Microscopic and macroscopic atomization characteristics of a
pressure-swirl atomizer, injecting a viscous fuel oil
SeyedMohammadAliNajaﬁ, Pouria Mikaniki, Hojat Ghassemi ⁎
School of Mechanical Engineering, Iran University of Science and Technology (IUST), P.O.B. 16765-163, Tehran, Iran
Received 29 October 2018
Received in revised form 30 March 2019
Accepted 4 April 2019
Available online 18 April 2019
Combustion of heavy fuels is one of the main sources of greenhouse gases, particulate emissions, ashes, NO
.Gasiﬁcation is an advanced and environmentally friendly process that generates combustible and clean gas
products such as hydrogen. Some entrained ﬂow gasiﬁers operate with Heavy Fuel Oil (HFO) feedstock. In this
application, HFO atomization is very important in determining the performance and efﬁciency of the gasiﬁers.
The atomization characteristics of HFO (Mazut) discharging from a pressure-swirl atomizer (PSA) are studied
for different pressures difference (Δp) and temperatures in the atmospheric ambient. The investigated parame-
ters include atomizer massﬂow rate ( _
m), discharge coefﬁcient (C
), spray coneangle (θ), breakup length (L
unstable wavelength of undulations on the liquid sheet (λ
), global and local SMD (sauter mean diameter) and
size distribution of droplets. Thecharacteristicsof Mazut sheet breakup are deducedfrom the shadowgraph tech-
nique. The experiments on Mazut ﬁlm breakup were compared with the predictions obtained from the liquid
ﬁlm breakup model. Validity of the theory for predicting maximum unstable wavelength was investigated for
HFO (as a highly viscous liquid). A modiﬁcation on the formulation of maximum unstable wavelength was pre-
sented forHFO. SMD decreases bygetting far from the atomizer. Themeasurement for SMDand θwere compared
with the available correlations.The comparisonsof the available correlations with the measurements ofSMD and
θshow a good agreement forBallester and Varde correlations, respectively. Theresults show that the experimen-
tal sizing data could be presented by Rosin-Rammler distributions very well at different pressure difference and
© 2019 The Chemical Industry and Engineering Society of China, and Chemical Industry Press Co., Ltd.
All rights reserved.
Heavy fuel oil
One of the essential sources of energy used in modern civilization is
fossil fuels, includes HFOs. Even with the reﬁnery's progress, these fuels
are still being produced and play an important role in energy production
in the world . This black fuel is very thick and viscous which at cooler
temperatures becomes semi-solid is often known as bunker fuel oil .
HFOs have been used as the replacement of good quality oils for more
energy reserves and chip resources .
Combustion of HFO causes environmental problems and also due
to the physical characteristics of HFOs, cannot be used in combined cy-
cles of high-efﬁciency power plants. HFO gasiﬁcation process has been
introduced to reduce emissions and also to use in thepowergeneration
cycle . HFO in this process converted to the clean gases. The products
of gasiﬁcation mainly consist of hydrogen and carbon monoxide, called
“syngas”which can be used as the raw material for the synthesis of liq-
uid fuels or chemicals. The gasiﬁcation can also becombined with a fuel
cell or power plant system in the integrated gasiﬁcation combined cycle
Fuel atomization is vital to process in determining the performance
of combustors or gasiﬁers, especially when HFO is using because of dif-
ﬁcult atomizing. Inappropriate performance of atomizer in a gasiﬁer can
lead to the production of pollutants. Nitrogen pollutants produced in an
entrained ﬂow gasiﬁer was studied by . They found that most of the
nitrogen pollutants formed during the devolatilization, which is pre-
cisely about injection performance. Also, they showed the impact of
air to fuel ratios on nitrogen pollutant production. Gasiﬁers can operate
only in some speciﬁc air to fuel ratios . Use twin-ﬂuid (air-assist or
air-blast) atomizer for heavy fuels is common , but these atomizers
disturb the air to fuel ratio, a parameter which should be intensely con-
trolled for the gasiﬁcation process. Gasiﬁers can operate using a PSA,
which do not change the air-to-fuel ratio . For this reason,it is inter-
esting to study the injection of heavy fuels using PSA .
Many experimental measurements and numerical simulations have
been done for low viscosity liquids in PSA, but few studies have been
carried out investigating spray of HFO as a high viscosity liquid. As a
result, the review of HFO spray is necessary for PSA especially in the
application of gasiﬁers.
Chinese Journal of Chemical Engineering 28 (2020) 9–22
E-mail address: email@example.com (H. Ghassemi).
1004-9541/© 2019 The Chemical Industry and Engineering Society of China, and Chemical Industry Press Co., Ltd. All rights reserved.
Contents lists available at ScienceDirect
Chinese Journal of Chemical Engineering
journal homepage: www.elsevier.com/locate/CJChE
Proper atomization in entrained ﬂow gasiﬁer ensures faster evapo-
ration of fuel and better mixing between the fuel and the air, thereby
resulting in higher efﬁciency. PSA is widely used in combustors as a
fuel atomizer. The atomizer discharges the liquid sheet ﬁlm in the gas-
iﬁer. The quality of spray in PSAs could be evaluated by investigation
the characteristics such as atomizer mass ﬂow rate _
m, discharge coefﬁ-
, breakup length L
, maximum unstable wavelength λ
angle θ, and SMD and droplet distribution.
Over the years, many experimental and numerical studies have been
devoted to analyzing the performance of PSAs with liquid biofuels, but
there is a lack in the literature for a spray of HFOs. Therefore, an exper-
imental study was conducted on an HFO spray, Mazut 280.
In this study, the performance characteristics of the PSA for HFO for
use in gasiﬁer have been obtained. HFO viscosity is highly temperature
dependent; for this reason, the effects of temperature and Δpon spray
behavior are investigated. In this paper, measurement characteristics
,θ, global SMD, SMD at different axial location,
and droplet distribution. Some researchers started the research in this
Liu et al.  investigated the air core formed in PSA experimentally
and theoretically. Several types of atomizers with different geometries
and different liquid viscosities were studied to obtain the effects on
the air core size. Rashad et al.  investigated the impact of PSA geo-
metric ratios on θand SMD by the injection of water. They have reported
optimal ratios to achieve the best performance (lower SMD) using PSA.
Ferreira et al.  determined the main parameters of twin-ﬂuid
nozzles with an internal mixing chamber for HFO. They studied the
effect of interaction air and liquid ﬂows at the interior chamber on the
atomization. They examined different air channel diameters and liquid
ports for different air and liquid mass ﬂow rate. They have been
shown under certain conditions the ﬂuid discharged to the inner cham-
ber is choked. The sonic condition was achieved for different air and
liquid mass ﬂow rates as a function of the air central channel diameter.
The best results of atomizing ﬂuid ﬂow rates have been obtained in
choked conditions. The pressure range was from 0 to 0.7 MPa. They
showed kinematic viscosity reached values 1.546 × 10
at 50 °C
and 0.255 × 10
at 80 °C. They recommended that for generating
aﬁne spray, the viscosity is to be around 4.5 × 10
requires heating the HFO to 130 °C.
Liu et al.  investigated the air core formed in PSA experimentally
and theoretically. Several types of atomizers with different geometries
and different liquid viscosities were studied to obtain the effects on
the air core size. They found that a decrease in the liquid viscosity and
a decrease in the swirl chamber length cause the air core size to be in-
creased. Also, they found that an increase in the liquid viscosity would
lead to a decrease in critical swirl chamber length (when the swirl
chamber length is greater than a critical length, the air core will disap-
pear). They presented new correlations with a better agreement with
the experimental data at a broader range of liquid viscosity. They used
kerosene RP-3 for their experiments.
Suzuki et al.  investigated the spray of PSA (Delavan oil burner
nozzle 60°A-0.85) for viscous liquids in high-pressure nitrogen and
room temperature. They used water, diesel fuel, palm methyl ester,
and silicone-oil for the atomization. The volume ﬂow rate was 50 to
, and the ambient gas pressure difference was 0.1 to
1.0 MPa. They used a laser diffraction method for measuring SMD and
the PIV technique for the measuring spray ﬂow ﬁeld. Properties of liquids
tests were ranging as; density was 825.2 to 995.65 kg·m
, surface tension was 20.1 to 72.28 mN·m
Payri et al.  studied the effect of pressure and temperature on the
thermodynamic properties of diesel and biodiesel fuels. The tempera-
tures range was 25 °C to 68 °C and pressure difference were 15 to
180 MPa. They showed that biofuel injection pressure affects sound
speed, and could change the dynamic of some injection elements. Also,
they showed density values increased with pressure and decreased
with temperature. The fuel density was 812 to 825 kg·m
was 2.06 to 2.34 mm
, and the surface tension was 0.0205 to
Zhang et al.  measured the ﬁlm thickness of a PSA injecting water
with the image-processing method. They calculate the liquid ﬁlm thick-
ness at different sections of downstream ﬂow. The results showed that
the ﬁlm thickness decreases with increasing distance. Ding et al.  in-
vestigated the instability and droplet size distribution of liquid–liquid
coaxial PSA. They analyzed the effect of inner and outer Δpon the
spray characteristics of atomization. They have investigated the inner
and outer injection pressure on the surface wave characteristics by
using a laser reﬂection setup. They were shown at the same Δp,the
temporal instability of a liquid sheet doesn't change spatially and
wave frequency increases as the Δpincreases. The results showed the
inner spray is more unstable and more accessible to break up rather
than the outer spray and the coaxial spray.
Kim et al.  studied the effect of geometric parameters on the ﬁlm
thickness and air core formation in a PSA injecting water. They observed
a good agreement between the analytical method and experimental
data of liquid ﬁlm. Also, they observed the proper ratio of L/Din the
swirl chamber. Δpwas 0.2 to 0.8 MPa. Liu et al.  investigated the
behavior of the kerosene droplets produced by a PSA. They used by
Fuel-PLIF and LIF/Mie laser sheet-imaging methods for the atomization
process. They analyzed the spray pattern, droplet size spatial distribu-
tion, mean droplet size, and distribution index with different Δp.They
measured mean droplet size at various locations (20–70 mm) down-
stream the atomizer. Water was used as the simulant of LOX and the
fuel was kerosene. The pressure difference for each atomizer varied
from 0.1 MPa to 0.6 MPa. Mlkvik et al.  studied spraying viscous
liquids by using four types of internal-mixing twin-ﬂuid atomizers
with different Δpand various gas-to-liquid ratios. Different atomizers
were studied in the view of the ﬂow ﬁeld, stability, and droplet sizing.
The injected liquid density was 1185 to 1242 kg·m
, the viscosity
was 60 to 308 kg·m
, and the surface tension was 74.25 to
. Broniarz-Press et al.  analyzed the effect of oriﬁce
shape (conical and plain) and Δpon the water atomization process for
As HFO atomizing is very difﬁcult and a challenging topic [14,23], in
this study, an atomizer was tested in the application of HFO entrained
ﬂow gasiﬁer and studied to obtain reliable data for further researches.
The entrained ﬂow gasiﬁers require a smaller fuel droplet size than
the ﬂuidized bed gasiﬁer. Therefore, in the present study, the essential
characteristicsof a PSA were investigated. The main features of an atom-
izer, which are required in the design of a gasiﬁer chamber, C
the average diameter of the droplets. The C
is necessary for the deter-
mination of an atomizer _
m,theθplays an important role in deﬁning the
chamber cross-section area, the L
and mean diameter of droplets are
very effective in designing the length of the chamber and the lifetime
of the droplet presence in the chamber to evaporate. Therefore, in this
study, these speciﬁcations have been empirically investigated in order
to evaluate an atomizer in the application of an entrained ﬂow gasiﬁer.
Also, the present study was focused on comparing the effect of Δpand
temperature (difference of viscosity) of HFO, Mazut with PSA. Although
PSAs are widely investigated, the applications using HFOs are discussed
very limitedly. As we know the behavior are very different for such a
fuel, therefore more detailed measurements and investigations are
The atomizer was tested and investigated on a cold test bench at at-
mospheric condition injecting HFO (Mazut) in various Δpand tempera-
tures. The test fuel injector is a hollow-cone PSA with 0.3 mm oriﬁce
diameter (Steinen, oil burner nozzle, standard H, 1.00 GPH, 60° ).
Physical properties for Mazut were determined as a function of temper-
ature and are presented in Table 1. Density and viscosity were measured
10 S.M.A. Najaﬁet al. / Chinese Journal of Chemical Engineering 28 (2020) 9–22
according to ASTM D 4052 and ASTM D 7042 standards,and surface ten-
sion was predicted by Riaziapproach  for Mazutwhich is mentioned
in . Warm Mazut was delivered by a gear pump from a 4-litter tank
through a pipeline and was injected vertically into a collecting vessel
and a pump back into the tank can be seen in Fig. 1.A PCO 1200hs digital
high-speed camera with resolution 1280 × 1024 pixels with shadow-
graph technique was used to take pictures of the liquid ﬁlms, exposure
time was 5 μs and the frame rate was 1357 fps (for droplets velocimetry).
A macro lens (AF Zoom Nikon 100 mm f/2.8D) was used in conjunction
with the camera. The measurement technique was similar to . The
minimum measurable droplet diameter range is about 20 μm, because
of the camera resolution limit. θ,L
, SMD, and droplet size distribution
were obtained by image processing.
The liquid temperature probe was placed about 5 cm before the
injector entrance and the insulation was performed correctly. All mea-
surements have been conducted after ensuring the steady state condi-
tion. Ambient pressure was about 87 kPa and the temperature was in
the range of 20–30 °C. The probes for ambient pressure and temperature
were at almost 5 m distance far enough from the setup to minimize air
pressure and temperature change because of the injection.
3. Results and Discussion
For the measurements, the heavy oil temperatures were 100, 110
and 120 °C. The Δpselected for injection were 0.7, 0.9, 1, and 1.2 MPa.
In the case of liquid temperature change, two limitations exist, liquid
viscosity and boiling. The liquid viscosity is very high at low tempera-
tures. By lowering Mazut temperature below 100 °C, high viscosity
leads to failure in appropriate atomization. Also, it was generally ob-
served that heating Mazut to the temperatures higher than 130 °C
leads to the boiling or fast evaporation of some species that results to
asigniﬁcant difference in the performance of the injector. Therefore, the
temperatures of 100, 110 and 120 °C were selected for the measurement.
In the case of the injection pressure change, as the design pressure
for the injector is 1 MPa. Therefore a range around this value was se-
lected for the experiments. Increasing injection pressures to values
higher than 1.2 MPa leads to a weak change in the injector performance
and the injection pressures lower than 0.7 MPa does not end to a ﬁne
spray. So the range 0.7–1.2 MPa was selected for the investigation.
The increments on liquid temperature and injection pressure were
selected for reporting data in a way that the differences in the spray
characteristics could be detected concisely and completely.
3.1. Macroscopic properties
Fig. 2 shows the spray patterns of the PSA for a series of increasing
temperature with constant Δp= 1.2 MPa. At very low temperature,
the liquid emerges as a round jet. High viscosity causes the ﬂuid in the
rotating chamber to not swirl well at a low temperature. At T=100°C
with Δp= 1.2 MPa, the jet in Fig. 2a looks like a “twisting ribbon”
appear from the atomizer exit oriﬁce. The jet is created at the bottom
of the stream with three branches and due to instabilities on the
edges of the jet and “ﬁngers”could be observed. Fingers appear because
of the competition between the centrifugal force and the viscous force.
Eventually, the ﬁngers produce large droplets.By an increase in temper-
ature, Fig. 2b, at T= 110 °C, the ribbon becomes wider and forms three
edges and emerges from the hole as apyramid. The twisted ribbon con-
verted to a pyramid with branches exit near the oriﬁce. Horizontal
waves cause the ﬁlm of the pyramid to break down. With the break-
down of the pyramid ﬁlm, due to the horizontal waves, there are
horizontal ligaments that later change to droplets. By increasing tem-
perature, Fig. 2c, at T= 120 °C, the injected ﬁlm reshapes to a smooth
conical sheet without branches because, as the viscosity is highly
reduced at this temperature. Fig. 2c also shows perforations which are
Physical properties of HFO (Mazut)
100 925.18 0.0100 0.0302
110 919.48 0.0078 0.0295
120 913.11 0.0063 0.0287
Fig. 1. Schematic of the PSA test rig.
11S.M.A. Najaﬁet al. / Chinese Journal of Chemical Engineering 28 (2020) 9–22
created in the liquid ﬁlm. They grow with increasing distance from the
atomizer. Perforations are ﬁnally broken and created vertical ligaments
and eventually form small droplets.
Fig. 3 shows images of a hollow conical liquid sheet of Mazut spray
discharging from the PSA at different ﬂow conditions with constant
temperature. In a PSA, the breakup of the conical liquid sheet occurs
via unstable growth of waves on the surface of the liquid spray ﬁlm
are seen in the images. In the case of lower Reynolds number (Re =
, the ratio of inertial forces to viscous forces, deﬁned by liquid
properties, characteristic length of the injector oriﬁce diameter and
sheet velocity at the tip of the injector), due to low Δpor high viscosity,
the liquid in the rotating chamber does not spill well. High Reynolds
number causes a scattering of the droplets and eventually gives wider
θ. Fig. 3 shows the variation stage as a function of the Weber number
for Mazut 120 °C with different Δp. The Weber number is deﬁned as
, the ratio of inertial forces to surface tension forces,
based on the liquid properties, characteristic length of the sheet
thickness and sheet velocity at the tip of the injector.
θ/2 (spray half cone angle) is deﬁned as the angle between the
spray axis line and a line tangent to the spray cone at the tip of the
atomizer. θcan be obtained by measuring the angle between two
lines (tangential to spray surface at the tip of the injector) which is
depicted in Fig. 3a. The measurements have been conducted manually
by analyzing the raw images from the shadowgraph technique. Edge
of spray is found according to the sharp variation of black color inten-
sity between the inner and outer side of spray at the tip of the injec-
tor. The angle was determined at different Δpand temperature of
Mazut. It can be seen in Fig. 3 that by increasing Δp,θincreases.
The discharging Mazut ﬁlm at T= 120 °C has a smooth surface
with different We
. One of the most critical parameters of the liquid
ﬁlm is the breakup length L
. It is measured as the distance from
the atomizer exit to the location of sheet breakup along the spray
Fig. 2. Images of Mazut spray at a constantΔp= 1.2 MPa with variations of viscosity discharging from PSAapplying an exposuretime of 5 μs. (a) T= 100 °C, (b) T= 110 °C, and (c) T=
Fig. 3. Imagesof Mazut spray at T= 120 °C discharging fromPSA applying an exposure time of 5μswithdifferentΔp. (a) Onion stagewith We
= 4762, (b) Tulipstage with We
and (c) Fully developed spray We
12 S.M.A. Najaﬁet al. / Chinese Journal of Chemical Engineering 28 (2020) 9–22
axis as illustrated in Fig. 3b. The breakup length can be measured by
knowing the pixel width and length from the calibration and measur-
ing breakup length in terms of pixels. Breakup length uncertainty was
calculated to be in the range of 0.2%–1.0%.
Fig. 4 shows the images of Mazut spray at different Δpand temper-
ature. Increase in spray angle and a decrease in L
is evident by the
increase in Δpand increasing temperature. These images are the raw
images, which were taken using the shadowgraph technique and used
for measuring of θ,L
Fig. 5 shows the variation of L
in terms of Δp,T, and We
liquid ﬁlms discharging from the PSA. For each test condition, more
than 100 images of spray captured to estimate the average value of L
A reduction in the viscosity outcomes in an increase in the growth
rate of the most unstable modes, eventually causing lower L
experimentally measured value of L
decreases with increasing We
shown in Fig. 5. The results have followed a trend, which is shown in
Fig. 5c by a hatched zone.
Fig. 6 shows the variation of θ/2 with Δp,T,Re, and We
the liquid ﬁlms discharging from the PSA. The measurements of Rizk
and Lefebvre  and Chen et al.  show a clear increase in the
angle with Δp. No obvious effect on spray angle was observed for injec-
tion pressure change by Dodge and Biaglow  and Decroso . This
variety of trends for Δpeffects on θindicates that the effect of Δpis not
always the same, but might depend on some other parameters, such as
the atomizer geometry or the liquid. An increase of θwith an increase in
Δpis apparent in Fig. 6a, although the magnitude of this increment is
not the same for all cases. A rise in the HFO temperature results in a
considerable increase in θ, Fig. 6b. This trend is in agreement with the
previous study by Ballester and Dopazo .
The half cone angle as a function of Reynolds number, Re, and Weber
are shown in Fig. 6c and d, respectively. The results show
that by increasing Reynolds number and Weber number, θincreases,
as the inﬂuence of temperature on viscosity for HFO is remarkable.
The difference between Fig. 6c and d can be valuable by notice to the
Fig. 4. Mazut spray images at different injection pressure and temperature.
13S.M.A. Najaﬁet al. / Chinese Journal of Chemical Engineering 28 (2020) 9–22
point that temperature dependencyof Re comes from μ
, while for We
it comes from σ
, the former variation is much severed (from 6.9
to 1.1 × 10
while later ranges from 3.14 to 3.26 × 10
Therefore, because of the strong temperature dependency of μ
comparison to σ
(We) for HFO, Re can be a better choice among dimen-
sionless numbers to present a trend for θvariations in different temper-
atures (the trend is shown in Fig. 6c by a hatched zone).
The discharge behavior of the Mazut for the PSA is determined from
the measurements of _
mwith Δpand temperature (viscosity) and is
shown in Fig. 7. Mass ﬂow rate and injection pressure difference were
measured directly with the maximum relative uncertainty of 0.9%–1%
and 0.8%–1.4% respectively.
As it was mentioned previously, Mazut viscosity decreases with an in-
crease in the liquid temperature. Viscous forces make ﬂuid movements
difﬁcult. Therefore an increase in mass ﬂow rate is expected by increasing
the liquid temperature, which is also reported by . This is also ap-
proved by comparing measurements when the liquid temperature in-
creased from T=100°CtoT=110°C.However,thereisadecreasein
the mass ﬂow rate by increasing the liquid temperature from T=110°C
to T= 120 °C. This phenomenon can occur because at around 120 °C,
some volatile petroleum components (Mazut is a multi-component
fuel) begin to boil and separate from the liquid. The presence of vapor
gases in the stream can interfere with the ﬂuid ﬂow and therefore
reduce the amount of discharge.
Fig. 5. Effect of the pressure difference, temperatu re and We
number on breakup length.
Fig. 6. Effects of Δp,T,Re, and We
on spray cone angle.
14 S.M.A. Najaﬁet al. / Chinese Journal of Chemical Engineering 28 (2020) 9–22
The discharge coefﬁcient of the atomizer, C
, is expressed as Eq. (1):
The values of C
as a function of the Δp, temperature, and Re number
are shown in Fig. 8. The curves in the graph are nearly horizontal in
Fig. 8a and c is conﬁrming the inviscid theory  which indicated
that Δpand Re should not inﬂuence the C
dent of Δpand Re number.The variation of the C
with fuel temperature
is depictedin Fig. 8b. The variations of C
and θwith temperatureshould
be attributed to viscosity variations. Because viscosity changes over 30%
within the experiments temperature range, but density and surface
tension change only less than 6%. All the curves in Fig. 8b exhibit the
maximum in C
which is characteristic of PSAs as the ﬂuid viscosity
decreases. At low temperatures, the liquid friction viscosity prevents
central air core from forming; for this reason, the liquid exits as a full
jet. As temperature increases, the liquid in the tangential hole or holes
of the swirl chamber causes the developed of the air core, and the liquid
emerges as an empty annular jet. The consequence is the reduction of
the effective exit area for higher temperatures. In Fig. 8b, shown C
follows a decreasing trend with viscosity; this behavior is also observed
in the previous study by Ballester and Dopazo . The value of C
varies from 0.48 to 0.52 for different temperature and Δp.
The PSA forms a conical liquid sheet, the effects of the surrounding
gas and the liquid viscosity form and grow waves on the liquid sheet.
By growing the wavelength, at a wavelength (called the maximum
unstable wavelength λ
), the liquid sheet breaks to the ligaments and
eventually droplets produce, shown in Fig. 3b. λ
is another macroscopic
spray feature that was measured and discussed in this section.
From the conservation of mass and geometry of conical liquid sheet,
the liquid sheet thickness at the oriﬁce exit can be calculated by
the axial velocity of the liquid sheet at the oriﬁce exit,
according to Eq. (2).
is estimated according to the measurements of θand U,itfollows
Uthe liquid sheet velocity (which is along the conical surface), was
measured from the high-speed records (droplets were tracked in
sequent images) and consequently, the liquid sheet thickness can be
calculated from Eqs. (2) and (3). more details of the formulation can
be found in .
A theoretical relation for the estimation of t
in a PSA in terms of
the atomizer geometrical parameters and ﬂow conditions is proposed
by Rizk and Lefebvre . The investigation revealed that the theoreti-
cal relation could not predict t
accurately when the air-core is not
formed completely, which is the case at T= 100 °C and low-pressure
The liquid sheet Weber number, We
and Reynolds number, Re for a
PSA can be expressed as Eqs. (4) and (5) by Sivakumar et al. :
Fig. 7. Effect of temperature and pressure on Mazut mass ﬂow rate.
Fig. 8. Effect of the pressure difference, temperature and Reynolds number on C
15S.M.A. Najaﬁet al. / Chinese Journal of Chemical Engineering 28 (2020) 9–22
Senecal et al.  carried out the linear stability analysis of liquid
sheet of thickness t
=2hmoving with velocity Uthrough a quiescent
ambient gas and where ρ
is the liquid density, d
diameter of oriﬁce,
σis surface tension and μ
is liquid viscosity. The liquid sheet is charac-
terized by Senecal et al.  in terms of aerodynamic Weber number,
expressed as Eq. (6):
is the density of the ambient gas. According to the theory of
ﬁlm breakup, the instability proliferates on the surface of the liquid
ﬁlm and λ
can be calculated using Eq. (7) :
for long wavelengthðÞ
for short wavelengthðÞ
The relative velocity between the liquid ﬁlm and the surrounding
environment causes instability in the liquid ﬁlm. The instability created
on the surface increases with increasing distance from the atomizer
until it breaks up the liquid ﬁlm and generates ligaments. Instabilities
on the ligament cause breakup and create the droplets this process is
named primary breakup. As the distance increases, aerodynamic force
breaks up the primary droplets and generates smaller droplets, this
process is known as secondary atomization. Fig. 9a shows a hollow
cone liquid ﬁlm of PSA and primary and secondary breakup processes
. The process is important because of controlling the droplets
diameter on downstream. Senecal et al.  modeled the liquid ﬁlm
breakup in 2D. They showed that based on the gas Weber number
), the linear instability in liquid ﬁlms could be classiﬁed into long
wavelength and short wavelength with a critical Weber number
) of 1.69. The long wavelength instability occurs for We bWe
and the short wavelength occurs for We NWe
To measure λ
experimentally, shadowgraph method was used. The
images were calibrated to deﬁne the pixel length for each frame. There-
fore, by measuring the length of λ
in pixels, it can be converted to the
length of λ
in meters (the procedure is similar for measuring L
wavelength was measured for different Weber numbers. Typically,
wavelengths were obtained by averaging 25–35 images of a particular
condition. Theoretically predicted λ
 compared with the measure-
ments as is shown in Fig. 10. Wavelength average relative uncertainty
was in the range of 2.7%–4%. The comparison reveals the fact that the
theory can also be reliable for heavy viscous liquids but with underesti-
mation. Two curves are plotted in Fig. 10 that is ﬁtted to the experimen-
tal measurements. The overall format of the formulation in Eq. (7) was
preserved for the curve ﬁt. The differencebetween theoretical equations
(Eq. (7)) and the ﬁtting for the experiments, shown in Fig. 10 is only
changing the coefﬁcient of 2 with 2.8 for long wavelength and 3/2
with 2.2 for short wavelength.This correction is a signiﬁcant conclusion
as the waves are a representation of the instabilities and lead to the
Fig. 9. a) Liquid ﬁlm, primary and secondary breakup processes, b) Determined axial locations for SMD measurements.
Fig. 10. The comparison between the theoretically predicted most unstable wavelength
) and the experimentally measured one for the Mazut PSA.
16 S.M.A. Najaﬁet al. / Chinese Journal of Chemical Engineering 28 (2020) 9–22
primary and secondary breakup. Therefore, it can be expected that SMD
variations should follow the SMD trends for conventional liquids.
3.2. Microscopic properties
It can be seen obviously in Fig. 3c that after breakup only the liga-
ments and droplets atthe focus zone were captured. Overlapping occurs
rarely, and this is because of thefollowing reasons: a) usinga macro lens
with the beneﬁts of limited depth of ﬁeld and focus on close objects,
b) hollow conical spray, and c) dilute spray.
Axial measurements were made throughout the spray at various
Weber and Reynolds numbers (i.e., at different Δpand temperatures).
As the viscosity decreases with temperature increase, the different
liquid temperature can be representative of the viscosity change.
Droplets diameters were measured using the shadowgraph method.
To measure the droplets diameter, by image processing, the pixel area is
determined by calibration, and the diameter size of each droplet d
determined by the number of pixels that occupied by the droplet.
SMD or D
can be obtained by, SMD ¼PI
the total number of droplets in each image. The SMD can be deﬁned
as the diameter of a droplet having the same volume/surface area
ratio as the entire spray. SMD relative uncertainty was calculated to be
in the range of 4%–6%.
SMD was obtained for different temperatures and Δpfrom the
breakup to Z= 70 mm and summarized in Fig. 11. Also, an increase in
Weber number leads to a decrease in SMD. In overall, higher tempera-
tures (lower viscosity) and Δplead to lower SMD.
There are dozens of empirical correlations published for SMD in the
literature. Available SMD calculations on pressure-swirl injectors for
HFO atomization was compared in . In the present study, four
best-predicted correlations [39–42] are compared with experimental
measurements. It is depicted in Fig. 12a that Ballester correlation 
is the best correlation in SMD prediction for Mazut atomization
in T= 120 °C. Some best-ﬁtted available correlations on θ[28,35,43,
44] were also evaluated using HFO injection in T=120°C.Itisshown
in Fig. 12b that Varde  correlation was almost successful in
predicting θfor HFO.
In Fig. 9b some selected axial locations are determined. SMD mea-
surements were applied to these locations. Fig. 13 shows SMD measure-
ments at different axial locations, Δp, and liquid temperatures. Results
show the ﬁnest spray atomization can be obtained around Z=
70 mm, atomization is completed at the Δp=1.2MPaandT=120°C.
SMD decreases with an increase in axial location. Less viscosity (Mazut
with higher temperature) forms smaller droplet size in the spray at the
same axial location. Droplet sizes at Zlarger than 70 mm at Δp=
1.2 MPa and T= 120 °C are less concerned for HFO gasiﬁer applications
because droplets there will interact with the gasiﬁer gas ﬂow and quickly
evaporate and change to smaller droplets. Therefore, droplet size at Z
smaller than 70 mm is of great importance. As shown in Fig. 13 at Z=
70 mm the atomizer produces droplet with SMD about 100 μmat
Δp= 1.2 MPa and T= 100 °C and SMD about 70 μmattheΔp=
1.2 MPa and T= 120 °C. Measurements show an increase in Δpfrom
0.7 to 1.2 MPa leads to the production of smaller droplets and decrees
in SMD of about 9% when T= 110 °C and decrees in SMD of 16% when
T= 120 °C. An increase in temperature from 110 °C to 120 °C produced
smaller droplets and resulted in a decrease in SMD of about 11% when
Δp= 0.7 MPa, about 6% when Δp= 9 MPa, about 3% when Δp=
1.0 MPa, and about 2% when Δp=1.2MPa.
In the process of gasiﬁcation, fuel droplets evaporate and then the
reactions take place in the gas phase. As the theoretical results show,
the evaporation time depends on droplets size. As the diameter de-
creases, the evaporation time decreases, which is following the d
. In HFO gasiﬁer chamber, evaporation of large droplets generally
Fig. 11. Global SMD variations among Δp,T,We
17S.M.A. Najaﬁet al. / Chinese Journal of Chemical Engineering 28 (2020) 9–22
takes a long time. As a result, an increase in the number of large droplets
reduces the gasiﬁer efﬁciency. In fact, larger droplets determine the
evaporation time; therefore,it is crucial to identifyand resolve the drop-
let distribution . Since the proportion of large droplets may be
different even in the case of the same SMD. Therefore, a detailed inves-
tigation of cumulative droplet volume, spray global probability density
function (PDF), and relative number count of droplets is necessary.
The histogram plots of droplet size distribution at different axial
locations (40 mm, 50 mm,60 mm and 70 mm) for different Δpand dif-
ferent temperature are shown in Figs. 14 and 15. SMD for each location
is noted in these ﬁgures. In the ﬁgures, the histograms represent the cu-
mulative volume versus the droplet diameter. At Z=40mmforΔp=
1.2 MPa and T= 120 °C, the peaks of the histogram locate cumulative
volume of droplets reaches 25% at 35 μm and 22% at 65 μm. At the Z=
60 mm axial location, the cumulative volume of smaller droplets
(Db110 μm) increase. The maximum cumulative volume of the
droplets was 32% at 35 μmand24%at55μm. It indicates that more
small droplets were created. At the downstream of the spray (Z=
0.7 mm), the cumulative value reaches 42% at 35 μm.
At the Z=40mmregionforΔp=1.2MPaandT= 110 °C, the max-
imum of cumulative volume of droplets reaches 35% at 35 μm. At the Z=
60 mm axial location the value was 32% at 35 μm. At the downstream of
the spray (Z= 70 mm), the value reaches 32% at 35 μm.
The distribution is advantageous to understand the effect on tem-
perature and Δpin spray formation. To represent droplet sizing, some
distribution functions are used, the most well-known distribution is
Rosin–Rammler, as shown in Eq. (8):
where drepresents the droplet diameter (μm) and Y
the volume frac-
tion of the droplets with a diameter greater than d,dis the size constant,
and nis the size distribution parameter. Cumulative volume curves are
Fig. 12. The comparison of SMD and θ/2 variations for differentinjection pressures, measured data and the correlations on PSA, T= 120 °C.
Fig. 13. SMD for different Δpand Tof different axial location.
18 S.M.A. Najaﬁet al. / Chinese Journal of Chemical Engineering 28 (2020) 9–22
typical Rosin–Rammler distributions, the corresponding data for
different Δpand temperature are shown in Fig. 16. In addition, the cor-
responding constants (nand d) are calculated by curve ﬁtting of the
measurements and noted in these ﬁgures. The results show that the
experimental data could be presented by Rosin-Rammler distributions
very well. It can be also concluded that d(mean diameter) decreases
with increase in temperature and injection pressure. That is rational be-
cause of lowering viscosity and increasing momentum, respectively.
Fig. 14. Relative number count for different axial location (40 mm, 50 mm, 60 mm and 70 mm) for Δp= 1.2 MPa with different temperature (T= 100 °C, T= 110 °C, T=120°C).
Fig. 15. Relative number count for different axial location (40 mm, 50 mm, 60 mm and 70 mm) for Δp= 1.0 MPa with different temperature (T= 100 °C, T= 110 °C, T=120°C).
19S.M.A. Najaﬁet al. / Chinese Journal of Chemical Engineering 28 (2020) 9–22
In the numerical simulation of some applications, which liquid injec-
tion is a part of the simulation, knowing the droplet diameter distribu-
tion can facilitate the simulation of the injector. Some droplets with
speciﬁed diameters can be injected directly to the domain that would
lead to the saving in CPU time (omitting the modeling of atomization).
So reporting SMD distributions like Rosin-Rammler can be highly bene-
ﬁcial in some numerical simulations.
A global PDF gives the distribution of droplet size in overall spray at-
omization. This function is given as Eq. (9):
is the mean droplet size of size class i,N
is the number of drop-
lets in size class i, and ΔDis the width of the size classes. Global size
measurement of droplets is done (ΔD=10μm). Fig. 17 shows spray
global PDF versus the droplet diameter.
Fig. 17 represents the overall distribution of droplets diameters, but
Figs. 14 or 15 represent the local distribution at different axial locations.
Therefore, for instance, the PDF for Δp=1.2 MPa and T= 120 °C shown
in Fig. 17 is the summation of data presented inFig. 14 for Δp=1.2 MPa,
T= 120 °C and Z= 40, 50, 60, 70 mm. Comparing these two data, one
can conclude that the results for Fig. 17 is closer to a log-normal distri-
bution which is desired for droplet diameter distribution using PSA.
Droplets mean diameters can be deﬁned according to the different
applications. The deﬁnition and the formulation for different mean di-
ameters are presented by Lefebvre and McDonell . The correspond-
ing data for different pressure difference and temperature of Mazut
injection is presented in Table 2.
Mazut sprays of PSA were measured using the shadowgraph tech-
nique to study the effects of liquid viscosity and Δpon atomization.
Fig. 16. Cumulative droplet volume and Rosin-Rammler Distribution and experiment data for different pressure and temperature Mazut.
Fig. 17. Spray global Probability Density Function and the global mean droplet sizes.
20 S.M.A. Najaﬁet al. / Chinese Journal of Chemical Engineering 28 (2020) 9–22
The experimental measurements include _
,θ, and global and
local SMD. The maximum unstable wavelength was measured experi-
mentally and was compared with previously reported liquid ﬁlm
breakup models. The corresponding result shows that the measured
data follows the same trend as the model but with about 30% underes-
timation (for a highly viscous liquid). The model was modiﬁed by
changing the coefﬁcients from 1.5 and 2 to 2.2 and 2.8 respectively for
short and long wavelength. It can be concluded from the measurements
that an increase in Weber number leads to a decrease in SMD. The ﬁnest
droplets were observed at an axial distance of 70 mm in the experimen-
tal range for T= 120 °C and Δp= 1.2 MPa. At this distance, the tested
atomizer can produce droplets with SMD less than 80 μmatΔp=
0.7 MPa for liquids with dynamic viscosity 0.0063 (kg·m
The comparisons of the available correlations with the measure-
ments of SMD and θshow a good agreementfor Ballester and Varde cor-
relations respectively. Spray quality enhancement by increasing
injection pressure and temperature is revealed by results of cumulative
droplet volume, spray global PDF and relative number count of droplets.
 M. Vaezi, M. Passandideh-Fard, M. Moghiman, etal.,Gasiﬁcation of heavy fuel oils: A
thermochemical equilibrium approach, Fuel 90 (2) (2011) 878–885.
 M. Peters, W. Francis, Fuel and Fuel Technology, Pergamon Press, 1980.
 A. Saario, A. Rebola, P. Coelho, et al., Heavy fuel oil combustion in a cylindrical labo-
ratory furnace: measurements and modeling, Fuel 84 (4) (2005) 359–369.
 Z. Wan g, M. Liu, X. Cheng, et al., Experimental study on oxy-fuel combustion of
heavy oil, Int. J. Hydrog. Energy 42 (31) (2017) 20306–20315.
 M. Meratizaman, S. Monadizadeh, A. Ebrahimi, etal., Scenario analysisof gasiﬁcation
process application in electrical energy-freshwater generation from heavy fuel oil,
thermodynamic, economic and envir onmental assessment, Int. J. Hydrog. Energy
40 (6) (2015) 2578–2600.
 A. Di Carlo, E. Bocci, V. Naso, Process simulation of a SOFC and double bubbling ﬂu-
idized bed gasiﬁer power plant, Int. J. Hydrog. Energy 38 (1) (2013) 532–542.
 Z. Chen, S. Yuan,Q. Liang, et al., Distribution of HCN, NH
in an entrained
ﬂow gasiﬁer, Chem. Eng. J. 148 (2–3) (2009) 312–318.
 M.A. Salam, K. Ahmed, N. Akter, et al., A review of hydrogen production via biomass
gasiﬁcation and its prospect in Bangladesh, Int. J. Hydrog. Energy 43 (32) (2018)
 Z. Li, Y. Wu, H. Yang, et al., Effect of liquid viscosity on atomization in an internal-
mixing twin-ﬂuid atomizer, Fuel 103 (2013) 486–494.
 L. Wang, C.L. Weller,D.D. Jones, et al., Contemporary issuesin thermal gasiﬁcation of
biomass and its application to electricity and fuel production, Biomass Bioenergy 32
(7) (2008) 573–581.
 P. Mikaniki, S.M. Najaﬁ, H. Ghassemi, Experimental study of a heavy fuel oil atomi-
zation by pressure-swirl injector in the application of entrained ﬂow gasiﬁer, Chin.J.
Chem. Eng. 27 (4) (2019) 765–771.
 Z. Liu, Y. Huang, L. Sun, Studies on air core size in a simplex pressure-swirl atomizer,
Int. J. Hydrog. Energy 42 (29) (2017) 18649–18657.
 M. Rashad, H. Yong, Z. Zekun, Effect of geometric parameters on spray characteris-
tics of pressure swirl atomizers, Int. J. Hydrog. Energy 41 (35) (2016) 15790–15799.
 G. Ferreira, J.A. Garcia, F. Barreras, et al., Designoptimization of twin-ﬂuid atomizers
with an internal mixing chamber for heavy fuel oils, Fuel Process. Technol. 90 (2)
 T. Suzuki, H. Nishida, N. Hashimoto, et al., Hollow-cone Spray of Viscous Liquid in
High-pPressure Gas Environment-Experimental Investigation for the Application
of New Liquid Fuels to Gas -turbine, ICLASS 2012, 12th Triennial International
Conference on Liquid Atomization and Spray Systems, Heidelberg, Germany, 2012
 R. Payri, F. Salvador, J. Gimeno, et al., The effect of temperature and pressure on
thermodynamic properties of diesel and biodiesel fuels, Fuel 90 (3) (2011)
 X. Zhang, C. Shen, P. Cheng, et al., An image-processing based method for the mea-
surement of the ﬁlm thickness of a swirl atomizer, J. Vis. 20 (1) (2017) 1–5.
 J.-W. Ding, G.-X. Li, Y.-S. Yu, The instability and droplet size distribution of liquid-
liquid coaxial swirling spray: An experimental investigation, Exp. Thermal Fluid Sci.
82 (2017) 166–173.
 S. Kim, T. Khil, D. Kim, et al., Effect of geometric parameters on the liquid ﬁlm thick-
ness and air core formation in a swirl injector, Meas. Sci. Technol. 20 (1) (2008),
 C. Liu, F. Liu, J. Yang, et al., Experimentalinvestigations of spraygenerated by a pres-
sure swirl atomizer, J. Energy Inst. 92 (2) (2019) 210–221.
 M. Mlkvik, P. Stahle, H.P. Schuchmann, et al.,Twin-ﬂuid atomization of viscous liq-
uids: The effect of atomizer construction on breakup process, spray stability and
droplet size, Int. J. Multiphase Flow 77 (2015) 19–31.
 L. Broniarz-Press, S. Wlodarczak, M. Matuszak, et al., The effect of oriﬁce shape and
the injection pressure on enhancement of the atomization process for pressure-
swirl atomizers, Crop Prot. 82 (2016) 65–74.
 M.Huth,A.Heilos,Fuelﬂexibility in gas turbine systems: iMPact on burner
design and performance, Modern Gas Turbine Systems, Elsevier 2013,
 Steinen, Steinen nozzle catalog, [cited 2019 17 January]; Available from https://
 M.R. Riazi, T.E. Daubert, Characterization parameters for petroleum fractions, Ind.
Eng. Chem. Res. 26 (4) (1987) 755–759.
 A. Azimi, A. Arabkhalaj, R.S. Markadeh, et al., Fully transient modeling of the heavy
fuel oil droplets evaporation, Fuel 230 (2018) 52–63.
 M. Ochowiak, M. Matuszak, S. Wlodarczak, et al., The concept design and study of
twin-ﬂuid effervescent atomizer with air stone aerator, Chem. Eng. Process. Process
Intensif. 124 (2018) 24–28.
 N. Rizk, A. Lefebvre, Prediction of velocity coefﬁcient and spray cone angle for sim-
plex swirl atomizers, Int. J. Turbo Jet Engines 4(1–2) (1987) 65–74.
 S. Chen, A. Lefebvre, J. Rollbuhler, Inﬂuence of geometric features on the perfor-
mance of pressure-swirl atomizers, J. Eng. Gas Turbines Power 112 (4) (199 0)
 L. Dodge, J. Biaglow, Effect of elevated temperature and pressure on sprays from
simplex swirl atomizers, J. Eng. Gas Turbines Power 108 (1) (1986) 209–215.
 S. DeCorso, Effect of ambient and fuel pressure on nozzle spray angle, Trans. ASME
79 (3) (1957) 607–615.
 J. Ballester,C. Dopazo, Discharge coefﬁcient and sprayangle measurements for small
pressure-swirl nozzles, Atomization Sprays 4 (3) (1994) 351–367.
 B.R. Munson, T.H. Okiishi, W.W. Huebsch, et al., Fluid Mechanics, Wiley, Singapore,
 D. Sivakumar, S. Vankeswaram, R. Sakthikumar, et al., Analysis on the atomization
characteristics of aviation biofuel discharging from simplex swirl atomizer, Int. J.
Multiphase Flow 72 (2015) 88–96.
 N. Rizk, A.H. Lefebvre, Internal ﬂow characteristics of simplex swirl atom izers, J.
Propuls. Power 1 (3) (1985) 193–199.
 P. Senecal, D.P. Schmidt,I. Nouar, et al., Modeling high-speed viscous liquid sheet at-
omization, Int. J. Multiphase Flow 25 (6–7) (1999) 1073–1097.
 A. Saha, J.D. Lee, S. Basu, et al., Breakup and coalescence characteristics of a hollow
cone swirling spray, Phys. Fluids 24 (12) (2012), 124103.
 A. Lefebvre, Atomization and Sprays, Combustion: An International Series, Hemi-
sphere Pub. Corp, 1989.
 A. Radcliffe, The performance of a type of swirl atomizer, Proc. Inst. Mech. Eng. 169
(1) (1955) 93–106.
 A. Jasuja, Atomization of crude and residual fuel oils, J. Eng. Power 101 (2) (1979)
 J. Ballester, C. Dopazo, Drop size measurements in heavy oil sprays from pressure-
swirl nozzles, Atomization Sprays 6 (4) (1996) 377–408.
Different mean diameters and their applications for different pressure differences and temperatures of Mazut
Mean diameter Name Length Surface area Volume Volume–length Sauter mean diameter
Application Comparisons Surface area controlling Volume controlling, hydrology Evaporation, molecular diffusion Mass transfer, reaction
T= 100 °C Δp= 1.0 MPa 71.44 81.11 89.57 100.30 109.24
Δp= 1.2 MPa 56.67 65.13 73.65 83.96 94.17
T= 110 °C Δp= 0.7 MPa 65.91 71.99 77.75 84.44 90.67
Δp= 0.9 MPa 54.43 60.54 67.13 74.55 82.53
Δp= 1.0 MPa 48.59 54.48 61.36 68.96 77.86
Δp= 1.2 MPa 52.03 57.32 62.88 69.12 75.67
T= 120 °C Δp= 0.7 MPa 51.67 58.18 64.95 72.82 80.95
Δp= 0.9 MPa 53.47 58.88 64.62 71.04 77.82
Δp= 1.0 MPa 52.16 57.32 62.76 68.83 75.23
Δp= 1.2 MPa 53.67 58.29 62.91 68.11 73.28
21S.M.A. Najaﬁet al. / Chinese Journal of Chemical Engineering 28 (2020) 9–22
 W.v. Ohnesorge, Formation of drops by nozzles and the breakup of liquid jets, Z.
Angew. Math. Mech. 16 (4) (1936) 355–358.
 E. Giffen, The Atomisation of Liquid Fuels, Chapman & Hall, 1953.
 K. Varde, Spray cone angle and its correlation in a high pressure fuel spray, Can. J.
Chem. Eng. 63 (2) (1985) 183–187.
 K. Annamalai, I.K. Puri, Combustion Science and Engineering, CRC press, 2006.
 A. Miskam, Z. Zainal, I. Yusof, Characterization of sawdust residues for cyclone gas-
iﬁer, J. Appl. Sci. 9 (12) (2009) 2294–2300.
 A.H. Lefebvre, V.G. McDonell, Atomization and Sprays, CRC press, 2017.
22 S.M.A. Najaﬁet al. / Chinese Journal of Chemical Engineering 28 (2020) 9–22