Download as pdf or txt
Download as pdf or txt
You are on page 1of 9

Flame downwash behavior in horizontal jet fires with

crossflow: Experiment and physical model

Jiang Lva, Longhua Hua,*, Xin Lia, Hongyu Lua, Yuxuan Maa,
Xiepeng Suna, Suk Ho Chungb
a
State Key Laboratory of Fire Science, University of Science and Technology of China, Hefei, Anhui, 230026, China
b
Clean Combustion Research Center, King Abdullah University of Science and Technology, Thuwal, Saudi Arabia
______________________________________________________________
Abstract

A flame downwash phenomenon was investigated for horizontal jet fire with crossflow perpendicular to the fuel
discharge direction. Such a jet fire configuration involves complex flow and air-fuel mixing interactions due to
relatively different directions of the fuel jet, buoyant flow, and cross-flow, for which a systematic data does not
exist yet. Experiments were conducted with various nozzle diameters (8 ~ 20 mm) with propane fuel and varying
fuel flow rate. The crossflow air speed, generated by a wind tunnel, varied from 1.76 to 4.56 m/s. Results showed
that the length of flame downwash first increased and then decreased with the increase in the fuel supply flow rate.
The flame downwash length became smaller with the increase in the crossflow air speed at relatively small fuel
flow rate and became larger with the increase in the crossflow air speed at relatively large fuel flow rate. A
dimensional analysis was performed on flame downwash behavior of the horizontal jet fires based on the analysis
of controlling mechanisms, from which three characteristic parameters were identified. First, the dimensionless
mass flow rate (𝑆𝑚̇f⁄𝜌∞ )/ [𝑈c × (𝑈c2 ⁄𝐺 )2 ] represents the fuel flow rate normalized by the characteristic air
entrainment, which influences the total flame length. Second, the momentum flux ratio of jet to crossflow R
represents the relative drag force of negative pressure zone in the leeward side, which determines the amount of
fuel or total flame length trapped along the leeward side to produce the downwash flame. Third, the competition
of the jet momentum to flame buoyancy flux (𝑚̇f 𝑈j /𝜌∞ )3⁄4 ⁄(𝑚̇f 𝐺/𝜌∞ )1⁄2 represents the characteristic length in
the jet discharged direction where the flame motion turns from horizontal (jet momentum controlled) to vertical
(buoyancy controlled) direction. The measured length of flame downwash considering the nozzle diameter, fuel
flow rate, and crossflow air speed was satisfactorily correlated by the proposed mechanisms/parameters.

Keywords: Horizontal jet fire; Crossflow; Flame downwash length; Momentum flux ratio of jet-to-crossflow; Dimensionless fuel
flow rate
______________________________________________________________

*Corresponding author: hlh@ustc.edu.cn

1
1. Introduction pressure zone will also be formed on the leeward side
when a horizontal jet encounters a crossflow and the
Jet fire is an important subject in combustion injected fuel can be trapped to the leeward side of the
science and fire research related to risk assessment of nozzle, having a downwash flame. Note that for a
thermal hazard. Numerous studies have been vertical upward jet-crossflow configuration, the flame
conducted to investigate flame geometry for vertical downwash length can be appreciably influenced by
[1-3] and inclined [4-9] jets in still air. the buoyant force (opposite to the orientation of
In practical environments, there frequently exists downwash) generated by the entire flame, acting on
wind or crossflow [10]. When the jet momentum is the flame body attached to the nozzle (downwash part)
relatively small under a crossflow, a fundamental and inducing upward draft by air entrainment, such
phenomenon called downwash flame [11-15] was that the relative importance of buoyancy is varied
observed with the flame attached on the nozzle along with the fuel flow rate/momentum. However,
leeward side in a vertically upward jet fire. for a horizontal jet-crossflow configuration, the effect
There are still very limited studies [12, 13, 15] in of gravity on flame downwash can be appreciably
quantifying the flame downwash length. Majeski et al. suppressed (buoyancy is perpendicular to the
[12] measured the flame downwash lengths by downwash direction) in controlling the downwash
employing a vertical cylindrical nozzle with the length, in contrast to that in a vertical jet. It involves
diameter D = 22.1 mm at a constant velocity of fuel more complex flow and air-fuel mixing interactions
jet Uj under the crossflow speed Uc varying from 2.0 due to relatively different directions of the fuel jet,
to 10.0 m/s, and proposed a non-dimensional buoyant flow, and cross-flow, for which a systematic
correlation: data does not exist yet. This kind of configuration
could be encountered practically in a flare with
𝐿d
= 13.6ln (
1
) − 1.29 (1) horizontal fuel injection experiencing a strong wind at
𝐷 𝑅0.5 the offshore drilling platform. In such a case, a flame
downwash could cause excessive temperature near the
where R is the momentum flux ratio of jet to crossflow nozzle and potentially damaging the nozzle.
𝜌f 𝑈j2 In this work, experiments were performed to
defined by 𝑅 = , 𝜌f and 𝜌∞ are the fuel density
𝜌∞ 𝑈c2 explore flame downwash length influenced by nozzle
and ambient air density, respectively. Shang et al. [13] diameter, fuel flow rate, and crossflow air speed for
conducted experiments with varying fuel jet velocity. horizontal jets. A physical model on the flame
Later, Li et al. [15] found that the length of flame downwash behavior in horizontal jet fires was
downwash first increased to a maximum value then proposed to interpret the evolution of flame
declined with increasing initial velocity of fuel jet due downwash length combining the interaction between
to the competition between pressure difference and jet crossflow momentum, fuel jet momentum, and
momentum. In summary, the momentum flux ratio of buoyancy.
jet to crossflow, which determines the relative drag
force in the leeward negative pressure zone, is 2. Experiment
important in controlling the flame downwash length.
However, there is another basic jet-crossflow Experiments were performed in a wind tunnel (72
configuration, which has not been studied yet, that is, m (L) × 1.5 m (W) × 1.3 m (H)), as shown in Fig. 1.
a horizontal jet with a crossflow. One can expect that, The crossflow velocity (Uc: 1.76 ~ 4.56 m/s) was
similar to flame downwash in a vertical jet, a negative measured by an anemometer (±0.01 m/s accuracy)
Honeycomb structures 3.0
2.9 Probe 1
Probe 2
Wind speed (m/s)

2.8
Probe 3
2.7
2.6
2.5
2.4
Probe 2.3
2.2 Uc=2.58 m/s
· Nozzle
2.1
2.0
0 10 20 30 40 50 60
Time (s)
Crossflow Uc ·
1.3 m

1.76 - 4.56 m/s


15 cm Ld
·
L
d

Image processing
procedure
Upward view
Wind tunnel: Total length 72 m Propane
Fig. 1. Schematic of experimental setup.

2
Table 1.
Summary of experimental conditions.
Momentum flux
Nozzle diameter, Fuel flow rate Fuel jet velocity, Jet exit Reynolds Crossflow
ratio of jet to
D [mm] [L/min] Uj [m/s] number, ReD velocity, Uc [m/s]
crossflow, R
8 0.27 - 6.63 470.0 - 11750.0 0.0051 - 21.3169
12 0.8, 1.5, 2.5, 5.0, 0.12 - 2.95 313.3 - 7833.3 1.76, 2.58, 0.0010 - 4.2107
16 0.07 - 1.66 235.0 - 5875.0 0.0003 - 1.3323
10.0, 15.0, 20.0 3.57, 4.56
20 0.04 -1.06 188.0 - 4700.0 0.0001 - 0.5457

installed at the wind tunnel portal. The turbulence 3. Results and discussion
fluctuations (u'rms/Uc) of local crossflow was less than
5%. 3.1 Flame downwash variation for horizontal
Four 60 cm long stainless-steel cylindrical tubes jet fires
having diameters of 8, 12, 16, and 20 mm were used Figure 2 depicts a sequence of typical flame
as the fuel nozzle to ensure the fully developed
turbulent pipe flows. The distance between the nozzle Cross- Jet exit
and wind tunnel outlet was 0.4 m to reduce the flow
complex effect of wall constraint. At this position, the

·
mf = 2.45 × 10-5 kg/s
flow characteristics were confirmed to be the same as
that at the wind tunnel outlet. The nozzle was
positioned horizontally along the center of the wind
tunnel. Ld =17.08cm
Propane was used as fuel and its flow rates were
·
-5
regulated by a mass flow meter (precision: 0.01 L/min) mf = 4.60 × 10 kg/s
in the range of 0.8 ~ 20.0 SLPM (0.245 ~ 6.13 × 10−4
kg/s), corresponding to Uj ranged from 0.04 to 6.63
m/s. The Reynolds number (ReD = fUjD/f) of the Nozzle Ld =22.08cm
fuel jet ranged from 188 to 11,750, where f is the
·

mf = 7.67 × 10-5 kg/s


dynamic viscosity. The flame downwash behavior
occurs when the momentum flux ratio of jet to

Increasing fuel flow rates


crossflow R ranges from 0.000457 to 2.37. Table 1
provides all the test conditions involving nozzle Ld =27.88cm
diameter, fuel flow rate, fuel jet velocity, jet exit
·

Reynolds number, crossflow velocity and momentum mf = 1.53 × 10-4 kg/s


flux ratio of jet to crossflow. The ambient temperature
𝑇∞ during the experiments was 293 ~ 295 K. Each
condition was repeated three times and reproducibility Ld =34.11cm
was satisfactory.
·

The CCD camera (25 fps, 3 megapixels) was mf = 3.06 × 10-4 kg/s
located below the nozzle zone with vertically upward
view to record flame images for 30 s with total 750
frames. Each image was converted into grayscales to
Ld =36.73cm
extract the luminosity information based on the RGB
value of the image. Then, a program based on the
·

mf = 4.60 × 10-4 kg/s


OTSU algorithm [16], ensuring the maximum
likelihood of local contrast between the flame and the
background, provided the optimal criterion to
distinguish the flame from the background, that is, Ld =26.77cm
“flame” (binary value of 1) or “no-flame” (binary
·

value of 0) for each pixel point of the image, such that mf = 6.13 × 10-4 kg/s
a binary image can be obtained for each image. The
flame image probability distribution contours (the
bottom inset contour in Fig. 1) can be achieved by Ld =21.11cm
averaging 750 binary images, to determine the
averaged presence probability value for each pixel
point. The flame dimension (Ld) is defined as the 0 5 10 15 20 25 30 35 40 [cm]
contour at 50% presence probability value level, as Fig. 2. Flame images showing flame downwash behaviors
applied in previous studies [6-9, 17-19]. with increasing fuel flow rates (D = 20 mm, Uc = 1.76 m/s).

3
pictures showing the flame downwash behavior at For a specified fuel flow rate (fuel jet velocity), Ld
various fuel flow rates 𝑚̇f (fuel jet velocities) for first extends and then declines with increasing Uc.
fixed Uc = 1.76 m/s and D = 20 mm. Due to the Firstly, as Uc increases, the effect of negative pressure
negative pressure zone and the generated recirculation will be stronger that the reverse vortex becomes
vortex at the leeward side, the flame edge crawls intensive, which can bring more fuel to the leeward
toward the upstream region from the nozzle exit. The side of the nozzle as such Ld increases. However, as
presence of a recirculation vortex transports part of Uc is relatively strong, the flame/flow turbulence
the fuel to the leeward side and the fuel burns in the increases, which enhances entrainment/mixing of air
wake zone, producing a flame attached on the cylinder with fuel hence Ld decreases.
wall, called the downwash flame. It is obvious that the Based on the flame downwash length and fuel jet
length of flame downwash first increased and then velocity, the fuel residence time for available fuel-
declined with the increase in the fuel supply flow rate, preheating to be only 0.008 - 4.02 s. Meanwhile, the
as illustrated in Fig. 2. contact area between the downwash flame with the
The data of flame downwash length (Ld) in the circular nozzle can be regarded as a very thin line and
present horizontal jet fires is shown in Fig. 3 against the cross flow would cool the nozzle wall. Therefore,
fuel flow rate under several crossflow air speeds for the effect of heating may be negligible. Concerning
various nozzle diameters. For a specified Uc, Ld the non-monotonic behavior of the flame downwash
presents a non-monotonic trend with increasing fuel length with fuel flow rate, the laminar to turbulent
flow rate (or fuel jet velocity), which first increases transition Re of 2300 is marked in Fig. 3. Considering
and then declines. For relatively low jet velocities, as the effect of nozzle diameter and crossflow air speed,
the fuel flow rate increases, more fuel can be brought the increasing and then declining behavior cannot be
fully explained based on the transition of the fuel flow
into the leeward side of the recirculation vortex and
through the nozzle. The effect of Richardson number
thus the length of flame downwash could extend.
Ri will be explained later.
However, as the fuel flow rate increases to a relatively
To better understand the flow/combustion
large level, the fuel jet momentum counteracts the mechanisms controlling the flame downwash
crossflow momentum (the momentum flux ratio of jet behavior in the horizontal jets, CFD simulations were
to crossflow increases). Note that the crossflow performed using the commercial code of Fluent 6.3.26
momentum generates the “reverse flow effect” at the for D = 12 mm. The modified standard k-ε turbulence
leeward side by forming a negative pressure region, model, eddy-dissipation-concept (EDC) model along
trapping fuel in this region. Hence, the length of flame with a one-step reaction model for the species and the
downwash declines as the fuel flow rate is further discrete ordinate radiation model were adopted, which
increased. have been successfully applied to the simulations of
0.6
D = 8 mm 100
Re=2300 Re=2300 D = 12 mm
0.5
Crossflow
laminar turbulent 10
0.4 laminar turbulent speed (Uc)
1.76 m/s
1
0.3 2.58 m/s
Flame downwash length (Ld) [m]

3.57 m/s
0.2 4.56 m/s 0.1
Richardson number

Ri
0.1 0.01

0.0 0.001
0.6
Re=2300 D = 16 mm D = 20 mm 100
0.5
laminar turbulent
10
0.4
1
0.3

0.2 0.1
Re=2300
0.1 laminar turbulent 0.01

0.0 0.001
0.0000 0.0002 0.0004 0.0006 0.0000 0.0002 0.0004 0.0006
Fuel flow rate mf [kg/s]
Fig. 3. Flame downwash length with fuel flow rate under several crossflow air speeds and nozzle diameters.

4
Fig. 4. CFD simulation results of flow field for flame downwash behavior of horizontal jet fires in crossflow for D = 12 mm.

5
35 the downwash length as the momentum flux ratio of
Experimental jet to crossflow increases.
30 range in [12] Nozzle Figure 5 compares the measured data of the flame
diameter(D) downwash length for the present horizontal jets with
25 8 mm the proposed model for vertical jets by Majeski (Eq.
12 mm
Ld (1)), where R was the controlling parameter [12]. It
20 16 mm
20 mm
shows that the model developed for the vertical jets
D could not describe the data of Ld for the present
15
horizontal jets, which exhibits the non-monotonic
10 behavior. Note that Majeski model was derived from
y = 13.6x - 1.29 R ranging from 0.0157 to 1.574 for a specified
5 diameter of D = 22.1 mm and a specified fuel flow
rate. For large fuel flow rates (corresponding to small
0 ln(1/R0.5)), the trends of the present data and Majeski
-1 0 1 2 3 4
model are consistent, while the effect of the nozzle
ln(1/ R0.5) diameter cannot be captured since only one diameter
Fig. 5. Comparison of experimental data for the length of was tested in developing the model. Note that the
flame downwash in horizontal jet fires with model by
Majeski [12] for D = 22.1 mm (Eq. (1)).
model under-predicted for the present data, which can
be understood based on the buoyancy effect for the
jet fires [20-25]. For the grid dependence check, vertical jet experiment, even though the effect of the
26,795,096 grid cells having the largest cell size of 2 buoyancy was not considered in the model. For small
× 2 × 2 mm and the smallest cell size of 0.5 × 0.5 × fuel flow rates (large ln(1/R0.5)), the trends are
0.5 mm, 3,337,000 grid cells with the largest cell size reversed.
of 4 × 4 × 4 mm and the smallest cell size of 1 × 1 × As mentioned previously, the non-monotonic
1 mm, together with 975,982 grid cells with the behavior of Ld with fuel flow rate cannot be fully
largest cell size of 6 × 6 × 6 mm and the smallest cell explained based on the laminar to turbulent transition
size of 1.5 ×1.5 ×1.5 mm were tested. The simulation of the fuel flow in the nozzle. Although the buoyancy
results of the most drastic change of the velocity in the effect can be suppressed for the present horizontal jet,
entire region for 3,337,000 and 26,795,096 grid cells the buoyancy effect cannot be ruled out for small fuel
are similar. Therefore, the 3,337,000 grid cells were flow rate (Reynolds number). In this regard, 𝑅𝑖 =
used for every simulation cases, ensuring the accuracy (𝜌∞ −𝜌b )𝑔𝐷 (𝑇 −𝑇 )𝜌 𝑔𝐷
2 ≈ f ∞ ∞2 is considered based on the
of simulation. Note that the refinement region 𝜌f 𝑈𝑗 𝑇∞ 𝜌f 𝑈𝑗
𝜌∞ −𝜌𝑏 𝑇𝑓 −𝑇∞
includes 0.06 × 0.06 m square area around the outlet approximation ( ≈ ), which represents the
𝜌∞ 𝑇∞
of the horizontal jet. More information about the
simulation can be found in the Supplementary relative importance of jet momentum to buoyancy.
material. The mean flame temperature Tf for propane of 1200K
The simulation results of flow field for flame [7] is used here. Note that the Richardson number
downwash behavior are shown as Fig. 4 at various (logarithmic plot in Fig. 3) decreases very rapidly
fuel flow rates and crossflow air speeds. In the (since proportional to 𝑈j−2 ). This implies that
increasing regime for the length of flame downwash although the effect of buoyancy (gravity) can be small
with fuel flow rate (corresponding to the relatively for large fuel flow rate regime in the horizontal jet, the
small momentum flux ratio of jet to crossflow), the buoyancy effect could be pronounced in the small fuel
flow field patterns change slightly with the flow rate regime.
momentum ratio and the streamlines deflected a large Based on above experimental observation and the
angle from the nozzle exit to the leeward side, which discussion on the effect of Ri, a new physical model
implies that the fuel is carried to the leeward side right is required to describe the flame downwash behavior
after it comes out of the nozzle. Therefore, the flame in horizontal jets.
downwash length gets longer with the increase in the
fuel flow rate. In this regime, it exists a critical state 3.2 Dimensional analysis on flame downwash
as shown in the inset in Fig. 4, where the clockwise behavior in horizontal jet fires
vortex is observed near the nozzle exit.
In the decreasing regime for the flame downwash Figure 6 depicts a physical model of flame
downwash phenomenon for the horizontal jets in
length with fuel flow rate (relatively large momentum
crossflow, which involves a three-dimensional
ratio), a stagnation point appears in the near jet wake,
complex flow-interaction behavior. The fuel first
as shown in the inset figure. The deflection angle of
ejected along the jet (x-) direction due to initial jet
streamlines from the nozzle exit decreases with momentum and the crossflow (perpendicular to jet in
increasing momentum flux ratio of jet to crossflow. y-direction) would carry the fuel to the downstream
This implies that it is more difficult for the fuel to be toward the leeward side, meanwhile the buoyancy in
trapped to the leeward side, leading the decrease of z-direction drives flame to upward positive z-

6
direction, washing away from the flame attached base the characteristic horizontal length in the jet
zone. Due to the negative pressure zone at the nozzle discharged direction where the flame motion majorly
leeward side induced by the crossflow, the fuel is turns from horizontal (jet momentum controlled) to
trapped into the nozzle leeward side and forming a vertical (buoyancy controlled). Here, 𝑄̇ is heat release
downwash flame. In these processes, the flame rate related to 𝑐𝑝 𝑚̇f ∆𝑇. This competition could be
downwash length is expected to be controlled by the 3⁄4
(𝑚̇f 𝑈j ⁄𝜌∞ )
following physical mechanisms/parameters: represented by 𝐿m = , which represents
(𝑚̇f 𝐺 ⁄𝜌∞ )1⁄2
First, the initial fuel flow rate 𝑚̇f characterizes the
the length scale for the horizontal distance a flame can
magnitude of combustion and determines the total
reach under no crossflow condition [6, 28] and is
flame length. When the fuel is completely consumed
related to the inverse of the square root of Ri.
at flame tip, the total amount of air entrainment should Therefore, based on the above physical analysis,
be proportional to S𝑚̇f , where S is the stoichiometric the following relations can be proposed for Ld:
ratio of entrained air to fuel mass. And the
characteristic flow rate of air entrainment 𝜌∞ 𝑈c × 𝐿d 𝑆𝑚̇f⁄𝜌∞
(𝑈c2 ⁄𝐺 )2 can be applied to normalize 𝑚̇f [26], where ~fn { , 𝑅} (2)
𝐿m 𝑈c ×(𝑈c2 ⁄𝐺 )2
𝐺 = (𝑇f − 𝑇∞ )𝑔⁄𝑇∞ represents the overall flame
buoyancy induced by average flame temperature rise. We normalize the flame downwash length by the
Second, the relative magnitude of the characteristic parameter 𝐿m , because it is a horizontal characteristic
momentum in each direction determines the behavior flame length in the jet direction, in a same horizontal
of flame. There are two characteristic momenta that direction as the flame downwash. Ld should increase
control the flame downwash length: jet momentum with the increasing normalized fuel mass flow rate
𝜌f 𝑈j2 and the crossflow momentum 𝜌∞ 𝑈c2 . The (𝑆𝑚̇f ⁄𝜌∞ )/ [𝑈c × (𝑈c2 ⁄𝐺 )2 ] and the decreasing R.
negative pressure generated by the crossflow tends to For correlating the flame downwash length and
induce the fuel to the leeward side to form a decoupling this problem, it is verified that the
downwash flame. While the jet momentum carries the normalized flame length has a power function relation
fuel to move along the nozzle axis direction, which with the normalized fuel mass flow rate
makes the length of flame downwash to decline. (𝑆𝑚̇f ⁄𝜌∞ )/ [𝑈c × (𝑈c2 ⁄𝐺 )2 ]. Then, the exponent of
Therefore, the relative drag force at negative pressure (𝑆𝑚̇f ⁄𝜌∞ )/ [𝑈c × (𝑈c2 ⁄𝐺 )2 ] can be found by
zone of the leeward side can be represented by their plotting the normalized flame downwash length
momentum flux ratio, determining how much (ratio) versus normalized fuel mass flow rate for the similar
of the total flame length could be dragged along the R. It is found that the exponential of the normalized
leeward side to produce the downwash flame. fuel mass flow rate is 0.21 for best fit. Then, the power
Finally, the competition of jet momentum flux function relation of R is further proposed where the
𝐺0 = 𝑚̇f 𝑈j with flame buoyancy flux 𝐵0 = best power value is found to be −0.42. Finally, the best
𝑄̇ 𝑔⁄𝑐𝑝 𝑇∞ ~𝑚̇f ∆𝑇⁄𝑇∞ 𝑔 = 𝑚̇f 𝐺 [27], which decides fit function, found as shown in Fig. 7, well represents

Flame buoyancy
f U j2

Ld
Plume mass flow rate
when the fuel is
·
completely burned Smf

Characteristic air Relative dragging force at z


entrainment flow2 x
negative pressure zone
rate U c (U c G )  U c2
2
f U j2
R=
 U c2 y
Fig. 6. Physical model of flame downwash behavior in horizontal jet fires with crossflow.

7
160 4. Concluding remarks
140
This work explored the flame downwash behavior
120 R2 = 0.97 in horizontal jet fires under crossflow. The
y = 17.6 x dimensional analysis was derived from the
100
dominating physical mechanisms of flame downwash
Ld 80 Nozzle behavior. Major findings include:
Lm 60
diameter (D)
8 mm (1) The flame downwash length exhibits a non-
40 12 mm monotonic behavior, which first increases then
16 mm decreases with the increase of the fuel flow rate.
20
20 mm (2) A dimensional analysis was performed on
0 flame downwash behavior of the horizontal jet
0 1 2 3 4 5 0.216 7 8 9 fires based on the analysis of dominating
  mechanisms, from which three characteristic
 Smf    R −0.42 parameters were identified. First, the
(
U U 2 G  )
2

 c c  dimensionless mass flow rate


Fig. 7. Representation of flame downwash length in (𝑆𝑚̇f⁄𝜌∞ )/ [𝑈c × (𝑈c2 ⁄𝐺 )2 ] represents the
horizontal jet fires at various conditions based on the fuel flow rate normalized by the characteristic
proposed physical model and controlling non-dimensional
air entrainment, which influences the total
parameters.
flame length. Second, the momentum flux ratio
the data of flame downwash length in horizontal jet of jet to crossflow R represents the relative drag
fires under various conditions: force of negative pressure zone in the leeward
side, which determines the amount of fuel or
total flame length trapped along the leeward
𝐿d 𝑆𝑚̇f⁄𝜌∞ 0.21
= 17.6 [ ] 𝑅−0.42 (3) side to produce the downwash flame. Third, the
𝐿m 𝑈c ×(𝑈c2 ⁄𝐺 )2
competition of the jet momentum to flame
buoyancy flux (𝑚̇f 𝑈j /𝜌∞ )3⁄4 ⁄(𝑚̇f 𝐺/𝜌∞ )1⁄2
Note that the non-monotonic behavior shown in represents the characteristic length in the jet
Fig. 3 disappears by using the physical parameters discharged direction where the flame motion
discussed above. The correlation for the flame turns from horizontal (jet momentum
downwash behavior of the horizontal jet fires (Eq. (3)) controlled) to vertical (buoyancy controlled)
shows that LHS ~ Ld ×g0.5 (with the power of g of 0.5 direction.
from Lm ∝ G−0.5) and RHS ~ g0.42 and thus Ld ~ g−0.08; (3) The proposed non-dimensional formula
while Shang’s correlation for the vertical jet fires represents the experimental results of flame
shows that Ld ~ g−0.75 [13]. This implies that the effect downwash lengths well based on the derived
of the buoyancy on flame downwash behavior for the characteristic parameters. The buoyancy effect
vertical jet fires is much greater than the horizontal jet on flame downwash shows relatively weaker
fires. For vertical jet fires under crossflow, the dependence on the gravity for the horizontal jet
buoyancy is in the same direction as the fuel jet. The than that of the vertical jet in crossflow.
buoyancy is driving the flow upward, suppressing the However, the consideration of buoyancy for
negative pressure zone. For the horizontal jet fires, we the horizontal jets is still important in
can also anticipate an upward flow induced by predicting the non-monotonic behavior of the
buoyancy, however, passing around the nozzle. Such flame downwash length.
flow could also interact complicatedly with the vortex
induced by crossflow at the leeward side.
Acknowledgements
Nevertheless, such flow interaction in the negative
This work was supported by key international
pressure zone should be essentially different from that (regional) cooperative research project of National
of vertical jet fires with crossflow. As the nozzle Natural Science Foundation of China (NSFC)
diameter is relatively small, we can expect that such (52020105008). SHC was supported by King
buoyancy effect could be less important than that for Abdullah University of Science and Technology
a vertical jet, as also verified by the correlation power (KAUST).
dependence on the gravity. Note that even though the
power dependence of Ld on the gravity is much Supplementary material
smaller in the present horizontal jets as compared to
the vertical jets, the consideration of buoyance Supplementary material associated with this article
(gravity) is still important in predicting the non- can be found, in the following link,
monotonic behavior with the fuel flow rate. https://www.editorialmanager.com/proci/download.a

8
spx?id=289736&guid=09079d7e-c247-4cfe-a8c2- [17] E.E. Zukoski, B.M. Cetegen, T. Kubota, Visible
baad107d53b5&scheme=1. structure of buoyant diffusion flames, Proc. Combust.
Inst. 20 (1985) 361-366.
References [18] C.C. Liu, L.Y. Huang, T.D. Deng, H. Jiang, P.Z. Wu, B.
Liu, J. Deng, Z.M. Luo, Influence of bottom wall on
[1] M.A. Delichatsios, Air entrainment into buoyant jet characteristics of jet diffusion flames under cross-wind,
flames and pool fires, Combust. Flame 70(1) (1987) 33- Fuel 288 (2021) 119661.
46. [19] Z.H. Wang, K.B. Zhou, L. Zhang, X. Nie, Y.Q. Wu, J.C.
[2] M.A. Delichatsios, Transition from momentum to Jiang, A.S. Dederichs, L. He, Flame extension area and
buoyancy-controlled turbulent jet diffusion flames and temperature profile of horizontal jet fire impinging on a
flame height relationships, Combust. Flame 92 (4) (1993) vertical plate, Process. Saf. Environ. Protect. 147 (2021)
349-364. 547-558.
[3] J.G. Quintiere, B.S. Grove, A unified analysis for fire [20] C.C. Liu, L.Y. Huang, T.D. Deng, H. Jiang, P.Z. Wu, B.
plumes. Proc. Combust. Inst. 27(2) (1998) 2757-2766. Liu, J. Deng, Z.M. Luo, Influence of bottom wall on
[4] H.A. Becker, D. Liang, Effect of burner orientation and characteristics of jet diffusion flames under cross-wind,
ambient airflow on geometry of turbulent free diffusion Fuel 288 (2021) 119661.
flames, Proc. Combust. Inst. 18 (1981) 1061-1071. [21] J. Ju, X.J. Duan, B. Sarkodie, Y.J. Hu, H. Jiang, C.Z. Li,
[5] N. Peters, J. Göttgens, Scaling of buoyant turbulent jet Numerical simulation of flow field and residence time
diffusion flames, Combust. Flame 85 (1-2) (1991) 206- of nanoparticles in a 1000-ton industrial multi-jet
214. combustion reactor, Chin. J. Chem. Eng.
[6] X.L. Zhang, L.H. Hu, X.C. Zhang, F. Tang, Y. Jiang, Y.J. https://doi.org/10.1016/j.cjche.2021.12.008.
Lin, Flame projection distance of horizontally oriented [22] C.B. Jang, S.W. Choi, J.B. Baek, CFD modeling and
buoyant turbulent rectangular jet fires, Combust. Flame fire damage analysis of jet fire on hydrogen pipeline in
176 (2017) 370-376. a pipe rack structure, Int. J. Hydrogen Energ. 40 (2015)
[7] S.X. Liu, M.A. Delichatsios, L.H. Hu, Flame profile 15760-15772.
parameters of horizontal turbulent jets: Experiments, [23] H. Mashhadimoslem, A. Ghaemi, A. Palacios, A.H.
similarity analysis and an integral model, Combust. Behroozi, A new method for comparison thermal
Flame 207 (2019) 1-9. radiation on large-scale hydrogen and propane jet fires
[8] J. Lv, X.L. Zhang, S.X. Liu, H.Y. Lu, Y.X. Ma, L.H. Hu, based on experimental and computational studies, Fuel
Flame morphology of horizontal jets under sub- 282 (2020) 118864.
atmospheric pressures: Experiment, dimensional analysis [24] C. David, F.E. Thomas, CFD for simulation of
and an integral model, Fuel 307 (2022) 121891. crosswind on the efficiency of high momentum jet
[9] S.X. Liu, L.H. Hu, An experimental study on flame turbulent combustion flames, J. Environ. Eng. 134
envelope morphologic characteristics of downward- (2008) 561-571.
orientated buoyant turbulent jet fires, Proc. Combust. Inst. [25] H. Mashhadimoslem, A. Ghaemi, A.H. Behroozi, A.
37 (2019) 3935-3942. Palacios, A new simplified calculation model of
[10] L.H. Hu, A review of physics and correlations of pool geometric thermal features of a vertical propane jet fire
fire behaviour in wind and future challenges, Fire Saf. J. based on experimental and computational studies,
91 (2017) 41-55. Process. Saf. Environ. Protect. 135 (2020) 303-314.
[11] R.F. Huang, J. M. Chang, The stability and visualized [26] H.Y. Lu, M.A. Delichatsios, X. Li, S.X. Liu, J. Lv, L.H.
flame and flow structures of a combusting jet in cross Hu, Flame geometry of downward buoyant turbulent jet
flow, Combust. Flame 98(3) (1994) 267-278. fires under cross flows: Experiments, dimensional
[12] A.J. Majeski, D.J. Wilson, L.W. Kostiuk, Predicting the analysis and an integral model, Proc. Combust. Inst.
length of low-momentum jet diffusion flames in 38(3) (2021) 4917-4925.
crossflow, Combust. Sci. Technol. 176 (2004) 2001- [27] M.A. Delichatsios, J.P. Zhang, Ground wind generated
2025. near the base by the massive convective column of very
[13] F.J. Shang, L.H. Hu, X.P. Sun, Q. Wang, A. Palacios, large-scale mass fires. Fire Saf. J. 111 (2020) 102914.
Flame downwash length evolution of non-premixed [28] P.N. Papanicolaou, E.J. List, Statistical and spectral
gaseous fuel jets in crossflow: Experiments and a new properties of tracer concentration in round buoyant jets,
correlation, Appl. Energy 198 (2017) 99-107. Int. J. Heat Mass Transf. 30 (1987) 2059-2071.
[14] D.B. Mosiria, R.F. Huang, C.M. Hsu, Characteristics of
backward-inclined non-premixed jet flames in
crossflow, Exp. Therm. Fluid Sci. 98 (2018) 429-444.
[15] X. Li, L.H. Hu, F.J. Shang, Flame downwash transition
and its maximum length with increasing fuel supply of
non-premixed jet in cross flow, Energy 164 (2018) 298-
305.
[16] N. Otsu, A threshold selection method from gray-level
histograms, IEEE Trans. Syst. Man. Cybern. Syst. 9
(1979) 62-66.

You might also like