Sage Journals: Discover world-class research

Abstract

Achieving full premixing and complete evaporation in lean prevaporized premixed combustors is challenging as it depends on the spray injector characteristics, prevaporisation strategy, and flow conditions. This article experimentally explores the stability and structure of a turbulent swirling n-heptane spray flame under various degrees of prevaporization. The results show that preheating the air to 343 K and 393 K has little effect on the lean blow-off velocity, while recessing the fuel injection significantly decreases the lean stability limit. To correlate these limits, various attempts to define a Damköhler number were made, but unlike previous studies with no prevaporisation, the difficulty in defining laminar flame speed in the present case does not allow a single correlation to work for all degrees of prevaporization. Four stable cases that differ in equivalence ratio, air preheat temperature, and fuel injection recess are investigated using one-dimensional PDA, OH* chemiluminescence and CH $_{2}$ O-planar laser-induced fluorescence (PLIF). Cases without fuel injection recess or air preheat exhibit a conical-shaped heat release zone near the shear layers. Preheating the air to 393 K reduced the Sauter mean diameter, increased prevaporization, and enabled a second heat release zone downstream of the fuel injection. Recessing the fuel injection by 25 mm reduced droplet velocities and led to a semi-spherical instead of a conical heat release zone. The CH $_{2}$ O-PLIF signal without injection recess was high along the central axis and its distribution resembled that observed for spray jet flames. In contrast, with recessed spray injection, CH $_{2}$ O was mainly found outside the central recirculation zone and only appeared inside during lean blow-off; similar to previous work with premixed flames. These findings show that different methods of prevaporization, which only differ by subtle changes in droplet characteristics, strongly impact flame stability. The present data can be used for turbulent flame modelling focusing on sprays and finite-rate kinetics.

Keywords

Blow-off lean premixed prevaporized preheat spray combustion swirl

Introduction

The combustion of liquid fuels is often more complex than that of gaseous fuels because of the presence of droplets in the mixture of air and fuel vapour. When a single liquid droplet burns in air, the flame is a non-premixed flame that burns at stoichiometry in the gaseous phase surrounding the droplet.¹ Non-premixed flames, however, are often undesirable because of their high NO $_{x}$ emissions. A well-known strategy to reduce emissions from liquid-fuelled flames is the lean premixed prevaporized (LPP) combustion concept.² LPP combustion aims to completely vaporize the fuel and mix it with air at a lean equivalence ratio before combustion. This technique can reduce NO $_{x}$ emissions effectively, but it brings additional challenges such as instabilities and lean blow-out (LBO).³ The methods used for prevaporization and their effect on the LBO limit, the characteristics of the spray, and the structures of the flame will be explored systematically in this article in an effort to produce datasets for modellers and for improving our understanding.

While many lab-scale LPP experiments can achieve full prevaporization and premixing, these phenomena are often not achieved in practice even in gas turbines that claim to use LPP.⁴ One reason behind this difficulty is that full prevaporization and full premixing have opposite requirement in gas turbine combustion – the former requires the droplets to be near a heat source (i.e. the flame) while the latter needs the initial spray to be sufficiently far from the flame.⁵ As a result, many LPP combustors may actually be burning with a spray flame at various degrees of prevaporization. In these flames, some fuel will have enough time to evaporate and mix with air, which may result in a local equivalence ratio within the nominal flammability limits. Meanwhile, the remaining fuel droplets will produce a non-premixed flame. In addition to these two combustion modes, the presence of droplets has been shown to affect many flame characteristics, such as the flame speed and its propagation mechanism^6,7, ignitability⁸, NO $_{x}$ emissions⁹ and temperature profiles.¹⁰

Given the effect of droplets on the critical characteristics of the flame and the two different combustion modes, it is not surprising that the presence of droplets can affect flame stability. In premixed flames, LBO is determined by the competition between chemical kinetics and fluid mechanics time scales, which can be described by the Damköhler number.¹¹ For a spray flame, these factors contribute to LBO as well, but it is further complicated by the effect of droplets. In one of the earlier studies, Mizutani and Nakajima¹² found that adding kerosene droplets to a propane–air mixture while keeping the overall equivalence ratio fixed extended the lean stability limit and increased the burning velocity. Others have tried to understand the parameters governing the LBO limit using different properties of the fuel, such as volatility, but the results are less conclusive. Grohmann et al.¹³ tested the LBO limit of a dual-swirl burner for four different liquid fuels and found that larger fuel droplet and slow-vaporizing fuels are associated with wider lean stability limit. In contrast, at similar air inlet temperatures, both Burger and Yates¹⁴ and Rock et al.¹⁵ found that fuels which vaporize easily were the most resistant to lean blow-off.

Studies have also examined LBO of spray flames with various the degree of prevaporization. Both Grohmann et al.¹³ and Rock et al.¹⁵ found that preheating the air improved lean stability limit. The Sydney Piloted Needle Spray Burner can increase prevaporization by recessing the fuel injection upstream of the burner exit, which also was found to increase blow-off velocity.^16,17 This finding differs from when gaseous fuel was injected through the central pipe, which had an optimal recess distance that significantly extended the stability limit by increasing the inhomogeneity in the mixing of fuel and air.¹⁸ Another study on the Sydney burner found that as the jet velocity increased and the acetone flame approached blow-off, the flame appeared to transition from a more non-premixed flame to a more premixed flame.¹⁹ These studies suggest that air preheat, atomization recess, and air velocity are important parameters for prevaporization and the LBO limit.

The objective of this article is to examine the stability and structure of turbulent swirl-stabilized n-heptane spray flames with different amount of prevaporization, achieved by controlling the air preheat temperature and the recessing the fuel injection upstream of the burner exit. The lean stability limits at various air inlet temperature and fuel recess will be assessed. Then, four stable cases that differ in global equivalence ratio, air preheat temperature, and fuel recess distance respectively will be selected for further studies using direct photographs, 1D-PDA, OH* chemiluminescence, and CH $_{2}$ O-PLIF (planar laser-induced fluorescence). The NO $_{x}$ emission is also measured, but because it is less relevant for flame stability and structure, it is presented in Supplemental Materials.

Methods

Experimental conditions

Diagrams of the burner used for the experiments are shown in Figure 1. The fuel was supplied from a pressurised tank kept at 5 bar using nitrogen gas. The fuel flow rate was controlled by a needle valve and was calibrated by weighing the mass of the fuel sprayed for 5 min using a stop watch with accuracy of 0.01 s and a scale with accuracy of 0.001 g. The fuel passed through a 7-micron in-line particulate filter and was then injected into the centre of the burner using an hollow-cone atomizer from the Lee Company (IAZA1200147K, cone angle of 50 $^{\circ}$ , orifice diameter of 0.102 mm). This atomizer was chosen for these experiments because it has a small head diameter of only 3.4 cm, which allows the air to flow closely to the base of the spray and premix with the fuel.

Figure 1.

Diagrams of the burner with dimensions in millimetre (mm). The recess of the fuel and inner air (i.e. length of the premixer) can be adjusted by changing the height of the tubes in the centre. Two locations of fuel and inner air injections are used in this work: no recess (a) and 25 mm recess (b). A close-up of the cross section of the burner exit is shown in (c), and a top view of the premixer and burner exit is shown in (d).

The burner provides two separate streams for air injection. The two streams originated from a central compressed-air system with particulate and coalescent filters. The central air was then split into two streams at a T-junction. Each stream of air had its own mass flow controller, in-line heaters, and PID temperature controllers to set the flow rate and temperature. The temperatures of the two air streams were the same for all cases in this study. The inner air was injected in the axial direction without swirl. In contrast, the outer air was injected tangentially at the base of a cone-shaped premixer, which provided the swirl needed to stabilize the flame.

The relative positions of the fuel injector, inner air nozzle, and outer air annulus can be easily adjusted, which allows the degree of prevaporization to be varied. In this article, two configurations were used. Figure 1(a) shows the first configuration where the inner nozzle is flush with the exit plane, and the tip of the atomizer is 5 mm upstream of the burner exit. In the second configuration, the fuel nozzle and inner air nozzle were moved upstream of the burner exit by 25 mm, as shown in Figure 1(b). The atomizer is still 5 mm upstream of the inner air nozzle exit. This method of recessing the fuel will have some effects on the airflow since the inner and outer air can now begin to mix more upstream. This problem could have been avoided if only the atomizer was recessed while the inner air pipe exit stayed flush with the burner exit, but doing so resulted in large amounts of fuel hitting and accumulating on the inner wall of the inner air pipe during test runs, which was not desirable. Therefore in this work, recessing the spray involved moving both the fuel and inner air nozzles upstream. In all cases, the burner was enclosed by four quartz plates with the dimension of 150 mm $\times$ 80 mm $\times$ 3 mm for optical access.

Four stable flames were selected for detailed measurements, and their conditions are listed in Table 1. The fuel flow rate was 0.102 g/s for all cases. Case A is the base case that has no preheating and no recess (the burner geometry shown in Figure 1(a)). Case B also has no preheating or recess, but it has higher air velocity and lower global equivalence ratio, $ϕ_{g}$ , than Case A does. Case A+Preheat has the same flow rates and air splits as Case A but both streams of air are preheated to 393 K. Case A+Recess has the same $ϕ_{g}$ as Case A and Case A+Preheat, but the inner air and fuel pipes are moved 25 mm upstream (the burner geometry shown in Figure 1(b)). The air split of Case A+Recess is also different because the flame is unstable at the air split of Case A. The outer air velocity, inner air velocity and bulk velocities are denoted by $U_{o}$ , $U_{i}$ and $U_{b}$ , respectively. They are calculated using the equations below

U_{o} = \frac{4 {\dot{m}}_{o}}{(d^{2} - d_{o}^{2}) π ρ (T)}

(1)

U_{i} = \frac{4 {\dot{m}}_{i}}{(d_{i}^{2} - d_{f}^{2}) π ρ (T)}

(2)

For cases without recess

U_{b} = \frac{4 ({\dot{m}}_{o} + {\dot{m}}_{i})}{(d^{2} - d_{o}^{2} + d_{i}^{2} - d_{f}^{2}) π ρ (T)}

(3)

For cases with recess

U_{b} = \frac{4 ({\dot{m}}_{o} + {\dot{m}}_{i})}{d^{2} π ρ (T)}

(4)

where

{\dot{m}}_{i}

is the inner air mass flow rate,

{\dot{m}}_{o}

is the outer air mass flow rate and

ρ

is the density. The definition and value of the various diameters are illustrated in Figure 1(c).

Table 1.

The conditions for the four main flames explored. The fuel flow rate is 0.102 g/s for all cases. $U_{o}$ , $U_{i}$ and $U_{b}$ refer to the outer air velocity, inner air velocity and bulk air velocity, respectively. ${\dot{m}}_{i}$ is the inner air mass flow rate, and ${\dot{m}}_{o}$ is the outer air mass flow rate. The inner and outer air streams have the same temperature.

Case	$T_{air}$ [K]	Recess [mm]	$ϕ_{g}$	$U_{o}$ [m/s]	$U_{i}$ / $U_{o}$	$U_{b}$ [m/s]	${\dot{m}}_{i}$ / ${\dot{m}}_{o}$
A	293	0	0.60	9.8	0.72	9.0	0.28
B	293	0	0.44	13.4	0.75	12.3	0.29
A+Preheat	393	0	0.60	13.1	0.72	12.0	0.28
A+Recess	293	25	0.60	-	-	7.5	0.87

Experimental procedure and diagnostics

Droplet size and axial velocity were measured using a 1D LDA and PDA system from DANTEC. The transmitting probe consisted of two 514 nm laser beams. One of the two beams has been phase shifted by 40.00 MHz in a Bragg cell to allow measurement of negative velocity. The beams intersect in the vertical direction (axial) to create a measurement volume in the burner. The collection probe was placed at 70 $^{\circ}$ off-axis from the forward scattering direction so that refraction is the dominant light scattering mode. A photo of the setup and more details about the system have been discussed by Wang.²⁰ The system can measure droplets of 0–160.49 $μ$ m in diameter. For each position, data was either collected for 30 s or 20,000 samples. The validation rate was above 90%. Sampling rate was above 10 kHz at dense locations (e.g. just downstream of the burner exit), but the rate decreased as the measurement location moves away from the burner exit. The transmitting and receiving probes were mounted on a 3D traverse to easily measure multiple locations.

High-speed OH* chemiluminescence images were taken at 5 kHz by a Photron SA1.1 monochrome high speed CMOS camera with 1024 $\times$ 1024 pixel $^{2}$ coupled with an IRO intensifier from LaVision that has a spectral range of 190–800 nm. A Cerco 2178 UV lens and a $310 \pm 10$ nm Andover filter were placed on the intensifier. The camera was gated at $180 μ$ s, and the aperture was f/4 for all cases. For each flame, images were recorded for 1 s, which gives 5000 images. Inverse Abel Transform was then applied to the averaged image to approximate the locations of heat release (HR) zone qualitatively.

The low-speed CH $_{2}$ O-PLIF measurement taken at 10 Hz used a 355 nm laser beam with a pulse energy of about 100 mJ from a low-speed Nd:YAG laser (Continuum Surelite, Electro-optics, INC) to excite multiple transitions within the 4 $_{0}^{1}$ band of the $A^{2} A^{1}$ $\leftarrow$ $X_{1} A^{1} 4^{1}$ system of CH $_{2}$ O. The laser beam was directed towards the burner using two dichroic mirrors. Then it passes through a 38 mm cylindrical concave lens and a 500 mm spherical biconcave lens to form a well-expanded laser sheet of $52 \times 0.25$ mm $^{2}$ . The fluorescence signal is captured by an intensified CCD Nanostar camera from LaVision gated at 100 ns with a Zeiss 100 mm f/2 lens and a multi-pass band filter (FF01-CH2O-50). Photos and more details on the laser system have been discussed by Ciardiello.²¹ When postprocessing the CH $_{2}$ O-PLIF images, the laser background and flame background were subtracted from each image. Since doing so meant subtracting the dark background twice, the dark background was then added to each image. A median filter with a kernel size of 5 $\times$ 5 pixels $^{2}$ was then applied to remove the outliers. Laser profile was not corrected because the laser sheet was well expanded as discussed by Pathania et al.²²

Simulation of droplet evaporation

The evaporation of a heptane droplet is simulated to compare with the experimental data and to better understand prevaporisation. Details about the numerical method can be found in previous work.^23,20 The code simulates a stationary heptane droplet in stationary air at 1 bar without considering any chemical reactions. The initial temperature of the fuel droplet is 293 K for all cases, and the initial temperature of the air is either 293 K or 393 K. Since the simulations are not the main focus of this paper, the results are presented in Supplemental Materials and are only briefly discussed in the main body of this article.

Results and discussion

Lean stability limit

The stability limits of the burner with and without fuel injection recess and at air preheat temperatures of 293 K, 343 K and 393 K are presented in Figure 2 for a range of inner air flow rates. For each combination of recess and air preheat temperature, the blow-off condition was found by first igniting the flame at flow rates far from blow-off. Then, with the fuel flow rate fixed at 0.102 g/s and inner air flow rate fixed at a specific value, the outer air velocity was increased in increments of approximately 0.3 m/s every 20–30 s until the flame extinguished. The procedure was repeated for at least three times for each condition. The maximum standard deviation of the blow-off velocity is 8%, though it is below 3% on average.

Figure 2.

The blow-off bulk velocity, $U_{b, LBO}$ , is plotted as a function of the air velocity at the inner air nozzle exit, $U_{in}$ (left), and the stable region being below the line. The blow-off global equivalence ratio, $ϕ_{g, LBO}$ , is plotted as a function of the equivalence ratio based on inner air, $ϕ_{in}$ , and the stable region being above the line. The conditions of the four stable flames examined in detail in this paper are indicated with filled markers. Case A and Case A+Preheat have the same $ϕ_{in}$ and $ϕ_{g}$ .

The effects of the flow and mixture conditions on the LBO of the flame are shown in Figure 2. The left subplot shows the bulk velocity at LBO, $U_{b, LBO}$ , as a function of air velocity at the inner air nozzle exit, $U_{in}$ . As means of evaluating the impact of the injected mixture characteristics, the right subplot presents the same results but in terms of the global equivalence ratios at LBO, $ϕ_{g, LBO}$ , versus the equivalence ratio calculated by considering only the inner air, $ϕ_{in}$ . For all combinations of air preheat temperature and burner geometry, having a higher $U_{in}$ or lower $ϕ_{in}$ resulted in higher $U_{b, LBO}$ . This trend is likely due to the effect of swirl. Results from a previous study using a burner with similar geometry (i.e. non-swirling, axially-injected inner flow and tangentially injected outer flow) suggest that there exists an optimum level of swirl for stability.²⁴ Thus, when the velocity of the non-swirling inner air is increased, more swirling outer air can be added before passing the optimum level of swirl and blowing off.

Increasing the air preheat temperature led to higher $ϕ_{g, LBO}$ , but it is also important to look at $U_{b, LBO}$ because $U_{b, LBO}$ takes the effect of air preheat on density into account, whereas $ϕ_{g, LBO}$ doesn’t. When the LBO limit is plotted in terms of $U_{b, LBO}$ , the differences due to air preheat appear smaller. Air preheating only slightly decreased $U_{b, LBO}$ when the fuel injection was not recessed, while it had a negligible effect when fuel injection was recessed. In contrast, recessing the fuel injection significantly reduced $U_{b, LBO}$ . The opposite trend was observed in the experiments of the Sydney Needle Spray Burner.^16,17 These contrasting findings could be due to differences in the burner geometry, since the Sydney burner is a pilot-stabilized jet flame with a fuel needle set at a recessed position, whereas the present burner is swirl-stabilized and also recesses the fuel atomization with the inner air nozzle, which affects the aerodynamics and point of attachment of the flame.

To further understand the blow-off limit and compare it with existing studies in the literature, we can try to correlate these limits with global correlations. It is also vital to explore these correlations for LBO because this is what engineers would attempt to do when trying to design a new burner based on LPP ideas. Therefore, it is useful to see not only how it works, but also how one might estimate the parameters going into it. One such correlation is the one first proposed by Radhakrishnan et al.²⁵ and originally used for turbulent bluff body premixed flames, but also applied for in studies with heptane spray flames.^26,27 By definition, blow-off is expected to occur when the inverse of the Damköhler number (Da) of the flow exceeds a critical value, $R$ :

\frac{1}{Da} = \frac{τ_{c}}{τ_{e}} = {[\frac{ν U_{LBO}}{L / S_{L}^{2}}]}^{1 / 2} > R

(5)

where

U_{LBO}

is the velocity at LBO and

L

is the characteristic length of the recirculation zone that is taken here as the outer diameter of the outer annulus (20 mm).

τ_{c}

is a characteristic chemical time,

τ_{e}

is the characteristic flow time,

ν

is the kinematic viscosity, and

S_{L}

is the laminar flame speed. Following the literature reviewed by Cavaliere et al.,²⁶ the kinematic viscosity for spray flames was evaluated at a temperature halfway between the reactant and adiabatic flame temperature. The value of

S_{L}

that must be input in the correlation that then defines the chemical timescale needs discussion. For fully premixed systems, the definition is unambiguous. For spray flames from pressure atomisers, which gives a conventional non-premixed system, using

S_{L}

evaluated at stoichiometry gave a reasonable degree of collapse of the blow-off data.^26,27 In the present case, where we have significant prevaporization, various methods of calculating

S_{L}

are used and compared.

In first method of calculating the flame speed, the mixture is evaluated at stoichiometry; this approach was taken in previous studies with spray flames.^26,27 The critical value of 1/Da as a function of $ϕ_{in}$ in this method is summarized in Figure 3(a). The critical value found in the present work is 0.68, with a relatively low standard deviation of 0.15. The difference between the minimum and maximum values relative to the mean is about twice of that found in experiments from Yuan et al.,²⁷ showing the greater variation between LBO events in the present flame. The two common methods of increasing prevaporisation – that is, increasing air temperature and recessing fuel injection – led to a lower critical value. This trend is the opposite of that observed by Cavaliere et al.,²⁶ who compared LBO of premixed and nonpremixed methane flames as well as heptane spray flames. In their study, the premixed flame had the highest critical value, and the spray flame had the lowest. The contrasting results between this experiment and previous studies could be associated to the evaluation of $S_{L}$ at stoichiometry, which may not be appropriate for spray flames with high degree of prevaporisation.

Figure 3.

The critical 1/Da for the blow-off correlation first proposed by Radhakrishnan et al.²⁵ The flame speed used to calculate 1/Da is evaluated at stoichiometry (a), at the global equivalence ratio $ϕ_{g}$ (b) or using equation (6) (c). The stable regions are below the lines. The conditions of the four stable cases explored in detail in this study are indicated with filled markers.

At the limit of full prevaporization and premixing, it would be incorrect to take the stoichiometric condition to evaluate the flame speed. Instead, $S_{L}$ should be evaluated at the global equivalence ratio, $ϕ_{g}$ . Still, when the critical value of 1/Da is evaluated using this second approach, as presented in Figure 3(b), the range of 1/Da becomes significantly wider, hence less representative to mark LBO. This occurs due to the low $ϕ_{g}$ that results in very low $S_{L}$ values in some cases. Yet, the flames investigated in the present work have a considerable amount of fuel vapour due to the prevaporisation strategies used, while also containing droplets of a range of sizes at the flame front, which need to be accounted for at the evaluation of $S_{L}$ .

Finally, a third method of calculating $S_{L}$ is considered, which uses equation (6). This correlation for the flame speed in sprays was first proposed by Ballal and Lefebvre,²⁸ and it considers the effect of evaporation rate in addition to the chemical reaction rate.⁶ It is important to consider droplet effects in the estimation of flame speed because droplets are known to have a first-order effect on flame speed and the flame’s resilience to stretching.^7,29,30 In equation (6), $α_{g}$ is the thermal diffusivity of the gas phase, $S_{L, g}$ is the gaseous $S_{L}$ evaluated at $ϕ_{g}$ , $λ_{eff}$ is an effective evaporation constant that is obtained from Myers and Lefebvre³¹ at 1200 K. $C_{1}$ is $d_{20} / d_{32}$ , and $C_{3}$ is $d_{30} / d_{32}$ , where $d_{32}$ is the SMD at the start of the premixer, $d_{20}$ is the mean surface area diameter and $d_{30}$ is the mean volume diameter. These values of the droplet diameter are estimated using the 1D PDA data. $Ω$ is the fraction of fuel in the gaseous phase, which is here estimated using the diameter and velocity data from the PDA measurement as well as the $d^{2}$ -law that has been shown to give accurate estimates.³²

S_{L} = α_{g} [\frac{C_{3}^{3} (1 - Ω) d_{32}^{2} α_{g}}{C_{1} λ_{eff}} + \frac{α_{g}^{2}}{S_{L, g}^{2}}]^{- 0.5}

(6)

The critical 1/Da using the third method is shown in Figure 3(c). The flame speed is slightly higher than when it is evaluated based on

ϕ_{g}

, but the difference is small. As a result, the critical 1/Da is similar to the value in Figure 3(b). These values are significantly greater than the results from studies that have also used equation (6). For instance, spherically propagating flames were found to extinguish at critical values of between 0.2 and 0.6.³³

The attempts in this section show that the LBO limit of spray flames with prevaporization could not be easily predicted with global correlations. The complexity of prevaporization and presence of a polydisperse spray makes it especially difficult to evaluate flame speed. Furthermore, these results indicate that local droplet-induced effects – thus far not accounted for in LBO-prediction correlations – must also contribute significantly to flame extinction and must be accounted for in future works.

Direct flame photos

To better understand the effect of prevaporization, four stable flames that differ in global equivalence ratio, air preheat temperature, and fuel injection recess distance respectively are investigated in more detail in the remaining of this paper. The four flame conditions are summarized in Table 1, while their photographs in the visible spectrum are shown in Figure 4. Imaged in the visible spectrum or to the naked eye, lean premixed hydrocarbon flames are characterised by a typical blue hue while soot has a distinct orange colour. As a first indication of the characteristics of each of the flames investigated in detail here, one can infer their degree of prevaporization and premixing based on direct photographs. Case A, being the base case of this study, has no air preheat or fuel injection recess. Even though it has a lean global equivalence ratio of 0.6, it is still marked by significant amount of soot, which is formed at around 1 $d$ downstream of the injection location. When the air velocity is increased from Case A to Case B, the global equivalence ratio reduced to 0.44, which seems to reduce soot significantly. Less soot was also observed when the equivalence ratio remained at 0.6 but the air was preheated to 393 K in Case A+Preheat. Lastly, when the fuel injection is recessed by 25 mm in Case A+Recess, no orange hue is seen in the long-exposure photo, suggesting that Case A+Recess is the closest to a prevaporized and premixed flame. It is also worth noting that some lift-off of the flame is also observed in this case because the inner air pipe is recessed with the fuel pipe, in contrast to the others (A, B, A+preheat) where the flame attached to exit of the inner air pipe. This lack of attachment point could have contributed to the narrower lean stability of the cases with recess, discussed in the previous section. Overall, one can infer that the degree of prevaporization and premixing seems to increase from Cases A to B, to A+Preheat and A+Recess, respectively. The characteristics of the spray in the next section will be discussed in this context.

Figure 4.

Direct photos of four stable cases. The top photos are taken at short exposure (shutter speed of 4 ms, f/5.6). The bottom photos are taken at long exposure (shutter speed of 769 ms, f/5.6).

These photos can be compared against that of other confined, turbulent, swirling, spray flames. DLR has a confined, swirl-stabilized burner with a pre-filming airblast atomizer.¹³ Their reference flames look less sooty, wider, and shorter than the more non-premixed cases presented in this paper. The flames also seem to be unattached, like Case A+Recess. Their flames do not have droplets near the central axis, whereas the burner in this article does, which will be shown in the next section on droplet velocity and size. This difference in fuel distribution may have contributed to the different flame appearance. The Cambridge swirl flame burner is another confined spray burner.^26,27 In addition to using swirl to stabilize, these flames are also stabilized by a bluffbody in the centre of the burner exit. The atomizer is placed in the centre of the bluffbody, while the air flows around the outside of the bluffbody. As a result, these flames have an inner cone that aligns with the spray cone, and an outer cone along the shear layer and is attached to the corner of the bluff-body. This double cone structure is not observed in the flames in the current work. Their bluffbody likely also acts as a physical barrier between the fuel and air stream, limiting the mixing between the two streams. This might have caused the flames to appear sooty like the less-premixed cases in this article.

Droplet velocity and size

Results and analysis from the PDA measurements are presented in this subsection to understand droplet size and velocity in the four stable cases. Figure 5 presents the normalized average velocity, RMS of the normalized fluctuating velocity and the SMD ( $d_{32}$ ) at three normalized axial distances of $z / d$ , where $z$ is the axial distance, and $d$ is the outer diameter of the outer annulus (20 mm). For all three axial positions, measurements were conducted from $x / d = - 0.75$ to $x / d = 0.75$ in steps of 0.05. If less than 50 samples were measured in 30 s, it is assumed that droplet presence at that location is negligible, and therefore the figure will not have a data point at that location. Note that the velocity measurements for different flames are normalized by different values, as listed in the caption of Figure 5. The velocity measurements are normalized by $U_{o}$ (outer air velocity) for Cases A, B, and A+Preheat and by $U_{b}$ (bulk air velocity) for Case A+Recess.

Figure 5.

The distribution of normalized average velocity, normalized RMS of the fluctuating velocity and SMD of droplets at $z / d = 0.55, 1.0, 1.5$ . The velocity measurements are normalized by the average outer air velocity of $U_{o} = 9.8, 13.4$ and $13.1$ m/s for Cases A, B and A+Preheat, respectively. Case A+Recess is normalized by its bulk air velocity of $U_{b} = 7.5$ m/s.

Case A has the highest average velocity near the central axis at all three heights, which is likely because the droplets here are being carried by the inner air jet. As one moves away from the central axis, the velocity decreases and reaches a minimum at around $x / d = \pm 0.5$ , which should be the approximate location of the recirculation zone. The SMD of droplets at all locations measured are approximately between 30 and 40 $μ$ m. A more detailed discussion of the droplet profiles can be found in Supplemental Materials.

Case B has higher air velocities than Case A does, which decrease the residence of the droplets and should result in less prevaporization. Given that the normalized droplet velocity and size profiles are similar to those observed for Case A, however, this effect on prevaporization should be small.

The air preheat temperature is increased by 100 K in Case A+Preheat. Preheating the air increases the bulk air velocity that in turn increases the droplet velocity, as evident from the comparison of droplet velocity PDFs in the left column of Figure 6. The higher air velocity reduces the residence time for the droplet, which alone should decrease prevaporization, but the higher air temperature also speeds up droplet evaporation. The net result is an increase in prevaporisation. This is apparent from its significantly smaller SMD at all locations. The PDFs of droplet diameter also peaks at a smaller diameter than the case without preheat, which indicates that preheating the air results in a higher fraction of small droplets.

Figure 6.

The probability density function (PDF) of droplet velocity (left) and diameter (centre) as well as the liquid volume contained in droplets of a particular diameter, $V_{d}$ , normalized by the total liquid volume (right), $V_{tot}$ , at $z / d = 0.55$ and $x / d = 0$ for Cases A, A+Preheat and A+Recess.

Recessing the atomization in Case A+Recess increases the droplet residence time and allows more fuel to evaporate before reaching the flame. Droplets formed from pressure atomizers usually have high initial velocity. When the atomizer is recessed, the droplets will have slowed down more by the time they reach the burner exit. This lower droplet velocity also contributes to a higher residence time and more prevapoization. Although its SMD values in Figure 5 at most locations are similar to those in Case A, the enhanced prevaporization is still evident: at $z / d = 1.5$ , droplets are only found for a span of $x / d = 0.75$ , whereas droplets are present for a span of $x / d = 1.15$ in Case A, which suggests that the droplets do not penetrate as far with recessed atomization. The droplets at $z / d = 1.5$ also have slightly higher SMD than the droplets upstream. The increase in SMD value as one moves downstream could indicate that most smaller droplets have already evaporated or been consumed.

The effect of fuel recess is also evident from the right column of Figure 6, which shows the liquid volume contained in droplets of different diameter ( $V_{d}$ ) normalized by the total liquid volume ( $V_{tot}$ ). These graphs highlight that the large droplets, which are too low in number to be noticeable in PDFs of droplet diameter, are nonetheless important because they carry the most amount of fuel and take the longest to evaporate completely. Case A has more fuel contained in large droplets than Case A+Recess. The outliers with large diameter can also be observed in Figure 7 that plots the velocity and diameter of every droplet in the sample at $x / d = 0$ . Case A has several droplets with diameter of above $100 μ$ m, which both Cases A+Recess and A+Preheat have very few of. Figures 6 and 7 together show that increasing the air preheat temperature decreases the SMD values by shifting the peak of the diameter PDF to the smaller side, whereas recessing the fuel injection results in lower droplet momentum and fewer droplets with extremely large diameter.

Figure 7.

The velocity and diameter of every droplet in the sample at $z / d = 0.55, 1.0, 1.5$ and $x / d = 0$ for Cases A, A+Preheat and A+Recess.

The droplet size measurements agree with the results from the simulation of droplet evaporation at conditions of Cases A, A+Preheat and A+Recess, which are presented in Supplementary Materials. The droplets simulated at Case A conditions have the largest diameters and supports the hypothesis that Case A is the closest to a non-premixed flame. The droplets at the conditions of Case A+Preheat have significantly lower SMD values than Cases A and A+Recess, which is also found in the PDA measurements. Recessing the fuel injection decreases droplet size in a different way: by giving the evaporation a head start and increasing the droplet residence time. These factors slightly decreases the droplet diameter, but they are not as effective preheating the air. This finding is consistent with the experimental observation that Cases A and A+Recess have similar SMD values.

HR location

OH* chemiluminescence is used qualitatively to identify the HR locations and flame shapes. The average OH* chemiluminescence after applying inverse Abel transform are presented in Figure 8. The HR zone for Case A is a hollow cone that corresponds to the shear layers and sits in the outer air annulus. Its shape resembles that of other swirling spray flames on burners of similar geometry.³⁴

Figure 8.

Mean images of OH* chemiluminescence after inverse Abel transform for the four stable cases.

When the air velocity is increased in Case B, the HR zone remained cone-shaped but the cone is narrower than in Case A. The narrow cone corresponds to the narrower flame photographed in Figure 4 and is presumably caused by the higher air velocities. Case B also has less OH* signal around the central axis than Case A has because the higher flow inner air flow rates leads to lower residence time and less chemical reactions occurring around the central axis.

Increasing prevaporization altered the HR zones significantly. Case A+Preheat still has a conical HR zone near the shear layers but it has a second HR region that resembles the shape of an elongated Bunsen flame on the central axis. The OH* signals from the two HR zones are similar in strength. Given that this case has smaller droplets, one possible explanation for the inner HR zone is that this case has enough prevaporization and premixing between the fuel vapour and the inner air stream to achieve significant HR downstream of the inner air exit. On the other hand, the remaining droplets still burn near the shear layer as they do in Cases A and B, and thus the outer conical HR zone is still observed. It’s worth noting that dual reaction zones have been observed before using OH-PLIF for ethanol spray flames on two versions of the Sydney piloted spray burner.^19,17 They saw an inner reaction zone due to the premixing between fuel vapour and carrier air, and an outer reaction zone due to the pilot heat, which has some similarity with the dual HR zones of Case A+Preheat.

The HR zone for Case A+Recess is different from the other three cases. The conical HR zone near the shear layer is not observed; instead, the flame has a circular-shaped HR zone significantly thicker than other cases just downstream of the burner exit. The large difference observed between this case and the cases without recess could be due to spray momentum. The droplets in Case A+Recess have the lowest velocities and thus get trapped by the swirl more easily, which results in a thicker HR zone downstream of the burner exit. Additionally, recessing the fuel injection promotes premixing and prevaporisation by increasing the droplet residence time. As a result, OH* signal is observed near the central axis downstream of the inner air exit, like in Case A+Preheat.

CH $_{2}$ O-PLIF

CH $_{2}$ O-PLIF is useful for identifying regions of preheat and partially burnt products. This section compares the CH $_{2}$ O-PLIF images of the four stable cases and during the blow-off of some cases. Figure 9 shows three instantaneous images of CH $_{2}$ O-PLIF for each case. The majority of images have high-intensity spherical dots. These are observed even after the flame has blown off but the spray is still on, such as in the last image of Figures 10 and 11. This issue has been observed before in the CH $_{2}$ O-PLIF measurement of heptane spray flames³⁵ and could be due to stimulated Raman scattering.¹⁷ Given the purpose of the CH $_{2}$ O-PLIF measurement here is to compare different cases, the instantaneous images can still be used to compare the flames qualitatively because these spherical dots do not drastically alter the overall shape of the CH $_{2}$ O-PLIF zones. The average images, however, are not meaningful due to this interference from the droplet and will not be shown.

Figure 9.

Instantaneous images of CH $_{2}$ O-PLIF for the four stable cases. Each image has been normalized by its peak value.

Figure 10.

Sequence of CH $_{2}$ O-PLIF images for Case A before the flame blows off.

Figure 11.

Sequence of CH $_{2}$ O-PLIF images for Case A+Recess during blow-off transient.

CH $_{2}$ O-PLIF signals are observed around the central axis for Case A. This makes sense given this is the region where the fuel droplets begin oxidizing. But it is odd that the strongest signals seem to appear further downstream. Given that CH $_{2}$ O is usually found in fuel-rich regions, unless there is significant amount of partially oxidized fuel leaving the burner, part of the signals could be due to soot or PAH fluorescence. Because CH $_{2}$ O- and PAH-PLIF have overlapping fluorescence spectra, their signals are difficult to distinguish for sooty spray flames.^36,37 Therefore, the regions that correspond to the orange sooty regions in the direct photos – the most intense regions – are likely contaminated, which is approximately downstream of $z / d = 1$ .

Case B also has the strongest CH $_{2}$ O-PLIF signals downstream of the fuel injection. The shape is narrower, which is consistent with the narrower flame appearance in the direct photo and the conical HR zone from OH* chemiluminesence. The CH $_{2}$ O layer for Case A+Preheat looks similar to the layers in Cases A and B. Less bright fragments are observed though, because the flame is less sooty. The shapes of the CH $_{2}$ O layer for Cases A, B and A+Preheat resemble those observed in the combustion of fuel jet, in which CH $_{2}$ O is usually found downstream of the fuel injection point.^17,36,38,39

Case A+Recess has a different CH $_{2}$ O-PLIF structure than the other three cases. While CH $_{2}$ O is still observed immediately downstream of the burner exit, no CH $_{2}$ O signal is observed near the central axis further downstream. CH $_{2}$ O is also found in the side recirculation zone, suggesting that there is a build up of preheated gas and partially burnt products. The CH $_{2}$ O profiles for Case A+Recess resembles those from confined swirling premixed flames, which typically find CH $_{2}$ O outside the central recirculation zone.⁴⁰

Given that recessing the fuel had large effect on the LBO limit, we can compare the CH $_{2}$ O-PLIF during blow-off for Cases A and A+Recess. Figure 10 shows the evolution of CH $_{2}$ O-PLIF right before Case A blows off in steps of 100 ms. From Figure 10(a) to (c), the shape still resembles that of the stable case. But the CH $_{2}$ O-PLIF signals disappears between Figure 10(c) and (d), which could not be captured because of the low speed of the laser system. Case A+Recess shows a different pattern. While the start of the sequence in Figure 11(a) resembles the stable flame of Case A+Recess, strong CH $_{2}$ O signals are observed in the central recirculation zone in Figure 11(b). The signal gradually diminishes and the flame disappears in the last image of this sequence.

Even though the images contain interference, it is clear that the change in CH $_{2}$ O-PLIF signals is different during the LBO transient of Cases A and A+Recess. Comparing the sequences for the two cases shows that the blow-off process for the recessed case is much longer. Previous studies^41,42 have observed a key difference in CH $_{2}$ O-PLIF between non-premixed and premixed flames during LBO transients: wide and relatively uniform CH $_{2}$ O signals are observed for premixed flames but not for non-premixed flames. This difference is also observed in the current work. Case A exhibits features of a non-premixed flame, but the appearance of CH $_{2}$ O in the recirculation zone as the Case A+Recess approaches blow-off is typical of premixed flames.

Conclusions

This study investigated the stability and structure of turbulent swirling n-heptane flames with different degrees of prevaporization, which was varied by changing the air preheat temperature and the recess distance of the fuel injection. Recessing atomization by 25 mm and decreasing the inner air velocity both reduced the blow-off velocity. In contrast, preheating the air to 343 K or 393 K had little to no effect on the LBO limit. A correlation based on the Da number was applied to the data, but it was not able to accurately predict the blow-off limit. The results also did not agree with the trends from previous studies that used the same correlation. These findings suggest that the lean stability limit of spray flames subjected to prevaporization cannot be predicted using such simple correlations.

To understand how different variables affect the behaviour of the flame, four stable flames that differed in global equivalence ratio, air preheat temperature and fuel injection recess distance were studied in more detail using diagnostics. The PDA results showed that preheating the air decreased the SMD by shifting the droplet diameter PDF to the smaller side. Although the SMD values remained similar when fuel injection was recessed, fewer droplets with extremely large diameters were still found in the sample. Despite their reduced number, those are expected to heavily impact the flame front locally due to their size and significant amount of fuel released, which may partly explain the results concerning the Da-based correlation for LBO prediction.

All cases without recessed atomization had conical HR zones near the shear layers as revealed by OH* chemiluminescence. When the air was preheated to 393 K, an additional HR zone downstream of the fuel injection was enabled by the increased prevaporization and premixing between the fuel vapour and the inner air stream. Also, due to increased prevaporization, an HR zone further downstream of fuel injection was observed with recessed fuel injector. Additionally, the droplets in the recessed case had less momentum by the time they reached the burner exit, thus getting trapped by the swirl more easily and, in turn, resulting in another thick HR zone close to the burner exit. Regardless of the air preheat temperature, strong CH $_{2}$ O-PLIF signals were found downstream of the fuel injection point in all cases without recessed nozzle, a feature characteristic of non-premixed flames. When the atomizer was recessed, however, the CH $_{2}$ O-PLIF signals of the stable flame and blow-off transient resembled those of a premixed flame.

These experiments show that different methods of prevaporisation, which differ by small differences in droplet profiles, significantly affect the flame structures and their blow-off mechanisms. Ultimately these parameters affect the stability limit of the burner and contribute to the difficulty in predicting the blow-off conditions of prevaporised flames. The data are useful for validating turbulent combustion theories as they focus on challenging-to-model phenomena and will be available in the Cambridge Swirl Flame Database.⁴³

Supplemental Material

sj-zip-1-scd-10.1177_17568277231159173 - Supplemental material for Stability and structure of lean swirling spray flames with various degrees of prevaporization

Supplemental material, sj-zip-1-scd-10.1177_17568277231159173 for Stability and structure of lean swirling spray flames with various degrees of prevaporization by Jiayi Wang, Pedro M. de Oliveira, Rohit Singh Pathania, Ingrid El Helou and Epaminondas Mastorakos in International Journal of Spray and Combustion Dynamics

Supplemental Material

sj-sj-pdf-2-scd-10.1177_17568277231159173 - Supplemental material for Stability and structure of lean swirling spray flames with various degrees of prevaporization

Supplemental material, sj-sj-pdf-2-scd-10.1177_17568277231159173 for Stability and structure of lean swirling spray flames with various degrees of prevaporization by Jiayi Wang, Pedro M. de Oliveira, Rohit Singh Pathania, Ingrid El Helou and Epaminondas Mastorakos in International Journal of Spray and Combustion Dynamics

Footnotes

Acknowledgements

The authors kindly acknowledge Dr. Jili Wei, Frederick Bertani, and Manar Almazrouei for their assistance during the PDA measurement campaign.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article.

ORCID iDs

Jiayi Wang

Pedro M. de Oliveira

Epaminondas Mastorakos

Supplemental material

Supplemental material for this article is available online.

References

Giusti

Sidey

Borghesi

et al. Simulations of droplet combustion under gas turbine conditions. Combust Flame 2017; 184: 101–116.

Lefebvre

Ballal

. Gas turbine combustion: alternative fuels and emissions. 3rd ed. CRC Press Taylor & Francis Group, 2010.

Xiao

Huang

. Lean blowout limits of a gas turbine combustor operated with aviation fuel and methane. Heat Mass Transf 2016; 52: 1015–1024.

Bolszo

Mauzey

McDonell

et al. Experimental Investigation of Liquid Fuel Atomization, Mixing and Pollutant Emissions for a 30 kW Gas Turbine Engine. In Proceedings 18th ILASS- Americas Conference. Irvine, CA, 2005.

Palies

. Premixed combustion for combustors. In Stabilization and Dynamic of Premixed Swirling Flames. 2020. pp. 57–103.

Neophytou

Mastorakos

. Simulations of laminar flame propagation in droplet mists. Combust Flame 2009; 156: 1627–1640.

de Oliveira

Mastorakos

. Mechanisms of flame propagation in jet fuel sprays as revealed by OH/fuel planar laser-induced fluorescence and OH* chemiluminescence. Combust Flame 2019; 206: 308–321.

de Oliveira

Allison

Mastorakos

. Ignition of uniform droplet-laden weakly turbulent flows following a laser spark. Combust Flame 2019; 199: 387–400.

Baessler

Mösl

Sattelmayer

. NO

_{x}

emissions of a premixed partially vaporized kerosene spray flame. J Eng Gas Turbine Power 2007; 129: 695–702.

10.

Ahmadi

Chrigui

Sadiki

et al. Effect of Fuel-Evaporation Zone Length on Spray Combustion in a Gas Turbine Combustion Chamber. In 23rd Annual Conference on Liquid Atomization and Spray Systems. Brno, Czech Republic.

11.

Shanbhogue

Husain

Lieuwen

. Lean blowoff of bluff body stabilized flames: scaling and dynamics. Prog Energy Combust Sci 2009; 35: 98–120.

12.

Mizutani

Nakajima

. Combustion of fuel vapor-drop-air systems: Part I – open burner flames. Combust Flame 1973; 20: 343–350.

13.

Grohmann

Rauch

Kathrotia

et al. Influence of single-component fuels on gas-turbine model combustor lean blowout. J Propuls Power 2018; 34: 97–107.

14.

Burger

Yates

. Influence of fuel physical properties and reaction rate on threshold heterogeneous gas turbine combustion. Proceedings of ASME Turbo Expo 2012; 2: 63–71.

15.

Rock

Chterev

Emerson

et al. Liquid fuel property effects on lean blowout in an aircraft relevant combustor. J Eng Gas Turbine Power 2019; 141.

16.

Lowe

Kourmatzis

Masri

. Turbulent spray flames of intermediate density: stability and near-field structure. Combust Flame 2017; 176: 511–520.

17.

Singh

Juddoo

Kourmatzis

et al. Heat release zones in turbulent, moderately dense spray flames of ethanol and biodiesel. Combust Flame 2020; 220: 298–311.

18.

Meares

Masri

. A modified piloted burner for stabilizing turbulent flames of inhomogeneous mixtures. Combust Flame 2014; 161: 484–495.

19.

Masri

Gounder

. Turbulent spray flames of acetone and ethanol approaching extinction. Combust Sci Technol 2010; 182: 702–715.

20.

Wang

. Lean Premixed Prevaporised Combustion for Microturbines. PhD Thesis, University of Cambridge, 2022.

21.

Ciardiello

. Transient Phenomena in Annular Combustors. PhD Thesis, University of Cambridge, 2021.

22.

Pathania

Skiba

Ciardiello

et al. Blow-off mechanisms of turbulent premixed bluff-body stabilised flames operated with vapourised kerosene fuels. Proc Combus Inst 2021; 38: 2957–2965.

23.

Wang

Mastorakos

. Autoignition of isolated n-heptane droplets in air and hot combustion products at microturbine conditions. Combust Theory Modell 2022; 26: 541–559.

24.

Sapra

Chander

. Lean blow-off characteristics of a tangential entry type dual swirling free and impinging flame. Fuel 2021; 295: 120598.

25.

Radhakrishnan

Heywood

Tabaczynski

. Premixed turbulent flame blowoff velocity correlation based on coherent structures in turbulent flows. Combust Flame 1981; 42: 19–33.

26.

Cavaliere

Kariuki

Mastorakos

. A comparison of the blow-off behaviour of swirl-stabilized premixed, non-premixed and spray flames. Flow, Turbul Combust 2013; 91: 347–372.

27.

Yuan

Kariuki

Mastorakos

. Measurements in swirling spray flames at blow-off. Int J Spray Combust Dyn 2018; 10: 185–210.

28.

Ballal

Lefebvre

. Flame propagation in heterogeneous mixtures of fuel droplets, fuel vapor and air. Symposium (International) on Combustion 1981; 18: 321–328.

29.

Greenberg

Silverman

Tambour

. On droplet enhancement of the burning velocity of laminar premixed spray flames. Combust Flame 1998; 113: 271–273.

30.

Dvorjetski

Greenberg

. Influence of non-unity lewis numbers and droplet loading on the extinction of counter-flow spray diffusion flames. Proc Combus Inst 2000; 28: 1047–1054.

31.

Myers

Lefebvre

. Flame propagation in heterogeneous mixtures of fuel drops and air. Combust Flame 1986; 66: 193–210.

32.

Law

Sirignano

. Unsteady droplet combustion and droplet heating-II: conduction limit. Combust Flame 1977; 28: 175–186.

33.

de Oliveira

Sitte

Mastorakos

. Polydispersity effects in low-order ignition modeling of jet fuel sprays. Combust Sci Technol 2022; 194: 258–271.

34.

Providakis

Zimmer

Scouflaire

et al. Characterization of the coherent structures in swirling flames stabilized in a two-staged multi-injection burner: influence of the staging factor. Comptes Rendus - Mecanique 2013; 341: 4–14.

35.

Yuan

Kariuki

Dowlut

et al. Reaction zone visualisation in swirling spray n-heptane flames. Proc Combus Inst 2015; 35: 1649–1656.

36.

Idicheria

Pickett

. Formaldehyde Visualization Near Lift-off Location in a Diesel Jet. SAE Technical Paper Series 2006.

37.

Berlman

. Handbook of Fluorescence Spectra of Aromatic Molecules. 2nd ed., Academic Press, 1971.

38.

Cung

Zhu

Moiz

et al. Characteristics of formaldehyde (CH2O) formation in dimethyl ether (DME) spray combustion using PLIF imaging. SAE Int J Fuels Lubricants 2016; 9: 138–148.

39.

Skeen

Manin

Pickett

. Simultaneous formaldehyde PLIF and high-speed schlieren imaging for ignition visualization in high-pressure spray flames. Proc Combus Inst 2015; 35: 3167–3174.

40.

Subash

Liu

et al. Flame investigations of a laboratory-scale CECOST swirl burner at atmospheric pressure conditions. Fuel 2020; 279: 118421.

41.

Kariuki

Dowlut

Yuan

et al. Heat release imaging in turbulent premixed methane-air flames close to blow-off. Proc Combus Inst 2015; 35: 1443–1450.

42.

Ciardiello

Pathania

Helou

et al. Lean blow-off investigation in a linear multi-burner combustor operated in premixed and non-premixed modes. Appl Energy Combust Sci 2022; 9: 100041.

43.

Sidey

JAM

Giusti

Benie

et al. The Swirl Flames Data Repository, 2017. http://swirl-flame.eng.cam.ac.uk.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

6.58 MB

0.00 MB

0.28 MB

Stability and structure of lean swirling spray flames with various degrees of prevaporization

Abstract

Keywords

Introduction

Methods

Experimental conditions

Experimental procedure and diagnostics

Simulation of droplet evaporation

Results and discussion

Lean stability limit

Direct flame photos

Droplet velocity and size

HR location

CH 2 O-PLIF

Conclusions

Supplemental Material

sj-zip-1-scd-10.1177_17568277231159173 - Supplemental material for Stability and structure of lean swirling spray flames with various degrees of prevaporization

Supplemental Material

sj-sj-pdf-2-scd-10.1177_17568277231159173 - Supplemental material for Stability and structure of lean swirling spray flames with various degrees of prevaporization

Footnotes

Acknowledgements

Declaration of Conflicting Interests

Funding

ORCID iDs

Supplemental material

References

Supplementary Material

CH $_{2}$ O-PLIF