Experimental study of local extinction mechanisms on a...

18th International Symposium on the Application of Laser and Imaging Techniques to Fluid Mechanics・LISBON | PORTUGAL ・JULY 4 – 7, 2016

Experimental study of local extinction mechanisms on a spray jet flame

A. Verdier*, J.Marrero Santiago, A.Vandel, G.Godard, G.Cabot, M.A. Boukhalfa and B.Renou CORIA-UMR6614

Normandie Université, CNRS, INSA et Université de Rouen 76800 Saint Etienne du Rouvray

France * Correspondent author: [email protected]

Keywords: High-speed OH-PLIF, Spray jet flame, Flame extinctions, PDA

ABSTRACT

This paper presents high-speed (HS) images of OH-PLIF collected at a repetition rate of 10 kHz along the entire

length of an n-heptane spray jet flame. The experimental set-up is composed of an annular non-swirled air co-flow

that surrounds a central hollow-cone spray injector, leading to a stable flame with well-defined boundary

conditions. The experiments include accurate measurements of droplet size (PDA), droplet and carrier phase

velocity (PDA) and two-dimensional flame structure (OH-PLIF). The polydisperse spray distribution yields small

droplets along the centerline axis while the majority of the mass is situated as big droplets along the spray borders.

The flame structure presents a classical shape, with an inner wrinkled partially premixed flame front and an outer

diffusion flame front. Although the High Speed (HS)-OH-PLIF images are only qualitative, they are found to be a

sufficient spatial and temporal resolution to relay the dynamics of extinctions events. Applying data processing

tools and coupling with aerodynamics results allowed to highlight different extinctions mechanisms. In the inner

reaction zone the measured speed, as well as the turbulent kinetic energy, showed that the large turbulence scales

played a significant role in the dynamics of extinction. However; the locally extinction in the outer reaction zone can

be attributed to the big droplets that present a skin term for the flame.

1. Introduction

Spray combustion involves many complex physical phenomena, including atomisation,

dispersion, evaporation and combustion, which generally take place simultaneously or within

very small regions in the combustion chambers. Although numerical simulation is a valuable

tool to tackle these different interactions between liquid and gas phases, the method needs to be

validated through reliable experimental studies. Therefore, accurate experimental data on flame

structure and on liquid and gas properties along evaporation and combustion steps are needed.

These experiments also allow to provide a physical understanding of fuel droplet interaction

with the flame structure in real and representative two-phase flow configurations, in the

perspective of moving forward into more complex configurations (aeronautical, gas turbine, …).


The spray jet flame is a canonical configuration which presents the essential feature of very well

defined boundary conditions. The flame topologies are representative of those obtained in real

burners with 3D complex flow motions (swirl or bluff-body), including a large distribution of

droplet sizes and different combustion regimes. The simulation of these geometries is a complex

task since it requires at least an accurate prediction of the fuel vapour and the thermal budget

between the droplet and its gaseous surrounding through the evaporation model, and a detailed

description of the combustion reactions able to predict the different modes of combustion (Ma,

Naud et al. 2015). The spray jet flame has already been experimentally investigated including

studies on the flame structure (Cessou and Stepowski 1996; Friedman and Renksizbulut 1999;

Marley, Welle et al. 2004; Cléon, Honoré et al. 2015; Correia Rodrigues, Tummers et al. 2015) or

the stabilisation of the edge-flame (Marley, Welle et al. 2004; Cléon, Honoré et al. 2015).

However, one of the fundamental aspects to consider in turbulent two-phase combustion is the

strong interactions between the chemical reaction, the turbulence field and the droplets which

occur during the vaporisation and combustion processes. For instance, the turbulence can

amplify local heat releases which greatly influence the flame by momentum, heat and mass

transfer. The rate of change in the spatial flame structure due to scalar dissipation rates and local

variations in the mixing zone can induce extinctions events (Kaiser and Frank 2009). With the

improvement of the high repetition rates, the high speed diagnostics can temporarily resolve the

scalars in turbulent flows. These efforts have yielded a greater understanding of dynamic

turbulent flame behaviour through tracking the temporal evolution of transient phenomena such

as local extinction, auto-ignition and turbulence-chemistry interactions (Boxx, Stöhr et al. 2009;

Boxx, Arndt et al. 2010; Abram, Fond et al. 2013; Slabaugh, Pratt et al. 2015). Besides, the

previous work on the extinction phenomena, in a turbulent spray jet flame the situation is more

complex and an understanding of extinction mechanisms need to be studied.

The present experimental study focuses on the detailed characterization of extinction

mechanisms along the combustion phases in an n-heptane spray jet flame by applying high

speed optical diagnostics. The local properties of flow as spray droplet dispersion, size, velocity

and the carrier phase velocity are obtained by Phase Doppler Anemometry (PDA) and by

Particle Tracking Velocity (PTV). High Speed OH-PLIF (HS-OH-PLIF) is used to time-resolve the

flame dynamics and in particular the transient phenomena as local extinction. Three mechanisms

of extinctions in three different zones are identified and further analysed in detail using a

specific methods for processing images.


2. Experimental approach

2.1. Experimental facility

Experiments are carried out in an atmospheric and open burner based on the geometry of the

gaseous KIAI burner (Cordier, Vandel et al. 2013) (Fig. 1). The fuel injection system is composed

of a simplex fuel injector (Danfoss, 1.35 kg.h-1, 80°, hollow cone) and an external annular, non-

swirling air co-flow, with an inner and outer diameter of 10 and 20 mm respectively. Air and

liquid fuel (n-heptane) mass flow rates are controlled by thermal and Coriolis mass flow

controllers. The inlet conditions of air and fuel are 6 g.s-1 (T=298 ± 2 K) and 0.28 g.s-1 (T=298 ± 2

K) respectively, which leads to an air bulk velocity of 19.9 m.s-1. In Fig. 1, X and Z represent the

radial and axial coordinates, respectively.

Fig. 1: Detail of the injection system and typical flame picture when the flow is illuminated by a Nd:YAG laser sheet

2.2 Optical diagnostics

A schematic of the high-speed OH-PLIF imaging set-up is shown in Fig. 2. In this paper, OH-

signal is used as a marker of the reaction zone so a breakage in the OH profile is deemed to mark

local extinction. A Nd-YAG-laser operating at 532 nm (with a power of 104 W) is used to pump a

tunable dye laser (Sirah Credo). The resultant output pulse energy is 380 µJ per shot in the probe

volume. The excitation wavelength is tuned to the Q1(5) transition of the ∑

← ∏ band of OH at λ =282.665 nm. The beam is then expanded to 35mm in height

using a diverging cylindrical lens before being focused into a sheet using a focused with a 200

mm focal length. Due to the low energy delivered, the detection system consists of a CMOS-

camera Photron Fastcam SA5 mounted with an external image intensifier (High Speed IRO,

ZX

Air inlet

Fuelinlet


Lavision) equipped with UV lens (f/2). The camera has been run at a repetition rate of 10 kHz

with an array of 896 x 848 pixels and an image resolution of 500 μm/pixel. Intensifier delay (2900

ns) and gate (200 ns) are set to optimize the signal to noise ratio. Background noise arising from

elastic scattering by the droplets is reduced with two high-pass optical filters (Schott WG295).

The camera on-board memory can hold over 7500 frames, corresponding to an acquisition time

of 750 ms. The OH-PLIF signals are collected using a broadband collection strategy from 308 to

330 nm with a band-pass filter (Schott UG11). Moreover, the laser sheet profile is taken into

account and corrected by filling the test facility with a homogeneous mixture of air/acetone in

the quartz plate. Besides the presence of droplets in the field of view, the signal to noise ratio is

really good and varied between 100 and 150.

Fig. 2 : Optical arrangement for high-speed OH-PLIF measurements.

Droplet size and velocity are characterised by a commercial PDA system (DANTEC) operating

in DUAL mode with 50° front-scattering probes. An argon laser provides green (514.5 nm) and

blue (488 nm) beams. Beam spacing is 50 μm; the focal length of the transmitting lenses is 350

mm, and the focal length of the receiving optics is 310 mm. The used aperture mask allows a

detection diameter range of 139 μm. The measurement volume can be approximated by a

cylinder of 120 μm in diameter and 200 μm in length. At each measurement location, data

sampling is limited to 40,000 droplets or to 30s of measuring time, allowing converged statistics

of size-classified data. Due to the spray structure and particle concentration distribution, the

measurements are not possible below z = 10 mm. Besides the spray characterization, the carrier

phase velocity is also investigated by seeding the co-flow with 2.5 μm olive oil droplets.

FILTER

(UG11 / WG295x2)

282,65 nm

PU

MP

LA

SE

RN

d-Y

AG

53

2n

m

DYE LASER

SIRAH

CREDO

CAMERA

Intensifier


2.3 Data processing

The objective of this study is to understand the extinction mechanisms in a spray jet flame. Due

to the huge amounts of data resulting from high-speed imaging, it is necessary to use an

automatic image-processing to extract several results. A pre-processing is applied to normalize

raw images and can be divided into two steps. First, it is the background correction which allow

to subtract the reflexions and electronic noise from the camera intensifier by averaging 500

images, recorded when the experiment is switched off but with the laser beam. Due to the non-

homogeneous energy distribution in the laser beam, the signal fluorescence is artificially

modulated. It is possible to have access to the mean laser sheet profile, to correct these spatial

energy variations, with filling a region of acetone vapor and by averaging 1000 images. Finally,

the pre-processing steps can be summarized with the following equation:

(1)

Where is the corrected images, is the raw images, is the mean laser profile

and is the mean backroung image.

An image-processing tool is developed to automate the flame front detection with a Non Linear

Diffusion (NLD) filter to reduce the level of noise and to enhance gradients in the images. The

method is based on the original appoach formulated by Perona and Malik (Perona and Malik

1990) and has several advantages: (i) Noise is smoothed locally within regions defined by object

boundaries, whereas little or no smoothing occurs between image objects. (ii) Local edges are

enhanced because discontinuities, such as boundaries, are amplified (Malm, Sparr et al. 2000;

Hartung, Hult et al. 2009).

A traditional way to smooth an image is to convolve it with a Gaussian Kernel and resolve the

solution of the linear diffusion equation:

(2)

Perona and Malik (Perona and Malik 1990) proposed to exchange the constant scalar diffusion

in Eq. (2) for a scalar-valued function of the gradient of the gray image. The diffusion

equation is now written by the following equation:


[ | | ]

(3)

The diffusivity function g is a monotonically decreasing function which is responsible for the

edge preservation. Perona and Malik (Perona and Malik 1990) proposed two different diffusivity

functions:

| |

| |

(4)

| | (

| |

)

(5)

It is found that the NLD filters use with Eq. (4) demonstrate a good ability to preserve the flame

front edges without artificial shifting in position (Han, Cai et al. 2014). In the present

experimental study a NLD filter with a diffusivity function equal to Eq. (4) is chosen and results

are represented in Fig. 3.

Fig. 3 (a) Image corrected by pre-processing. (b) Image filtered by NLD. (c) Comparison between raw and NLD

profiles.

Numerically, a semi-implicit iterative scheme is implemented and executed in parallel in Matlab

software. It is an excellent compromise between computational cost and filter performance. Then

active contour models are used to obtain the curves that most accurately describe the flame front.

These curves are obtained using the concept of geodesic path computation. The level set

representation is employed; even complex curve evolutions including those with topological

changes in their structures can be handled. The geodesic active contour model is introduced by

Caselles et al. (Caselles, Catté et al.).

(a) (b) (c)


This approach is equivalent to find geodasic distance of two points. The idea behind the level set

method is to imbed a curve within a surface and to solve the Eq. (6). The first step is to specify an

initial guess for the contour, which is then moved by image driven forces boundaries of the

desired objects. In our models, two types of forces are considered. The internal forces are

designed to keep the model smooth during the deformation process, while the external forces,

which are computed from the underlying image data, are defined to move the model toward an

object boundary or other desired features within the image.

| | (

| |) | |

(6)

Fig. 4: (a) Representation of Φ (Eq. 6) and initial contour for the level set. (b) Image filtered by NLD and final

contour obtained by level set method.

A representation of Φ with the initial contour is plotted in Fig. 4a. As described below, the final

contour (Fig. 4b) is obtained when the function Φ is equal to zero. From the extract contour,

plenty of information can be obtained as the curvature.

Moreover, to improve the droplet-flame front interactions, a Particle Tracking Velocimetry (PTV)

is used, in the corrected images, to determine the velocity and the trajectory associated to

individual droplet in the spray jet flame. The method works in a Langrangian frame of reference.

PIV calculates the associated velocity to group tracer particles at fixed spatial positions, while

PTV, depending on the experimental conditions, allows the determination of the velocity

associated to individual particles trajectories. In order to obtain these results, the PTVlab (Matlab

software) is used (Brevis, Niño et al. 2011). Fig. 5 shows the procedure to extract the velocity and

(b)(a)


trajectory for each droplets contained in the Region Of Interest (ROI). To optimize the simulation

two ROIs are chosen: one between the two reaction zones and a second below the flame leading

edge. Droplets are extracted by image processing from the corrected images.

Fig. 5 : (a) Image corrected by pre-processing. (b) Droplet extraction from image corrected. (c) Droplet velocity after

simulation (PTVlab) in one of ROI.

3. Results and discussion

3.1) Global analysis of flame shape

Fig 6: (a) Instantaneous OH-PLIF image with diameter contour lines. (b) Mean OH-PLIF image with droplet velocity

colored by the droplet diameter

An instantaneous OH-PLIF image of the spray jet flame (Fig 6(a)) illustrates the double flame

structure. The spray jet flame is lifted and stabilized about 25 mm downstream the injector as

shown in Fig 6(b). It exhibits two branches corresponding to an inner and an outer flame front

sharing a common flame base.

(a) (b) (c)

(a) (b)

Innerreaction

zone

outerreaction

zone

outerreaction

zone


This flame structure results from the spray heterogeneity in size (Marley, Welle et al. 2004;

Cléon, Honoré et al. 2015; Correia Rodrigues, Tummers et al. 2015), where the large droplets

spread into the outer part of the air co-flow and the ambient air (red lines), and the small

droplets are mainly located in the centre. Indeed, in this region, the very small droplets (blue

lines) represent only a small quantity of the liquid mass and undergo a slow evaporation,

leading to fuel lean mixture, below the lean flammability limit. As shown by the droplet velocity,

the small droplet (blue vectors) mainly stay in the cold central zone and never cross the flame

front. On the contrary the red vectors in Fig 6(b) indicate that large, inertial droplet move to the

external part of the spray, crossing the leading edge and reaching the hot region between the

inner and outer reaction zones.

The inner flame structure is strongly wrinkled and located along the shear layer created by the

air co-flow discharging into the ambient air. Indeed, the mean location of the inner flame is

placed over a high turbulent kinetic energy region of the NRC flow (Fig. 8(b)). It is characterised

by an intense mixing between air co-flow and small droplets at downstream locations, due to

high level of turbulent kinetic energy. In addition to turbulence, the strong variations of

equivalence ratio explain the highly wrinkled shape of the inner reaction zone and its fluctuating

behavior.

Additionally, the inner flame is characterised by a strong OH gradient indicating that

combustion occurs in a partially premixed regime with a flame propagation mechanism. Fuel

droplets are still visible in the vicinity of the inner reaction zone and may have crossed it.

Indeed, these droplets correspond to the large droplets moving to the outer part of the spray as

shown by droplet velocity (red vectors Fig. 8(c)). In the region between the flame fronts, where

weak OH signal is still visible, the high surrounding temperature, due to the combustion,

imposes a fast droplet evaporation. The strong thermal exchange induces an augmentation of the

gaseous fuel mass fraction. This fuel reservoir will react further in the diffusion-like outer flame

front. The outer reaction zone is less wrinkled, more stable, thicker than the turbulent inner

reaction zone, and characterised by a smoother OH gradient.

In the combustion regime, usually a reaction progress variable ( ) is used to describe the

progress of combustion in a flame front. In the fresh gas, the progress variable is conventionally

put to zero. In the burnt gas, it equals to unity. Across the flame, the intermediate values

describe the progress of the reaction to turn into burnt gas the fresh gas penetrating the flame

sheet. A progress variable can be set with the help of any quantity like temperature, reactant

mass fraction, provided it is bounded by a single value in the burnt gas and another one in the

fresh gas.


For instance, c can be written in the following way :

It is seen that c is a normalization of a scalar quantity. The index denotes the burnt gas and

the fresh gas. Experimentally, the mean progress variable is obtained from the binearisation of

the mean of all instantaneous OH-PLIF images (Erreur ! Source du renvoi introuvable.).

Fig. 7: Left side: Mean OH-PLIF image. Right side: Mean progress variable < >

Figure. 7 (right side) indicates that for radial distances smaller than X=4 mm, the droplets are

always present in fresh gas. However, regardless of axial station, at X=20 mm, the droplets will

be in burnt gas.

3.1) Zones of local extinctions in a spray jet flame

The following section concerns the analysis of OH-PLIF results. The high repetition rate is used

to understand the physical phenomena involved in flame dynamics. The field of view is now

equal to 35*35 mm² and chosen to have a good spatial resolution. It is worth noting that PLIF-

OH is used throughout this study as a marker of reaction zone so that a breakage in an otherwise

continuous OH profile is deemed to mark local extinction. The interval between two consecutive

instantaneous OH-PLIF images is equal to 0.1 ms and gives a good time resolution to study the

flame dynamics.

Figure 8(a) shows the mean OH-PLIF with the different extinction zones. These points represent

the first position (X & Z) of extinction. The next point is necessarily obtained after a stable

reaction condition that means a continuous inner reaction zone. 100 extinctions are detected on

1000 instantaneous HS-OH-PLIF. The local extinctions, in the inner reaction, appear in two

regions represented by red deltas and blue triangles. This bimodality implies that they are


different extinctions mechanisms. The red deltas are present between X=10 and X=15 mm for

different heights upper than 27mm. This zone is a mixing zone enhanced by the turbulence

induced by the inner shear layer as shown in Fig. 88(b) by the TKE. However, a second local

extinction zone, identified by the blue triangles, close to the leading edge is observed. The flame

stabilization is mainly controlled by the available amount of fuel vapour, and located around

stoichiometric conditions, where the flame speed is fastest and where the turbulence is low.

Furthermore, this zone is characterised by large droplet ( ) with a mean velocity

equal to . A key contribution here lies in concluding that there are two extinction zones

in the inner flame front with two different mechanisms. One is due to the flow aerodynamic and

the second is due to the large droplets may cross the flame base.

Fig. 8: (a) Mean OH-PLIF with extinction points. (b) Turbulent kinetic energy with air velocity vield. (c) Mean

diameter field and mean droplet velocity.

Frequent events of local extinction or ignition are seen at various locations in the flame. In

summary, three regions of extinction can be observed in the spray jet flame and defined here for

subsequent analysis with more details:

(i): Extinctions in the inner reaction zone close to the leading edge. These mark the extinctions

due to the droplets crossing the flame base. These large droplets are expected to greatly

influence the flame by momentum, heat and mass transfer.

(ii): Extinctions in the inner reaction zone but far away the flame base. When a vortex interacts

with the flame front, a thinning of the flame front is observed and can induce a local extinction.

(iii): Due to the trajectory and velocity of large droplets, they can cross the flame front close to

the leading edge and then the outer reaction zone. Due to a local rich fuel vapour zone around

the droplet and due to a low droplet temperature, the reaction zone is disturbed, which results in

a local extinction in the wake of the droplets.

(a) (b) (c)


3.2) Impact of aerodynamic on local extinctions

Fig. 9 : Selected sequence of HS-OH-PLIF images - time separation = 0.1 ms - Extinction due to the aerodynamic.

Red lines present the flame contour close to the extinction.

Fig. 9 presents a sequence of HS-OH-PLIF with an extinction mechanism due to the aerodynamic

(from a stable reaction condition). The local extinction appears at X=12mm and Z=28mm. A gap

is produce in the initial inner reaction zone and causes a hole which separates the inner flame

front in two reaction zones. Then due to the air velocity field, the hole, with the two reaction

zones, are convected to the top. Due to the turbulent air flow, the hole propagation rate is not


constant and can induce sometimes the closure of the extinction. It is worth noting that the flame

front (OH signal) upstream the hole is much thinner that the initial inner reaction zone. This

means that the flame is subjected to a strong stretching, which can cause the local flame

extinction. The OH signal seems becoming less intense at the end of the extinction sequence,

presumably due to increased heat losses associated with increasing scalar dissipation rate.

(Rehm and Clemens 1999) has been previously observed a correlation between the flame front

and vorticity field locations. Note that the extinction does not appear in a region where the flame

front curvature is important. It can be concluded that the curvature is not the first parameter to

flame stretch and doesn’t cause later the flame front extinction, compared to the strain rate. The

vortex dissipation rate seems relevant for the understanding of the extinction mechanisms and is

now discussed. From the tomographic images (seeding the air co-flow by olive oil Fig. 10) the

large turbulence scales with vortices are observed. Through an autocorrelation method close to

the mixing layer X=13mm and Z=30mm (Eq.7 and 8), an indication of coherence of order is

provided.

(7)

∫

(8)

Fig. 10: Autocorrelation coeficient with an image of air co-flow seeding by olive oil.

Figure. 10 presents the solution of Eq.8 with an image of coflow seeding by olive oil. The form of

the autocorrelation function is such that it decreases rapidly, after which it decays more slowly.

The integration domain for the determination of the integral length as a representative length


scale of the turbulence can be specified by a frequence way. It is a measure of the longest

correlation time between the flow velocity(vorticity) at two points in the flow field and is equal

in this case to 82Hz.

This means that at a frequency equal to 82Hz, the vortices disturbed the flame front by a high

strain rate and may cause an insufficient momentum transfer to entrain the required air for

sustaining an inner reaction zone. The flame becomes too stretched and the local mixture is

outside the required condition to keep a stable flame front, which causes a local extinction of

inner reaction zone. To validate the hypothesis that it is the large vortices which induce the local

extinctions, an extinction occurrence frequency analysis is done in the inner reaction zone. A

probe volume located in X=16mm and Z=30mm is used to determine the temporal apparition of

extinction. 𝑓 represents the distance between the probe volume and the flame front. The specific

value of 𝑓 represents the position of instantaneous flame front, whereas the negative and

positive values concern the fresh and burnt gases, respectively. The specific values of 𝑓

corresponds to the distance between the probe volume and the outer reaction zone and is

considered here as extinctions. Note that with these previous values of 𝑓 it is possible to show

the low fluctuations of the outer reaction zone. Using a power spectral density function (PSD), a

peak at 78Hz is obtained. This analysis allows concluding that the large turbulence scale

(vortices) modified the flame behavior until sometimes extinction. This analysis allows

concluding that the large turbulence scales (vortices) modify the flame behavior until sometimes

extinguish the reaction zone as illustrated in Fig. 11(c).

Fig. 11: (a) Extinction occurrence temporal analysis. (b) Extinction occurrence frequency analysis. Profile at X=16mm

and Z=30mm. (c) Schematic representation of the extinction due to aerodynamic.

(a) (b)

Hig

h

TK

E

Hig

h

TK

E

Extinction

Innerreaction

zone

Vortices

(c)


3.3) Impact of droplets on local extinctions

Fig. 12: Sequence of HS-OH-PLIF - time separation = 0.1 ms. Extinctions in the inner reaction zone close to the

leading edge. Red lines present the flame contour close to the extinction

PTV and PDA results give a very good agreement notably in the region below the stabilisation

point and in the region between the two reaction zones. As already seen in the previous section,

small droplets mainly stay in the cold central zone, where they evaporate slowly and depend on

air velocity (Fig. 8(b)) due to their low Stokes number and never cross the flame front. Small

droplets can arrive between the two reaction zones by the hole induces by local extinction in the

inner reaction zone. On the contrary, large, inertial droplets move to the external part of the

spray, crossing the flame base and reaching the hot region. The droplets which arrive and cross

the flame base have the following characteristics:

2 < 6

��~

𝑝 𝐾 (Verdier 2016)


Nevertheless, the flame induces dramatically change in the characteristics of droplets. When

they reaching the immediate vicinity of the flame base (high temperature), their temperature

increase and the strong thermal exchange imposes a fast evaporation regime resulting in a

augmentation of the gaseous fuel mass fraction and so a droplet diameter diminution. It is worth

noting that the flame decelerates the droplet as shown by (Ma and Roekaerts 2016). The mean

characteristics of droplets present between the two reaction zones are:

< 5

��~7

𝑝 𝐾 (Verdier 2016)

Figure. 12 presents a HS-OH-PLIF image sequence, where a big droplet (yellow circle) arrives

with a relative velocity in the leading edge. After the passage of the drop a local extinction,

located at the place where the drop is passed, is visualised. Unlike the rapid extinction

convection that occurs far away in the inner reaction, the closure between the two reaction zones,

to obtain a stable reaction condition, is reach more quickly in this case. This shorter time is due to

local conditions close to the stoichiometry with low turbulence. The hole in the reaction zone

indicates a local quenching due to large heat loss caused by droplet evaporation (Mercier, Orain

et al. 2007). Furthermore, the n-heptane vapour around the droplet induces locally a rich mixture

which can be above to the fuel flammability. Furthermore, the n-heptane vapour around the

droplet induced locally a rich mixture which can be superior to the fuel flammability limit. These

droplet-flame behaviors are observed also in the outer reaction zones.

Fig. 13 shows four different sequences to highlight the droplet impact on the outer reaction zone.

Further downstream, droplets that have survived the flame brush continue through the outer

reaction zone, where they disturb the flame front. In most of the cases (#1, #2 and #4), there are

not extinction in term of discontinuity in the outer OH-PLIF signal. However, the combustion

doesn’t occur in the wake of droplet, this means that the local conditions behind the droplet is

too rich to burn. The major difference between the inner reaction and the outer reaction zone is

the droplets velocity and their temperature. With a lower velocity in the hot region, the time to

cross the outer reaction is longer than for the inner reaction zone.


Fig. 13: 4 different sequences of HS-OH-PLIF - time separation = 0.1 ms. Extinctions in the inner reaction zone close

to the leading edge.

4. Acknowledgements

The financial support provided by ANR TIMBER is gratefully acknowledged.

5. Conclusion

In this paper, a detailed experimental study of turbulence-droplet-chemistry interactions events

in an n-heptane spray jet flame is presented. Although the information is qualitative, the

sequences of High Speed OH-PLIF possess sufficient spatial and temporal resolution allowing

follow the evolution of transient phenomena as local extinctions. The flame exhibits a double

structure with inner and outer reaction zones, where fuel droplets are still present and disturb

the reaction. Droplet size, droplet and carrier phase velocities are preliminary characterised by

PDA, which give information about turbulence field. Two kinds of extinction mechanism in

three different zones, two in the inner reaction (one close the leading edge and a second far away

#1

#2

#3

#4


the flame base), and in the outer reaction zone are identified. In the inner reaction zone far away

the flame base, the temporal evolution of extinction events could thus be correlated to the

behaviour of turbulent velocity field. Indeed, the turbulent flow generates vortices, close to the

shear layer, which greatly influence the flame by momentum, heat and mass transfer. Due to the

large turbulent scales, the flame becomes too stretched and the local mixture is outside the

required condition to keep a stable flame front, which causes a local extinction. The extinction

propagation rate is mainly turbulent and depends on the flow field. The first extinction

mechanism is due to the strong turbulence-chemistry interactions and appear for axial stations

upper to Z=27mm. The second mechanism is related to the droplet-chemistry interactions and

arrives in the flame base and in the outer reaction zone. The big droplets, which have a lower

temperature than in the flame, act as a temperature sink for the flame front which finally

extinguishes due to the cooling effect. In addition, due to the evaporation process, there is a fuel

vapour surrounding the droplet which increases the local equivalence ratio. In the wake of

droplet, it may be possible that the mixture is too rich to burn.

6. References

Abram, C., B. Fond, et al. (2013). "High-speed planar thermometry and velocimetry using

thermographic phosphor particles." Applied Physics B 111(2): 155-160.

Boxx, I., C. Arndt, et al. (2010). "High-speed laser diagnostics for the study of flame dynamics in

a lean premixed gas turbine model combustor." Experiments in Fluids: 1-13.

Boxx, I., M. Stöhr, et al. (2009). "Sustained multi-kHz flamefront and 3-component velocity-field

measurements for the study of turbulent flames." Applied Physics B: Lasers and Optics

95(1): 23-29.

Brevis, W., Y. Niño, et al. (2011). "Integrating cross-correlation and relaxation algorithms for

particle tracking velocimetry." Experiments in Fluids 50(1): 135-147.

Caselles, V., F. Catté, et al. "A geometric model for active contours in image processing."

Numerische Mathematik 66(1): 1-31.

Cessou, A. and D. Stepowski (1996). "Planar Laser Induced Fluorescence Measurement of [ OH]

in the Stabilization Stage of a Spray Jet Flame." Combustion Science and Technology

118(4-6): 361-381.

Cléon, G., D. Honoré, et al. (2015). "Experimental investigation of structure and stabilization of

spray oxyfuel flames diluted by carbon dioxide." Proceedings of the Combustion Institute

35(3): 3565-3572.

Cordier, M., A. Vandel, et al. (2013). "Laser-Induced Spark Ignition of Premixed Confined

Swirled Flames." Combustion Science and Technology 185(3): 379-407.


Correia Rodrigues, H., M. J. Tummers, et al. (2015). "Spray flame structure in conventional and

hot-diluted combustion regime." Combustion and Flame 162(3): 759-773.

Friedman, J. A. and M. Renksizbulut (1999). "Investigating a methanol spray flame interacting

with an annular air jet using phase-Doppler interferometry and planar laser-induced

fluorescence." Combustion and Flame 117(4): 661-684.

Han, Y., G.-B. Cai, et al. (2014). "A conditioned level-set method with block-division strategy to

flame front extraction based on." Chinese Physics B 23(5): 058901.

Hartung, G., J. Hult, et al. (2009). "Flame front tracking in turbulent lean premixed flames

using stereo PIV and time-sequenced planar LIF of OH." Applied Physics B 96(4): 843-862.

Kaiser, S. A. and J. H. Frank (2009). "Spatial scales of extinction and dissipation in the near field

of non-premixed turbulent jet flames." Proceedings of the Combustion Institute 32(2):

1639-1646.

Ma, L., B. Naud, et al. (2015). "Transported PDF Modeling of Ethanol Spray in Hot-Diluted

Coflow Flame." Flow, Turbulence and Combustion: 1-34.

Ma, L. and D. Roekaerts (2016). "Modeling of spray jet flame under MILD condition with non-

adiabatic FGM and a new conditional droplet injection model." Combustion and Flame

165: 402-423.

Malm, H., G. Sparr, et al. (2000). "Nonlinear diffusion filtering of images obtained by planar

laser-induced fluorescence spectroscopy." Journal of the Optical Society of America A

17(12): 2148-2156.

Marley, S. K., E. J. Welle, et al. (2004). "Effects of leading edge entrainment on the double flame

structure in lifted ethanol spray flames." Experimental Thermal and Fluid Science 29(1):

23-31.

Mercier, X., M. Orain, et al. (2007). "Investigation of droplet combustion in strained counterflow

diffusion flames using planar laser-induced fluorescence." Applied Physics B 88(1): 151-

160.

Perona, P. and J. Malik (1990). "Scale-space and edge detection using anisotropic diffusion." IEEE

Transactions on Pattern Analysis and Machine Intelligence 12(7): 629-639.

Rehm, J. E. and N. T. Clemens (1999). "The large-scale turbulent structure of nonpremixed planar

jet flames." Combustion and Flame 116(4): 615-626.

Slabaugh, C. D., A. C. Pratt, et al. (2015). "Simultaneous 5 kHz OH-PLIF/PIV for the study of

turbulent combustion at engine conditions." Applied Physics B 118(1): 109-130.

Verdier, M. S., Vandel, Saengkaew, Cabot, Grehan, Renou (2016). "Experimental study of local

flame structures and fuel droplet properties of a spray jet flame." Proceedings of the

combustion Institute Accepted for oral presentation.

Experimental study of local extinction mechanisms on a...

Documents

Transcript of Experimental study of local extinction mechanisms on a...