Materials Science and Engineering A 428 (2006) 4753

Numerical modeling of the thermal behavior during the LENS process

Riqing Ye a , John E. Smugeresky b , Baolong Zheng a , Yizhang Zhou a , Enrique J. Lavernia a,
a

Department of Chemical Engineering and Materials Science, University of California, Davis, CA 95616, United States
b Sandia National Laboratories, Laser Engineered Net-Shaping, Livermore, CA 94551-0969, United States
Received 23 October 2005; accepted 14 April 2006

Abstract
Laser Engineered Net-Shaping (LENS ) is an emerging manufacturing technique that ensures significant reduction of process time between
initial design and final components. The fabrication of fully dense parts with appropriate properties using the LENS process requires an in-depth
understanding of the entire thermal behavior of the process. In this paper, the thermal behavior during LENS was studied, both numerically
and experimentally. Temperature distribution and gradient in the fabricated part were obtained by finite element method (FEM) simulation. The
numerical results are in good agreement with the experimental observations. The numerical method may be used to optimize process parameters
and predict the thermal response of LENS fabricated components.
2006 Elsevier B.V. All rights reserved.
Keywords: Laser Engineered Net-Shaping; Thermal behavior; Finite element analysis

1. Introduction
Recent advances in software and hardware have prompted the
development of manufacturing methods for fabricating component parts directly from a computer aided design (CAD) model.
Among available manufacturing methods, Laser Engineered
Net-Shaping (LENS ), an additive manufacturing process, has
engendered increasing interest as both complete part fabrication as well as a cost saving repairing technology. The LENS
process can be effectively used to fabricate complex threedimensional solid metallic parts directly from a CAD model.
Moreover, high strength and ductility can be achieved through
LENS processing [13].
The first practical LENS facility, which incorporates the
features from stereo-lithography and laser welding, emerged at
Sandia National Laboratories. The concept of LENS can be
traced back to the 1980s [4,5]. It was not until mid 1990s, after
significant advances in graphic software (CAD) and completely
controlled translation stages were released, that the current technology allowed. LENS is a process which builds solid metallic
parts by injecting powder into a laser-generated molten zone.
The fabrication process has several steps: First, a solid model is
sliced electronically into a sequence of layers of a given thick-

Corresponding author. Tel.: +1 530 752 0554; fax: +1 530 752 8058.
E-mail address: [email protected] (E.J. Lavernia).

0921-5093/$ see front matter 2006 Elsevier B.V. All rights reserved.
doi:10.1016/j.msea.2006.04.079

ness. Second, to build a metallic part, a solid substrate is used as

a base and the laser beam is focused on the substrate to create a
molten pool into which powder is simultaneously fed. Third, the
substrate is moved beneath the laser beam, to trace out a pattern
defined by the electronic slice of the model, melting material is
added to the surface as a narrow strip of added material. Starting
from the bottom of the part, one layer is produced at a time. After
formation of a layer, the powder feeding nozzle and laser beam
assembly is moved in the z-direction. Accordingly, the part is
then built line by line and layer by layer.
There has been an increasing interest in investigating the
LENS process. Many experimental studies have been performed in an effort to provide fundamental insight into the
influence of process parameters on product properties [1,69].
In particular, a number of these studies have attempted to clarify
thermal measurement issues that emerge from the understanding
of the interaction between the laser beam and the metal powder
component. These thermal measurements can be classified into
two categories: contact temperature sensor measurements and
non-contact measurements. For the contact temperature sensor
measurements, thermocouples were used to measure temperatures during the deposition. The transient temperature of a
specified spot during fabrication was plotted [7]. In the noncontact measurements side, a high-speed imaging technique
was used to study the deposition process and observe the focal
molten zone/powder interaction at Sandia National Laboratories [68]. Wei et al. utilized a ThermavizTM two-wavelength

48

R. Ye et al. / Materials Science and Engineering A 428 (2006) 4753

imaging pyrometer system to obtain the thermal images of the

LENS processing zone [9]. The molten pool length, width, area
and volume scales as well as the temperatures surrounding the
molten zone were obtained. The temperature gradients and cooling rates in the vicinity of molten pool were determined. These
thermal images help to understand laser/powder interaction and
morphology of molten pool. It should be pointed out that all
the above work is mainly focused on the characters inside the
molten pool and its immediate area. Temperature distributions
outside this local area are difficult to capture experimentally
and calculation to simulate them would help in understanding
thermal histories. On the computational side, some preliminary
finite element modeling of the LENS process has also been
published. Hofmeister et al. treated the problem as the multidimensional moving boundary problem [6]. They simulated
LENS processing by finite element method (FEM) with element birthing. This method has a clear physical meaning and
provides vivid thermal images of the LENS process. The preliminary temperature field in the fabricated wall was obtained
[6].
Despite impressive progress, there are many unanswered
questions, especially problems regarding the entire thermal
behavior of the LENS process. Since it is very difficult to
measure temperature during the LENS process, only limited
information is available on the characteristics of the entire temperature field of the processed part. Complete understanding
of the LENS thermal behavior is critical for fabricating fully
dense parts with appropriate properties. Numerical simulation
methods have the potential to provide detailed information of
the thermal behavior. The element birthing method can solve
the moving boundary problem. At each incremental step, the
boundary conditions are changed and the computed data should
be stored. It needs a very powerful processor, with large RAM
and hard-drive space. It is difficult to perform the FEM code
when the mesh is refined. Another method used to solve the
moving boundary problem is the fixed boundary method developed by Takeshita and Matsunawa [10] and Costa et al. [11].
By an approximate immobilization transformation and assumptions, the moving boundary problem can be reduced into the
fixed boundary problem. This method can refine the mesh of the
geometrical structures and execute large finite element analyses.
Their papers describe a numerical fixed boundary model that
can be used to investigate the entire thermal behavior during the
entire LENS process. The objective of this paper is to provide
physical insight into the experimentally observed phenomena in
the LENS process.
The finite element method is used to study the entire heat
transfer response during the LENS process. Moreover, the
LENS experiments were done to verify the numerical simulations. In the simulation process, a thin wall part deposited on
the substrate is discretized by using cubic solid elements. The
problem is solved as an initial boundary value problem. This
transient problem was integrated through time. At each incremental step, the temperatures of the nodes where the laser beam
is focused are set as the melting point temperatures, and the temperature distribution in this transient process is then calculated
via appropriate heat transfer formulae. The results of the previ-

ous calculation step are set as the initial conditions for the next
calculation step. The new cycle starts after the updated temperature field is determined.
2. Fundamental equations and boundary conditions
2.1. Fundamental governing equations
For an arbitrary system, the heat equilibrium equation in the
body is given as [12]:
U =

q + r
x

(1)

where is the material mass density, U is the internal energy

per unit mass assumed to be a function of temperature only (U
means rate of change of U per time), q is the heat flux vector
per unit of current area crossing surface from environment to
body and r is the heat flux per unit volume generated in the
body. The above equation is rewritten by multiplying an arbitrary variational temperature field, T, and integrating over the
volume:

U T dV =
T
q dV +
T r dV
(2)
x
V
V
V
where the middle term can be rewritten as:






T
T
q dV =
q T q
dV
x
x
x
V
V

(3)

Using the divergence theorem, the volume integral is converted to surface integral:

(4)
q T dV = q T dS
x
V 
S
As a result, the thermal energy balance equation yields the
form:








T

TU dV
T r dV (5)
q dV = T q dS +
x
V
V 
S
V
in which, q = q n , where n is the unit outward normal to the
surface.
It is observed in experiments that the heat flux is proportional
to the temperature gradient. Fouriers law of heat conduction has
the form:
q = k

T
x

(6)

where k is the material thermal conductivity matrix, usually is

a temperature-dependent anisotropic property. Then, rewriting
Eq. (3), it becomes:








T  T
T U dV +
T r dV
k
dV = T q dS +
x
x
V
V 
S
V
(7)
In the finite element analysis process, Tn is the nodal temperature degrees of freedom, the geometric model is discretized

R. Ye et al. / Materials Science and Engineering A 428 (2006) 4753

49

to finite elements and the temperatures in the elements can be

interpolated from the nodal temperatures,
T = NT n

(8)

where N is the interpolation function, the variational temperature

field, T is given as:
T = NT n

(9)

It is noted that T is independent. Introducing the interpolation

into the heat balance Eq. (7) leads to:




N n  N m
N n U dV +
k
dV T m

x

x
V
V




n
=
N q dS +
N n r dV
(10)
S

in which Tm is the temperature at node m. For the heat transfer

process, we have:
U = cT

(11)

where c is the specific heat of the material. Substituting the

time integration operator into Eq. (10) leads to the following
fundamental equation:
[C]{T } + [K]{T } = {Q}
where
[C] =

V

cN n N m dV

(12)

(13)

N n
N m
k
dV
x
x
V 




{Q} = N n q dS +
N n r dV

[K] =

(14)
(15)

where {T} is the vector of nodal temperature degree of freedom

and Nn and Nm are the interpolation functions for the node n and
m, respectively. This set of equations represents the fundamental equations for heat transfer process at each incremental step.
In this paper, the FEM software, ABAQUS [13] is employed
to calculate the temperature field. For transient analysis in each
step, the equations need to be integrated over time. The backward difference algorithm is used as time integration operator
in ABAQUS program. A detailed description can be found in
manual of ABAQUS [13].
2.2. Boundary conditions and assumptions
In the simulation, the fabrication of a thin wall (0.25 mm) is
studied. Fig. 1 is the schematic diagram of the fabricated part
and substrate. A Cartesian coordinates system is used in the
simulation process.
It is found in experiments that complete melting of the powder occurs in the laser focal zone [9]. It is also observed that
the intergranular melting only occurs into underlying layer at
a fraction of grain size, which means the heat-affected zone is
relatively small [1,6,8]. Moreover, the powder particles do not
become molten until they are injected into the molten pool in

Fig. 1. Geometric model used in FEM (the thickness of thin-walled wall is

0.25 mm, the size of substrate is 100 mm 10 mm 100 mm).

the deposition region. For the low laser power conditions considered (laser power was 240 W), the morphology and size of
molten pool are stable. It is reasonable to assume a stationary
state of heat dissipation [14]. So, the geometric model used in
the simulation has the same outline as the finished part. The part
is discretized into equal size elements. The size of element is set
via the size of molten pool [15].
In the LENS process, the thin wall sample is deposited on
a substrate. In general, the material of the substrate is the same
as that of the part to be formed. The transfer of heat by conduction from molten zone to fabricated wall and substrate governs
the fabrication process [10,1517]. Ignoring the convective heat
transfer in the molten zone, it is justified that the temperatures
of the nodes where the laser beam is focused can be set as the
melting point temperatures during the simulation. The part to be
formed can be set as adiabatic [15]. The heat transfer is assumed
to occur only by conduction through the formed sample and into
the substrate.
During the fabrication process, the temperature in the substrate should vary with time. It is found that the substrate reaches
steady state when the build part is several millimeters tall [6].
In the calculation, the thickness of the substrate is 10 mm.
Therefore, in the present work, the temperature on the bottom
of substrate was set as a constant (323 K).
When the first molten pool is placed on the substrate with an
initial temperature at all the nodes, the element is held on the set
temperature in the incremental time step. We assume the initial
temperature in the part to be formed is zero. The temperature
field in formed wall and substrate is calculated via fundamental
governing equations. The molten pool moves to the next element
in the x-direction. The results of the previous calculation step
are set as the boundary conditions for the current calculation
step. This process was repeated until the molten pool moves
to another end of the line. Then molten pool is returned to the
beginning location, but on the next upper layer. The new cycle
starts until the wall is complete. The preceding assumptions are
summarized as:
(1) the thermal conductivity, specific heat and other materials
parameters are temperature-independent;
(2) only heat conduction in wall and substrate is considered;
(3) the quasi-steady state is established, i.e. the molten pool
does not change size with time;

50

R. Ye et al. / Materials Science and Engineering A 428 (2006) 4753

Table 1
Processing parameters and physical properties of AISI 316 stainless steel
Sample/substrate traveling velocity (mm/s)
Fabricated wall length (mm)
Fabricated wall width (mm)
Fabricated wall height (mm)
Substrate length/width (mm)
Substrate height (mm)
Specific heat (J/(kg K))
Thermal conductivity (W/(m K))
Density (kg/m3 )

5
11
0.25
6.5
100
10
500
16.3
8000

(4) the temperatures of the element nodes where the laser beam
is focused are set as the melting point temperatures;
(5) the part to be formed is adiabatic.
3. Results and discussions
In the presented study, AISI 316 stainless steel powder is
used to form the thin-walled sample. The substrate is the same
material as the wall sample. The thermo-mechanical property
data used in the finite element analysis were obtained by experimental observations and reference book [18,19], which is shown
in Table 1. It should be noted that thermo-physical parameters
vary with ambient temperatures. In this fabrication process, the
heat transfer properties vary briefly in the area surrounding the
molten zone, and attain a steady value far from molten pool.
Therefore, it is reasonable to assume the thermo-physical properties are constant in numerical modeling, this assumption was
justified in refs. [11,15].
The completed fabricated thin wall part has the geometrical size of 0.25 mm thick 11 mm long 6.5 mm high, which
is shown in Fig. 1. An eight-node solid cubic element is
chosen to discretize the geometric model. The thin wall is

Fig. 2. 2D view mesh of wall.

placed on the center of the substrate. The substrate is 100 mm

long, 100 mm wide and 10 mm thick. The velocity of sample/substrate is set as 5 mm/s. In the experiment, each layer
height is 0.13 mm. We assume that the liquid/solid interface is
around 0.20.3 mm from center of molten pool for the low laser
power (240 W) condition, so in the wall, the size of element
is chosen as 0.5 mm 0.25 mm 0.13 mm. In the substrate,
adaptive refined meshing is used to discretize the substrate. In
the sample/substrate contact interface and its surrounding area,
refined mesh is used to eliminate the oscillation. A 2D schematic
representative of the thin wall and substrate surrounding the wall
with element mesh is shown in Fig. 2.
The side view image of the temperature distribution in the
thin wall when the wall is formed to 1.3 mm tall is shown in
Fig. 3. The value of distance from molten pool plus 0.25 mm
(one-half of the size of element) is equal to the distance from
center of molten pool. It is found that the temperatures drop
sharply around the area of the molten pool. As the distance from
molten pool increases, the temperatures decrease gradually. It
predicts that the heat is quickly dissipated from molten pool and
the temperature drops to about 256 C (529 K) when the distance
is larger than 7 mm from the molten zone. The thin wall remains
at a high temperature during the fabrication process. As a result,
rapid cooling would lead to refined grains giving high hardness
and low porosity microstructure. This result coincides with the
experimental observations [7].
Experiments have been done to verify the numerical simulations. A ThermavizTM two-wavelength imaging pyrometer
system (Stratonics, Laguna Hills, CA) was employed to obtain
thermal images of the LENS processed part. The system can
provide information about solidification during processing. Its
disadvantage is that the accuracy of temperature measurement
depends on the resolution of CCD video camera. The experimental setup is shown in Fig. 4. This system can record a temperature
range from 1500 to 2500 K. The advantage of this system is that
it can provide a two-dimensional and contact-less image of the
temperature field. It can provide us with the understanding of the
size of molten pool and its temperature distribution plus thermal
history of part. More details about the experiments are described
elsewhere [9]. The experimental data for AISI 316 stainless steel
for laser power of 240 W are shown in Fig. 5. It is found that
the molten pool is superheated. It is shown that the temperature
inside the molten pool decreases sharply from the center to the
liquid/solid interface. The temperature also cools down rapidly
outside of the molten pool. From experimental observations, the
molten pool is very small, appropriately 0.5 mm long. It indicates that our assumption for size of molten pool is appropriate
for the laser beam localized at such laser power. From Fig. 5,
it is reasonable to set the laser focal zone nodes temperature

Fig. 3. Image of temperature distribution near to the molten pool (when the wall is formed to 11 mm long (x-direction) and 1.3 mm tall (z-direction)).

R. Ye et al. / Materials Science and Engineering A 428 (2006) 4753

Fig. 4. Schematic configuration used in LENS experiments.

as the melting point temperatures, considering the temperature

continuity and other heat dissipation. However, this temperature
measurement system cannot provide the information when the
distance from molten pool is large. It is because the camera can
only record temperatures greater than 1500 K. The combination
of experimental measurements and numerical simulations can
provide more detailed information of the LENS process.
In the present study, two factors significantly affect the magnitude and distribution of temperature field in the LENS process: (a) morphology of molten pool and (b) heat transfer modes,
such as conduction, radiation, etc. It is found in experiments that
the molten pool is not spherical, but approximates an elliptical
shape [6,9]. It is appropriate to use cubic element for the simulation. The heat conduction, radiation and convection play a role
in the fabrication process. However, it is found that the heat conduction dominates in the deposition processing [9,17,20,21]. As
a result, the assumptions and results of the present study are in
general agreement with the experimental observations. It shows

Fig. 5. Temperature distributions around the molten zone indicate that the
molten pool size is around 0.5 mm (0.25 mm 2).

51

Fig. 6. Temperature distributions of the layer beneath the molten zone (where
the wall is formed to 11 mm long (x-direction) and 1.3 mm tall (z-direction)).

the potential of numerical modeling for investigating the thermal

behavior in the area beyond molten pool.
Fig. 6 shows the temperature distribution of the layer beneath
the molten zone. It is seen from Fig. 6 that temperature values drop rapidly in the molten pool/solid contact area, then,
temperature drops to a nominal value of 529 K (256 C) before
increasing. The reason is that the heat transport is a transient process and the energy is transferred by conduction to surrounding
area. For this case, the laser focal zone just left the previous
layer of the wall. When the laser beam moves to the other end of
the wall referred to x = 0, the absorbed energy remains somewhat
concentrated in the preceding focused zone, heat received in last
step results in temperatures surrounding this area still cooling,
before again being heated as the laser beam moves from the left
to right forming the next layer.
As the wall height increases, the nominal wall temperature
increases. Fig. 7 shows the temperature distribution when the
molten pool is in the center region of the wall at a height
of 2.6 mm. As the thin wall part is built up, the new layer is
deposited on the increasingly warmer layers and as the molten

Fig. 7. Temperature distributions of the layer left to the molten zone (molten
zone is in the center of wall, x = 3 mm, z = 2.6 mm).

52

R. Ye et al. / Materials Science and Engineering A 428 (2006) 4753

Fig. 8. Temperature distributions around molten zone: (A) molten zone is at

the left side of wall and (B) molten zone is in the center of wall with steeper
temperature gradient; illustrating heat flow is dominated by conduction rather
than by convection.

pool moves farther away from the heat sink of the substrate, the
temperature drops gently. The temperature is now at a nominal
value of about 900 K, as appeared to 529 K in Fig. 4, at a height
of 1.3 mm.
The temperature distribution in the z-direction was found to
depend on location relative to the edge in the thin wall. Fig. 8
shows the temperature distribution in z-direction for molten pool
at different places. The temperature at the edge of the wall (the
start of a deposit layer) is slightly higher than the temperature
at the center of wall (one-half way across the deposit layer) in
z-direction. The heat transfer is a time- and distance-dependent
process controlled by conduction. For this case, the heat around
the molten pool in the center of the wall is transferred to the surrounding regions in all directions, so the temperatures decrease
very quickly. As to the heat transfer of molten pool at the left
side of wall, the heat flux only occurs in z-direction and increasing x-direction. As expected, the temperature at the edge of the
thin wall is slightly higher than that at the center of wall in zdirection. Moreover, the temperature in z-direction is lower than
that in x-direction at the same distance, as seen from Figs. 68.
The reason is the substrate beneath the wall is the sink of heat
flux. The temperature in z-direction drops faster.
In this study, a two color high-speed thermal imaging technique has been used to examine LENS processing. However,
the temperature gradient is underestimated when the digital camera does not have enough pixels. Hofmeister et al. was first to
use high-speed thermal imaging for AISI 316 stainless steel.
Their gradients in the previous layer beneath the laser are about
400 K/mm [6]. As mentioned above, this value underestimates
the temperature gradients ahead of molten pool. Wei et al.
obtained the temperature field of the molten pool and its surrounding area for the case of AISI 316 stainless steel in the
LENS process by using the ThermavizTM system. It is found
that the gradients in the front of molten pool are about 750 K/mm
[9].
The numerically calculated temperature gradient in the xdirection in the area below the laser focal zone is shown in
Fig. 9 indicating that the highest temperature gradient around

Fig. 9. Numerically calculated temperature gradient for the layer beneath the
molten zone (where the wall is formed to 11 mm long (x-direction) and 1.3 mm
tall (z-direction)).

the molten pool is 588 K/mm. The numerical results are in the
middle of experimental measurements of 400 and 750 K/mm
[6,9]. It shows that the gradient values from numerical study
coincide with the experimental results very well. It is also seen
from Fig. 9 that the temperature gradient in the vicinity of the
molten pool is substantially higher than that far from molten
pool where the gradient values decrease with increasing the distance from molten pool, which indicates that the heat-affected
region only exists in the vicinity of the local laser focal zone.
4. Conclusions
For AISI 316 stainless steel thin wall fabricated in the LENS
process, numerical simulation was performed to study the entire
thermal behavior in process. Temperature distribution and gradient in the fabricated part were obtained from the results of
finite element method simulation. The conclusions are summarized as: (1) the temperature gradient near the molten pool is of
the order of 5 102 K/mm; (2) it is found that the temperatures
change sharply around the area of the molten pool and decrease
gradually when distance from molten pool increases. The numerical results coincide with the experimental data; (3) numerical
simulations have the potential to give the information of entire
thermal behavior, especially to give the temperature information
of the area out of the field of viewing for a pyrometer.
Acknowledgements
The work is supported by the National Science Foundation
(NSF DMR00-76498 and NSF DMI-0423695) and in part by
the U.S. Department of Energy (DE-AC04-94AL85000).
References
[1] C.L. Atwood, M.L. Griffith, L.D. Harwell, D.L. Greene, D.E. Reckaway, M.T. Ensz, D.M. Keicher, M.D. Schlienger, J.A. Romero, M.S.

R. Ye et al. / Materials Science and Engineering A 428 (2006) 4753

[2]
[3]
[4]
[5]
[6]
[7]

[8]

[9]

[10]

Oliver, F.P. Jeantette, Laser Spray Fabrication for Net-Shape Rapid Product Realization LDRD, Sandia Report, SAND99-0739, Sandia National
Laboratories, 1999, pp. 331.
G.K. Lewis, E. Schlienger, Mater. Des. 21 (4) (2000) 417423.
J.W. Sears, Int. J. Powder Metall. 37 (2001) 2930.
C.O. Brown, E.M. Breinan, B.H. Kear, Method for fabricating articles
by sequential layer deposition, US Patent (1982).
J. Mazumder, W.M. Steen, J. Appl. Phys. 51 (1980) 941947.
W.H. Hofmeister, W. Melissa, J. Smugeresky, J. Philliber, M. Griffith,
M. Ensz, JOM-e 51 (7) (1999).
M.L. Griffth, M.E. Schlienger, L.D. Harwell, M.S. Oliver, M.D. Baldwin, M.T. Ensz, M. Essien, J. Brooks, C.V. Robino, J.E. Smugeresky,
W.H. Hofmeister, M.J. Wert, D.V. Nelson, Mater. Des. 20 (23) (1999)
107113.
W.H. Hofmeister, D.O. MacCallum, G.A. Knorovsky, Ninth International
Conference on Computer Technology in Welding, Detroit, MI, 1999, pp.
187196.
W. Wei, Y. Zhou, R.Q. Ye, D. Lee, J.E. Craig, J.E. Smugeresky, E.J. Lavernia, International Conference on Metal Powder Deposition for Rapid
Manufacturing, San Antonio, TX, 2002.
K. Takeshita, A. Matsunawa, Metall. Mater. Trans. 32B (2001) 949959.

53

[11] L. Costa, R. Vilar, T. Reti, R. Colaco, A.M. Deus, I. Felde, Mater. Sci.
Forum 473474 (2005) 315320.
[12] F.P. Incropera, D.P. DeWitt, Fundamentals of Heat and Mass Transfer,
second ed., John Wiley & Sons, New York, 1985.
[13] Heat transfer and thermal-stress analysis with ABAQUS, Hibbitt, Karlsson, and Sorensen, Inc., Pawtucket, Rhode Island, 2001.
[14] G. Bruggemann, A. Mahrle, Th. Benziger, NDT&E Int. 33 (2000)
453463.
[15] B. Basu, J. Srinivasan, Int. J. Heat Mass Transfer 31 (1988) 23312338.
[16] B. Kim, T. Kim, J. Kim, K. Kim, H. Lee, Metall. Mater. Trans. 33A
(2002) 14491459.
[17] R. Dehoff, P.M. Sarosi, P.C. Collins, H.M. Fraser, M.J. Mills, MS&T
2005 Conference, Pittsburgh, PA, 2005.
[18] P.D. Harvey, Engineering Properties of Steel, American Society for Metals, Metals Park, OH, 1982.
[19] W. Wei, Y. Zhou, J. He, J.E. Smugeresky, E.J. Lavernia, International
Conference on Metal Powder Deposition for Rapid Manufacturing, San
Antonio, TX, 2002.
[20] W. Hofmeister, M. Griffith, M. Ensz, J. Smugeresky, JOM (2001) 3034.
[21] J.C. Rozzi, F.E. Pfefferkorn, F.P. Incroopera, Y.C. Shin, J. Heat Transfer
120 (1998) 899905.