Professional Documents
Culture Documents
com
ScienceDirect
Additive Manufacturing 5 (2015) 3139
Abstract
Additive manufacturing (AM) has several possible advantages over traditional manufacturing including increased design freedom, reduced
material usage, and shorter lead-times. A noteworthy capability of AM is the ability to monitor the process during material deposition and interrupt
the process during fabrication if necessary. Recently, such monitoring, feedback, and control have been made possible by implementing in situ
infrared (IR) thermography in powder bed fusion AM technologies. The purpose of the current research was to investigate the acquisition of
absolute surface temperatures using in situ IR imaging of the melted or solid surfaces layer-by-layer during fabrication within an electron beam
melting (EBM) system. The thermal camera was synchronized with the systems signal voltages of three synchronized events (pre-heating, melting,
and raking) to automatically capture images. To acquire absolute temperature values from the IR images, a calibration procedure was established
to determine the solid materials emissivity and reflected temperature or mean radiant temperature of the build chamber, which are necessary input
parameters for the IR camera. A blackbody radiator was fabricated via EBM and was used as a tool to determine the emissivity of Ti6Al4V
(determined to be 0.26 in the temperature range of the current study). Furthermore, a mathematical model was developed to determine the view
factors associated with the systems interior (e.g. heat shielding) that were used in calculating the mean radiant temperature of the manufacturing
environment (342 C). Experimental validation of the model was performed using a thermocouple embedded during fabrication that showed
a 3.77% difference in temperature. A temperature difference of 366 C (1038 C vs. 672 C) was observed when comparing uncorrected IR
temperature data with corrected temperature data. Upon validation of the IR parameters for a melted area, experimentation was conducted to also
determine powder emissivity (found to be 0.50). The thermal model presented here can be modified and implemented in other AM technologies
for consideration of radiation energy to acquire absolute temperatures of layered surfaces, leading to improved thermal monitoring and control of
the fabrication process.
2014 Elsevier B.V. All rights reserved.
Keywords: Electron beam melting; Infrared; Monitoring; Emissivity; Surface temperature
1. Introduction
Electron beam melting (EBM) is an additive manufacturing
(AM) process for direct-metal freeform fabrication that uses
metal powder as the precursor to build solid metal parts. Within
the past decade, the development of processing parameters for
http://dx.doi.org/10.1016/j.addma.2014.12.001
2214-8604/ 2014 Elsevier B.V. All rights reserved.
several alloys (copper, niobium, iron, TiAl, TiNb, and nickelbased superalloys) has made them suitable for fabrication by
EBM technology [1,9,13,1720]. Although EBM has been used
in the commercial fabrication of parts for the medical and
aerospace industries [8,24]; opportunities still exist to make AM
systems more repeatable and reproducible for the production of
high-quality products that may be qualified during fabrication
through the use of monitoring, feedback and control. Christensen
et al. [5] discussed qualification of EBM technology for orthopedic applications and a need for systematic process monitoring
32
2. Methods
2.1. Electron beam melting process
The EBM process uses a metal powder (typically Ti6Al4V,
ASTM F75 CoCr, or Grade 2 Titanium), as the precursor material to build solid parts in a layerwise fashion inside a vacuum
chamber (104 Torr). The Arcam A2 system follows four steps
in its fabrication process: (1) depositing a layer of metal powder
(typically in the range 0.050.20 mm thick) across the build platform using the machines raking mechanism, (2) preheating the
layer to 50% of the metals melting temperature (760 C for
Ti6Al4V) using the electron beam gun at a low beam current
(8.8 mA) and high scan speed (14,600 mm/s) to lightly sinter the
metal powder (also helps reduce the parts residual stresses), (3)
selectively melting the preheated powder by increasing the beam
power (up to 17 mA) and reducing the scan speed (500 mm/s) to
effectively reach the materials melting point, and (4) lowering
the build platform a height equal to one layer thickness. This
process is repeated until the part is complete.
2.2. Thermal imaging system installation
The IR camera utilized in this study was a FLIR SC645 (FLIR
Systems, Inc., Wilsonville, OR) with a resolution of 640 480
pixels and a temperature range of up to 2000 C. The distance
between the camera and the powder bed was 330 mm, resulting in a field of view of 274 mm by 206 mm (with a 25 lens),
and a spatial resolution of 175 m/pixel. The camera was positioned atop the build chamber and used a zinc-selenide (ZnSe)
viewing window. This window type was selected due to its
high transmittance of the IR wavelengths; it is the most widely
used type of window in IR imaging [10]. A mechanical shutter
mechanism was installed inside the chamber covering the ZnSe
window to protect it from metallization that may occur during
the melting of metal powder particles. The shutter also served to
ensure that images were not obscured from unwanted particles
on the windows surface. The angle between the surface normal
and the camera was 18 (with an incline allowance of up to
25 ). A computer-aided design (CAD) rendering of this setup
is shown in Fig. 1. Images of the build surface were captured
by the IR camera following the completion of the melt process
step. An automated system was developed using LabVIEW software (National Instruments, Austin, TX) to open the shutter for
1 second upon completion of the melt step; images were captured during the 1 s interval. This process was repeated for each
subsequent layer.
2.3. IR camera calibration
IR imagers are capable of providing quantitative thermal
images; however, their measurement accuracy is dependent upon
the effect of environmental conditions and surface properties
(e.g. reflectivity and emissivity). Measuring and accounting for
such conditions is critical to the calibration procedure of IR
cameras to obtain absolute temperature measurements. The radiant energy received from an object will be a function of its
33
temperature, spectral emissivity (a surfaces ability to emit radiation energy), reflections from its surroundings, and atmospheric
transmission. Fig. 2 is a schematic representation of the sources
that were measured by the IR camera. These sources include
the target, radiation from surroundings (e.g. heat shield surfaces), and emission and absorption from the environment (e.g.
viewing glass). The main sources of thermographic disturbance
are the ambient sources, or reflecting sources, and the targets
emissivity.
was placed inside the build chamber atop the build platform
and heated to 300 C. The emissivity value for the blackbody
(0.99), previously provided by the manufacturer, was assigned
to the IR camera. An average temperature for the surface of
the blackbody was recorded without the ZnSe window present,
assigning an external transmission of 100%. The window was
then installed and the external optics transmission parameter was
modified until the average temperature of the surface equaled
that of the temperature measured without the viewing window.
34
Fig. 3. CAD rendering of the Arcam A2 heat shield used (bottom view) showing thermocouple locations (a) and isometric view of heat shield assembly (b).
where FSj is the view factor between the target surface (S) and
each wall (j), representing the fraction of the radiation leaving the
surface s that strikes the j-th wall. Tj represents the temperature of
the j-th wall [6]. In Eq. (1), the basic formulation for determining
the view factor between two walls (F12 ) is:
cos 1 cos 2
1
F12 =
dA1 dA2
(2)
A1 A1 A2
r 2
where A is the area of each wall, r is the distance between the
two walls, and is the angle between each walls normal vector
and the vector pointing toward each wall.
After identifying the enclosure, the thermal state and radiation properties of each surface (shield walls) of the enclosure
were specified [23]. In the current analysis, the surfaces were
assumed to be opaque, diffuse, and gray; therefore, nontransparent (which is a reasonable assumption since the heat
shields are solid steel sheets). The surfaces were diffuse emitters and diffuse reflectors independent of wavelength (a common
assumption made in radiation problems to simplify analysis)
[4]. Also, each surface of the enclosure was assumed to have
isothermal properties and the incoming and outgoing radiations
were assumed to be uniformly distributed over each surface
[4]. The view factor for each wall of the enclosure was calculated independently by using published formulae and plots,
algebra (utilizing the basic formulation shown by Eq. (2)),
and then verified using MATLAB software (MathWorks, Natick, MA) [21]. Shield wall temperature measurements were
recorded during three different builds of Ti6Al4V using type
K thermocouples (Range: 270 to 1372 C, Model: HKQIN116U-26, Omega Stamford, CT) cemented (high-temperature
cement, Omegabond 400, Omega, Stamford, CT) to the shield
walls (Fig. 3). The use of only three thermocouples was limited
by the pass-through that was installed which allowed the installation of wiring inside the system while still maintaining an
appropriate vacuum environment for fabrication. The temperatures of the shield walls were measured during each build,
averaged for each wall, and used to calculate the mean radiant
temperature using the view factor model. For this experiment, it
was assumed that the left and right shields as well as the front
and rear shields had the same temperatures (since they were
exposed to the same thermal environment and the surface area
of each paired shield was the same).
2.3.3. Emissivity calculation
Any real surface has a distinct emissivity surface property (the
ability to emit radiation energy and denoted ), that is unitless
and varies between 0 and 1 [4,12]. The emissivity of a surface
is not constant; instead, it varies with temperature and surface
morphology. A common tool used in determining emissivity
is a blackbody cavity, or radiator. A blackbody cavity is one in
which radiation enters through an opening and undergoes multiple reflections within the cavity before exiting [4]. To determine
the materials (Ti6Al4V) emissivity, a rectangular prism with
an internal cavity was fabricated via EBM. Construction of these
samples in EBM was purposefully done to acquire the true emissivity value, since the unique EBM produced part surface is fairly
constant in every Ti6Al4V build. Once fabricated, the part was
removed from the machine for thorough cleaning of the internal
cavity. The powder inside the cavity was removed by directing
the hose of Arcams powder recovery system toward the cavitys
opening for 15 min. This is a system which uses a combination of pressurized air and metal powder (Ti6Al4V) to remove
sintered particles from a part. After the cavity was cleaned of all
sintered powder, it was used as a blackbody radiator.
35
Fig. 4. (a) CAD rendering (wire frame view) of blackbody radiator model and its internal cavity, (b) IR image of the blackbody radiator with software measurement
tools and cemented-on thermocouple.
4
Tr4 Tmr
4
TS4 Tmr
(3)
of the orifice. The resulting emissivity value was the value used
as the emissivity of melted (solid) Ti6Al4V. Additionally, a
thermocouple was cemented on the surface of the blackbody
radiator to be used as a validation tool. The validity of input
parameters was verified by comparing the IR cameras output
temperature measurement to that of the thermocouple. Fig. 4a
shows a CAD rendering of the blackbody radiator and Fig. 4b, an
IR image of the blackbody radiator at an elevated temperature.
Upon validation of the solid emissivity, a similar procedure that
will be discussed later (using a thermocouple) was implemented
to determine the emissivity of the Ti6Al4V powder.
2.4. Experimental IR parameter verication
To verify the aforementioned theoretical models, that is, the
radiation heat transfer theory used to derive the view factor
models and the blackbody radiator model, an experimental verification procedure was carried out. The experiment was designed
to compare the temperature measurements from a thermocouple
embedded during fabrication and the IR cameras temperature
measurements taken consecutively during the embedding process. To accomplish this, a solid cube was fabricated using
EBM on a thermocouple to make real-time temperature measurements while simultaneously acquiring IR images. Step one
in the experiment was to insert a type K thermocouple (Range:
270 to 1372 C, Model: HKQIN-116U-26, Omega Stamford,
CT) through a drilled hole in the center of the EBM start plate,
or build platform. The thermocouple contains an Inconel sheath
which protects the junction from damage that may be incurred
from the high-power electron beam. Fig. 5a is a CAD rendering of this setup. When the manufacturing process commenced,
the solid part was manufactured on the thermocouple, embedding it within the solid part as the layered process continued
(shown in Fig. 5b). Thermocouple temperature measurements
were recorded using National Instruments data acquisition software and a thermocouple input module (NI 9213, Austin, TX).
It is important to note that the tip of the thermocouple was only
partially exposed (2 mm) into the Z-axis build plane to prevent
36
Fig. 5. (a) CAD rendering of thermocouple placement through drilled hole within the build platform, (b) EBM-fabricated part with embedded thermocouple for
solid, and (c) embedded thermocouple within sintered powder of an EBM-fabricated part.
Fig. 6. IR image of EBM part with embedded thermocouple to verify solid emissivity measurements (a) and embedded thermocouple within sintered powder to
measure powder emissivity (b).
bending or breaking of the powder re-coater blades. Temperature measurements recorded at the thermocouple junction point
(about 0.5 mm from the thermocouple tip) were compared to
its counterpart IR measurements. The thermocouple junction
point was 0.15 mm in diameter which corresponds to three
EBM build layers of IR measurements. Therefore IR images captured at layers 19, 20, and 21 (representing layers closest to the
installed thermocouple) or corresponding Z-heights 1.33 mm,
1.40 mm, and 1.47 mm were used to compare IR temperature
measurements to thermocouple temperature measurements. It is
important to note that, for the experiment in Fig. 6a, the electron
beam scanned over the thermocouple, which was still exposed
at layer 19 and fully covered by solid Ti6Al4V at layers 20
and 21. Furthermore, temperature differences between layers
(Table 4) are most likely due to differences in scan direction from
layer to layer. The same validation procedures were followed
using a hollow cube to determine the emissivity of the sintered
powder (experimental setup shown in Fig. 5c). For the experiment shown in Fig. 5c, the temperature of the thermocouple was
used as an accurate representation of the powder temperature
and the emissivity of the surrounding powder was adjusted until
the IR temperature agreed with that of the thermocouple.
Upon determining the IR parameters, cylinders were fabricated using the standard processing parameters to fabricate
parts in EBM using Ti6Al4V. The average temperature from
a melted cylinder was taken at every other layer and recorded
until build completion. Additionally, the EBM system automatically logs temperature measurements every 10 layers using
a single thermocouple that is located below the build platform. Both temperature measurements were plotted against the
corresponding layer and compared. To prevent reduction in
transmittance, the ZnSe glass was cleaned every three builds
using Kimwipes (Kimberly-Clark Global Sales, Inc., Roswell,
GA) and isopropyl-alcohol.
3. Results and discussion
3.1. External optics transmission
Table 1 summarizes the parameters assigned to the camera
and the recorded temperatures in testing the external optics
transmission. The temperature measurements without a window showed the ideal temperature, or the temperature without
emission. The temperature with the ZnSe window installed was
37
Table 1
External optics testing of ZnSe window IR image parameters assigned.
Experiment
Emissivity
Avg. temp. ( C)
Std. dev. ( C)
No window
ZnSe window
ZnSe window (adjusted)
0.99
0.99
0.99
1.00
1.00
0.94
309
298
309
5
2
2
Table 2
Mean radiant temperature calculations.
View factors
T 4 Fsj =
Tmr (K) =
Tmr
( C)
6.1E+09
4.3E+10
3.0E+10
4.3E+10
2.0E+10
1.4E+11
4
T4 Fs j
T 4 Fsj =
615
342
Table 3
Emissivity data of Ti6Al4V (solid).
Sample
Temperature ( C)
Emissivity ()
EBM 1 (solid)
EBM 2 (solid)
Yang et al.
Yang et al.
Yang et al.
Yang et al.
Yang et al.
Yang et al.
Yang et al.
177
198
165
250
350
450
550
700
750
0.26
0.26
0.25
0.27
0.23
0.23
0.24
0.25
0.27
38
Table 4
Thermocouple and IR simultaneous temperature measurements for solid part.
Layer number
Thermocouple ( C)
IR camera ( C)
% Difference
19
20
21
730
717
712
703
689
685
3.72
3.84
3.76
Average
3.77
layers. As can be seen in the table, the IR camera and thermocouple measurements agree within an average difference of
3.77%.
Emissivity of sintered powder was measured using the methods described in Section 2.2 and IR temperature data was taken
from the region of interest described in Fig. 6b. The mean radiant temperature determined in Section 3.2 was used for all IR
temperature calculations. Table 5 shows the results obtained
where the original IR temperature was corrected using different emissivity values until it was in close agreement with the
thermocouple measurement that was taken simultaneously. The
resulting emissivity value for Ti6Al4V sintered powder was
determined to be 0.50. Powder emissivity will tend to vary from
the particle size, shape, and packing used, thus the emissivity determined is specific to Ti6Al4V powder used in this
study (purchased from Arcam AB). With results provided in
Table 5, it is now possible to determine temperature of the powder throughout processing. It was found that powder temperature
upon deposition was 430 C. Such thermal gradient between
the recently melted part and newly deposited powder can cause
excess cooling that leads to thermal stresses and undesirable
microstructures (e.g. martensitic phase in Ti6Al4V). By
using this information, it may become possible to optimize the
process by adding a powder heating system that reduces thermal
gradients due to powder deposition.
Table 5
Thermocouple and IR simultaneous temperature measurements for sintered powder.
Layer number
IR camera, C (uncorrected)
Thermocouple ( C)
Emissivity
IR camera, C (corrected)
% Difference
19
20
21
994
996
993
680
678
676
0.50
0.50
0.50
678
680
677
0.25
0.29
0.15
Average
0.23
Fig. 7. EBM processing temperature measurements for temperature measurements using default IR camera parameters, corrected parameters, and thermocouple data
from the bottom of the build platform.
39