mm, represents the best zone to transform a spark into a successful ignition,
showing points over 80% of probability.
Fig. 8 Left: Airflow turbulent kinetic energy (m2s-2). Right: ignition probability map overlapped to the air mean
velocity magnitude (ms-1).
The location and shape of the lines recall the outer recirculation zone (ORZ). Furthermore, the
maximum steepness of the probability field appears to follow the mean airflow velocity contours
where velocities and turbulence relax just outside the swirling air jet, showing a big correlation
18th International Symposium on the Application of Laser and Imaging Techniques to Fluid Mechanics・LISBON | PORTUGAL ・JULY 4 – 7, 2016
between these parameters and ignition. At axial stations close to z=5 mm and for r<5 mm, 1 over
10 sparks will succeed to generate a stabilised flame by a different mechanism than the others.
The air velocity here is negative due to the inner recirculation zone (IRZ) and the flame kernel
will be directed into the spray cone where it will certainly find regions rich in fuel but also a
strong shear and velocities that will intensely stretch it and convect it upwards into poor regions.
Figures 9 to 12 present kernel propagation sequences recorded with the high-speed Phantom
V2512 camera. Although images are acquired every 250 µs, each image here is 500 µs after the
previous one because no such temporal resolution is needed for this purpose. It is worth noting
that the camera is recording at an angle that enables the perception of the azimuthal kernel
displacement but privileges the movement perception in the XZ plane. The spark saturates some
pixels that present high intensities in all images before 3 ms after the shutter opening, and stays
visible in the other ones. Two points from the ignition interrogation mesh have been selected to
be analysed here (red dots in Fig. 8). Figure 9 shows the development and movement of a kernel
for a successful ignition at x=40; z=40 mm. This point has an ignition probability of 67%. The first
image corresponds to the flame kernel 160 µs after the spark breakdown (shutter opening). It
seems to be cut at the left because the camera has no optic access to the region close to the quartz
window (it is hidden by the burner column). The following images show the flame kernel
thermal expansion and global downward movement, trapped by the ORZ and moving towards
the spray through the slow velocity region. After 4 ms, punctual growth of pixel intensities
reveals an increase in heat release.
Fig. 9 Successful ignition. Evolution of the kernel position for laser focusing point at x=40 mm, z=40 mm.
Figure 10 presents an unsuccessful ignition for the same point (x=40; z=40 mm). Since the first
image one can see the thermal expansion divides the flame kernel in two parts of similar sizes.
0,16 ms 0,66 ms 1,16 ms 1,66 ms 2,16 ms
2,66 ms 3,16 ms 3,66 ms 4,16 ms 4,66 ms
x=40 mm
z=40 mm
Successful
Ignition
18th International Symposium on the Application of Laser and Imaging Techniques to Fluid Mechanics・LISBON | PORTUGAL ・JULY 4 – 7, 2016
The part that goes upwards keeps moving upwards, diminishing, and disappears by 2,66 ms.
The kernel that moves downwards, having approximately the same surface and volume, lives
longer but it is too small to survive. It seems to relight at 4,66 ms but it ends by extinguishing. If
compared to the previous figure (Fig. 9), the intensity at this time is much lower. In addition,
after 2,66 ms it appears to move slightly upwards.
Fig. 10 Missed ignition. Evolution of the kernel position for laser focusing point at x=40 mm; z=40 mm.
Figure 11 shows a successful ignition for another point (x=20; z=40 mm) with a much lower
ignition probability (6%). Here, the flame kernel is submitted to much higher air ascending
velocities than when placed at the other point. The flame kernel seems not to move too much
during the first 2 ms, although it experiences a stretching and a loss in emission intensity. It
starts to divide in two parts. Only the lower one will survive and light the flame, moving
downwards and towards the spray with a strong azimuthal component.
Fig. 11 Successful ignition. Evolution of the kernel position for laser focusing point at x=20 mm; z=40 mm.
0,16 ms 0,66 ms 1,16 ms 1,66 ms 2,16 ms
2,66 ms 3,16 ms 3,66 ms 4,16 ms 4,66 ms
x=40 mm
z=40 mm
Missed
Ignition
0,16 ms 0,66 ms 1,16 ms 1,66 ms 2,16 ms
2,66 ms 3,16 ms 3,66 ms 4,16 ms 4,66 ms
x=20 mm
z=40 mm
Successful
Ignition
18th International Symposium on the Application of Laser and Imaging Techniques to Fluid Mechanics・LISBON | PORTUGAL ・JULY 4 – 7, 2016
Figure 12 shows two series of images for two missed ignitions at the same point (x=20; z=40 mm),
where kernels are clearly convected upwards losing any chance to find propitious conditions to
survive. The sequences show that the quicker a kernel moves upwards, the quicker it diminishes.
The present kernel evolution sequences are representative of the majority of the observed cases.
It is only at small values of r, in the IRZ, that the kernel is sucked downwards towards the spray
and reconvected upwards very violently by the air co-flow, rarely being successfully ignited.
Fig. 12 Missed ignitions. Evolution of the kernel position for laser focusing point at x=20 mm; z=40 mm.
If the flame kernel survives enough time without being convected upwards and finds propitious
regions to grow and produce more and more chemical reactions, the complete burner ignition
takes some time and passes through different phases. These phases are captured by the high-
speed images and by the pressure signal from the chamber, and are illustrated in Fig. 13. They
can be classified into the following:
Kernel growth and movement
Developed kernel settling in the ORZ
Spray light-around
Maximum heat release
Flame attachment to nozzle
Flame lifting
The pressure signal shows strong pressure variations due to the spark shockwave and noise
from 0 ms to 15 ms after breakdown. During this time, the flame kernel moves through the
chamber and grows into a little weak flame that often touches the opposite wall to descend into
0,16 ms 0,66 ms 1,16 ms 1,66 ms 2,16 ms
x=20 mm
z=40 mm
Missed
Ignition
0,16 ms 0,66 ms 1,16 ms 1,66 ms 2,16 ms
x=20 mm
z=40 mm
Missed
Ignition
18th International Symposium on the Application of Laser and Imaging Techniques to Fluid Mechanics・LISBON | PORTUGAL ・JULY 4 – 7, 2016
the ORZ at about 20 ms after breakdown. At this moment, pressure starts to increase in the
chamber and the spray light-around starts, following an anti-clockwise movement imposed by
the swirling motion. A pressure peak is reached when the whole spray has been lighted, 42 ms
after breakdown. This moment is coupled to the maximum heat release observed in the images
and may be caused by fuel vapour accumulation. This stage is followed by a noisy conical flame
attached to the spray cone that generates strong pressure oscillations and becomes unstable to
disappear at about 70 ms after breakdown. After this, the pressure signal presents oscillations
with lower frequencies and decreasing amplitudes. When the conical intense flame disappears, it
leaves a lifted, lean and stable blue flame, as the one shown in Fig. 1 (here, the last image at 124
ms). Characteristic times vary from one event to another but Fig. 13 is representative of the
process.
Fig. 13 Evolution of the pressure signal inside the chamber during an ignition event. Images of the different ignition
phases at instants from 1 to 125 ms after the energy deposition.
1 ms 21 ms 33 ms 42 ms 54 ms
63 ms 69 ms 73 ms 98 ms 124 ms
18th International Symposium on the Application of Laser and Imaging Techniques to Fluid Mechanics・LISBON | PORTUGAL ・JULY 4 – 7, 2016
4. Acknowledgements
The financial support provided by ANR TIMBER is gratefully acknowledged.
5. Conclusion
An ignition study has been carried out starting with the characterisation of airflow velocity and
fuel droplet size-classified velocity. The ignition probability distribution within the burner has
been presented and discussed, as well as the propagation of the flame kernel and its different
possible pathways to evolve into a stabilised flame. The different phases through which the
system passes during the ignition process from a kernel to a stable flame have been identified.
PDA results show how big droplets experience higher slip velocities than small droplets and that
slip velocities are higher out of the air co-flow jet, towards the start of the ORZ. Ignition
probability appears to be higher towards the ORZ and close to the burner walls, where air
velocity and turbulent kinetic energy present low values. High-speed visualisation of flame
kernel propagation confirms that the ORZ is a privileged region for ignition. The flame kernel is
more likely to survive if convected downwards. High air velocities will lead to the extinction of
the flame kernel. This work provides a useful database for numerical simulations and new
insight on laser induced ignition in two-phase flows.
6. References
Andrew P Wandel, Combustion and Flame, (2014). Cardin C , Bruno Renou, Gilles Cabot, Abdelkrim Mourad Boukhalfa, Combustion and Flame,
160 (2013) 1414–1427 Cléon G, D. Honoré, C. Lacour, A. Cessou, Proceedings of the Combustion Institute, 35 (2015)
3565-3572. Cordier M, A. Vandel, G. Cabot, B. Renou, A.M. Boukhalfa, Combustion Science and
Technology, 185 (2013) 379-407. Cordier M, PhD Thesis report, INSA de Rouen, (2013). Correia Rodrigues H, M.J. Tummers, E.H. van Veen, D.J.E.M. Roekaerts, Combustion and Flame,
162 (2015) 759-773. Correia Rodrigues H, M.J. Tummers, E.H. van Veen, D.J.E.M. Roekaerts, International Journal of
Heat and Fluid Flow, 51 (2015) 309-323. Correia Rodrigues H, Mark J. Tummers, Eric H. van Veen, Dirk J.E.M. Roekaerts, Combustion
and Flame, (2014). Garcia Rosa N, PhD Thesis report, Université de Toulouse, (2008). Hannebique G, PhD Thesis report, Université de Toulouse, (2013). Hwan Lee S, Hyungrok Do, Jack J. Yoh, Combustion and Flame, 165 (2016) 334-345.
18th International Symposium on the Application of Laser and Imaging Techniques to Fluid Mechanics・LISBON | PORTUGAL ・JULY 4 – 7, 2016
Kourmatzis A, P.X. Pham, A.R. Masri, Combustion and Flame, (2015). Lawes M, Y. Lee, A. S. Mokhtar and R. Woolley, Combustion Science and Technology, 180 (2).
pp. 296-313. Letty C, Epaminondas Mastorakos, Assaad R. Masri, Mrinal Juddoo, William O’Loughlin,
Experimental Thermal and Fluid Science, 43 (2012) 47–54. Linassier G, PhD Thesis report, Université de Toulouse, (2012). Morkous S Mansour, Issam Alkhesho, Suk Ho Chung, Experimental Thermal and Fluid Science,
(2015). Mulla IA, Satyanarayanan R. Chakravarthy, Nedunchezhian Swaminathan, Ramanarayanan
Balachandran, Combustion and Flame, 000 (2015) 1–16. Nicolia C, Denet B., Haldenwang P., Combustion and Flame, 000 (2015) 1–14. Paulhiac D, PhD Thesis report, Université de Toulouse, (2015). Providakis T, PhD Thesis report, Ecole Centrale Paris, (2013). Sánchez AL, Javier Urzay, Amable Liñán, Proceedings of the Combustion Institute, 35 (2015)