Surface & Coatings Technology 280 (2015) 370–377

Contents lists available at ScienceDirect

Surface & Coatings Technology

journal homepage: www.elsevier.com/locate/surfcoat

Numerical investigation on effects of nanoparticles on liquid feedstock

behavior in High Velocity Oxygen Fuel (HVOF) suspension spraying
Ebrahim Gozali a, Mahrukh Mahrukh b, Sai Gu b,⁎, Spyros Kamnis c,⁎
a
School of Engineering Sciences, University of Liverpool, Liverpool L69 3BX, UK
b
Cranfield University, Cranfield, Bedford, Bedfordshire MK43 0AL, UK
c
Monitor Coatings Limited, 2 Elm Road, North Shields Tyne & Wear, NE29 8SE, UK

a r t i c l e i n f o a b s t r a c t

Article history: Suspension feedstock in high velocity oxy-fuel flame jets has opened a new area of research with great potential
Received 5 September 2014 for advanced coatings. Understanding the suspension behavior in such a multidisciplinary process is a key factor
Revised 17 July 2015 in producing repeatable and controllable coatings. In this study, the effects of solid nanoparticles, suspended in
Accepted in revised form 6 September 2015
liquid feedstock, on suspension fragmentation, vaporization rate and gas dynamics are investigated in the High
Available online 11 September 2015
Velocity Oxygen Fuel (HVOF) suspension spraying process by numerical modeling. The model consists of several
Keywords:
sub-models that include pre-mixed combustion of propane–oxygen, non-premixed ethanol–oxygen combus-
Nanoparticles tion, modeling aerodynamic droplet break-up and evaporation, heat and mass transfer between liquid droplets
Suspension and gas phase. Moreover, the thermo-physical properties of suspension (mixture of solid nanoparticles and liquid
HVOF suspension spraying process solvent) are calculated from theoretical models. The results show that small droplets carrying high nanoparticle
Thermal spray concentrations develop higher surface tension and result in less fragmentation. The recommended ethanol drop-
Numerical modeling let size at high nanoparticle loadings is found to be 50 μm due to the high evaporation rate in the mid-section of
the nozzle. For larger droplets, severe fragmentation occurs inside the combustion chamber (CC) while complete
evaporation takes place in the free jet region outside the gun.
© 2015 Published by Elsevier B.V.

1. Introduction constraints imposed by the experiments. Most importantly, in such a

High Velocity Oxygen Fuel (HVOF) suspension spraying is based on
the conventional high velocity oxy-fuel (HVOF) thermal spraying pro-
cess and was developed with the aim of spraying submicron or nano-
particle suspensions with hypersonic speed to deposit thin and very
dense coatings [1–3]. HVOF suspension spraying uses a liquid solvent
as a carrier fluid to process nano-scale materials, in which the coating
material is in the form of a suspension. A suspension is a heterogeneous
mixture containing solid particles and a solution or solvent (water, eth-
anol, or isopropanol) [4]. Although a fairly wide variety of coating mor-
phologies can be obtained, the size of microstructural features within
the coatings is typically governed by that of the feedstock. Therefore,
understanding liquid feed stock behavior in such a multidisciplinary
system is important for its further development.
The HVOF suspension spraying process involves complex stages of
droplet fragmentation, liquid droplet evaporation, nanoparticle ag-
glomeration and nanoparticle/gas heat transfer coupling. The lack of ex-
perimental data and quantitative analyses of the internal regions of the
torch makes modeling and numerical methods valuable tools for under-
standing the overall physics of the process by overcoming the technical
⁎ Corresponding authors.
E-mail address: Spyros@monitorcoatings.com (S. Kamnis).
multidisciplinary process chemico-physical parameters are closely
linked and make effective control of the process very challenging [5,6].
Composition of suspension feedstock, for example percentage of nano-
particles in the liquid solvent, is one of the most important parameters
that affects the system outcome.
In the HVOF suspension spraying process the concentration of
suspended solid nanoparticles in the solvent may vary from one case to
another, depending on the application. The thermo-physical properties
of suspension such as density, viscosity, specific heat, thermal conductiv-
ity and surface tension do not remain constant but differ according to
nanoparticle concentration [7–13]. In addition, rate and final location of
droplet evaporation in a thermal spray torch is critically governed by
the physical properties of the solvent. This implies that assigning pure
solvent properties or averaging solid and liquid properties in the suspen-
sion analysis may produce unrealistic results leading to a high level of er-
rors in numerical simulations. This study will highlight this problem by
providing a quantitative comparison between the vaporization rate of a
typical homogeneous liquid solvent (ethanol) and non-homogeneous
liquid suspension (ethanol carrying nanoparticles).
It must be noted that numerical analysis of the HVOF suspension
spraying process has rarely been documented in the literature. Dongmo
et al. [14] investigated the process by using the TopGun-G torch (GTV
Verschleißschutz GmbH, Betzdorf, Germany). This modeling contains

http://dx.doi.org/10.1016/j.surfcoat.2015.09.012
0257-8972/© 2015 Published by Elsevier B.V.
 E. Gozali et al. / Surface & Coatings Technology 280 (2015) 370–377
an overall discussion of the HVOF suspension spraying process in which
both liquid ethanol droplets (300 μm) and solid titania particles (0.5–
50 μm) are injected from the gun inlet with identical mass flow rates
as discrete phases. The results showed that solid particles and liquid
droplets moved with different velocities in the domain due to the differ-
ences in their properties. In another research study from this group, the
combustion chamber of the gun was modified to a conical shape
resulting in an increase in the process efficiency and avoiding nanopar-
ticles contact with the combustion chamber walls [6]. However, in these
studies the change in the liquid feedstock properties due to the nano-
particle inclusion was ignored.
The aim of this work is to analyze the influence of nanoparticle
concentration and droplet diameter on the vaporization rate, the
secondary breakup of the liquid phase, and the HVOF suspension
spraying gas dynamics. The numerical analysis for this study consists
of modeling pre-mixed (oxygen/propane) and non-premixed
(oxygen/ethanol) combustions, interaction between gas and liquid
phases (two-way coupling), secondary break-up and vaporization
of droplets. Moreover, for modeling the suspension, the thermo-
physical properties are calculated from the theoretical models. This
work is based on and continues the numerical analysis of the con-
ventional HVOF thermal spray process as described in earlier studies
[15,16]. Extensive validation of the combustion, discrete phase and
flow model has been performed in earlier studies and therefore for
brevity is not repeated here [15–19]. In addition, the details of vali-
dated discrete phase break up sub-models employed in this study
can be found elsewhere [19,20]. Direct quantitative validation of
the flow field within the gun is not possible with the current
experimental equipment so the validation relies more on qualitative
data.
2. Model description
The first part of the model simulates the temperature and velocity
fields of the HVOF suspension spraying flame jet using the example of
an industrial DJ2700 torch (Wohlen, Switzerland) as shown in Fig. 1.
Working conditions for the simulations are summarized in Table 1
[21]. The Realizable k–ε model is applied here to simulate the gas
phase turbulence and the gas flow is assumed to be compressible.
Fig. 1. Schematic diagram of the axisymmetric computational domain and boundary
conditions.
371
Table 1
Operating parameters for the HVOF suspension spraying process.
Working conditions
Propane Flow rate: 0.003526 kg/s Temperature: 300 K
Oxygen Flow rate: 0.01197 kg/s Temperature: 300 K
Atmosphere Pressure: 101,325 Pa Temperature: 300 K
Internal wall boundary – Temperature: 300 K
Furthermore the eddy dissipation model is employed to model pre-
mixed (oxygen/propane) combustion with hyper-stoichiometric oxy-
gen mass fraction.
Discrete phase modeling (DPM) is used in this study to model the
trajectory of the droplet/particle phase. The suspension properties un-
dergo significant changes from the injection point to the landing loca-
tion. The first stage is the aerodynamic break-up of droplets, as the
slow moving droplets are entrained into the jet and accelerate in the
high velocity gas stream. Depending on their size, the thermophysical
properties of the liquid, and their interaction with the surrounding
gas, droplets undergo severe deformation and then break up into small-
er droplets. For the secondary breakup simulation, the Taylor Analogy
Breakup (TAB) model is applied. The TAB model is suitable for droplets
with low Weber numbers (We b100) and the model is, also, well
adapted to the conditions of spraying and validated in earlier work
[19]. A phase change will occur for liquid ethanol droplets after evapo-
ration. Then the ethanol gas has to mix with excess oxygen, which re-
mains after complete propane oxidation, due to molecular transport
(diffusion) prior to reaction. Similar to the pre-mixed combustion, the
eddy dissipation model is used to model this non-premixed combus-
tion. The propane and oxygen flow rates indicated in Table 1 provide
maximum flame temperature and velocity for DJ2700 torch (Sulzer-
Metco) with hyper-stoichiometric oxygen mass fraction. For the
HVOF suspension spraying, the oxygen mass fraction must be
hyperstoichiometric in relation to complete ethanol combustion be-
cause the oxygen remnant will be used for ethanol combustion as diffu-
sion flame. Mass flow rate of liquid ethanol droplets is 0.0001 kg/s,
which is selected according to the results in the previous study [21].
The dissolved powder content is not included in this study and will be
examined in a future paper instead the thermophysical properties of
the suspension (mixture of liquid ethanol and titania powders) is calcu-
lated by theoretical models which in detail will be described in the fol-
lowing section.
Detailed descriptions of the gas phase, discrete phase, breakup, com-
bustion models and droplet heat and mass transfer with continuous
phase can be found elsewhere [15,16,19,21–24].
2.1. Effect of suspension
To model the suspension, the effect of nano-size particles on the
liquid phase is considered as the change in the liquid bulk density,
viscosity, specific heat, thermal conductivity, and surface tension. The
density of suspension is defined as (Eq. (1)) [7]:
ρsusp ¼ ð1−C Þρl þ Cρp ð1Þ
where C is the volume concentration of solid particles in suspension, ρl
is the density of liquid ethanol and ρp. is the density of solid titania — in
this study the titania powders are considered to have a density of
4230 kg/m3. Since the dissolved powder content is not charged into
the suspension and instead its thermophysical properties are calculated
by the theoretical models, discussing the type and size distribution of
the nanoparticles are of less significance and it will be neglected in
this work.
For modeling the viscosity of suspension, Einstein's formula is re-
stricted for the low volume concentrations {μsusp = μ(1 + 2.5C)} and
372 E. Gozali et al. / Surface & Coatings Technology 280 (2015) 370–377

it was modified by Brinkman for higher concentrations (Eq. (2)) along the axis. In the Rosin–Rammler size distribution, the mass fraction
[26]: of droplets diameter greater than ‘d’ is given by:

μl n
μ susp ¼ ð2Þ Y d ¼ e−ðd=dÞ ð6Þ
ð1−C Þ2:5

where μl is the viscosity of liquid ethanol. where d is the size constant ‘Mean Diameter’ and n is the size distribu-
Specific heat is defined as [9]: tion parameter ‘Spread Parameter’ [28].
Without Droplets (Case-1) in the text and figures refers to a case in
csusp ¼ Ccp þ ð1−C Þcl ð3Þ which droplets are not injected and only the gas dynamics of the
HVOF suspension spraying process is analyzed.
here cp is the specific heat capacity of titania powder (3780 J/kg-K), and
cl is the specific heat capacity of liquid ethanol. 3.1. Gas dynamics and vaporization rate of droplets
Experimental analyses have shown that thermal conductivity (ksusp)
of nano-fluids increases with nanoparticle concentration. This also de- 3.1.1. Injection of droplets of constant sizes
pends on size, shape and temperature of suspended particles [9,13]. Three cases (Case-2.1, Case-2.2, and Case-2.3) with diameters of 50,
For thermal conductivity of spherical nanoparticles the Bruggeman 150, and 300 μm, respectively, are selected to analyze the HVOF suspen-
model gives better prediction [27] than other models with no limitation sion spraying gas dynamics and vaporization rate of the suspension.
on the concentration. Droplets in Case-2.1 are injected into the HVOF suspension spraying
flow at a velocity of 15 m/s, whereas in the two latter cases the injection
velocity is 30 m/s. The initial mass flow rate for all cases is specified to be
1
 k pffiffiffiffi
ksusp ¼ ð3C−1Þkp þ ð2−3C Þkl þ l Δ; ð4Þ 0.1 g/s. These values for injection velocity and initial mass flow rate are
4 4 selected based on our previous parametric investigation which was
"  2 # aimed to explore optimum operating parameters for the HVOF suspen-
kp  k 
Δ ¼ ð3C−1Þ 2 2
þ ð2−3C Þ þ 2 2 þ 9C þ 9C 2 p
; ð5Þ sion spraying process [21]. The numerical results revealed that injection
kl kl velocity increases to some extent with droplet diameter and this is
probably due to fact that higher injection force is required for larger
here kp is the thermal conductivity of titania powder (10.4 W/m-K) and droplets in order to penetrate them into the HVOF jet. The initial tem-
kl is the thermal conductivity of the base fluid. perature of the droplets is assumed to be 300 K. Each case is simulated
It has been reported [11,12] that surface tension of an ethanol based with four different solid particle concentrations; namely 0, 5, 15, and
suspension is almost the same as pure ethanol for low particle concen- 25 wt.%.
trations (up to 3 wt.%). However, a ten percent increase was observed Fig. 2(a–c) shows the temperature and velocity profile of the HVOF
for a suspension with higher volume concentrations (N10 wt.%). This suspension spraying flow including the vaporization rate of the suspen-
is due to an augmentation in the Van der Waals forces between nano- sion with a diameter of 50 μm along the centreline axis. From this figure,
particles at the interface of liquid and gas, which leads to higher surface the cooling effect due to the liquid ethanol droplet evaporation for the
tension in the suspension. In this study, a ten percent increase in surface cases at the ends: 0 wt.% and 25 wt.% are found to be approximately
tension is considered for suspension with high concentrations (15 and 700 K and 300 K close to the nozzle entrance, respectively, compared
25 wt.%). to Without Droplet Case-1 (Fig. 2-a). Thus the maximum temperature
The suspension properties are calculated from commonly used the- difference between the cases is 400 K. This, also, occurs after
oretical models and temperature dependent pure liquid properties are convergent-divergent section but with opposite trend. However, the ve-
incorporated [25] and curve-fitted procedure is applied in their temper- locity difference between the cases is not considerable because the
ature range as shown in Table 3. amount of cooling is not high enough to change the pressure and the ve-
locity. Fig. 2-c shows that the gas temperature cooling in the HVOF sus-
3. Numerical results and discussion pension spraying process is directly related to the droplet vaporization
rate, which means the lower the evaporation rate, the less gas temper-
This section addresses how the HVOF suspension spraying gas dy- ature cooling and vice versa. Moreover, Fig. 2-c shows the highest va-
namics, rate of evaporation and secondary break up of liquid droplets porization rate of 1.83 × 10−7 kg/s for the homogenous droplets,
are affected by the addition of nanoparticles into the liquid feedstock. while the addition of 25 wt.% nanoparticles into the base fluid (ethanol)
A surface type injection scheme is used in this study. For this scheme, reduces it by 20%. As can be seen from the aforementioned figure, the
three different constant diameters and Rosin–Rammler distribution maximum rate of evaporation occurs inside the combustion chamber
are considered, where each case (liquid droplet) is investigated with for all the cases, while the final location of evaporation stretches to
four different nanoparticle concentrations (0, 5, 15, and 25 wt.%). In the gun exit when droplets are loaded with higher concentrations. The
this scheme, streams of liquid droplets are released axially from a sur- reason for this is that at the beginning dense droplets with higher spe-
face and the solver simulates the interaction of droplets with the com- cific heat are slow to vaporize, then due to the growing higher relative
bustion gases along the central axis of the torch as droplets move velocities between the phases they disintegrate and evaporate with
delay compared to those with less density and lower specific heat.
Table 2 This can be supported with the vaporization laws which are used to
Injection diameters for the HVOF suspension spraying process. model droplets evaporation and mass transfer to the continuous
phase. Detail description of these laws and theory behind them can be
Injection
diameters found elsewhere [24,29,30].
However, when droplets are injected with a diameter of 150 μm and,
Constant 50 μm 150 μm 300 μm
diameters Case-2.1 Case-2.2 Case-2.3 again, four different nanoparticle concentrations, the maximum tem-
Rosin–Rammler 30 μm, 40 μm, 130 μm, 140 μm, 280 μm, 290 μm, perature difference between the cases at the ends: 0 wt.% and 25 wt.%
distribution 50 μm, 60 μm, 150 μm, 160 μm, 300 μm, 310 μm, is almost 200 K before the convergent–divergent section, see Fig. 2-d.
70 μm 170 μm 320 μm A significant difference in the evaporation rate of droplets is observed
Case-2.4 Case-2.5 Case-2.6
inside the combustion chamber for the studied cases. The highest and
 E. Gozali et al. / Surface & Coatings Technology 280 (2015) 370–377 373

Table 3
Thermophysical properties of pure liquid and suspension.

Property Mass fraction Temperature range

K
0 wt.% 5 wt.% 15 wt.% 25 wt.%

Density ρsusp = aT + b ρsusp = aT + b ρsusp = aT + b ρsusp = aT + b 300–360

kg/m3 a = −0.97 a = −0.92 a = −0.83 a = −0.73
b = 1077 b = 1234 b = 1550 b = 1865
Viscosity μsusp = aT + b μsusp = aT + b μsusp = aT + b μsusp = aT + b 273–348
kg/m-s a = −1.72 × 10−5 a = −1.96 × 10−5 a = −2.59 × 10−5 a = −3.54 × 10−5
b = 6.36 × 10−3 b = 7.23 × 10−3 b = 9.55 * 10−3 b = 1.31 × 10−2
Specific heat csusp = aT + b csusp = aT + b csusp = aT + b csusp = aT + b 300–360
J/kg-K a = 11.02 a = 10.47 a = 9.37 a = 8.27
b = −873.8 b = −641.1 b = −175.7 b = −289.7
Thermal conductivity ksusp = aT + b ksusp = aT + b ksusp = aT + b ksusp = aT + b 273–373
W/m-K a = −2.64 × 10−4 a = −3.04 × 10−4 a = −4.27 × 10−4 a = −6.07 × 10−4
b = 2.46 × 10−1 b = 2.85 × 10−1 b = 4.15 × 10−1 b = 6.90 × 10−1
Surface tension σsusp = aT + b σsusp = aT + b σsusp = aT + b σsusp = aT + b 283–323
N/m a = −8.32 × 10−5 a = −8.32 × 10−5 a = −9.16 × 10−5 a = −9.16 × 10−5
b = 4.68 × 10−2 b = 4.68 × 10−2 b = 5.15 × 10−2 b = 5.15 × 10−2

Pure liquid properties are taken from Perry's Chemical Engineering Handbook [25] and curve-fitted in their temperature range. Suspension properties then are calculated from commonly
used theoretical models in which temperature dependent pure liquid properties are incorporated.

lowest rates 1.54 × 10−7 and 9.52 × 10−8 kg/s stand for droplets with
0 wt.% and 25 wt.% concentrations respectively (see Fig. 2-f). Compared
to Case-2.1, 15% decrease in the evaporation rate of droplets is noted for
the Case-2.2 which shows that increasing the size of injection droplets
can reduce vaporization of liquid feedstock.
In the last case (Case-2.3) where the droplet diameter is raised to
300 μm, the maximum gas temperature difference along the centerline
between the cases (0 wt.% and 25 wt.%) reaches 460 K in the combus-
tion chamber, while it reaches almost 100 K inside the barrel, and at
the gun exit the temperature difference among all cases is negligible.
The velocity difference for different concentrations is not considerable
in Case-2.3. A similar trend is observed for droplet vaporization rate
compared to the above case (Case-2.2) but the difference in the evapo-
ration rate for varied nanoparticle loadings is further reduced. This is
due to the increase in droplets Weber number which intensifies drop-
lets deformation and leads to severe atomization and decreases the va-
porization process.
3.1.2. Injection of droplets of various sizes
In this section, droplets with Rosin–Rammler diameter distribution
and four different nanoparticle concentrations (0, 5, 15, and 25 wt.%)
are injected into the HVOF suspension spraying flow from an inlet
surface (see Table 2). In Case-2.4, five streams of droplets with varied di-
ameters 30, 40, 50, 60, and 70 μm are injected at the same time from an
inlet surface. The case is examined with four different nanoparticle con-
centrations. Fig. 3-a shows a comparison of temperature difference be-
tween Without Droplets (Case-1) and Case-2.4. As can be seen, the
temperature difference between (Case-1) and Case-2.4 with high nano-
particle loading (25 wt.%) is 145 K, while this is 850 K close to the nozzle
entrance for Case-2.4 with low particle loading (0 wt.%). This is a further
temperature loss for droplets with low particle loading as compared
with Case-2.1 which confirms that the cooling effect for varied diameter
droplets is higher than for the constant diameter droplets. Fig. 3-b
shows that the reduction in the gas velocity is more for this case as com-
pared with Case-2.1. This is 165 and 220 m/s for the cases at the ends (0

Fig. 2. Comparison of the gas temperature, velocity fields, and, rate of evaporation experienced by the droplets injected from a surface with constant diameters of (a–c) 50 μm and (d–f)
150 μm, and having various solid nanoparticle concentrations.
374 E. Gozali et al. / Surface & Coatings Technology 280 (2015) 370–377
Fig. 3. Comparison of the gas temperature, velocity fields, and, rate of evaporation experienced by the droplets injected from a surface with various diameters of (a–c) 30 μm, 40 μm, 50 μm,
60 μm, 70 μm, and (d–f) 130 μm, 140 μm, 150 μm, 160 μm, 170 μm, and having various solid nanoparticle concentrations.
and 25 wt.%) compared to Case-1 and Case-2.4. The reason for this is the
higher cooling effect and the variation in droplet size from 30 μm to
70 μm. In this case every droplet has a different evaporation rate and
moves with different relative velocity inside the torch. The evaporation
rate for the ends cases (0 and 25 wt.%) is calculated to be 1.2 × 10−7 and
1.36 × 10−7 kg/s along the centerline axis. The cause of fluctuation in
the rate of evaporation is due to different sizes of droplets carrying dif-
ferent masses. The extra loss or cooling of the gas temperature, as drop-
lets of various diameters are injected, reveals that the rate of
vaporization is also dependent on the droplets injection diameter. For
smaller droplets (30 μm, 40 μm), higher evaporation occurs inside the
combustion chamber while larger diameter droplets (60 μm, 70 μm)
are evaporated in the barrel.
When droplets are injected with diameters 130, 140, 150, 160
and 170 μm (Case-2.5), the trend for the gas temperature cooling is
changed. Here the temperature difference between 0 wt.% and
25 wt.% nanoparticles loading is 550 K in the middle of the combus-
tion chamber and it decreases to 100 K along the barrel centerline
axis (Fig. 3-d). In Fig. 3-e the velocity field does not experience any
significant changes and shows the same pattern as observed for the
Case-2.2. Calculated vaporization rate for the sub cases (0 and
25 wt.%) of Case-2.5 is 1.40 × 10− 7 and 7.67 × 10− 8 kg/s respectively
(Fig. 3-f). Further analysis reveals that there is a decrease in the over-
all evaporation rate along the gun axis by 9% for 0 wt.% nanoparticle-
loading and 19% for 25 wt.% concentration compared with Case-2.2
(constant diameter). The final location of droplets evaporation
moves from the C–D nozzle toward the middle of barrel as the nano-
particle concentration is increased from 0, to 5 wt.%. Furthermore,
the droplets leave the gun without complete evaporation when the
concentration of nanoparticles increases from 15 to 25 wt.%.
In Case-2.6, when a range of droplets diameters from 280, 290,
300, 310, and 320 μm are injected into the HVOF suspension spraying
flow, no significant variations are observed as compare with
constant diameter injection (300 μm) Case-2.3 which confirms that
all these droplets have a similar trend of gas cooling and rate of
evaporation.
In summary, droplet evaporation depends principally on two pa-
rameters i.e., nanoparticles concentration, and droplets diameter. As
the droplets size and nanoparticles concentration increase, the rate of
evaporation decreases. Also an increase of vaporization rate is observed
for varied diameters 30 to 70 μm (Case 2.4) compared to constant diam-
eter 50 μm (Case-2.1). However, for the second and third set of varied
diameters 130 to 170 and 280 to 320 μm, the effect is opposite due to in-
creased droplet size. It can be concluded that increasing nanoparticle
concentration in suspension droplets with various and large diameters
would decrease the amount of droplets evaporation.
3.2. Secondary break-up
For this analysis, droplet streams with constant diameters are select-
ed and their diameter reduction and final location of evaporation are
highlighted. Furthermore, for Rosin–Rammler distribution, a set of five
streams (only one case) is investigated.
3.2.1. Injection of droplets of constant sizes
A comparison of droplets diameter reduction is shown in Fig. 4(a–b)
for the two sub cases of the Case-2.1. This demonstrates secondary break
up and reduction in diameter of 50 μm droplets with two different solid
nanoparticles concentrations. It shows that droplets experience a sharp
decrease in their diameter from 45 to 20 and 20 to 10 μm approximately
for all nanoparticles concentration variations. In Case-2.1, We reach a
peak value of almost 12 due to the growing relative velocities between
the droplets and gas phase in the combustion chamber, where the main
atomization occurs. However, the value of We remains below 14 for all
the nano-loadings. Hence the break up type is vibrational. In this case,
even the addition of the solid nanoparticles into the droplets does not
affect the disintegration process. Therefore, the vaporization of the
droplets is the dominant factor and controls the process when droplets
are injected with small and constant diameter of 50 μm.
In contrast, as the diameter of the droplets is raised to 150 μm (Case-
2.2), the We varies between 25 and 30 based on the nanoparticle
concentration in the droplets and severe fragmentation occurs in the
 E. Gozali et al. / Surface & Coatings Technology 280 (2015) 370–377 375

Fig. 4. Comparison of the droplets (a–b) diameter reduction, with a constant diameter of 50 μm, and having 0 and 25 wt.% nanoparticle concentrations.

middle of the combustion chamber. This is also reflected by the reduc-
tion of droplets diameter from 150 to 35 and 35 to 13 inside the com-
bustion chamber, see Fig. 4(c–d). For all the studied concentrations,
droplets are evaporated in the middle of barrel and release solid nano-
particles, but with the highest concentration (25 wt.%) a delay occurs
in the rate of evaporation and cause the droplets to reach the nozzle out-
let. Similar results are observed for larger droplets (300 μm) while the
We increases from 50 to 70 for Case-2.3. The above analysis indicates
that the fragmentation process is dominant for large droplets (150
and 300 μm) while the rate of evaporation is higher for small droplets
(50 μm).
Finally, Fig. 5(a–d) compares the evaporation rate of droplets with
diameters of 50 and 150 μm, and with 0 and 25 wt.% nanoparticle
concentrations as they are injected from an inlet surface. In this figure,
significance of fragmentation and evaporation in the HVOF suspension
spraying process is evident. It shows that larger droplets (150 μm)

Fig. 5. Comparison of evaporation rate of the droplets with constant diameters of (a–b) 50 μm and (c–d) 150 μm, each with 0 and 25 wt.% nanoparticle concentrations.
376 E. Gozali et al. / Surface & Coatings Technology 280 (2015) 370–377
with high concentration (25 wt.%) leave the gun without complete va-
porization, which can lead to serious consequences in real applications
and can create defects in coatings. It can be inferred that by increasing
the concentration of nanoparticles in the base fluid, the rate of evapora-
tion decreases causing the delay in the complete vaporization of the
droplets. However, simulation results reveal that droplets with the
smaller diameter are evaporated in the middle of the barrel and conse-
quently release nanoparticles. This may suggest that smaller droplets
can be used in applications where nanoparticles with a high melting
point are required for coatings. It should be noted in the present work
only the employed models are validated with experimental data and
the results are predicted by the models. Therefore, experimental studies
are demanded to validate the findings.
3.2.2. Injection of droplets of various sizes
Considerable reduction in diameter is observed for Case-2.4, when
droplet streams with five different diameters (30, 40, 50, 60 and
70 μm) and four different concentrations are injected into the gun
(Fig. 6(a–b)). The degree of disintegration is dependent on the initial
size of droplets. For example, the breakup in the small droplets of
30 μm leads to a reduction from 20 to 14 and 14 to 7 μm, while in the
large droplets of 70 μm the reduction is from 66 to 30 and 30 to
24 μm. In this case, the fragmentation is higher for the droplets greater
than 50 μm and lower for those droplets less than 50 μm. Droplets frag-
mentation is observed at different positions and it is hardly distinguish-
able at higher positions (N0.02 m) in the graph due to the simultaneous
injection of droplets with varied sizes and overlapping at higher posi-
tions. The value of the Weber number stays below 14 for different con-
centrations and the breakup type is vibrational as observed for the
constant diameter droplets of 50 μm (Case-2.1).
When droplets with various sizes from 130 to 170 μm are injected
(Case-2.5), 71% reduction is noticed in their size for the four different
concentrations. For these cases, the We fluctuates between 26 to 35
and severe fragmentation is observed inside the combustion chamber.
The maximum reduction of 85% in the droplets diameter is observed
as they are injected in a range of 280 to 320 μm. Moreover, We increases
from 55 to 80 proving that the large droplets experience intensive frag-
mentation inside the combustion chamber. When the Rosin–Rammler
diameter distribution is applied, more droplets discharge through the
torch without prior evaporation and this complication becomes intense
with increasing nanoparticle concentration in the droplets.
In summary, if injection of droplets with small sizes (diameter
≤50 μm) is favored in an application, extra liquid surfactant should be
added to the suspension to reduce the droplets surface tension and in-
tensify the fragmentation process. However, with regard to droplets
Fig. 6. Comparison of the droplets (a–b) diameter reduction, with various diameters of 30 μm, 40 μm, 50 μm, 60 μm, 70 μm and having 0 and 25 wt.% nanoparticle concentrations.
atomization in the HVOF suspension spraying process, larger droplets
undergo severe break up with less evaporation rate when carrying
higher nanoparticle concentrations. To increase the effectiveness of
the HVOF suspension spraying process when injecting large droplets,
an optimum addition of nanoparticles into the solvent is required. In
the group type injection, proper selection of the injection angle, initial
velocity and diameter distribution are demanded to achieve optimum
results.
It should be noted that in the original design of the DJ 2700 gun, the
gas carrier tube is located at the center of the back wall and it is
surrounded by annular oxygen/fuel (O/F) inlets. It creates a recircula-
tion zone close to the back wall at the injection area and the particles
start to spread out near the nozzle throat where the flame reaches the
axis of the torch and interacts with the droplets. Likewise when the
droplets are injected at an angle of 45°, they are directed toward the
core of the combustion chamber core where they have a close interac-
tion with the flame, which makes the evaporation process more effec-
tive in this injection scheme. In this method one has to control the
injection parameters to avoid the droplets colliding with the combus-
tion chamber walls. Moreover, even though the employed models in
the present work can be applied to liquid-fuelled HVOF guns, the nu-
merical model must include further evaporation from a combustion-
driven liquid fuels such as kerosene or liquid propane and its interac-
tion. Therefore, it is highly unlikely these results match with that of
liquid-fuelled HVOF torches.
4. Conclusion
The effects of the solid nanoparticles on the liquid droplets in the
suspension are considered as the change in the liquid bulk thermo-
physical properties. Then effects of suspension on the gas dynamics, va-
porization rate, and secondary break up in the HVOF suspension
spraying process are investigated. The main findings are:
• There is a considerable difference in the final location of evaporation
of homogeneous and non-homogeneous droplets. This is an impor-
tant factor in the numerical analysis of suspension as nanoparticles
are created in a computational domain as new entities.
• If a high concentration loading in a suspension is favored in an ap-
plication, the use of smaller droplets (up to 50 μm) give a better re-
sult as they experience high evaporation in the mid-section of the
nozzle.
• If larger droplets (150 and 300 μm) are injected into the gun, frag-
mentation is the dominant factor which controls the process. With
increasing solid particle concentrations in the droplets the final
 E. Gozali et al. / Surface & Coatings Technology 280 (2015) 370–377
location of evaporation moves toward the gun exit and in the ex-
treme cases the droplets leave the gun without complete evapora-
tion.
• Large droplets undergo severe fragmentation inside the combus-
tion chamber which causes evaporation to take place in front of
the convergent–divergent nozzle even with the highest nanoparti-
cle loading.
Acknowledgment
The authors would like to acknowledge the financial support by the
UK Engineering and Physical Sciences Research Council (EPSRC) project
grant: EP/K027530/1 and the financial support for the research student-
ship from the School of Engineering in Cranfield University, and Xi'an
Jiaotong-Liverpool University. Moreover, we acknowledge Dr. A.M.
Hilmarsson-Dunn for her contribution in editing the text.
References
[1] G. Bolelli, V. Cannillo, R. Gadow, A. Killinger, L. Lusvarghi, J. Rauch, M. Romagnoli,
Surf. Coat. Technol. 204 (2010) 1163.
[2] E. Bemporad, G. Bolelli, V. Cannillo, D. De Felicis, R. Gadow, A. Killinger, L. Lusvarghi,
J. Rauch, M. Sebastiani, Surf. Coat. Technol. 204 (2010) 3902.
[3] N. Stiegler, D. Bellucci, G. Bolelli, V. Cannillo, R. Gadow, A. Killinger, L. Lusvarghi, A.
Sola, J. Therm. Spray Technol. 21 (2012) 275.
[4] M. Gell, E.H. Jordan, M. Teicholz, B.M. Cetegen, N. Padture, L. Xie, D. Chen, X. Ma, J.
Roth, J. Therm. Spray Technol. 17 (2008) 124.
377
[5] J. Rauch, G. Bolelli, A. Killinger, R. Gadow, V. Cannillo, L. Lusvarghi, Surf. Coat.
Technol. 203 (2009) 2131.
[6] E. Dongmo, R. Gadow, A. Killinger, M. Wenzelburger, J. Therm. Spray Technol. 18
(2009) 896.
[7] D.A. Drew, S.L. Passman, Theory of Multicomponent Fluids, Springer, Berlin, Heidelberg,
New York, 1999 121.
[8] S.M.S. Murshed, K.C. Leong, C. Yang, Int. J. Therm. Sci. 47 (2008) 560.
[9] Y. Xuan, W. Roetzel, Int. J. Heat Mass Transf. 43 (2000) 3701.
[10] J.W. Gao, R.T. Zheng, H. Ohtani, D.S. Zhu, G. Chen, Nano Lett. (2009) 4128.
[11] S. Tanvir, L. Qiao, Nanoscale Res. Lett. 7 (2012) 226.
[12] M. Moosavi, E.K. Goharshadi, A. Youssefi, Int. J. Heat Fluid Flow 31 (2010) 599.
[13] R.H. Chen, T.X. Phuoc, D. Martello, Int. J. Heat Mass Transf. 53 (2010) 3677.
[14] E. Dongmo, A. Killinger, M. Wenzelburger, R. Gadow, Surf. Coat. Technol. 203 (2009)
2139.
[15] S. Kamnis, S. Gu, Chem. Eng. Process. 45 (2006) 246.
[16] S. Kamnis, S. Gu, Chem. Eng. Sci. 61 (2006) 5427.
[17] T. Furuhata, S. Tanno, T. Miura, Y. Ikeda, T. Nakajima, Energy Convers. Manag. 38
(1997) 1111.
[18] E. Brinley, K.S. Babu, S. Seal, J. Miner. Met. Mater. Soc. 59 (2007) 54.
[19] N. Zeoli, S. Gu, S. Kamnis, Int. J. Heat Mass Transf. 51 (2008) 4121.
[20] B.E. Gelfand, Prog. Energy Combust. Sci. 22 (1996) 201.
[21] E. Gozali, S. Kamnis, S. Gu, Surf. Coat. Technol. 228 (2013) 176.
[22] S. Kamnis, S. Gu, T.J. Lu, C. Chen, Comput. Mater. Sci. 43 (2008) 1172.
[23] N. Zeoli, S. Gu, S. Kamnis, Comput. Chem. Eng. 32 (2008) 1661.
[24] E. Gozali, M. Mahrukh, S. Kamnis, S. Gu, J. Therm. Spray Technol. 23 (2014) 940.
[25] R.H. Perry, D.W. Green, McGraw-Hill, 1997. p. 160.
[26] H.C. Brinkman, J. Chem. Phys. 20 (1952) 571.
[27] B.X. Wang, L.P. Zhou, X.F. Peng, Int. J. Heat Mass Transf. 46 (2003) 2665.
[28] M. Li, P.D. Christofides, Chem. Eng. Sci. 61 (2006) 6540.
[29] R.S. Miller, K. Harstad, J. Bellan, Int. J. Multiphase Flow 24 (1998) 1025.
[30] S.S. Sazhin, Prog. Energy Combust. Sci. 32 (2006) 162.