JTTEE5 18:753–768

DOI: 10.1007/s11666-009-9309-2
1059-9630/$19.00
The Author(s). This article is published with open access at Springerlink.com

Peer Reviewed
Modeling and Control of High-Velocity
Oxygen-Fuel (HVOF) Thermal Spray:
A Tutorial Review
Mingheng Li and Panagiotis D. Christofides

(Submitted January 6, 2009; in revised form February 17, 2009)

This work provides a tutorial overview of recent research efforts in modeling and control of the
high-velocity oxygen-fuel (HVOF) thermal spray process. Initially, the modeling of the HVOF thermal
spray, including combustion, gas dynamics, particle in-flight behavior, and coating microstructure evo-
lution is reviewed. The influence of the process operating conditions as predicted by the fundamental
models on particle characteristics and coating microstructure is then discussed and compared with
experimental observations. Finally, the issues of measurement and automatic control are discussed and
comments on potential future research efforts are made.

Keywords control, HVOF, modeling

counterparts in terms of microhardness, elastic modulus,
wear resistance, and thermal conductivity.
Modeling and control play an important role in the
design and operation of the HVOF thermal spray process,
partly because of its inherent complexity. Generally, the
1. Introduction relationship among processing conditions, particle char-
acteristics, and the resulting coating properties is highly
High-velocity oxygen fuel (HVOF) is a thermal spray nonlinear and might not be thoroughly revealed by
technique that was introduced in 1980s (Ref 1). It is a experimental studies. As a result, the optimization of
particulate deposition process in which microsized parti- HVOF thermal spray based on the conventional design of
cles of metals, alloys, or cermets are propelled and experiments (DOE) methodology could be limited, espe-
heated in a sonic/supersonic combusting gas stream and cially when the HVOF system is significantly modified
are deposited on a substrate at high speeds to form a (Ref 5). Mathematical modeling is an excellent comple-
layer of coating. Compared with plasma spray, coatings ment to the experimental approach. It provides a funda-
sprayed by the HVOF process have outstanding charac- mental understanding of the underlying momentum and
teristics including higher density, bond strength, and heat-transfer mechanisms, which in turn, might be used to
toughness, as a result of the significantly higher particle guide system design and operation (Ref 6, 7). Moreover,
velocity at impact and relatively lower particle temper- the compensation of feed disturbances and process vari-
ature. High-velocity oxygen fuel sprayed functional ability during real-time process operation becomes
coatings are now widely used in various industries to essential in order to fabricate coatings of a consistent
enhance performance, extend product life, and reduce quality. This motivates the development and implemen-
maintenance cost. Representative examples include WC/ tation of real-time control systems in HVOF thermal
Co-based wear-resistant coatings for drilling tools, YSZ- spray to suppress variations in the particle temperature
based thermal barrier coatings for turbine blades, and and velocity at the point of impact on the substrate. Such
nickel-base corrosion-resistant coatings for chemical efforts have been facilitated by recent advances in the gas
reactors. More recently, the HVOF process has been and particle velocity and temperature measurements via
used for processing nanostructured coatings using real-time diagnostic techniques (Ref 8).
microsized nanocrystalline powder particles (Ref 2-4),
which exhibit superior properties over conventional
2. Modeling of HVOF Thermal Spray

Mingheng Li, Department of Chemical and Materials Engineer- 2.1 Multiscale Character of HVOF Thermal Spray
ing, California State Polytechnic University, Pomona, CA 91768; Figure 1 shows a schematic of a representative HVOF
Panagiotis D. Christofides, Department of Chemical and Biomo-
thermal spray system (Sulzer Metco Diamond Jet Hybrid
lecular Engineering, University of California, Los Angeles, CA
90025; and Department of Electrical Engineering, University of 2700, Sulzer Metco, Westbury, NY). In this process, the
California, Los Angeles, CA 90025. Contact e-mails: pdc@seas. premixed fuel gas (typically propylene or hydrogen) and
ucla. edu and [email protected] oxygen are fed to the air cap, where they react to produce

Journal of Thermal Spray Technology Volume 18(5-6) Mid-December 2009—753

 Extended Air Cap
Peer Reviewed

Cooling Water Inlet Cooling Water Outlet

Air

Powder 12o 2o

10.94 mm
7.16 mm
and N2

Oxygen
and Fuel

15.6 mm 7.95 mm 54.0 mm

Fig. 1 Diamond Jet Hybrid thermal spray gun

high-temperature combustion gases. The exhaust gases, The gas flow in HVOF thermal spray is essentially a
together with the air injected from the annular inlet orifice, compressible reacting flow process featured with turbu-
expand through the nozzle to reach supersonic velocity. lence and subsonic/sonic/supersonic transitions (Ref 13).
The air cap is cooled by both water and air to prevent it A comprehensive description of this process requires
from melting. The powder particles are injected at the time-consuming direct numerical simulations. To simplify
central inlet nozzle using nitrogen as the carrier gas. the simulation, the Reynolds-Averaged Navier-Stokes
Consequently, rapid momentum and heat transfer between (RANS) equations are usually used so that the small-scale
the gas and the powder particles lead to acceleration and turbulent fluctuations might not be directly solved. To
heating of the particles. The molten or semimolten parti- convert the Navier-Stokes equations into the ensemble-
cles are carried toward the substrate by the expanding gas averaged form, the Boussinesq hypothesis (Ref 14) is
jet. The particles hit the substrate, cool, and solidify, employed to represent the Reynolds stresses (with fluc-
forming a thin layer of coating material with low porosity. tuation terms) with the mean velocity gradients (without
The multiscale character of a typical HVOF thermal fluctuation terms). Specifically, the governing equations
spray process is shown in Fig. 2 (Ref 9). The microstruc- include the conservations of mass, momentum, energy,
ture of HVOF sprayed coatings results from the defor- species transport, turbulent kinetic energy, dissipation
mation, solidification, and sintering of the deposited rate, and so forth (Ref 11). The equations written in
particles, which are dependent on the substrate properties Cartesian tensor notation are:
(e.g., substrate temperature) as well as the physical and Continuity:
chemical state (e.g., temperature, velocity, melting ratio,
and oxidant content) of the particles at the point of impact @q @
on the substrate. The particle in-flight behavior, however, þ ðqvj Þ ¼ 0 ðEq 1Þ
@t @xj
is coupled with the gas dynamics, which are directly
related to various processing conditions such as gas flow where q is the density, t is the time, and vj are the velocity
rate, fuel/oxygen ratio, spray distance, and so forth. While components in each of the xj directions.
the macroscopic thermal/flow field can be readily Momentum balance:
described by continuum-type differential equations gov-
erning the compressible two-phase flow, the process of @ @ @p
ðqvi Þ þ ðqvi vj Þ ¼

particle deposition is stochastic and discrete in nature and @t @xj @xi
is best described by stochastic simulation methods (Ref 9). 

@ @vi @vj 2 @vl
þ l þ
dij
@xj @xj @xi 3 @xl
2.2 Gas Dynamics
@
The gas and particle dynamics in the HVOF thermal þ ð
qv0i v0j Þ; i ¼ 1; 2; 3 (Eq 2)
@xj
spray are coupled with each other. To simplify the anal-
ysis, the assumption of one-way coupling is usually made where p is the pressure, l is the molecular viscosity, dij is
when the particle loading, which is defined as the ratio of the Kronecker delta, and
qv0i v0j is the Reynolds stress
mass flow rate of particles to that of gases, is low (about term representing the effect of turbulence. Based on the
4-5%). Under such an assumption, the existence of par- Boussinesq approximation,
ticles has a minimal influence on the gas thermal and flow


dynamics, while the particle in-flight behavior is derived @vi @vj 2 @vl

qv0i v0j ¼ lt þ
qk þ lt dij
from the particle momentum and heat-transfer equations @xj @xi 3 @xl
(Ref 10, 11). The modeling of the two-way coupling of the  
gas and particulate phases is also possible at the expense where lt is the turbulent viscosity [lt ¼ qcl k2 =
] and k is
of a higher computational cost (Ref 12). the turbulence kinetic energy (k = ½v02 i ).

754—Volume 18(5-6) Mid-December 2009 Journal of Thermal Spray Technology

 Peer Reviewed
Gas mass flow rate Particle mass flow rate
Particle size and shape Operating
Substrate properties Fuel/oxygen ratio Injection velocity and angle conditions
Nozzle design Spray distance and angle

macroscale
Composition
Velocity
Pressure
Particle Temperature Process
Density Gas dynamics inflight dynamics
behavior Melting
Velocity
Oxidation

microscale
Temperature

Coating
Coating porosity, hardness, deposition efficiency
microstructure

Fig. 2 Multiscale character of the HVOF thermal spray process (Ref 9, 27)

Turbulence model (using the k-
model as an example): Ideal gas law:

 
@ @ @ lt @k p ¼ qRT ðEq 7Þ
ðqkÞ þ ðqvi kÞ ¼ lþ
@t @xi @xj rk @xj
where R is the specific gas constant, or the molecular gas
þ Gk þ Gb
q

YM (Eq 3) constant divided by the molecular weight of the gas.
In the modeling of HVOF thermal spray, it is very
and

  important to note that because of the high operating
@ @ @ l @
temperature, combustion products will dissociate into a
ðq
Þ þ ðqvi
Þ ¼ lþ t number of species with low molecular weight, for example,
@t @xi @xj r
@xj
OH and H and so forth (Ref 5, 13). The predicted tem-


2
þ C1
ðGk þ C3
Gb Þ
C2
q (Eq 4) perature could be significantly higher if such a dissociation
k k is not taken into account. Depending on the computa-
where
is the turbulence dissipation rate, Gk and Gb are tional time, single or multistep reduced reaction chemistry
the generations of turbulence kinetic energy caused by the models may be used. In the literature, different
mean velocity gradients and buoyancy, respectively, and approaches have been used to model the reaction rate,
YM is the contribution of the fluctuating dilatation in including: (a) infinitely fast reaction rate (e.g., Ref 10, 13,
compressible turbulence to the overall dissipation rate. 15, 16) or instantaneous chemical equilibrium at the
C1
= 1.44, C2
= 1.92, Cl = 0.09, rk = 1.0, and r
= 1.3. entrance of the combustion chamber, (b) finite reaction
Species transport: rate in Arrhenius form (e.g., Ref 17-20), (c) finite reaction
rate limited by turbulent mixing (e.g., Ref 11, 12, 21-25) or
@ @ @
ðqyi Þ þ ðqvj yi Þ ¼
ðJi Þ þ Ri ; i ¼ 1; . . . ; N
1 the reaction is faster than the mixing rate so that the
@t @xj @xj overall combustion process is limited by the latter.
ðEq 5Þ In many practical situations such as the HVOF thermal
spray, the eddy-dissipation model describes the limiting
where yi is the mass fraction of each species, Ji is the rate, and thus knowledge of accurate Arrhenius rate data
diffusion flux of species i calculated by Maxwell-Stefan is not needed (Ref 12). Based on the fact that the gas
equations, Ri is net rate of production of species i by residence time in the combustion chamber (convergent
chemical reaction, and N is total number of species section of the nozzle) is much longer than the one in the
involved in the reaction. subsequent sections, it is reasonable to assume that the
Energy balance: reaction occurs primarily in the combustion chamber fol-


@ @ @ @H lowing a global one-step equilibrium chemistry model,
ðqHÞ þ ðqvi HÞ ¼ C þ SH ðEq 6Þ which can be determined by minimizing the Gibbs free
@t @xi @xj @xj
energy under constant enthalpy and constant pressure
where C is the ratio of the effective viscosity and the using existing equilibrium codes (Ref 26). The pressure
Prandtl
P number, H is the total enthalpy defined by can be measured at the operating conditions. Alterna-
H¼ N i¼1 Hi yi and SH is the source term (e.g., heat gen- tively, if the mass flow rates of oxygen and fuel are
erated by the exothermic reaction). available, a previously developed approach (Ref 11) can

Journal of Thermal Spray Technology Volume 18(5-6) Mid-December 2009—755



be used to determine the chamber pressure. Specifically, dT ðc
1ÞM2 dA cM2 dx 1
cM2 dq
Peer Reviewed

with the given oxygen and fuel flow rates, a combustion ¼
k þ

T 1
M2 A 2 D ðc
1ÞM2 cp T
pressure is assumed, based on which a quasi-one-
ðEq 9Þ
dimensional model is used to calculate the equilibrium
composition and temperature at the combustion chamber. where A is the cross-sectional area perpendicular to the flow
The total mass flow rate at the throat of the nozzle is then direction, D is the diameter of the nozzle, c is the heat
calculated, based on which the pressure is adjusted until capacity pratio,
ffiffiffiffiffiffiffiffiffiffi and M is the mach number (note that
the discrepancy between the calculated and the specified v/M = a = cRT is the sonic velocity. At the throat of the
total mass flow rates falls below a user-specified tolerance nozzle, where the flow area reaches the minimum, the mach
(Ref 27). number is 1). k is the Darcy friction coefficient correlated as
The computational fluid dynamic (CFD) simulation is a function of the Reynolds number of the equivalent
able to provide the detailed temperature and velocity roughness of the wall by the Colebrook equation (Ref 30):
information of the gas field. For example, Fig. 3 shows the

temperature contour in the combustion chamber of the 1 e=D 2:51
pffiffiffi ¼
2 log10 þ pffiffiffi ðEq 10Þ
Diamond Jet 2700 gun (Ref 11). The function of air is k 3:72 Re k
clearly demonstrated; the hot flame is surrounded by
The term
(dq/dx) is the heat-removal rate per unit length
the cooling air around the torch wall, thus protecting
of the nozzle, which might be estimated by fitting exper-
the hardware from being overheated. Figure 4 shows the
imental data and model prediction (Ref 31). Equations 8
contour and centerline profile of static pressure in the
and 9 are integrated simultaneously using the Runge-
internal and external gas fields. The pressure gradually
Kutta method.
decreases from 6.2 bar in the combustion chamber to 0.6
For an even simpler analysis neglecting friction or cooling
bar (or a gage pressure of
0.4 bar) at the exit of the
(this is reasonable from the combustion chamber to the
nozzle, which is close to a
0.3 bar gage pressure mea-
nozzle throat in HVOF guns such as the Sulzer Metco
sured by the manufacturer (Ref 11). Because this pressure
Diamond Jet Hybrid 2700 (Sulzer Metco, Westbury, NY)
is less than the ambient pressure, the flow is underex-
and the Praxair-Tafa JP-5000 (Praxair Surface Technolo-
panded and a mach disk is formed downstream of the
gies, Indianapolis, IN)), the following isentropic equations
nozzle exit. The flow adjusts to the ambient pressure by
may also be used to relate the temperature, pressure, and
a series of shock waves, which are usually observed
density at two points along the nozzle (Ref 32):
experimentally.
For a quasi-one-dimensional analysis, the assumptions T2 1 þ ½ðc
1Þ=2M12
of instantaneous equilibrium at the entrance of the com- ¼ ðEq 11Þ
T1 1 þ ½ðc
1Þ=2M22
bustion chamber and frozen flow along the nozzle can be
used (Ref 28). The mathematical model accounting for  ½c=ðc
1Þ
wall friction and water cooling is described by (Ref 29): p2 1 þ ½ðc
1Þ=2M12
¼ ðEq 12Þ
p1 1 þ ½ðc
1Þ=2M22
2 2
dM 1 þ ½ðc
1Þ=2M
¼  ½1=ðc
1Þ
M2 1
M2 q2 1 þ ½ðc
1Þ=2M12

 ¼ ðEq 13Þ
dA dx 1 þ cM2 dq q1 1 þ ½ðc
1Þ=2M22

2 þ kcM2 þ
A D 1 þ ½ðc
1Þ=2M2 cp T
ðEq 8Þ

5
Static pressure (10 Pa): 4.4E-01 1.6E+00 2.8E+00 4.0E+00 5.1E+00 6.3E+00

Static Temperature (K)

Static pressure (10 Pa)

3.3E+03
2.7E+03
5

2.1E+03
6
1.5E+03
9.0E+02 4
3.0E+02
2

0
0 0.1 0.2 0.3
Axial distance along the centerline (m)

Fig. 3 Contour of static temperature in the combustion cham- Fig. 4 Contours of the static pressure in the Diamond Jet
ber (Ref 11) HVOF thermal spray gun (Ref 11)

756—Volume 18(5-6) Mid-December 2009 Journal of Thermal Spray Technology

  ½ðcþ1Þ=2ðc
1Þ mixing of cold carrier gas and the hot combustion flame in
A2 M1 1 þ ½ðc
1Þ=2M22

Peer Reviewed
¼ ðEq 14Þ the combustion chamber, which is multidimensional in
A1 M2 1 þ ½ðc
1Þ=2M12
nature.
Based on the aforementioned relationships, it can be
derived that the total mass flow rate is:
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2.3 Particle Dynamics
p0 cM
pr
2 ½ðcþ1Þ=ðc
1Þ The modeling of particulate phase in the HVOF ther-
m_ ¼ pffiffiffiffiffiffi Ath ðEq 15Þ
T0 Rg c þ 1 mal spray is typically based on the Lagrangian approach.
First, the average distance between individual particles in
where Ath is the cross-sectional area at the throat (where the HVOF thermal spray process can be estimated based
the area is the minimum), Rg is the molecular gas constant, on the analysis of Crowe et al. (Ref 38). Specifically,
M
pr is the average molecular weight of the combustion
 
products, and T0 and p0 are, respectively, the stagnation Ld p 1 þ j 1=3
temperature and stagnation pressure in the combustion ¼ ðEq 18Þ
dp 6 j
chamber. This equation clearly reveals that the mass flow
rate and the combustion pressure are not independent, where Ld is the distance between two particles and j is the
which explains why the combustion pressure should be ratio of particle loading to particle/gas density ratio. Based
solved using an iterative procedure when the flow rates of on a particle loading of 4% and a density ratio of 103 to
oxygen and fuel are provided. Table 1 shows a comparison 104, Ld/dp is about 20-50, which implies that the individual
between predicted and experimentally measured chamber powder particles are isolated from each other (Ref 37).
pressures under various operating conditions of a Dia- Therefore, it is reasonable to assume that particle coagu-
mond Jet thermal spray gun using the approach developed lation is negligible (Ref 11, 27), and thus the powder size
in Ref 27. The discrepancy is generally less than 6%, distribution does not change during flight (Ref 22). In the
which validates the assumption of chemical equilibrium in modeling of particle dynamics, it is usually assumed that
the chamber. the major force acting on a particle in the HVOF thermal
Equations 8 and 9 or 11 to 14 are only applicable in the spray process is the drag force, and other forces, such as
internal flow field. For a one-dimensional analysis of the the Basset history term, gravitational force, forces caused
external field, empirical formulas might be correlated by pressure gradients, and so forth, can be neglected
from experimental measurement (Ref 28, 33-36). Tawfik (Ref 21, 39). With these assumptions, the particle motion
and Zimmerman provided the following correlation for- along the axial direction in the Cartesian coordinates is
mulas for typical HVOF processing conditions (Ref 29): described by:
(
dvp 1    dxp
v 1;
 x  xc mp ¼ CD qg Ap vg
vp vg
vp ; ¼ vp
¼ 0:85 ðEq 16Þ dt 2 dt
ve 1
exp 1
x=xc ; x>xc
ðEq 19Þ
(
T
Ta 1;
 x  xc where mp, vp, dp, and xp are the mass, velocity, diameter,
¼ 1:25 ðEq 17Þ
Te
Ta 1
exp 1
x=x c
; x > xc and position of the particle, respectively. Ap is the pro-
jected area of the particle on the plane perpendicular to
where the subscripts ‘‘a’’ and ‘‘e’’ stand for ambient con- the flow direction. vg and qg are the velocity and density of
dition and nozzle exit condition, respectively, and xc is the the gas. CD is the drag coefficient, which is a function
potential core length (xc/De = 4.2 + 1.1 Me2 , Ref 29). of the local Reynolds number (Re) defined by Re =
Figure 5 shows the axial profiles of gas velocity and static (dp|vg
vp|qg)/lg, where lg is the gas viscosity (Ref 30). In
temperature along the centerline in a Metco Diamond Jet the processing of nanostructured coatings, the particles are
thermal spray process based on a simplified quasi-one- typically not spherical (Ref 16) and the corresponding
dimensional model (Ref 27, 37) and a CFD model shape factor can be taken into account in the CD calcu-
(Ref 11). It can be seen that the general trends are similar, lation in this case (Ref 40, 41). Note that the particles
while the centerline temperature derived from the one- experience forces from all three spatial directions and
dimensional model is higher than the one from the CFD equations similar to Eq 19 should also be solved in the
model. This is because the former is not able to handle the other two Cartesian coordinates.

Table 1 Comparison of computational and experimental results (Ref 21) for a Diamond Jet thermal spray gun using a
quasi-one-dimensional model (Ref 27)
exp calc
Case O2, scfh C3H6, scfh Air, scfh N2, scfh Pchamber , psia Pchamber , psia Error, %

1 635 185 790 29.4 69.7 68.7
1.4

2 635 185 395 29.4 59.7 56.1
6.0
3 879 185 795 29.4 76.7 77.6 1.2
4 347 110 632 29.4 44.7 44.8 0.2

Journal of Thermal Spray Technology Volume 18(5-6) Mid-December 2009—757

 2000 3000

Gas axial velcity in the centerline (m/s)

Gas temperature in the centerline (K)

1-D model 1-D model
Peer Reviewed

2-D CFD model 2-D CFD model

1750
2500
1500
2000
1250

1000 1500

750
1000
500
500
250

0 0
0 0.1 0.2 0.3 0 0.1 0.2 0.3
(a) Distance along the centerline (m) (b) Distance along the centerline (m)

Fig. 5 Comparison of (a) axial gas velocity and (b) static temperature along the centerline solved using a simplified quasi-one-
dimensional model (Ref 27, 37) and a CFD model (Ref 11)

Assuming spherical particles, the particle heating can melting, and fp is the melting ratio, or the ratio of the
be described by a partial different equation (Ref 42): melted mass to the total mass of the particle (0 £ fp £ 1).

 A numerical integration method is provided in Ref 27
@Tp ðr; tÞ 1 @ @Tp ðr; tÞ for the particle heating/cooling with phase transitions. As
qp cpp ¼ 2 r 2 kp ; 0  r  rp
@t r @r @r mentioned previously, because of the high turbulence in
B:C: the fluid flow, the particles are likely to be affected by the
instantaneous fluctuation in the fluid phase. However,
@Tp ð0; tÞ
¼0 (Eq 20) either the RANS equations or the quasi-one-dimensional
@r model solves only for the mean fluid velocity. In order to
@Tp ðrp ; tÞ account for the turbulent dispersion of particles, the sto-
kp ¼ hðTg
Tp ðrp ; tÞÞ
@r chastic tracking approach can be used in which a random
where rp is the radius of the particle, k is the thermal velocity fluctuation term is added to the mean gas phase
conductivity of the gas, and h is the heat-transfer coeffi- velocity (Ref 22). The fluctuation velocity component in
cient correlated by the Ranz-Marshall empirical equation each spatial direction is kept constant over the charac-
(Ref 30): teristic lifetime of the eddies and is updated in each time
interval. Each run of the above discrete random walk
kg h i model provides a snapshot profile of particle motion in the
h¼ 2 þ 0:6Re1=2 Pr1=3 ðEq 21Þ gas field. A statistical effect of the turbulence on particle
dp
dispersion may be obtained by computing the particle
where the Prandtl number (Pr) is calculated by Pr ¼ trajectories in this manner for a sufficiently large number
cpg lg =kg where cpg , lg, and kg are the heat capacity, vis- of independent simulations.
cosity, and thermal conductivity of the gas, respectively. Key conclusions from the modeling of particle
In general, particles in the HVOF thermal spray pro- dynamics (Ref 11, 22, 27, 46, 47) are summarized in the
cess experience melting and solidification (Ref 42). In such following paragraphs.
a case, Eq 20 should be modified to account for the Particles are affected by the gas field to different
propagation of the melting/solidification front toward/ extents depending on their size. Small particles may be
from the particle center. This is a Stefan problem with accelerated and heated up to very high velocities and
moving boundaries (Ref 25, 43). Further work on the temperatures. However, because of the entrainment of the
breakup, cooling, and solidification of metal particles is ambient air in the external field, the gas velocity and
also available in the literature (Ref 44, 45). For good heat- temperature decay downstream of the potential core
conducting particles, if the Biot number (Bi = hdp/6kp) is (Fig. 5). As a result, the velocities and temperatures of
less than 0.1, the internal particle temperature gradients small particles drop more sharply than those of larger
can be ignored. In such a case, Eq 20 can be simplified to particles because of their smaller momentum and thermal
the following form (Ref 27): inertias. In some cases, small particles may reach the
8 dTp
melting point in a short period of time and become fully
>
< mp cpp dt ; ðTp <Tm ; f ¼ 0Þ melted during flight. However, they may eventually be in a
hA0p ðTg
Tp Þ ¼ DHm mp dfdtp ; ðTp ¼ Tm Þ ðEq 22Þ liquid/solid or even solid state as they reach the substrate.
>
: dTp
mp cpl dt ; ðTp >Tm ; f ¼ 1Þ For particles of large sizes, the periods for acceleration and
heating are both longer, and the velocity (or temperature)
where A0p is the surface area of the particle, Tm is the profiles become nearly flat after their velocity and tem-
melting point of the particle, DHm is the enthalpy of perature are higher than those of the gas. A typical profile

758—Volume 18(5-6) Mid-December 2009 Journal of Thermal Spray Technology

 0.5 µm

Peer Reviewed
2000 1 µm 2500
5 µm
10 µm
20 µm

Axial temperature (K)

2000

Axial velocity (m/s)

1500 30 µm
gas
1500
1000
0.5 µm
1000 1 µm
5 µm
500 10 µm
500 20 µm
30 µm
gas
0 0
0 0.1 0.2 0 0.1 0.2
(a) Distance along the centerline (m) (b) Distance along the centerline (m)

Fig. 6 Profiles of (a) particle velocity and (b) temperature along the centerline (Ref 22)

0.05 For this reason, particles used in the HVOF thermal spray
coating process cannot be too small. This is true even for
the processing of nanostructured coatings, in which the
substrate
powders comprise micron-sized agglomerates with grain
size below 100 nm.
Radial position (m)

thermal spray gun

The particle velocity and temperature at impact are
strongly dependent on particle size and also on the tra-
0 jectory that the particle takes. The maximum of particle
velocity and temperature generally occur at median par-
ticle sizes. When the same size particles take different
trajectories (due to differences in the initial injection
location and random fluctuations in the gas phase), they
might achieve different velocities and temperatures upon
impact, see Fig. 8. This behavior is consistent with
−0.05 experimental observations of Zhao et al. (Ref 48).
−0.05 0 0.05 0.1 0.15 0.2 0.25 0.3
Axial position (m) It is also demonstrated in the modeling study that the
particle temperature can be strongly affected by the axial
Fig. 7 Trajectories of different size particles injected at differ- injection location and initial injection velocity while the
ent radial locations in the flow field (Ref 22) particle velocity is more robust with respect to these two
parameters. This is also validated by the studies of Hanson
et al. (Ref 49, 50). However, the axial injection location is
of particle velocity, temperature, and melting degree is not a variable to be adjusted in real time, and therefore its
shown in Fig. 6 (Ref 37). application in automatic control might be limited.
The effect of particle turbulence dispersion is shown in While mathematical modeling can provide general
Fig. 7 (Ref 22). This snapshot profile of the particle tra- trends and important insight into the behavior of the
jectories in the flow field is obtained by feeding 100 par- HVOF thermal spray process, a perfect match between
ticles to the HVOF thermal spray system from 5 uniformly model prediction and measurement is very difficult to
distributed radial locations in the carrier nitrogen stream obtain. For example, the CFD simulations typically
and 20 uniformly distributed sizes between 1 and 20 lm. It underpredict the particle temperature (Ref 6), partly
is seen that although most particles are highly concen- due to the negligence of exothermic particle oxidation
trated in the centerline inside of the HVOF torch, the (Ref 12) and the contribution of viscous dissipation to the
particles tend to expand toward the radial direction when two-phase heat-transfer coefficient (Ref 51). A particle
they approach the substrate, due to substantial radial gas oxidation model has been recently accounted for in the
velocities (Ref 22). Small particles are significantly influ- CFD model by Zeoli et al. (Ref 52).
enced by the gas flow pattern close to the substrate. They
are also sensitive to the fluctuations in the gas phase and
2.4 Powder Size Distribution
have more random trajectories. Some of them might even
follow the gas stream and will not be deposited on the It is very important to note that the powder particles
substrate. Note that submicron particles might be depos- used in the HVOF thermal spray process are typically
ited on the substrate by thermophoretic force; however, polydisperse, and therefore the powder size distribution is
this force is not as strong as the turbulent dispersion. an important parameter that should be accounted for in

Journal of Thermal Spray Technology Volume 18(5-6) Mid-December 2009—759

 Particle axial velocity at impact (m/s)
700 1600

Particle temperature at impact (K)

Peer Reviewed

600 1400

500
1200
400
1000
300
800
200

100 600

0 400
0 20 40 60 80 100 0 20 40 60 80 100
Particle size (µm) Particle size (µm)
(a) (b)

Fig. 8 (a) Axial particle velocity and (b) temperature upon impact as a function of particle size (Ref 22)

the evaluation of the average in-flight particle character- coating

 properties
 with volume-averaged powder proper-
istics. Two commonly used two-parameter powder size ties PP instead of a single particle size because larger
distribution functions in HVOF thermal spray are log particles have a stronger influence on coating properties
normal and Rosin-Rammler. A log normal powder size than smaller ones. The volume-averaged powder proper-
function has following form (Ref 53): ties are defined as:
" # R dub 1 3
1 ðln dp
lÞ2 d pdp PPðdp Þf ðdp Þdðdp Þ
f ðdp Þ ¼ pffiffiffiffiffiffi exp
ðEq 23Þ PP ¼ lb R6dub ðEq 27Þ
2prdp 2r2 1 3
dlb 6 pdp f ðdp Þdðdp Þ

where l and r2 are two dimensionless parameters corre- where dlb and dub are the lower and upper bounds in the
sponding to the mean and the variance of ln dp, which particle size such that the denominator is larger than 0.99.
obeys the normal distribution. For particles that are log
normally distributed, l and r can be determined using
(Ref 46, 47): 2.5 Particle Impact and Coating Growth
sffiffiffiffiffiffiffi!2 Splat formation and coating growth have been studied
p
3
ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi d90 extensively (Ref 6, 54, 55), and a comprehensive review
l ¼ ln d10 d50 d90
1:831 ln
d10 has been provided by Fauchais et al. (Ref 56) to summa-
sffiffiffiffiffiffiffi ðEq 24Þ rize the effect of particle properties (temperature, veloc-
d90 ity, molten state, oxidization state, etc.) and substrate
r ¼0:781 ln
d10 conditions (roughness, composition, crystallinity, etc.) on
particle flattening, splashing, and solidification. Most of
where d10, d50, and d90 are three characteristic diameters the available research work is based on fully melted par-
that can be obtained experimentally (Ref 2). ticles (which is typical of plasma spray processes), which
The Rosin-Rammler powder size distribution function might not reflect the fact that particles at impact on the
is: substrate in HVOF thermal spray may be in mixed melt-

 
n  ing states (fully melted, partially melted, or solid) due to
n dp n
1 dp different sizes and different trajectories in the gas flow
f ðdp Þ ¼

exp

ðEq 25Þ
dp dp dp field. This has been substantiated by both experimental
studies (Ref 57) and numerical simulations (Ref 11, 27).
or
Moreover, in the HVOF thermal spray processing of

n 
dp carbides with binding metals, such as the WC-Co powders,
Mdp ¼ exp

ðEq 26Þ only the metals may be in a molten state because the gas
dp
temperature in a conventional HVOF thermal spray pro-
where Mdp is the retained weight fraction (weight fraction cess is not high enough for melting carbides, which have

p is the mean
of particles with diameter greater than dp), d high melting points (e.g., 3143 K for tungsten carbide)
particle size, and n is the spread factor. n and d
p in the (Ref 58). Ivosevic et al. (Ref 42) studied the deformation
Rosin-Rammler distribution function are the slope and of HVOF sprayed polymer particles using the volume of
the inverse of the exponential of the intercept slope ratio fluid (VoF) method and found that when particles are
of the straight line represented by ln dp versus partially melted, the internal temperature distribution has
ln(
ln Mdp ) (Ref 22). a significant effect on the final shape of the splat. Their
Because of the polydispersity in the size distribution mathematical model successfully explained the experi-
and the strong effect of particle size on particle velocity mentally observed ‘‘fried-egg’’ shape of the splat, which
and temperature, it is more appropriate to relate the features a low-temperature, nearly hemispherical core

760—Volume 18(5-6) Mid-December 2009 Journal of Thermal Spray Technology

 Peer Reviewed
Particle velocity
Random number Gas dynamics
Particle size Particle temperature
generators Particle dynamics
Particle melting degree

Particle hitting
Splat formation
position

Growth rules

Coating growth

Fig. 9 Multiscale modeling of the HVOF thermal spray process (Ref 9)

located in the center of a thin disk (Ref 42). Therefore, to dp

simulate the coating growth in HVOF sprayed coatings, it
is necessary to account for partially or unmelted particles.
A rule-based stochastic simulation approach was Partially melted particle

developed for the formation of HVOF sprayed coatings in

which the particles of mixed melting states are accounted Unmelted part Melted part
for (Ref 59). The general procedure is shown in Fig. 9.
Specifically, it is assumed that the coating formation pro-
cess is a sequence of independent discrete events of each Dum Splat
individual particle landing on the previously formed
h

coating layer. The powder size and landing location are

derived from random number generators, which follow the
Dm
powder size distribution function and a uniform distribu-
tion function (due to the movement of the HVOF thermal
spray gun), respectively. The particle properties upon Fig. 10 Deformation of partially melted particle upon impact
on the substrate (Ref 59)
impact are then determined based on the gas and particle
models presented previously. The splat is formed and
added to the previously deposited coating layer following melted part will form a ring around this hemisphere,
appropriate rules discussed in the next paragraph. The whose flattening ratio can be calculated using the same
particle deposition is repeated until a desired coating formula, see Fig. 10. This rule is based on experimental
thickness is reached. The program then calculates the observations (Ref 42).
coating porosity, surface roughness, and deposition  When a particle hits the substrate, the melted part of
efficiency. the particle will fit to the surface as much as possible.
The rules governing the particle deformation and The splat will move forward until it is in close contact
coating growth are summarized (Ref 59): with the previously deposited coating surface.
 The deformation of fully melted particles follows the  If the unmelted part of a partially melted particle hits
law of Madejski (Ref 60): at the point of the previously deposited layer that is
formed by an unmelted particle, particle rebound
Ds might occur and consequently a hole is formed in the
n¼ ¼ nðRe; We; PeÞ ðEq 28Þ
dp center of the disk. Otherwise, it will attach on the
coating surface as a hemisphere.
where Ds is the diameter of the thin disk formed by the
particle and n is the flattening ratio that depends on  If the splat comes to a vertical drop during spreading,
several dimensionless parameters characterizing the the ratio of the splat that has not been settled down
impact and spreading processes: (a) the Reynolds will be calculated. If the step does not continue with a
number, which represents the viscous dissipation of the gap that can be covered by the splat, the splat will
inertia forces, (b) the Weber number, which quantifies break or cover the corner of the step according to the
the conversion of the kinetic energy into surface ratio and the height of the step. Otherwise, the gap
energy, and (c) the Peclet number, which expresses the will be covered by the splat and a pore will be formed.
freezing rate. For partially melted particles, this  If the splat encounters a dead end, it will first fill the
assumption is modified such that the unmelted part will available space, and then flow over the outer surface
form a hemisphere with the equivalent volume and the depending on the remaining volume.

Journal of Thermal Spray Technology Volume 18(5-6) Mid-December 2009—761

Peer Reviewed

Fig. 11 Simulated microstructure of coatings formed by fully melted particles and particles with mixed melting states (Ref 11)

The effect of melting state on the coating microstruc- example, residual stress is a key criterion of coating
ture is shown in Fig. 11 (Ref 11). When all the particles are property that may be affected by particle characteristics
fully melted, an ideal lamellar microstructure is formed. (Ref 63-66). Finite element analysis of residual stress of
However, under normal HVOF processing conditions, HVOF sprayed coating has been developed (Ref 67, 68).
many particles might be partially melted or even unmelted. However, the effect of partially molten particles on
When these particles are embedded in the coating, the residual stress is yet to be explicitly accounted for.
microstructure and resulting physical and mechanical
properties could be different. The fact that the particles are
not necessarily to be fully melted to achieve excellent
coating microstructure is consistent with practice. This is 3. Control of HVOF Thermal Spray
very important for the fabrication of nanostructured
coatings, since the nanostructure in the unmelted particles As pointed out by Moreau (Ref 69), some automatic
is preserved during flight. However, under certain condi- control systems have been developed and implemented in
tions where the particle melting ratio is very low, unmelted the thermal spray processing to regulate variables such as
particles may bounce off the substrate, resulting in a high- gas flow and material feed rates, substrate temperature,
porosity coating with a low deposition efficiency. and robot trajectory. However, the more challenging task
Several results from the stochastic simulations of is what and how control actions should be taken to com-
coating growth have been verified by experimental studies pensate for a drift in the particle characteristics observed
(Ref 59). For example, the simulation pointed out that a in the processing. This requires a fundamental under-
high melting degree (which occurs at an equivalence ratio standing of the relationship between manipulated inputs
of about 1.2, or a fuel-rich condition, where the gas and (i.e., key operating conditions) and controlled outputs
particle temperatures are the highest) and a high total (i.e., particle properties at the point of impact on the
mass flow rate would enhance a low coating porosity. This substrate) derived from off-line parametric analysis (e.g.,
is validated by the studies of NiWCrBSi coatings pro- plasma spray, Ref 70 and HVOF spray, Ref 49, 71) as well
cessing using a Praxair-TAFA JP-5000 HVOF thermal as the integration of real-time diagnostics and control
spray system (TAFA Inc., Concord, NH, acquired by algorithms in a feedback loop. Current research and
Praxair Surface Technologies, Indianapolis, IN) (Ref 61). development of online diagnostic and control systems are
Similar conclusions have been drawn in NiAl coatings mainly focused on plasma spray processes, for example,
processing using a UTP TopGun HVOF gun (UTP, Bad real-time diagnostics and control of particle velocity and
Krozingen, Germany) (Ref 62) and in steel coatings temperature in plasma spray using PID (Ref 72) and using
processing using a Praxair-TAFA JP-5000 HVOF gun artificial neural networks and fuzzy logic (Ref 73, 74). For
(Ref 50). The simulation also shows that a better parti- the HVOF thermal spray process, a feedback control
cle melting condition would enhance the deposition system is shown in Fig. 12. In such a closed-loop system,
efficiency, which is consistent with experimental studies the particle information from the online measurement
(Ref 50, 62, 63). system is compared with the desired set point. The dif-
A comparison of simulation results (Ref 11) with ference is sent to the feedback controller, which adjusts
experimental studies (Ref 50) shows that the porosity the processing parameters until the difference becomes
predicted by the model is higher than the experimentally zero. The development of such a feedback control system
measured value under similar operating conditions. One would require online measurement and diagnostic tech-
possible reason is that the CFD model usually underpre- niques, control-relevant parametric analysis as well as
dicts particle temperature (Ref 12). Further development controller design and implementation.
is still necessary to more accurately predict coating A variety of online sensing and diagnostic tools
microstructure and resulting properties from macroscopic have been developed for the thermal spray processes to
operating conditions using multiscale models. For directly measure gas enthalpy and particle characteristics.

762—Volume 18(5-6) Mid-December 2009 Journal of Thermal Spray Technology

 DISTURBANCES particle temperature (Ref 80). This system is able to

Peer Reviewed
analyze particle characteristics of the entire jet when only
OUTPUT the averaged particle information is provided.
INPUT
HVOF
The online diagnostic techniques supplemented with
PROCESS other measurement tools have significantly facilitated the
development of process maps for HVOF process optimi-
FINAL CONTROL
ELEMENT zation. These experimental investigations studied the
MEASUREMENT effect of operating parameters including gun type, fuel
CONTROLLER SYSTEM type, feedstock type and size, combustion pressure, fuel/
DECISION
INFORMATION
oxygen ratio, spray distance on the particle temperature,
velocity, melting ratio, oxidant content and the resulting
SET POINT
coating microstructure, porosity, hardness, wear abrasion,
Fig. 12 A schematic of feedback control of HVOF thermal and corrosion resistance (e.g., Ref 3, 28, 34, 48-50, 61, 62,
spray 65, 71, 81-92). The experimental studies have been sup-
plemented by CFD modeling efforts (Ref 10-13, 15-21,
23), which provide fundamental understanding of the gas
For example, DPV-2000 (Tecnar Automation, Saint-Bruno, dynamics, particle in-flight, and impact. The studies reveal
QC, Canada) is a commercial optical sensing device that that particle characteristics upon impact on the substrate
provides online measurement of individual particle char- (i.e., velocity, temperature, and melting degree) are key
acteristics including temperature (1000-4000 C, precision: parameters that affect coating porosity, bond strength, and
3%), velocity (5-1200 m/s, precision: 3%) and size residual stress. To develop a feedback control system for
(10-300 lm). The working principle of DPV-2000 is based the HVOF thermal spray process, it is necessary to choose
on two-wavelength pyrometry and a dual-slit optical the processing parameters that can be manipulated in real
device (Ref 75). Due to the presence of a photomask time to suppress the variation in these particle character-
between the two slits, a particle passing the measurement istics. It is widely acknowledged that the particle velocity
volume will generate a signal with two-peaks. This signal is and temperature (or melting degree) can be almost inde-
then transmitted through an optical fiber bundle to the pendently regulated by manipulating pressure in the
detection module, where the light passes through a combustion chamber and the fuel/oxygen ratio (Ref 27,
dichroic mirror and two interference band-pass filters 34), respectively. Both of them are functions of the flow
(around 790 and 990 nm, respectively) and is then imaged rates of fuel and oxygen (the dependence of chamber
on two photodetectors. The signals from the photodetec- pressure on gas flow rates is shown in Eq 15). Therefore,
tors are amplified, filtered, and transmitted to the control the manipulated input can be chosen to be the gas flow
module where the particle characteristics are calculated. rates. The effect of gas flow rates on particle velocity and
Specifically, the particle velocity is calculated from the temperature is shown in Fig. 13. The baseline conditions
distance between the image of the two slits in the mea- used in the analysis are the recommended processing
surement volume and the time-of-flight between the two conditions for WC-Co coatings (Ref 11). When the total
peaks of the particle signal. The temperature is deduced mass flow increases from 50 to 150% of its baseline value,
from the ratio of signals from the two photodetectors (or the velocity does not change much but the density
in-
energies radiated at the two wavelengths) with the creases linearly. As a result, the gas momentum flux qv2g
assumption that the particles behave as gray bodies. The is tripled, which implies a significant increase in the drag
diameter is calculated from the signal at one wavelength force for particle motion. This is because the drag force is
after a calibration procedure. The analysis rate of DPV- roughly proportional to qv2g if vp  vg. [Strictly speaking,
2000 can be up to 4000 particles/s, depending on the the drag force would not increase by as high as three times
spraying conditions. If supplemented with CPS-2000, the because of the change in CD. In fact, based on a drag
sensing system is able to characterize cold particles that formula CD = (24/Re) (1 + 0.15Re0.687) valid for Re < 1000
are otherwise not detected. The CPS-2000 is a cold par- (Ref 30), one can derive that FD = (1/2)CDqgAp(vg
vp)|
ticle sensor that employs a diode laser tuned on one of the vg
vp| = 3pldp (vg
vp)(1 + 0.15Re0.687). Therefore, the
pyrometerÕs wavelength to characterize the cold particles. drag force might be only doubled at most]. However, the
Similar techniques have been developed by Idaho gas temperature, difference of which from the particle
National Laboratory (Ref 76-79). The particle size and temperature provides the driving force for particle heat-
velocity are measured from a laser-particle sizing system ing, increases about only 4%. When the total mass flow
and a dual crossed-beam laser Doppler velocimeter. The rate is fixed, the gas temperature varies with the fuel/oxy-
temperature measurement is also based on the two-color gen ratio with an optimal value slightly larger than the
pyrometry. Both ensemble and single-particle techniques stoichiometric value, and a change in the fuel/oxygen ratio
are available to provide either time-resolved/spatially with a fixed total mass flow rate has little effect on the gas
averaged or spatially resolved/time-averaged information. pressure as well as on the gas momentum flux. Similar
Oseir Ltd. (Tampere, Finland) also developed a com- experimental studies are shown in Fig. 14 (Ref 71). In the
mercial product (SprayWatch) that uses a high-speed control of HVOF spray, it should be noted that the gas
charge-coupled device (CCD) camera to measure the temperature varies only about 12-13% in the equivalence
particle velocity and a two-color pyrometer to measure the ratio range of interest (0.55-1.8; the peak temperature

Journal of Thermal Spray Technology Volume 18(5-6) Mid-December 2009—763

 Gas properties at nozzle throat (gp/gp )

Gas properties at nozzle throat (gp/gp bl)

1.25

bl
1.6
Peer Reviewed

pressure pressure
density 1.2 density
1.4 temperature temperature
velocity 1.15 velocity
molecular weight molecular weight
1.2 1.1
momentum flux momentum flux
1.05
1
1

0.8 0.95

0.9
0.6
0.85

0.4 0.8
0.5 1 1.5 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
(a) Total mass flow rate (m/m )
bl
(b) Equivalence ratio φ/φ
bl

Fig. 13 Influence of (a) total mass flow rate and (b) equivalence ratio on the gas properties at the throat of the nozzle. Normalized based
on a baseline condition (Ref 37)

Fig. 14 (a) Velocity and (b) temperature of aluminum particles as a function of total mass flow rate and fuel/oxygen ratio using Praxair
HV-2000 spray gun with propylene as the fuel (Ref 71)

occurs at an equivalence ratio of around 1.2, corresponding flow rate of fuel, u1 (t), and oxygen, u2 (t). Owing to the
to a fuel rich condition), which suggests that the window for almost decoupled nature of the manipulated input/
particle temperature control in the HVOF thermal spray is controlled output pairs, two proportional integral (PI)
rather narrow. This is different from the plasma spray controllers were proposed in Ref 27 and 47 to regulate the
process where the particle temperature can be adjusted in a process. Specifically, the controllers have the following
wider range by manipulating the torch current (Ref 72). form:
Based on model predictions and available experimental
f_ i ¼ yspi
yi ; fi ð0Þ ¼ 0; i ¼ 1; 2
observations, the control problem for the HVOF process  
is formulated as the one of regulating the volume-based   1
u0i ¼ Kci yspi
yi þ fi þ u00i ; i ¼ 1; 2 ðEq 29Þ
averages of velocity and temperature (or melting degree) sci
of particles at impact on the substrate by manipulating the fu1 ; u2 g ¼ f ðu01 ; u02 Þ
flow rates of fuel and oxygen at the entrance of the HVOF
thermal spray gun. The volume-averaged instead of where yspi is the desired set-point value and yi is the value
number-averaged particle properties are chosen in the of the output obtained from the measurement system (y1
control formulation because large particles contribute is the volume-based average of particle velocity, and y2 is
more to the coating volume. The information of individual the volume-based average of particle temperature or
particle size should be available in order to calculate the melting degree), u01 is the combustion pressure and u02 is
volume-based particle characteristics upon impact. In the the equivalence ratio. f is the mapping between the flow
absence of particle size measurements, the control objec- rates and the chamber pressure as well as the equivalence
tive could be the number-averaged particle properties, ratio. Kci is the proportional gain and sci is the integral
similar to the work of Fincke et al. (Ref 72) in feedback time constant. If the gas phase measurement is available, a
control of plasma spray. The manipulation of combustion model-based scheme can be used to estimate the particle
pressure and equivalence ratio is realized by adjusting the properties through the dynamic particle in-flight model

764—Volume 18(5-6) Mid-December 2009 Journal of Thermal Spray Technology

 480 0.52 225 880

Peer Reviewed
Propylene flow rate (scfh)

Oxygen flow rate (scfh)

Particle velocity (m/s)

Particle melting ratio

470 210 800
0.5

460 195 720

velocity 0.48
450 melting ratio 180 640
propylene
oxygen

440 0.46 165 560

0 5 10 15 20 25 30 0 10 20 30
(a) Time (s) (b) Time (s)

Fig. 15 Profiles of (a) controlled outputs (average particle velocity and melting ratio) and (b) manipulated inputs (flow rates of
propylene and oxygen) under the request of 5% increase in particle velocity and 5% increase in melting ratio (Ref 37)

450.5 0.5 177 640

Propylene flow rate (scfh)

Oxygen flow rate (scfh)

Particle velocity (m/s)

Particle melting ratio

176 620
450

0.48 175 600

propylene
velocity
449.5 oxygen
melting ratio
174 580

449 0.46 173 560

0 5 10 15 20 25 30 0 10 20 30
(a) Time (s) (b) Time (s)

Fig. 16 Profiles of (a) controlled outputs (average particle velocity and melting ratio) and (b) manipulated inputs (flow rates of
propylene and oxygen) in the presence of 10% decrease in spray distance (Ref 37)

(Ref 93). Interested readers may also refer to literature for the particle temperature increases significantly. Under
the design of model-based feedback control systems feedback control, the manipulated inputs drive the process
employing nonlinear control techniques for various par- outputs to their original steady-state values in 10 s, which
ticulate processes (Ref 94-100). demonstrates the robustness of the controller.
Closed-loop simulations under the control scheme of
Eq 29 have been carried out to demonstrate the effec-
tiveness of the proposed control formulation (Ref 37). It is
assumed in the computer simulation that the responses of 4. Concluding Remarks
gas and particle dynamics to the change of gas flow rates
are very fast, which is reasonable for such a supersonic The HVOF thermal spray is an enabling technique for
flow. With this simplification, it has been demonstrated the processing of functional coatings. Mathematical mod-
that the feedback controllers are very effective with eling has been an excellent complement to experimental
respect to set-point changes in both particle velocity and studies to provide systematic understanding of the under-
temperature (i.e., 5% increase in both particle velocity lying physics of the process and to enhance coating perfor-
and melting degree). As seen in Fig. 15, both the flow mance through optimized system design and operation.
rates of oxygen and fuel increase in order to have a higher However, detailed model development is still needed in
particle velocity. However, the temperature increases and some areas, for example, the oxidation, melting, and solid-
exceeds its desired value because of the increased cham- ification of particles during flight, deformation of partially
ber pressure. As a result, the changing rate of oxygen flow melted and unmelted particles, and its effect on coating
becomes slower than that of fuel after a short time, which microstructure evolution and residual stress, especially
lowers the equivalence ratio and drives the temperature when the particles are of irregular shape (which is common
down to its set point. Figure 16 demonstrates the response in the processing of nanostructured particles). Further
of the feedback controller in order to maintain the same modeling efforts should also include the validation and
particle velocity and temperature levels in the presence of refining of mathematical models to better match experi-
a 10% decrease in the spray distance (process distur- mental measurements (e.g., gas-particle heat transfer
bance). The particle velocity does not change much while accounting for viscous dissipation).

Journal of Thermal Spray Technology Volume 18(5-6) Mid-December 2009—765

 In addition to modeling, feedback control systems have 13. D. Cheng, Q. Xu, G. Trapaga, and E.J. Lavernia, A Numerical
Peer Reviewed

been designed in which the particle velocity and temper- Study of High-Velocity Oxygen Fuel Thermal Spraying Process.
Part I: Gas Phase Dynamics, Metall. Mater. Trans. A, 2001, 32,
ature are controlled by adjusting the flow rates of oxygen p 1609-1620
and fuel, and their effectiveness and robustness have been 14. J.O. Hinze, Turbulence, McGraw-Hill, 1975
tested through computer simulations. Further research 15. W.L. Oberkampf and M. Talpallikar, Analysis of a High-Velocity
efforts should focus on the development of a demonstra- Oxygen-Fuel (HVOF) Thermal Spray Torch. Part 2: Computa-
tion prototype by integrating real-time diagnostics and tional Results, J. Therm. Spray Technol., 1996, 5, p 62-68
16. D. Cheng, Q. Xu, G. Trapaga, and E.J. Lavernia, The Effect of
feedback control systems in an experimental HVOF Particle Size and Morphology on the In-Flight Behavior of Par-
thermal spray process. ticles During High-Velocity Oxyfuel Thermal Spraying, Metall.
Mater. Trans. B, 2001, 32, p 525-535
17. C.H. Chang and R.L. Moore, Numerical Simulation of Gas and
Particle Flow in a High-Velocity Oxygen-Fuel (HVOF) Torch,
Acknowledgment J. Therm. Spray Technol., 1995, 4, p 358-366
18. B. Hassan, A.R. Lopez, and W.L. Oberkampf, Computational
Financial support from NSF and ONR is gratefully Analysis of a Three-Dimensional High-Velocity Oxygen Fuel
acknowledged. (HVOF) Thermal Spray Torch, J. Therm. Spray Technol., 1998, 7,
p 71-77
19. A.R. Lopez, B. Hassan, W.L. Oberkampf, R.A. Neiser, and T.J.
Open Access Roemer, Computational Fluid Dynamics Analysis of a Wire-
Feed, High-Velocity Oxygen Fuel (HVOF) Thermal Spray
This article is distributed under the terms of the Cre- Porch, J. Therm. Spray Technol., 1998, 7, p 374-382
ative Commons Attribution Noncommercial License 20. S. Gu, C.N. Eastwick, K.A. Simmons, and D.G. McCartney,
Computational Fluid Dynamic Modeling of Gas Flow Charac-
which permits any noncommercial use, distribution, and teristics in a High-Velocity Oxy-Fuel Thermal Spray System,
reproduction in any medium, provided the original J. Therm. Spray Technol., 2001, 2001(10), p 461-469
author(s) and source are credited. 21. G.D. Power, E.B. Smith, T.J. Barber, and L.M. Chiappetta,
Analysis of a Combustion (HVOF) Spray Deposition Gun.
Report 91-8, United Technologies Research Center, East Hart-
ford, CT, 1991
References 22. M. Li and P.D. Christofides, Computational Study of Particle
In-Flight Behavior in the HVOF Thermal Spray Process, Chem.
1. P. Fauchais, A. Vardelle, and B. Dussoubs, Quo Vadis Thermal Eng. Sci., 2006, 61, p 6540-6552
Spraying? J. Therm. Spray Technol., 2001, 10, p 44-66 23. S. Kamnis and S. Gu, Numerical Modelling of Propane Com-
2. M.L. Lau, H.G. Jiang, W. Nuchter, and E.J. Lavernia, Thermal bustion in a High Velocity Oxygen-Fuel Thermal Spray Gun,
Spraying of Nanocrystalline Ni Coatings, Phys. Stat. Solidi A Chem. Eng. Process., 2006, 45, p 246-253
Appl. Res., 1998, 166, p 257-268 24. S. Kamnis and S. Gu, 3-D Modelling of Kerosene-Fuelled HVOF
3. B.R. Marple, J. Voyer, J.F. Bisson, and C. Moreau, Thermal Thermal Spray Gun, Chem. Eng. Sci., 2006, 61, p 5427-5439
Spraying of Nanostructured Cermet Coatings, J. Mater. Process. 25. S. Kamnis, S. Gu, and N. Zeoli, Mathematical Modelling of
Technol., 2001, 117, p 418-423 Inconel 718 Particles in HVOF Thermal Spraying, Surf. Coat.
4. R.S. Lima, C. Moreau, and B.R. Marple, HVOF-Sprayed Coat- Technol., 2008, 202, p 2715-2724
ings Engineered from Mixtures of Nanostructured and Submi- 26. S. Gordon and B.J. McBride, Computer Program for Calculation
cron Al2O3-TiO2 Powders: An Enhanced Wear Performance, of Complex Chemical Equilibrium Compositions and Applica-
J. Therm. Spray Technol., 2007, 16, p 866-872 tions, NASA Reference Publication 1311, Lewis Research Cen-
5. D. Cheng, G. Trapaga, J.W. McKelliget, and E.J. Lavernia, ter, Cleveland, OH, 1994
Mathematical Modelling of High Velocity Oxygen Fuel Thermal 27. M. Li, D. Shi, and P.D. Christofides, Diamond Jet Hybrid HVOF
Spraying of Nanocrystalline Materials: An Overview, Model. Thermal Spray: Gas-Phase and Particle Behavior Modeling and
Simul. Mater. Sci. Eng., 2003, 11, p R1-R31 Feedback Control Design, Ind. Eng. Chem. Res., 2004, 43, p 3632-
6. J. Mostaghimi, S. Chandra, R. Ghafouri-Azar, and A. Dolatabadi, 3652
Modeling Thermal Spray Coating Processes: A Powerful Tool 28. W.D. Swank, J.R. Fincke, D.C. Haggard, and G. Irons, HVOF
in Design and Optimization, Surf. Coat. Technol., 2003, 163-164, Gas Flow Field Characteristics, Thermal Spray Industrial Appli-
p 1-11 cations, Proceedings of the Seventh National Thermal Spray
7. E. Dongmo, M. Wenzelburger, and R. Gadow, Analysis and Conference, Boston, MA, 1994, p 313-318
Optimization of the HVOF Process by Combined Experimental 29. H.H. Tawfik and F. Zimmerman, Mathematical Modeling of the
and Numerical Approaches, Surf. Coat. Technol., 2008, 202, Gas and Powder Flow in HVOF Systems, J. Therm. Spray
p 4470-4478 Technol., 1997, 6, p 345-352
8. E. Turunen, ‘‘Diagnostic Tools for HVOF Process Optimiza- 30. R.B. Bird, W.E. Stewart, and E.N. Lightfoot, Transport Phe-
tion,’’ Ph.D. dissertation, Helsinki University of Technology, nomena, John Wiley & Sons, New York, 1960
Espoo, Finland, 2005 31. H. Katanoda, Quasi-One-Dimensional Analysis of the Effects of
9. P.D. Christofides, N.H. El-Farra, M. Li, and P. Mhaskar, Model- Pipe Friction Cooling and Nozzle Geometry on Gas Particle
Based Control of Particulate Processes, Chem. Eng. Sci., 2008, 63, Flows in HVOF Thermal Spray Gun, Mater. Trans., 2006, 47,
p 1156-1172 p 2791-2797
10. X. Yang and S. Eidelman, Numerical Analysis of a High-Velocity 32. J.A. Roberson and C.T. Crowe, Engineering Fluid Dynamics, 6th
Oxygen-Fuel Thermal Spray System, J. Therm. Spray Technol., ed., John Wiley & Sons, New York, 1997
1996, 5, p 175-184 33. L.Y. Jiang and J.P. Sislian, ‘‘LDV Measurements of Mean
11. M. Li and P.D. Christofides, Multi-scale Modeling and Analysis Velocity Components and Turbulence Intensities in Supersonic
of HVOF Thermal Spray Process, Chem. Eng. Sci., 2005, 60, High-Temperature Exhaust Plumes,’’ AIAA Paper 93-3069, 1993
p 3649-3669 34. W.D. Swank, J.R. Fincke, D.C. Haggard, G. Irons, and R.
12. A. Dolatabadi, J. Mostaghimi, and V. Pershin, Effect of a Cylin- Bullock, HVOF Particle Flow Field Characteristics, Thermal
drical Shroud on Particle Conditions in High Velocity Oxy-fuel Spray Industrial Applications, Proceedings of the Seventh
Spray Process, J. Mater. Process. Technol., 2003, 137, p 214-224 National Thermal Spray Conference, Boston, MA, 1994, p 319-324

766—Volume 18(5-6) Mid-December 2009 Journal of Thermal Spray Technology

35. L.Y. Jiang and J.P. Sislian, Velocity and Density Measurements 60. J. Madejski, Solidification of Droplets on a Cold Surface, Int. J.

Peer Reviewed
in Supersonic High-Temperature Exhaust Plumes, AIAA J., 1998, Heat Mass Transfer, 1976, 19, p 1009-1013
36, p 1216-1222 61. L. Gil and M.H. Staia, Influence of HVOF Parameters on the
36. N.T. Lagen and J.M. Seiner, Evaluation of Water Cooled Super- Corrosion Resistance of NiWCrBSi Coatings, Thin Solid Films,
sonic Temperature and Pressure Probes for Application to 2000 F 2002, 420-421, p 446-454
Flows, NASA Technical Memorandum 102612, Langley 62. J.A. Hearley, J.A. Little, and A.J. Sturgeon, The Effect of Spray
Research Center, 1990 Parameters on the Properties of High Velocity Oxy-fuel NiAl
37. M. Li, D. Shi, and P.D. Christofides, Modeling and Control of Intermetallic Coatings, Surf. Coat. Technol., 2000, 123, p 210-218
HVOF Thermal Spray Processing of WC-Co Coatings, Powder 63. J.P. Sauer and P. Sahoo, HVOF Process Control Using Almen
Technol., 2005, 156, p 177-194 and Temperature Measurement, Thermal Spray 2001: New Sur-
38. C.T. Crowe, M. Sommerfeld, and Y. Tsuji, Multiphase Flows with faces for a New Millennium, K.A. Khor, C.C. Berndt, and E.F.
Droplets and Particles, CRC Press, Boca Raton, FL, 1997 Lugscheider, Eds., ASM International, (Materials Park, OH),
39. L. Pawlowski, The Science and Engineering of Thermal Spray 2001, p 791-796
Coatings, John Wiley & Sons, Chichester, England, 1995 64. O.C. Brandt, Mechanical Properties of HVOF Coatings,
40. S.A. Morsi and A.J. Alexander, An Investigation of Particle J. Therm. Spray Technol., 1995, 4, p 147-152
Trajectories in Two-Phase Flow Systems, J. Fluid Mech., 1972, 55, 65. T.C. Totemeier, R.N. Wright, and W.D. Swank, Microstructure
p 193-208 and Stresses in HVOF Sprayed Iron Aluminide Coatings,
41. G.H. Ganser, A Rational Approach to Drag Prediction of J. Therm. Spray Technol., 2002, 11, p 400-408
Spherical and Nonspherical Particles, Powder Technol., 1993, 77, 66. T. Lillo and R. Wright, Microstructure, Processing, Performance
p 143-152 Relationships for High Temperature Coatings, 22nd Annual
42. M. Ivosevic, R.A. Cairncross, and R. Knight, 3D Predictions of Conference on Fossil Energy Materials, Pittsburgh, PA, 2008
Thermally Sprayed Polymer Splats: Modeling Particle Accelera- 67. P. Bansal, P.H. Shipway, and S.B. Leen, Effect of Particle Impact
tion, Heating and Deformation on Impact with a Flat Substrate, on Residual Stress Development in HVOF Sprayed Coatings,
Int. J. Heat Mass Transfer, 2006, 49, p 3285-3297 J. Therm. Spray Technol., 2006, 15, p 570-575
43. S. Kamnis, S. Gu, T.J. Lu, and C. Chen, Computational Simula- 68. P. Bansal, P.H. Shipway, and S.B. Leen, Residual Stresses in
tion of Thermally Sprayed WC-Co Powder, Comput. Mater. Sci., High-Velocity Oxy-fuel Thermally Sprayed Coatings—Modelling
2008, 43, p 1172-1182 the Effect of Particle Velocity and Temperature During the
44. N. Zeoli and S. Gu, Computational Simulation of Metal Droplet Spraying Process, Acta Mater., 2007, 55, p 5089-5101
Break-up, Cooling and Solidification During Gas Atomisation, 69. C. Moreau, Towards a Better Control of Thermal Spray Pro-
Comput. Mater. Sci., 2008, 43, p 268-278 cesses, Thermal Spray: Meeting the Challenges of the 21st Century,
45. N. Zeoli, S. Gu, and S. Kamnis, Numerical Modelling of Metal C. Coddet, Ed., ASM International (Materials Park, OH), 1998,
Droplet Cooling and Solidification, Int. J. Heat Mass Transfer, p 1681-1693
2008, 51, p 4121-4131 70. M. Friis and C. Persson, Control of Thermal Spray Process by
46. M. Li and P.D. Christofides, Modeling and Analysis of HVOF Means of Process Maps and Process Windows, J. Therm. Spray
Thermal Spray Process Accounting for Powder Size Distribution, Technol., 2003, 12, p 44-52
Chem. Eng. Sci., 2003, 58, p 849-857 71. E. Turunen, T. Varis, S.-P. Hannula, A. Vaidya, A. Kulkarni,
47. M. Li and P.D. Christofides, Feedback Control of HVOF Ther- J. Gutleber, S. Sampath, and H. Herman, On the Role of Particle
mal Spray Process Accounting for Powder Size Distribution, State and Deposition Procedure on Mechanical, Tribological and
J. Therm. Spray Technol., 2004, 13, p 108-120 Dielectric Response of High Velocity Oxy-fuel Sprayed Alumina
48. L. Zhao, M. Maurer, F. Fischer, and E. Lugscheider, Study of Coatings, Mater. Sci. Eng. A, 2006, 415, p 1-11
HVOF Spraying of WC-CoCr Using On-Line Particle Monitor- 72. J.R. Fincke, W.D. Swank, R.L. Bewley, D.C. Haggard, M. Gevelber,
ing, Surf. Coat. Technol., 2004, 185, p 160-165 and D. Wroblewski, Diagnostics and Control in the Thermal Spray
49. T.C. Hanson, C.M. Hackett, and G.S. Settles, Independent Process, Surf. Coat. Technol., 2001, 146-147, p 537-543
Control of HVOF Particle Velocity and Temperature, J. Therm. 73. A.F. Kanta, G. Montavon, M.P. Planche, and C. Coddet, Plasma
Spray Technol., 2002, 11, p 75-85 Spray Process On-Line Control by Artificial Intelligence Meth-
50. T.C. Hanson and G.S. Settles, Particle Temperature and Velocity odology, Adv. Eng. Mater., 2007, 9, p 105-113
Effects on the Porosity and Oxidation of an HVOF Corrosion- 74. A.F. Kanta, G. Montavon, M. Vardelle, M.P. Planche, C.C.
Control Coating, J. Therm. Spray Technol., 2003, 12, p 403-415 Berndt, and C. Coddet, Artificial Neural Networks vs. Fuzzy
51. D.E. Rosner, Transport Processes in Chemically Reacting Flow Logic: Simple Tools to Predict and Control Complex Pro-
Systems, Dover Publications, New York, 2000 cesses—Application to Plasma Spray Processes, J. Therm. Spray
52. N. Zeoli, S. Gu, and S. Kamnis, Numerical Simulation of Technol., 2008, 17, p 365-376
In-Flight Particle Oxidation During Thermal Spraying, Comput. 75. DPV-2000 Reference Manual, Tecnar Automation
Chem. Eng., 2008, 32, p 1661-1668 76. J.R. Fincke, W.D. Swank, and C.L. Jeffrey, Simultaneous Mea-
53. E.L. Crow and K. Shimizu, Lognormal Distributions: Theory and surement of Particle Size, Velocity and Temperature in Thermal
Applications, Marcel Dekker, New York, 1988 Plasmas, IEEE Trans. Plasma Sci., 1990, 18, p 948-957
54. O. Knotek and R. Elsing, Monte Carlo Simulation of the 77. J.R. Fincke and R.A. Neiser, Advanced Diagnostics and Mod-
Lamellar Structure of Thermally Sprayed Coatings, Surf. Coat. eling of Spray Processes, MRS Bull., July 2000, p 26-31
Technol., 1987, 32, p 261-271 78. J.R. Fincke, D.C. Haggard, and W.D. Swank, Particle Temper-
55. S. Kamnis and S. Gu, Numerical Modelling of Droplet ature Measurement in the Thermal Spray Process, J. Therm.
Impingement, J. Phys. D: Appl. Phys., 2005, 38, p 3664-3673 Spray Technol., 2001, 10, p 255-266
56. P. Fauchais, M. Fukumoto, A. Vardelle, and M. Vardelle, 79. J.R. Fincke, W.D. Swank, R.L. Bewley, and D.C. Haggard,
Knowledge Concerning Splat Formation: An Invited Review, Control of Particle Temperature, Velocity, and Trajectory in
J. Therm. Spray Technol., 2004, 13, p 337-360 Thermal Spray Process, Thermal Spray 2003: Advancing the Sci-
57. D. Zhang, S.J. Harris, and D.G. McCartney, Microstructure ence and Applying Technology, C. Moreau and B. Marple, Eds.,
Formation and Corrosion Behaviour in HVOF-Sprayed Inconel ASM International, (Materials Park, OH), 2003, p 1093-1099
625 Coatings, Mater. Sci. Eng. A, 2003, 344, p 45-56 80. E. Hamalainen, J. Vattulainen, T. Alahautala, R. Hernberg,
58. V.V. Sobolev, J.M. Guilemany, J.C. Garmier, and J.A. Calero, P. Vuoristo, and T. Mantyla, Imaging Diagnostics in Thermal
Modelling of Particle Movement and Thermal Behavior During Spraying. ‘‘Spraywatch’’ System, Thermal Spray: Surface Engi-
High Velocity Oxy-Fuel Spraying, Surf. Coat. Technol., 1994, 63, neering Via Applied Research, Proceedings of the International
p 181-187 Thermal Spray Conference (Montreal, QC, Canada), 2000, p 79-83
59. D. Shi, M. Li, and P.D. Christofides, Diamond Jet Hybrid HVOF 81. E. Lugscheider, C. Herbst, and L. Zhao, Parameter Studies on
Thermal Spray: Rule-Based Modeling of Coating Microstructure, High-Velocity Oxy-fuel Spraying of MCrAlY Coatings, Surf.
Ind. Eng. Chem. Res., 2004, 43, p 3653-3665 Coat. Technol., 1998, 108-109, p 16-23

Journal of Thermal Spray Technology Volume 18(5-6) Mid-December 2009—767

 82. H.L. de Villiers Lovelock, P.W. Richter, J.M. Benson, and P.M. Thermally Sprayed Stainless Steel Coatings in 3.4% NaCl Solu-
Peer Reviewed

Young, Parameter Study of HP/HVOF Deposited WC-Co tion, Corros. Sci., 2005, 47, p 605-620
Coatings, J. Therm. Spray Technol., 1998, 7, p 97-107 91. E. Turunen, T. Varis, T.E. Gustafsson, J. Keskinen, T. Falt, and
83. V. Gourlaouen, E. Verna, and P. Beaubien, Influence of Flame S.-P. Hannula, Parameter Optimization of HVOF Sprayed
Parameters on Stainless Steel Coatings Properties, Thermal Nanostructured Alumina and Alumina-Nickel Composite Coat-
Spray: Surface Engineering via Applied Research, Proceedings of ings, Surf. Coat. Technol., 2006, 200, p 4987-4994
the International Thermal Spray Conference (Montreal, QC, 92. W. Tillmann, E. Vogli, I. Baumann, G. Matthaeus, and T.
Canada), 2000, p 487-493 Ostrowski, Influence of the HVOF Gas Composition on the
84. W.C. Lih, S.H. Yang, C.Y. Su, S.C. Huang, I.C. Hsu, and M.S. Thermal Spraying of WC-Co Submicron Powders (
8 + 1 lm) to
Leu, Effects of Process Parameters on Molten Particle Speed and Produce Superfine Structured Cermet Coatings, J. Therm. Spray
Surface Temperature and the Properties of HVOF CrC/NiCr Technol., 2008, 17, p 924-932
Coatings, Surf. Coat. Technol., 2000, 133-134, p 54-60 93. M. Li, D. Shi, and P.D. Christofides, Model-Based Estimation
85. S. Wirojanupatump, P.H. Shipway, and D.G. McCartney, The and Control of Particle Velocity and Melting in HVOF Thermal
Influence of HVOF Powder Feedstock Characteristics on the Spray, Chem. Eng. Sci., 2004, 59, p 5647-5656
Abrasive Wear Behaviour of CrxCy-NiCr Coatings, Wear, 2001, 94. T. Chiu and P.D. Christofides, Nonlinear Control of Particulate
249, p 829-837 Processes, AIChE J., 1999, 45, p 1279-1297
86. J.G. Legoux, B. Arsenault, L. Leblanc, V. Bouyer, and C. 95. T. Chiu and P.D. Christofides, Robust Control of Particulate
Moreau, Evaluation of Four High Velocity Thermal Spray Guns Processes Using Uncertain Population Balances, AIChE J., 2000,
Using WC-10%Co-4%Cr Cermets, J. Therm. Spray Technol., 46, p 266-280
2002, 11, p 86-94 96. N.H. El-Farra, T. Chiu, and P.D. Christofides, Analysis and
87. M.P. Planche, B. Normand, H. Liao, G. Rannou, and C. Coddet, Control of Particulate Processes with Input Constraints, AIChE
Influence of HVOF Spraying Parameters on In-Flight Charac- J., 2001, 47, p 1849-1865
teristics of Inconel 718 Particles and Correlation with the Elec- 97. A. Kalani and P.D. Christofides, Simulation, Estimation and
trochemical Behaviour of the Coating, Surf. Coat. Technol., 2002, Control of Size Distribution in Aerosol Processes with Simulta-
157, p 247-256 neous Reaction, Nucleation, Condensation and Coagulation,
88. Y. Qiao, T.E. Fischer, and A. Dent, The Effects of Fuel Chem- Comput. Chem. Eng., 2002, 26, p 1153-1169
istry and Feedstock Powder Structure on the Mechanical and 98. P.D. Christofides, Model-Based Control of Particulate Processes,
Tribological Properties of HVOF Thermal-Sprayed WC-Co Kluwer Academic Publishers, Particle Technology Series, Neth-
Coatings with Very Fine Structures, Surf. Coat. Technol., 2003, erlands, 2002
172, p 24-41 99. D. Shi, N.H. El-Farra, M. Li, P. Mhaskar, and P.D. Christofides,
89. K.A. Khor, H. Li, and P. Cheang, Significance of Melt-Fraction in Predictive Control of Particle Size Distribution in Particulate
HVOF Sprayed Hydroxyapatite Particles, Splats and Coatings, Processes, Chem. Eng. Sci., 2006, 61, p 268-281
Biomaterials, 2004, 25, p 1177-1186 100. D. Shi, P. Mhaskar, N.H. El-Farra, and P.D. Christofides, Pre-
90. P.H. Suegama, C.S. Fugivara, A.V. Benedetti, J. Fernández, dictive Control of Crystal Size Distribution in Protein Crystal-
J. Delgado, and J.M. Guilemany, Electrochemical Behavior of lization, Nanotechnology, 2005, 16, p S562-S574

768—Volume 18(5-6) Mid-December 2009 Journal of Thermal Spray Technology