REV.CHIM.(Bucharest) 69No. 9 2018 http:www.revistadechimie.ro 2571Researches Concerning the Establishment of a 3D Model of the Electric [623832]

REV.CHIM.(Bucharest) ♦69♦No. 9 ♦2018 http://www.revistadechimie.ro 2571Researches Concerning the Establishment of a 3D Model of the Electric
Arc Thermal Spraiyng Process and the Experimental Validation Using
the Fluent Simulation Program 6.2.16
CRISTIAN DOBRESCU, BOGDAN FLOREA, ALINA NECSULESCU, MIHAI BUTU*
University Politehnica of Bucharest, Romania, 313 Spl. Independentei, 060042, Bucharest, Romania
Using the experimental data and the data presented in other papers [1-10] presented in the specialty
literature, the paper present a 3D model of the electric arc thermal spraying process, used to obtain a series
of pseudo alloys, based on computational fluid dynamics (CFD). In this respect, a three-dimensional model
was used to calculate the flow of material inside and outside the spray gun. The gas particle interactions
with the material were computed using the interaction method between the air flow direction and the two
movement directions of the spraying wire, directions which determine and on which the loading effect
depends. The particle size distributions were experimentally measured, taking into account the particles
collected on the spraying distance. Subsequently, the validation of the model was performed on the basis of
comparisons between the predictions made and actual particle speed measurements.
Keywords: electric arc thermal spraying process, pseudo alloys, computational fluid dynamics (CFD)
Due to its low cost compared to the high deposit rate,
the electric arc spraying process is a widely used
technology for the thermal spraying of metals and alloys inorder to protect them against corrosion and wear [11-16].
The conventional electric arc thermal spraying installation
includes two wire coils directly connected to the anodeand the cathode of a current generator. The two wires are
guided to a point in the center of a gas jet, forming an
electric arc between them. The droplets resulting from the
melting of the 2 wires are sprayed onto the coating surface.
The principle behind the process is simple, but theproperties of the resulting coatings are significantly
dependent on the working parameters.
The effect of relevant working parameters such as gun
design, the nature and the pressure of the spraying gas are
presented in the specialty literature [16-20], but the process
knowhow is based mainly on practice and experiments.In this respect, only a small number of works refer to the
process of modeling electric arc thermal spraying, which
justifies tackling this aspect. Future improvements aredependent on understanding the various phenomena that
influence the electric arc thermal spraying process.
Experimental part
To establish the 3D model, the FLUENT software was
used, known under the commercial name of computerizedfluid dynamics or in short CFD, this software being used to
perform a three-dimensional simulation of the electric arc
thermal spraying process using the Tafa 9000 gun, similarto that used in experiments. To establish the model, the
notations in table 1 were used.
a) The influence of process parameters on establishing
the model
One of the simplest electric arc thermal spraying
systems that can be imagined consists of two wires
connected to a current generator and guided in an emitting
stream with a spray gun connected to an air compressor.However, several improvements have been suggested in
the specialty literature which mainly relate to the nature of
* email: [anonimizat] spraying gas, the particle spraying nozzle, and the
particle jet. Spray gas pressure is the main parameter for
controlling gas flow and speed. An increase in spraying
gas pressure results in an increase in spraying speed and adecrease in mean droplet size [21-25]. A high velocity of
the droplets also causes a decrease in the porosity of the
coating. A drop in droplet size causes an increase in particleoxidation due to the increase in the specific surface area.
However, a higher droplet speed also causes a decrease in
contact time, which leads to a decrease in particleoxidation. These two effects act inversely. The materials
used for the electric arc spraying process are metals
(copper, aluminum and zinc) and metal alloys (steels, etc.),materials that are extremely sensitive to oxidation.
Therefore, choosing the nature of the spraying gas is
important. Thus, in particular, the use of an inert gas, suchas carbon dioxide or nitrogen, is sometimes preferred [26].
With the exception of specific applications for which
oxidation has a positive effect (for example the hardeningof the coating layer), in all other cases, the interaction
between the hot material with the environment is a
permanent concern. A thermal spray layer consists ofoverlaying sprayed droplets, for example, in the form of
laminas, created on the substrate by rapid solidification
and flattening of the melted droplets [20]. It is well knownthat oxidation can reduce the adhesion to the substrate of
the sprayed droplets, as well as the cohesion between the
molten droplets, this being an inconvenience forapplications against wear and corrosion. Moreover, during
the spraying of alloys such as stainless steel, a significant
amount of addition elements (for example, chromium,etc.) may be lost due to the presence of oxygen that oxidizes
these elements. The use of carbon dioxide or nitrogen as
spraying gases can prevent or reduce these phenomena,which leads to a better adhesion of the coating and to a
decrease in porosity. The nozzle model used in experiments
influences differently the efficiency of the electric arcspraying process. Consequently, a second aspect taken
into account when establishing the model concerns the
design of the nozzle. First of all, because the system is

http://www.revistadechimie.ro REV.CHIM.(Bucharest) ♦69♦No. 9 ♦2018 2572
always supplied with compressed air, and the external jet
can be clamped at the outlet of the nozzle, thus
representing a significantly divergent component, this can
contribute to the increased spread of jet of particles, whichis detrimental to the deposited layer. In particular, these
nozzles allow an increase in the gas speed at the
intersection point of the spraying flows. With the use ofdifferent types of nozzles, a strong influence on the mean
droplet size and the properties of the coating layer has been
observed. For example, the use of a convergent / divergentnozzle causes a decrease in the mean droplet size
compared to that obtained using a cylindrical nozzle. This
effect is directly correlated with an increase in the speedof the spraying gas. At the same time, it can be observed
that a smaller droplet size has the effect of improving the
quality of the coating layers. Even if the use of an inert gasreduces the oxidation of the pulverized material, the
amount of air involved in the process is so significant that
oxidation can not be completely avoided. The third aspecttaken into consideration is the use of another type of gas
called secondary gas in order to reduce the proportion of
air in the jet. This method can also help reduce thedivergence of the spraying jet. The technique consists of
increasing the amount of sprayed inert gas using a
secondary inert gas injection in order to reduce the amountof air entrained in the spray jet. There are different methods
used for this secondary injection. Thus, a secondary
injection mounted on the top of the gun, just at its outlet,was used [26]. In this case, there are two different effects,
namely the protection of the material due to the presence
of inert gas around the spraying flows and the second, anincrease in the total gas flow from the outlet of the nozzle,
which produces an increase in gas speed in this moment.
Improvements made by using this method are undoubtedlyassociated with the combination of these two effects. The
second technique consists of a secondary injection placedin the direction of the compressed air at the outlet of the
spray gun.
b) Modeling the electric arc thermal spraying process
Numerous numerical studies are currently devoted to
modeling of plasma jet spraying or supersonic flame
thermal spraying (HVOF). Most authors use 2D modeling
that also applies to these processes. Several studies alsouse three-dimensional models with relatively simple grids
[27]. In most cases, a 2D model would not be able to
accurately represent the process of electric arc thermalspraying. Therefore, the FLUENT software (6.2.16) was
used for modeling as it is the world leader in CFDs, namely
numerical fluid mechanics (MFN), computational fluidmechanics and computational fluid dynamics (CFD). CFD
is a branch of fluid mechanics, which uses algorithms,
numerical methods, and computers to model and solvefluid flow problems. This software also allows the use of
unstructured networks, made up of tetrahedral cells.
Modeling is based on solving a set of differential
conservation equations supplemented with numerous
additional equations,
models , often semi-empirical, to
analyse turbulence, pressure, cavitation, heat exchange,dispersed phase transport. These equations are discretized
using different methods, such as finite difference method,
finite elements, finite volumes, or boundary elements. Thedomains to be modeled are divided into small areas,
resulting in meshing networks with many nodes. The
equations, written for each node, are assembled into aglobal system of equations, which is then solved. The main
field of application is the modeling of sub-and supersonic
turbulent flows in the aerospace field, but there are manyother fields of use, such as vehicle and construction
aerodynamics, chemical process optimization, weather
forecasting, geological prospecting, exhaust dispersion,medical or military applications.Table 1
NOTATIONS USED TO ESTABLISH THE
MODEL

REV.CHIM.(Bucharest) ♦69♦No. 9 ♦2018 http://www.revistadechimie.ro 2573The theoretical basis of almost all flow modeling are
the Navier-Stokes equations describing the flow of a single
fluid phase. Eliminating the terms describing viscosity fromthese equations, leads to a simpler model, Euler’s
equations. By eliminating the terms describing the
turbulence, the potential flow equations are reached. Finally,these equations can be linearized, because by solving them,
obtaining the fields of the various variables within the area
in which the fluids flow, is aimed, namely the way in whichthe fluids interact with the surfaces that delimit the field in
which these fluids flow. For this purpose, computers are
used, and therefore numerical fluids mechanics can alsobe considered a branch of computer-aided design (CAD),
respectively computer-aided engineering (CAE). The
stronger the computers used are, the better the solution is,or the shorter its computation time is. The modelled
solutions are validated by comparing them to the measured
values on experimental stands (for example, aerodynamictunnels), or under real conditions.
The first step in the modeling process with FLUENT
software is to design the various components of the electricarc spraying gun. Therefore, these elements were designed
in CAD, a meshing tool of the FLUENT software. Some
simplifications were then performed to improve the qualityof network nodes. This change was operated with small
adjustments to avoid producing flat cells in the space
reserved for wide guides. Figure 2 presents the differentelements of the Tafa model 9000 gun head, namely:
– an aluminum case attached to the thermal spraying
gun;
– a colored aluminum cap with four regular rectangular
(90 °) openings for the secondary gas flow;
– a primary convergence nozzle (wire position guide)
supported by the colored cap; two FeC-NiFe wires (used in
experiments) attached to the power connectors and
supported by the primary nozzle.example, with a smaller cross section), is recommended,
except for the blue cap that has the largest diameter. Figure
2 presents a series of minor changes made to the Tafa gundesign before digitalization. Figure 2 highlights the
available sections for the gas flow, using the cross shaped
primary nozzle. The left image shows the initial crosssection and the right one shows the modified cross section.
The change was made by inserting the rectangular blocks
positioned properly. It can be seen that the current changeis very small and does not affect the flow, but avoids the
presence of flat cells only at the tangential points (lines in
the three-dimensional representation). The quality ofnetwork nodes was improved, so convergence was made
easier. As far as the grid is concerned, it was generated in
CAD and contains four different volumes. Moreover, only aquarter of the geometry was considered due to the
presence of two symmetry planes. The first volume refers
to the part of the geometry located upstream from the exitof the green cap (meaning the inside of the gun). This
volume was meshed, having 0.6 mm mesh nodes, only for
tetrahedral cells, and for meshing large areas or smallareas, 0.3 mm wide mesh nodes were used.
The second volume refers to a small region around the
spraying wires which was meshed separately usingtetrahedric cells with 0.3 mm mesh nodes. The third
volume refers to the rest of the geometry up to the exit of
the aluminum casing: it was meshed using tetrahedralcells, but with 1 mm mesh nodes. The last volume refers
to the outer part that is meshed with prisms (extruded
triangles) having 2 mm mesh nodes in the core, and aroundthe core meshed with extruded rectangles having 2 mm
mesh nodes. Finally, the grid was formed using about
150,000 cells, the interior of the spray gun head, 13,000cells in the electric arc region, 30,000 cells in the third
volume, and 140,000 cells in the last part (that is, the main
part of the external field). The external field extends axially
210 mm from the exit of the green cover and extends
radially 50 mm, that is, with a diameter of 100 mm.
c) The equations that govern the modelling and the
boundary conditions for the 3D modelling
When modeling, the typical Reynolds-NavierStokes
equation set was considered. Most of the results were
calculated using standard k-ε turbulence models, but the
entire k-ε model was also tested because it is better suited
to round jet study. The energy equation was taken into
account concerning temperature (that is, the default optionwithin FLUENT). Moreover, the interactions between the
thermal (for example, in the solid sections) and liquid
problems were not considered in the present paperbecause they were not of interest and because they would
not lead to an important improvement of the process. The
main assumptions of the proposed model considered thefollowing:
– The flow characteristics for two different types of
k-ε
turbulence were considered.
-The stationary flux was considered, ignoring the
development of the electric arc (that is, ignoring any
electromagnetic model of the arc itself).
-The droplet size that is formed during the spraying of
the wires which form the electric arc, and their distribution,
has been determined according to the size distributionscorresponding to the collected particles.
When generating the mathematical model, the
following conservation equations were taken into account:
– Continuity:
(1)
Fig. 1. Design of the different
components of the Tafa
spraying gun head, model 9000
Fig. 2. Changes
made to the
design of the Tafa
spraying gun
head, before
meshing, in order
to avoid flat areas
Currently, for the primary nozzle there are different design
and cap color schemes depending on the material to be
sprayed. In the present modeling, a cross shape and a greencap were used for the primary nozzle. In fact, this
configuration is preferred for most laboratory applications.
The green cover has an output diameter of 7.62 mm. Usinga cap with a smaller diameter provides higher particle
speed, higher temperature and vice versa. Using of a long
primary nozzle, such as the one currently used (for

http://www.revistadechimie.ro REV.CHIM.(Bucharest) ♦69♦No. 9 ♦2018 2574Momentum equation:
(2)
(3)
(4)
Taking into account the Boussinesq hypothesis, the stress tensor is defined as:
In which the formula
, is valid for the k-ε type turbulence model.
The kinetic energy turbulence
The turbulence of the dissipation speed of kinetic energy:
In which G= µ2tS2 and S are the magnitude rate defined as:
In the case of the standard k-ε turbulence model, the following set of constants is used:
Energy:
(7.8)(5)
(7)(6)
(7.9)
(7.10)
(12)(11)
(14)(13)
Moreover, a value of 0.85 was used for the Prandtl
number turbulence
.
These equations were solved using the FLUENT software
solution base. Moreover, the SIMPLEC method from the
FLUENT software menu was activated to allow for thecorrection of the interaction between pressure and speed,
because it allows for faster convergence compared to
SIMPLE (for example, the possibility of using a 0.9 sub-relaxation factor when using SIMPLEC instead of a 0.3 one
when using SIMPLE, for the pressure correction equation).
The boundary conditions considered were:a)- A single entry was considered for compressed air.
For this entry, a MASS_FLOW_INLET boundary condition
was considered, allowing for the direct airflow to bespecified, as the upstream pressure is not imposed, but
represents a deduced variable. The turbulence wasdetermined to be 5% (k is calculated from 3/2 (I
tV)2 above
this limit) and å was prescribed by the hydraulic diameter
of the inlet section.
b)- The results showed that most of the compressed
airflow flows axially through the primary nozzle and only a
small part of the gas flows through the secondary path(around the primary nozzle), which has a much smaller
minimum cross section.
c)- A fixed pressure was considered for the boundary
conditions for the external domain for which the turbulence
of the kinetic energy and its dispersion rate were mentioned
in a similar manner to the one used for intake.
d)- The hot wire ends and the molten droplet injection
conditions were specified as follows: from the
thermodynamic tables, approximately 2400 kJ kg
-1 are

REV.CHIM.(Bucharest) ♦69♦No. 9 ♦2018 http://www.revistadechimie.ro 2575required to melt and heat the material of the spray wire up
to the boiling point (that is, about 3133 K, in the case of
pure iron), which means that the energy dissipated in theelectric arc is always sufficient to reach the boiling point of
the materials (for example, at 100 A, the dissipated power
of 3.1 kW is greater than the product between the speed ofthe material flow with 2400 kJ kg
-1, which is 1.75 kW in
this case). Thus, the droplet injection temperature was
considered to always be the boiling point of the materialsince the remaining power (which is 1.35 kW in the above
example), is considered as a thermal flux dissipated on
the surface of the spraying wire that is ending. In the presentcase, this available power would be sufficient to vaporize
about 30% of the materials.
d) Properties of the materials used in the model
Sutherland formula was used in order to estimate air
viscosity in relation to temperature. Thus, air viscosity wasdetermined according to:

(15)
Similarly, the thermal conductivity of the air was deduced
assuming a Prandtl number of 0.7. Thus, the thermalconductivity of the air was determined according to:

(16)
Because it is not the default option within FLUENT, some
C functions were written to obtain the desired formula. Atthe same time, the well-known correlation between the
Schiller and Naumann relations (see above C coding) was
used to express the aerodynamic coefficient of the dropletsrelative to the air particles based on the Reynolds number.
This correlation shows that:

(17)
In terms of heat exchange, the default equation (that is,
FLUENT Ranz-Marshall) has not been changed. Table 2
shows the properties of the material, taken into accountfor the secondary phase (for example, mainly molten iron
droplets).hour (m
3 / h), most of the time, from equation 18 results a
flow rate which is 20% lower than the one indicated by the
control system. This difference can be explained by somepossible leaks (lack of seals on the experimental setting)
and a flowmeter that was not calibrated. However, since
pressure is one of the most important parameters, it wasused as an input parameter for the model. For thermal
spraying, the FeC-NiFe alloy couple was used. The voltage
and the intensity of the current in the electric arc, werevaried at different levels. The experimental data was used
to establish a correlation between the quantity of sprayed
material, the intensity and the voltage of the current in theelectric arc. Finally, the equation for the type 1.2 mm
spraying wire used in the experiments, was derived:

(19)
where D represents the mass flow (kg / h) of the sprayed
material, I is the current intensity in the electric arc (A)
and the U the voltage (V). This equation was established
on the basis of more than 90 different sets of parameters
with a 5% error for the current intensity between 50 and300 A. The relationship (19) is much more realistic than
the reported data, which were obtained in accordance with
the non-operating conditions, that is, without the electricarc and which are therefore doubtful.
The numerical results were calculated for a supply
pressure of 4.15 bar, as well as for a current intensity in theelectric arc ranging from 100 to 250 A. The corresponding
wire flow varied between 2620 and 9350 g / h (calculated
from equation (19) for current intensities of 100 and 250 A.
The DPV diagnostic tool from the FLUENT program
menu is used intensively to characterize the electric arc
spraying process. This allows for the particle properties to
be measured during spraying, such as
speed, temperature
and diameter . The measurement principle is based on
detecting the radiation emitted by a hot particle, using a
sensor, the speed being the parameter deduced from the
time interval between two recorded signals for a singleparticle flying in front of a two-slot device. The temperature
is determined using the ratio between the intensity signal
from two different wavelengths, using the gray bodyhypothesis, since the particle diameter is deduced from
the signal strength and requires calibration in order to obtain
realistic data. Moreover, two different coupling methodswere considered for the discrete modeling phase. For the
first method, a one-way coupling was used, in which case
the airflow does not influence the loading layer, and in thesecond one, a two-way coupling model was used. When
using these models, a finite number of diameters are taken
into account and the particles are taken from all the surfaceelements of the source in question. For the two-way
coupling model, each particle is considered to be a flow
and is associated with a flow rate which depends on theparticle size distribution, which must be expressed as
weight percentages. In the present case, this was
established according to a Rosin-Rammler distribution [28]based on particle size distribution collected from the
spraying distance. In practice, DPV allows for the particle
diameter to be measured based on signal strength.However, a calibration is required in order to obtain realistic
results. This calibration was made based on the particles
collected at the spraying distance, considered to be 200mm. A correction factor is then applied to the initial DPV
diameter, so that the mean corrected diameter of the
particle diameter detected using DPV becomes identicalto the diameter measured for the collected particles. This
step was carried out based on image analysis, and over
100 different images were considered for each set of
Table 2
THE PROPERTIES USED FOR THE MOLTEN DROPLETS
( FeC-NiFe)
– Spraying parameters
The spray gun was supplied with compressed air with a
pressure between 2 and 6 bar during the experiments.
According to the numerical results (for example, FLUENTcalculations), corresponding to the airflow, the speed can
be calculated using the formula:

(18)
where P is the relative pressure in bars.
For example, for a pressure of 4.15 bar, corresponding
to the gas flow with a flow rate of 101.4 cubic meters per

http://www.revistadechimie.ro REV.CHIM.(Bucharest) ♦69♦No. 9 ♦2018 2576parameters. According to this method, it is expected for all
particles within the size distribution to be detected by the
DPV instrument. In other words, it is expected for the DPVnot to favor a particular class of dropets, such as large
ones or high-temperature ones. Table 3 shows the diameter
of the distribution particles resulting for different currentintensities, read using DPV , with diameter correction.in the electric arc of 100 A (that is, for a load of 2.62 kg/h ).
These characteristics determine droplet behavior inside
the air jet and the effect of the loading process. In order toestablish the standard
k-ε turbulence model, a Prandtl
number with a turbulence of 0.85 was used.
Figure 4 shows the pressure ranges computed inside
the spraying gun head and at the exit region, ranges close
to the two symmetry planes. It is noted that the pressure
drops along the nozzle axis (for example, the primarynozzle) and through the four openings in the green cap
(the secondary path around the primary nozzle). At the
same time, there is a negative pressure region on thespraying wires, up to the intersection point of the two wires
forming the electric arc, from where the pressure rises
again to the ambient pressure.
Table 3
THE DISTRIBUTION OF PARTICLE SIZE of PARAMETERS, AT
DIFFERENT ELECTRICAL CURRENT INTENSITIES (P = 4.15bar)
According to this type of distribution dimension, the
mass fraction of particles with a size greater than d is givenby the formula:

(20)
where d and n are the mean mass diameter, and
respectively, the parameter spread. Actually, d is defined
as the mass fraction of particles having a diameter greater
than d, which is 36.8%. Figure 3 presents the particle size
distribution (cumulated mass) with the particle diameter
at different current intensities.
Fig. 3. Granulometric distribution in terms of volume, of the
fraction of measurements with particle diameter at different
current intensities in the electric arc
Analyzing the diagram shows that there is a small
increase in the mean particle size, correlated with an
increase of the current intensity in the electric arc, thedifferences in the size distribution compared to the
parameters used being very small.
Results and discussions
a) Numerical results
The required computation time for the different cases
considered (0.333 million cells) was approximately 5 hours
for 3000 iterations, using a standard computer equipped
with an Intel® Core ™ i5 processor. Although the flow rateis well established after 600 iterations, the residue
continues to decrease afterwards. Finally, after 30000
iterations, all residues reach values between 3.0 x 10
-5 (the
turbulence kinetic energy) and 6.0 x 10-8, except in the
continuity case, for which the residue does not decrease
after 600 iterations. However, the airflow asymmetry onthe boundaries of the computing domain is about 3.0 x 10
–
6 times the mass flow flowing through the nozzle after
3000 iterations. Some of the main flow characteristicsinside and outside the spraying gun head are shown for a
relative supply pressure of 4.15 bar, and a current intensity
Fig. 4. Pressure fields computed inside the spray gun head and at
the exit region near the vertical plane (left) and on the horizontal
plane (that is the wire plane, right) for the pressure of 4.15 bar
As can be seen, the present three-dimensional model is
entirely justified, because of the complexity of the spraying
gun head which leads to a fairly different flow in the twosymmetry planes.
Figure 5 shows the fields corresponding to speeds,
located over an extended area of interest.
Fig. 5. The jet propulsion speed at the top and bottom part of the
symmetry planes and the jet divergence over the axial plane
extended by 100 mm through the aluminum casing in the exit plane
From the first representation it can be noted that the
propulsion speed of the jet increases along the primary
nozzle, and also, the air that passes through the secondarypath around the primary nozzle can be noted. There is a
region where the speed is very high, located at the
intersection point of the spraying wires. This high-speedregion forms a horizontal V in the vertical plane (that is,
perpendicularly on the wires). A strong difference between
the jet divergence in the vertical and horizontal plane, canalso be noted. It has been found that jet divergence is much
more pronounced in the spraying wires plane than in the
vertical plane. Figure 6 shows the droplet trajectories,viewed in the two (colored) symmetry planes, based on
the axial speed of the droplet, for the same case presented
above (that is, for a load of 2.62 kg / h). The extendedexternal range has a length of 200 mm. The divergence of
the particle jet in the vertical plane (top view, lateral view)

REV.CHIM.(Bucharest) ♦69♦No. 9 ♦2018 http://www.revistadechimie.ro 2577is smaller compared to the one observed in the wire plane
(below, top view). More specifically, the angle of the
divergent particle jet is 12 ° in the vertical plane and about
20 ° in the horizontal plane, resulting in an elliptical sprayingpattern, with a radius of 20 and 35 mm, at a spraying
distance of 200 mm. The speed range is between 50 and
220 m / s at a distance of 200 mm.
a1) The effect of the turbulence model on the numerical
results
The standard k-ε turbulence model is the best known
and used model. However, one of its main drawbacks refers
to computing the round jet type. Therefore, it was timely tocheck the influence of the turbulence model on the
numerical results. Surprisingly, it was found that the results
obtained using the
k-ε turbulence model, were slightly
different from those predicted by the standard model.
Figure 7 presents the influence of a current intensity of
100 A, in which case the jet core presents almost the samelength for the two models, and the predicted speed
becomes slightly higher for the standard model from a
distance of 120 to 200 mm. Similarly, predicting the dropletspeeds was not done separately for the two models. For
the smallest droplets, the predicted speed is 1 m / s, lower
than the one of the feasible model, which is 2 m / s, andhigher for larger droplets. Thus, it was established to take
into account only the results obtained with one of the
turbulence models, and therefore the
k-ε model was used.
In fact, disturbing the flow on the wires appears sufficient
in order to cancel the differences in the predictions obtained
with the two models. Moreover, the result obtained indicatesthat the new flow is not comparable to the flow emitted
through a circular nozzle (round jet).a2)- The effect of the loading process
The results presented in figure 7 were those obtained
for a current intensity in the electric arc of 100 A. Figure 8
shows the loading effect on the axial speed componentalong the central axis. There is no loading effect on the
flow inside the spray gun.
Fig. 6. Particle trajectories seen in the
vertical symmetry plane (above) and the
horizontal symmetry plane (below)
Fig.7. Computerized axial component of gas speed for two
different turbulence models (continuous line represents the k-ε
standard model, dotted line represents the k-ε feasible model)
Fig. 8. Loading effect on the axial component of the spraying air
speed (for the k-ε model).
However, there is no loading effect inside the air jet. At
the 30 mm axial distance from the nozzle exit, the axial
speed component is about 50 m/s less for an intensity of
250 A, than for the one-way model (no effect is seen onthe airflow particles), as well as for other intermediate
cases.
However, the difference between cases with different
loading speeds, decreases with the axial distance. For
example, the difference between the two extreme cases
is 5 m/s only at the axial position of 200 mm (130 m/s formodel compared to 125 m/s for 250 A). It can also be seen
that the maximum speed is about 600 m/s and this value
is reached only at the intersection point with the sprayingwires. This very high speed value is due to the very small
gap between the two spraying wires at this point (fig. 4 or
5).
At the same time, it is obvious that this distance
fluctuates over time, due to the periodicity of the electric
arc (due to the wire contact) and the variation of the wirelength (because the melted material is removed from the
wires and deposited). Thus, this maximum value should
not be given much attention, because it fluctuates overtime. Figure 9 shows a comparison between the
experimental speeds obtained for the droplets and the
predicted speeds, for a current intensity in the electric arcof 100 A. Figure 10 shows the similar data obtained at a
current intensity greater than 100 A, namely, 175 A.
Error deviations are the standard deviation and have a
difference of ± σ. In terms of measurements, error
deviations also have an x component, the reason being
that particles considered in the 20 µm class have diameters

http://www.revistadechimie.ro REV.CHIM.(Bucharest) ♦69♦No. 9 ♦2018 2578ranging from 17.5 to 22.5 µm. On the contrary, for the
model, a finite number of different diameters are
considered in FLUENT. For a given diameter, the speeds
computed are slightly higher than the measured ones. Thespeed difference is about 10 m/s for the smallest droplets
and increases up to 20 m/s for the largest droplets. Thus,
the curve slope indicates the speed, related to the dropletdiameter, speed which is higher for the experiments than
for the chosen model. The results indicate that the
correction factor, which was applied to the initial diameter,was too high. In other words, the results will be consistent
only if the experimentally considered diameter will be about
50% higher for each class of particles. In this case, theaverage diameter determined by the DPV device of FLUENT
software, does not correspond to that measured on the
basis of photographs of collected particles. It is noted thatthere are no significant differences between figures 9 and
10. In particular, the speed range is almost the same (with
the difference that the speed is less than 4 m/s for eachparticle class), so it can be said that the loading effect is
fairly weak, not only for the experiments, but also for the
chosen model. The differences between the experimentsand the chosen model are also similar. According to the
experiments, the average droplet speed (relative to a
number of fractions) is 120 m/s at 100 A, 113 m/s at 175 Aand 107 m/s at 250 A. Taking into account the fractions of
the same number, the average speeds of the model are
128, 125 and 122 m/s for current intensities in the electricarc of 100, 175 and 250 A. Several reasons, such as the
inaccuracy of the model, could explain the small differences
between modeling and experiments, but the mostimportant is, undoubtedly, the uncertainty of the diameter
of the particles detected using DPV . In particular, it is not
obvious that the detected particles correspond to the entirerange of diameters of the collected particles. For example,the smallest particles could be too cold and, as such, their
corresponding signal would be too low, especially at a
spraying distance of 200 mm. In this case, an error couldexist in the correction factor applied to the diameter, such
that the diameter of the detected particles could be more
important than the fact that it was considered to be aconstant factor.
a3) The effect of the supply pressure
The loading effect was analyzed for a supply pressure of
4.15 bar and it was shown that increasing the load tends to
slightly increase the size of the droplet and slightly decreasethe speed of the droplet. Experimentally, the main
parameter affecting the dimensional distribution of the
droplet and the deposit speed, is the supply pressure (and,subsequently, the air flow rate). The results obtained at a
supply pressure lower than 2.75 bar are shown in Figure
11. The current intensity in the electric arc was 100 A andthe average diameter of the collected particles was
substantially higher than that corresponding to a higher
pressure level, namely, the average diameter was 39 µm
instead of 18.8 (table 2). Consequently, the Rosin-Rammler
granulometric distribution in the mass fraction was fixed
using the values d = 66 mm and n = 3.1 instead of 33.4µm and n = 3.2. Thus, the average diameter of the particles
obtained at 2.75 bar was about twice as big as that obtained
at 4.15 bar, although these were perfoemed with the capon and with a different material. A decrease in the mean
droplet size resulted, as the spraying pressure was
increased.
Fig. 9. Comparison between the experimentally obtained speeds
and the predicted ones (I=100 A).
Fig. 10. Comparison between the experimental particle speeds
and the predicted ones (I=175 A).
Fig. 11. Comparison between the measured speeds of the particles
and the predicted ones, at a lower pressure level of 2.75 bar
and I = 100 A
According to the experimental results shown in Figure
11, a very small influence of the particle diameter on theparticle speed is observed, which means that droplet speed
is almost independent of droplet diameter. This behavior is
very different from that obtained at a higher pressure level,at which a strong influence of droplet size on droplet speed
was observed. Regarding the modeling results, a
significant influence of the droplet diameter can beobserved. Considering these observations, it follows that
the difference between experiments and modeling results
is significant for the smallest particles, while the results forthe largest particles present a smaller discrepancy. It is
possible that the smallest droplets are not detected by the
DVP for the lowest level of spraying pressure. In fact, smalldroplets have a lower speed and lower temperature under
low spraying pressure conditions (for example, longer
transport times give a higher cooling speed). The lowertemperature could explain why the smallest droplets are
not detected at the lowest spraying pressure. Some
additional experiments performed at different distancescould allow for this hypothesis to be confirmed, because,
in principle, the smallest droplets should have a higher

REV.CHIM.(Bucharest) ♦69♦No. 9 ♦2018 http://www.revistadechimie.ro 2579temperature at a distance of 100 mm, compared to a
distance of 200 mm.
Conclusions
The paper presents a three-dimensional model of the
electric arc thermal spraying process using the Tafa 9000type gun, similar to the type of thermal spraying gun used
in experiments. The elements of the head of the thermal
spraying gun were designed with CAD, used to generatethe node network, including the four distinct volumes. The
FLUENT CFD (commercial fluent design) program was
used to study the air-material flow inside and outside thespraying system. Two different types of
k-ε turbulence
models were tested, namely a standard one and a feasible
one, which is indicated for the study of round jets. The twomodels were used to establish whether similar results are
obtained, specifying that, although the spraying gun head
has a circular socket, its emission flow exhibits a stronglydifferent behavior compared to that of a round jet, due to
the complexity of the internal elements. Thus, in particular,
the external jet shows a significantly different divergencein the vertical and horizontal planes. Also, two different
models were used to consider a second phase, namely,
molten droplets sprayed onto the surface of the layer. Thefirst was a one-way model, in which the effect of the two
phases on the primary flow is negligible. The second model
was based on a two-way coupling, which allows theloading effect to be considered. For this model, the
distribution of the droplet sizes was established through
measurements, which confirmed that the actual meanparticle diameter was considered to be that determined
using the particles collected along the spraying distance
path. The mass flow of pulverized material, depending onthe current intensity of the electric arc, was determined by
a relationship derived from a series of experimental results.
The established model was validated using a series of
comparisons between droplet transport time and drop
speed, measured using the DPV diagnostic tool. Inconclusion, it can be said that the current model can be
used by designers of thermal spraying equipment in order
to improve their design and in order to test the influence ofparameters such as spraying gas pressure or loading effect.
Moreover, the provided functions, established by simulation,
can be useful to all users who wish to use the FLUENTsoftware in order to customize their models.
References
1. BUTNARIU I.,RUICA M.F.,RUICA M.I., Aspects of the thermal spray
process arc, Rev.Metal.Int. 15, no.9, 2010, p.51-54
2. BUTNARIU I.,FLORESCU D.I., Operating characterization of productsupport body couples- coating and superficial layer , Rev.Metal.Int.
15 , no.11, 2010, p. 82-86
3. BUTNARIU I., Research achievement and investigation performancecharacterization of composite materials, Rev.Metal.Int. 15 no.10, 2010,
p.72-76
4. BUTNARIU I.,RUICA M.F.,CERNAIANU-STOIANOVICI A.A., CHIVU E.,RUICA M.I., Studies and researches on achievement strength
parameters and technological parameters of metal layer obtained by
arc thermal spraying, Rev.Metal.Int. 15 , no.8 , 2010, p. 16-18
5. BUTNARIU I.,Mathematical models used for computer simulation
for solidification processes of steel in chill , Rev.Metal.Int. 15, no.11,
2010, p.79-816. BUTNARIU I., Possibilities to properties increase the main of
reinforced mechanical metallic composites , Rev.Metal.Int. 14, no.13
,2009, p.5-77. AMZA V .D., BUTNARIU I., Researches regarding the influence of the
physical-chemical factors of elaboration, casting and solidification on
the steel ingots structure homogeneity, Rev.Metal.Int., 15, no.10, 2009,
p.39-418. BUTNARIU I.,The processing technology of coating by thermal
spraying arc at plane bearing and trees from various parts ,
Rev.Metal.Int. 15, 2010, p. 26-29
9. BUTNARIU I.,CERNAIANU-STOIANOVICI A.A., Studies on materials
used in present for thermal spraying process and their properties,
Rev.Metal.Int. 17, no. 12, 2012, p.93-95
10. BUTNARIU I., ANGELESCU D., FLORESCU D., Studies and
researches on quality improvment of continously cast billets unalloyed
and low alloyed steels, Rev.Metal.Int. 16 , no.6, 2011, p.33-37
11. BUTNARIU I.,CERNAIANU-STOIANOVICI A.A., ANGELESCU D.,
Requirements in service imposed to materials for friction couplings,
Rev.Metal.Int. 16 , no. 6, 2011, p.38-41
12.BUTNARIU I., CERNAIANU-STOIANOVICI A.A., Comparative analysis
of materials used for antifriction in metalization with electric arc
upon request operating result, Rev.Metal.Int. 17, no.12,2012, p. 96-98
13. BUTNARIU, I., FLORESCU, D.I., Research on influence of structure
on physical and mechanical properties of steel , Rev.Metal.Int. 15 ,
no.11, 2010, p.87-9014. BUTNARIU I., Models used in reffining process of treated steels
with rare gases and synthetic slags , Rev.Metal.Int. 15 , no.10, 2010,
p.77-7915. BUTNARIU I.,CERNAIANU-STOIANOVICI A.A., RUICA M.F .,
Researches on growth properties of attrition of alloyed steel used in
metal parts used in high temperature corrosive environments,Rev.Metal.Int. 15, no.7, 2010, p.31-33
16. NITA P .S., BUTNARIU I., CONSTANTIN N., The efficiency at
industrial scale of a thermodynamic model for desulphurization ofaluminium killed steels using slags in the system CaO-MgO-Al
2O3-
SiO2, Revista de metalurgia 46 , Espana, no.1, 2010, p.5-14
17. BUTNARIU I.,CONSTANTIN N., FLORESCU D.I., Studies on theemergence tensions in coatings to the request operating,
Rev.Metal.Int. 15, no.8, 2010, p.47-50
18. BUTNARIU I.,CONSTANTIN N., FLORESCU D.I.,Development andexperimentation with multifactorial intervention models for
prevention and reduction events for circulation, Rev.Metal.Int. 15,
no.9, 2010, p.85-87
19. ANGELESCU M, BUTNARIU I., ANGELESCU D., Own experiments
and pollution analysis on flow technology components sequences inOLD 1 section, U.P .B Sci. Bull., Series B, vol. 78, no.2, 2016 p.223-232
20. ANGELESCU M, BUTNARIU I., ANGELESCU D., Experimental
research regarding pollution at ArcelorMittal Galati in the load andsteel casting preparation sectors , U.P .B Sci. Bull., Series B, vol. 78,
no.3, 2016 p.213-220
21. BUTNARIU I.,RUICA M.F . , Environment technology on the recycling
possibility through OXYCUP waste slag process, iron slam briquettes,
dross of steel and rolling mill, sterile siliceous, Rev.Metal.Int. 15
,no.13, 2009, p.8- 1022. BUTNARIU I., RUICA M.F., The influence of agglomerate
,s ferrous
structure on performance of the iron furnace technology development
process, Rev.Metal.Int. 15 ,no.13, 2009, p.11- 14
23. BUTNARIU I., RUICA M.F ., Researches concerning the influence
sinter in the blast furnace on melting zone position. Rev.Metal.Int.,
15 , no.5, 2010, p.52- 56, 2010
24. BUTNARIU I., RUICA M.F ., The influence of agglomeration process
optimization of iron ore the development of iron in blast furnace,
Rev.Metal.Int. 15 , no.6, 2010, p.22- 24
25. CONSTANTIN N., NICOLAE M.,GEANTA V .,BUTNARIU I,, Dynamic
Model for Determining Direct Reduction Degree of Iron Oxides in
Blast Furnace , Rev. Metalurgia (Bucharest) 1993, p.5-726. WANG , X., HEBERLEIN, J,PFENDER, E.,GERBERICH, W ., Effect of
nozzle confirugstion, gas pressure, and gas type on coating properties
in wire arc spray, J.Therm.Soray Technol. 8, no.4, 1999, p.56527. BOLOT, R., LI, J., CODDET, C., Some key advices for the modeling
of plasma jets using FLUENT,Thermal Spray connects : Explore its
surfanging potential.DVS, DE, 2005, p.136728. DJAMARANI, K.M.,CLARK, I.M.,Characterization of particle size
based on fine and coarse fractions, Powser Technol, 93, 1997, p.101.
29. BUTNARIU I., Processes and technologies in extractive metallurgy,Technical Publishing House, Bucharest, 1995

http://www.revistadechimie.ro REV.CHIM.(Bucharest) ♦69♦No. 9 ♦2018 258030. BUTNARIU I., GEANTA V .,Tehnologii speciale de elaborare si
rafinare a otelurilor, Ed. Universitatea Politehnica Bucuresti, 1993
31.BUTNARIU I., CONSTANTIN N.,Protectia mediului înconjurãtor șimicroclimat, Ed.Universitatea Politehnica București, 1994
32. BUTNARIU I., FLORESCU N.T., PAVELESCU D., RIZEA C.,
Experimental research on the refining system LF-VD steels, Sci. Bull.of V alachia University MATERIALS and MECHANICS, vol.13, No.10 ,
2015, p.91-94
33. BUTNARIU I., FLORESCU N.T., PAVELESCU D., RIZEA C., Modelsused refining process made of steel treated clay and insert gas, Sci.
Bull. of V alachia University MATERIALS and MECHANICS, vol.13 , No.10,
2015, p.99-10134. BUTNARIU I., BUTNARIU D., BUTNARIU I., Studies and researchers
concerning the presence of iron in secondary aluminium meltings
and valorification of some aluminium waste, Journal Materials ScienceForum Vol. 690 , 2011, p. 250-253, Trans.Tech. Publications, Switzerland,
Online doi: 10.4028/www.scientific.net/MSF .690.250.
35.BUTNARIU, I., FLORESCU, D.I., Analysis the main sources ofpollution in metallic ferrous materials industry , Rev.Metalurgia 64 ,
no 7, 2012, p. 9-1336. BUTNARIU I., BUTNARIU D., BUTNARIU I., Technological
researches concerning a decrease in the lossesdue to the oxidation
of remelting scrap from aluminium alloys, Journal Materials ScienceForum Vol . 630 , 2009, p.71-74 , Trans. Tech. Publications, Switzerland,
Online doi: 10428 /www.scientific.net/MSF .630.71
37. BUTNARIU I., Heat treatment influence over the austenitic steelproperties used as biomaterials, Journal Materials Science Forum ,
Vol. 426-432 , 2003, , p.3295-3298 ,Trans.Tech.Publications, Switzerland,
2003 , Online doi : 10428/www.scientific.net/MSF .426.3295.38. BUTNARIU I.,CONSTANTIN N., Integrated Technology for obtaining
composites by electric arc metalizing applications to special purpose
industrial equipment-acronym MCDS, MEdN-PNCDI-CEEX ProgramContract 289, 2006-2008
39. BUTNARIU, I., Considerations for selection of characterization
methods and procedures of layer deposited throw metal coating andquality control methods, Sci. Bull. of V alachia University MATERIALS
and MECHANICS , No.5, 2010, p.115-120
40. BUTNARIU, I., Setting regines and processing possibilities ofmaterials submitted by unconventional processes at reconstructing
references, Sci. Bull. of V alachia University MATERIALS and
MECHANICS , No.5, 2010, p.121-125.
Manuscript received: 27.02.2018