Documente Academic
Documente Profesional
Documente Cultură
a r t i c l e
i n f o
Article history:
Received 16 November 2008
Accepted 5 January 2009
Available online 22 January 2009
Keywords:
Selective laser sintering
Shrinkage compensation
Accuracy
Exposure strategy
Laser scanning
a b s t r a c t
Shrinkage in selective laser sintering process is primarily inuenced by material, process parameters and
the geometry of the fabricated part. The part inaccuracy due to this shrinkage is overcome by calibrating
and compensating it. Further improvements in accuracy of the part can be achieved by conducting more
studies to appreciate the nature of deviations, by subjecting calibration part to varying build conditions.
This paper presents the results of the experimental study carried out to understand shrinkage behavior in
selective laser sintered polyamide 12 parts. Due to the inherent nature of the process, the shrinkage
behavior is often inuenced by exposure strategies, part positioning, part orientation, and other compensations applied to the part. When fabricated under different settings, shrinkage behavior of calibration
parts tends to differ from usual. In the present work, the variations in error patterns due to different strategies in building the calibration part have been reported. The discussions presented in this paper on the
shrinkage behavior of plastic parts are likely to make the process of compensation efcient and hence aid
in improving the accuracy of the process.
2009 Elsevier Ltd. All rights reserved.
1. Introduction
Rapid prototyping (RP) is an additive manufacturing process for
low volume, high value, custom-designed parts. These parts can be
made out of common engineering thermo-plastics such as polyamides, ABS, polycarbonate, polyphenylsulfone (PPSF) to metal
parts such as titanium, stainless steel and tool steel [1]. Since the
delivery of the rst commercial machine in 1988, RP has grown
as integral part of the new product development process. The use
of RP has reduced time to market a product, cut trial costs, and improved product quality by giving design and manufacturing professionals a tool to quickly verify and ne-tune designs before
committing these to expensive tooling and fabrication. RP also
has some challenges that must be improved upon before it becomes rapid manufacturing (RM) for producing parts in small
batches or customized parts. One of the main challenges is part
accuracy. This is main concern of industries such as aerospace
and bio-medical which would like to use RP technology for producing directly usable products. The capability to produce a part in
hours without any tooling is a powerful advantage for many industries. With the stronger plastics and even metallic materials used
in some of the RP processes, parts can be produced that will withstand reasonable amount of stress and higher temperature ranges.
However the parts produced tend to warp and/or shrink from its
given dimensions, forcing the user to run several trials of a part
to reach its ideal dimension or settle for a slightly inaccurate part
* Corresponding author. Tel.: +91 11 26596083.
E-mail address: pmpandey@mech.iitd.ac.in (P.M. Pandey).
0261-3069/$ - see front matter 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.matdes.2009.01.009
resistant [3]. These parts can be made to bend, snap or bolt together
and form exible hinges [4]. However, parts produced by SLS are
poor in terms of accuracy due to the various errors accumulating
from data preparation stage to nishing stage. One of the main
sources of size and shape variations of the part is shrinkage during
processing. The following paragraphs present some of the previous
work carried out by researchers to study shrinkage in SLS process.
Wang et al. [5] studied the effect of process parameters on
shrinkage characteristics in SLS process. They found that percentage shrinkage increases with increase in the scanning speed and
hatch spacing, but decreases with increasing layer thickness, the
laser power, part bed temperature and delay time. They used neural networks for the study of effect of process parameters. Ragunath and Pandey [6] studied the effect of process parameters on the
process and material shrinkage. They found that scan length inuences shrinkage in the X direction. They also predicted that scaling
factors can have a linear relationship with scan length. They derived empirical relations for percentage shrinkage in terms of scan
length using Taguchi method. However, they used scaling factors
based on the maximum dimensions not on the individual scan
lengths while compensating using the model developed by them.
Hopkinson and Sercombe [7] investigated the effect of part height,
part position and build direction on the shrinkage during indirect
SLS of aluminum powder. They found that errors decrease with increase in nominal dimensions in inltrated state than in a green
state for all build condition. They found that error in Z direction
is more pronounced than in-plane errors due to phenomenon
called Z-growth whereby the heat from the laser penetrates beyond the down facing surface to bond unwanted particles. They
characterized tolerances achievable on different build orientations
and found that a tolerance of 0.1 mm is achievable for a 40 mm
part in the XY direction and 0.3 mm in Z direction. Zhu et al. [8]
studied the shrinkage behavior in metal powders. They quantied
two types of shrinkages namely thermal shrinkage and sintering
shrinkage. They found that in-plane shrinkage (X and Y shrinkage)
is very less compared to the shrinkage in the build direction. They
found that the sintering shrinkage is mainly caused by densication and is a kind of elastic compressive shortening. They suggested that thermal shrinkage due to cyclic heating can be
reduced by controlling process parameters. In their work, the thermal shrinkage increases with increase in laser power and shrinkage decreases with increase in scan speed and scan spacing. Ning
et al. [9] considered the effect of geometry on the shrinkage of
metallic parts. They introduced the speed compensation technique
based on the scan length. In their method, when building a part,
the laser scan speed is adjusted dynamically according to the scan
length which varies with geometric shape of the part. The different
scan speeds for the scan lengths are chosen based on the shrinkage
values at different speeds.
Ning et al. [10] conducted series of experiments for direct metal
laser sintering (DMLS) process to nd the effect of hatch length on
the material anisotropy, heterogeneity and part strength. They
concluded that short hatch lines cause serious shrinkage and the
part becomes less homogeneous. They proposed an algorithm to
nd out optimal hatch direction for a typical layer by considering
the shrinkage as a function of hatch length. Manetsberger et al.
[11] studied the effect of temperature, time and pressure on the
shrinkage of polymer parts. They used a thermal simulation as a
basis for shrinkage compensation in SLS process. They expressed
shrinkage values as a function of temperature and also showed a
linear dependency to the pressure applied. Jacobs [12] discussed
the effects of shrinkage variation on the accuracy of rapid tooling
inserts. His work mainly concentrated on random shrinkage and
found that standard deviation of random shrinkage is directly proportional to the mean process shrinkage. He also found that nonuniformity in shrinkage is mainly attributed to geometry of the
2947
2948
20
strips
Y
200
X
Fig. 2. Shrinkage calibration specimen.
Table 1
Process parameters set for the experiments.
Parameter
Value
Hatching
Contouring
36
4.5
0.3 103
10
0.7
NA
176
0.15 103
2949
0.3
0.25
0.2
0.15
a=0.6 mm
a=0.645 mm
0.1
0.05
20
40
60
80
100
120
140
160
180
200
than that for contouring), however, in this case caution must be observed in guaranteeing that there are no unsintered particles between the contour path and the hatching region. Thus, the beam
offset for hatching (dh) must be chosen in such a way, so as to form
a narrow overlapping regions between the contour path and the
hatching region (Fig. 3). The overlap should not be too wide
though, to prevent over-sintering.
on shrinkage pattern, two different curves are plotted for deviations per unit length vs nominal dimensions.
Fig. 4 shows deviations per unit lengths for specimens using
beam compensation adjustment values with 0.6 and 0.645 mm
and its building strategies are listed in Table 2. As discussed earlier
if shrinkage of beam compensation adjustment values is not properly estimated, it introduces errors in deviations per unit length
pattern especially for smaller dimensions.
Lc Lm
100
Lc
where Lc is the nominal dimension of the part and Lm is the measured dimension of the part after sintering and cooling. In the present work, for some specimens, no beam compensation is applied
while fabricating specimens. The beam compensation value is adjusted to nominal dimensions while calculating deviations per unit
length. The adjusted nominal dimensions is Lc + a, where a is the
beam compensation adjustment value and is most of the times
equal to the spot diameter.
4. Results and discussion
4.1. Effect of beam compensation on shrinkage
As discussed earlier the beam compensation adjustment value
is normally chosen equal to the spot diameter. In earlier approaches the shrinkage of this width of scan line is usually neglected. In the present work, experiments are performed to study
the effect of this small shrinkage that is being neglected in deviations per unit length calculation.
In order to estimate the shrinkage in beam compensation
adjustment value (a) for the hatch lines, a part is designed with
dimensions 25 0.6 6 mm. The dimensions of this part are chosen such that when fabricated, this part will have single line exposure of laser beam. The thickness of the single hatch line part is
fabricated and measured and its value is found to be 0.555 mm.
Thus it is found that 0.6 mm scan track produced 0.555 mm thick
part after shrinkage. The percentage shrinkage calculated for this
single scan track is 7%. The beam compensation adjustment value
(a) is calculated as 0.645 mm after compensating this shrinkage.
In order to nd out the effect of this beam compensation value
Table 2
Building strategies for part compared in Fig. 4.
Process variable
Considering
shrinkage in a
Without considering
shrinkage in a
X direction only
X direction
X direction only
X direction
0
Is OFF
NA
0.645 mm
0
Is OFF
NA
0.6 mm
Yes
Yes
Table 3
Building strategies for parts compared in Fig. 5.
Process variable
With contour
exposure
Without contour
exposure
X direction only
X direction
0.59 mm
Is ON
0.33 mm
0
X direction only
X direction
0
Is OFF
NA
0.645 mm
Yes
Yes
2950
0.65
0.6
0.55
0.5
0.45
0.4
Without contour exposure
With contour exposure
0.35
0.3
0.25
0.2
20
40
60
80
100
120
140
160
180
200
posed and in another specimen both contour and hatch lines are
exposed as shown in Fig. 3. The dimensions are measured and
the deviations per unit length is calculated and plotted against
nominal dimensions for both specimens (Fig. 5).
For the specimen with only hatching exposure, the shrinkage is
found to vary between 0.2% and 0.3%. For the specimen with both
contour and hatching exposure, the shrinkage is found to vary between 0.35% and 0.65%. It is observed that this higher percentage of
shrinkage is caused primarily due to the constraining effect of the
contour exposure. Contour boundaries arrest the expansion of the
layer occurring during sintering process [8]. In addition to this,
more noise in error pattern is observed with the contour exposed
part than that of hatch line exposed part. The contour lines are
scanned with a low laser power and high beam speed as compared
to the hatch lines. This causes non-uniform shrinkage between two
lengths of specimens. Also the constraining effect is not consistent
between different strips of specimen and the result is a random
noise found over quadratic curve tted to deviations per unit
length data as shown in Fig. 5.
4.3. Effect of inertia of scanning mirror on shrinkage
Usually there are variations in scanning speed due to acceleration and deceleration of scanning mirror during hatching exposure
at the boundaries of the layer which are to be eliminated to get a
uniform energy density. In order to avoid inconsistencies in the energy density of exposure, the length of scan during which acceleration and deceleration of galvano mirrors takes place, should be
compensated. The laser should be switched off while scanning
these lengths and hence no exposure on this compensated length.
This length of unexposed compensated region is independent of
Table 4
Building strategies for parts compared in Fig. 7.
Process variable
With skywriting
Without skywriting
Y direction only
Y direction
0
Is OFF
NA
0.645 mm
Yes
Y direction only
Y direction
0
Is OFF
NA
0.645 mm
No
2951
2
Without skywriting
With skywriting
1.5
1
0.5
0
-0.5
20
40
60
80
100
120
140
160
180
200
tween 0.2% and 0.3% for different strip lengths, where as the
shrinkage along Y direction varies between 0.4% and 0.35%. For
specimen oriented along Y direction, the deviations per unit length
increases steeply with nominal dimensions for strip lengths between 20 and 120 mm and decreases gradually after 120 mm. Also
expansion rather shrinkage is found for small strip lengths (20, 30
and 40 mm strips). As discussed earlier, there is a shrinkage
expansion behavior during time-history of sintering [15]. Between
the scan lengths 20 and 110 mm, the expansion dominates the
shrinkage. This expansion is basically the falling of powder particles while melting and it is inuenced by various forces like buoyant force, marangoni ow and gravity and a more detailed
explanation for this in-plane expansion is given by Zhu et al. [8].
There is a larger variation of shrinkage to the nominal dimension in Y direction specimen due to in-homogenous nature of the
shrinkage in Y direction inuenced by two things: (a) thermal gradients and (b) recoater movements.
Firstly, the thermal gradient within a build chamber and the
variations in temperature during the building and cooling processes differ with the different length of strips in specimen. The
reason for this lies in the variation in the energy density. It should
be noted that larger strip length has large delay time between consecutive exposures of the same point, whereas lower length has
smaller delay time [19]. And due to the very low thermal conductivity of the material used in the process, the part bed retains heat
over a relatively long period of time. Moreover, shrinkage in larger
strip length is very different from that occurring in smaller length
strips since shrinkage in large geometries tends to be retarded owing to internal stresses [11].
Table 5
Building strategies for parts compared in Fig. 9.
Process variable
X direction only
X direction
0
Is OFF
NA
0.645 mm
Yes
Y direction only
Y direction
0
Is OFF
NA
0.645 mm
Yes
2952
0.3
0.2
0.1
0
-0.1
-0.2
-0.3
-0.4
-0.5
-0.6
20
40
60
80
100
120
140
Nominal dimensions (mm)
160
180
200
(0.4 to 0.35%). Scan length for the Y direction part is very short
and is constant for all strip length in scanning parallel to X direction unlike scanning in Y direction where scan length is equal to
the strip lengths. While scanning parallel to X direction, the shrinkage is dominant than expansion for lower length of strips. Expansion in Y direction is arrested by the scan lines parallel to X
directions as shown in Fig. 11. As discussed already, the expansion
and non-uniformity in shrinkage in Y direction part is clearly visible in the second degree curve (Fig. 10) tted to deviations per unit
length data points for Y direction scanning.
Table 6
Building strategies for the specimens compared in Fig. 10.
Process variable
Scanning along X
Scanning along Y
X direction only
Y direction
0
Is OFF
NA
0.645 mm
Yes
Yes
Y direction only
Y direction
0
Is OFF
NA
0.645 mm
NA
Yes
0.8
0.6
0.4
0.2
0
Scanning parallel to X
Scanning parallel to Y
-0.2
-0.4
-0.6
20
40
60
80
100
120
140
Nominal dimensions (mm)
160
180
200
2953
Positioning error
compensated
Positioning error
uncompensated
X direction only
Y direction
X direction only
Y direction
0
Is OFF
NA
0.645 mm
0
Is OFF
NA
0.645 mm
Yes
Yes
No
Yes
1.8
1.6
Position error compensated
Position error uncompensated
1.4
1.2
1
0.8
0.6
0.4
20
40
60
80
100
120
140
160
180
200
2954
5. Conclusion
References
Our overall research goal is to understand the nature of shrinkage occurring in SLS process to improve the accuracy of the parts.
This paper focused on the shrinkage behavior by subjecting the
shrinkage calibration specimen to varying build conditions. Certain
compensations other than shrinkage are needed to get accurate
estimate of the shrinkage. Beam offset, inertia of scanning mirror
and positioning errors in hatch generations are found to delude
the shrinkage pattern. Moreover, exposure strategies and part orientation are found to inuence the accuracy of the part to be produced. The conclusions which can be drawn from the work
presented in this paper are highlighted below:
[1] Chua CK, Leong KF, Lim CS. Rapid prototyping: principles and applications. 2nd
ed. Singapore: World Scientic; 2003.
[2] Noorani RI. Rapid prototyping: principles and application. USA: Wiley; 2006.
[3] Hopkinson N, Hague RJM, Dickens PM. Rapid manufacturing: an industrial
revolution for the digital age. USA: Wiley; 2006.
[4] Pham DT, Dimov SS. Rapid manufacturing. London: Springer-Verlag; 2006.
[5] Wang RJ, Wang L, Zhao L, Liu Z. Inuence of process parameters on part
shrinkage in SLS. Int J Adv Manuf Technol 2007;33:498504.
[6] Ragunath N, Pandey PM. Improving accuracy through shrinkage modeling by
using Taguchi method in selective laser sintering. Int J Mach Tool Manuf
2007;47:98595.
[7] Hopkinson N, Sercombe TB. Process repeatability and sources of error in
indirect SLS of aluminium. Rapid Prototyping 2008;14(2):10813.
[8] Zhu HH, Lu L, Fuh JYH. Study on shrinkage behavior of direct laser sintering
metallic powder. Proc IMechE Pt B: J Eng Manuf 2006;220:18390.
[9] Ning Y, Wong YS, Fuh JYH. Effect of control of hatch length on material
properties in the direct metal laser sintering process. Proc IMechE Pt B: J Eng
Manuf 2005;219/1:1525.
[10] Ning Y, Wong YS, Fuh JYH, Loh HT. An approach to minimize build errors in
direct metal laser sintering. IEEE Trans Autom Sci Eng 2006;3(1):7380.
[11] Manetsberger K, Shen J, Muellers J. Compensation of non-linear shrinkage of
polymer materials in selective laser sintering. In: Proceedings of the SFF
symposium; 2003. p. 34656.
[12] Jacobs P. The effects of random noise shrinkage on rapid tooling accuracy.
Mater Des 2000;21:12736.
[13] Wang X. Calibration of shrinkage and beam offset in SLS process. Rapid
Prototyping 1999;5(3):12933.
[14] Shi Y, Li Z, Sun H, Huang S, Zeng F. Effect of properties of polymer materials on
the quality of selective laser sintering parts. IMechE Pt L: J Mater Des Appl
2004;218:24752.
[15] Lu PK, Li W, Lannutti JJ. Density gradients and the expansionshrinkage
transition during sintering. Acta Mater 2004;52:205766.
[16] Venuvinod PK, Ma W. Rapid prototyping laser based and other
technologies. London: Kluwer Academic; 2004.
[17] Jain PK, Pandey PM, Rao PVM. Experimental investigations for improving part
strength in selective laser sintering. Virt Phys Prototyping 2008;3(3):17788.
[18] EOS Gmbh. Basic training manual EOSINT P380, Munich, Germany; 2003.
[19] Jain PK, Pandey PM, Rao PVM. Effect of delay time on part strength in selective
laser sintering. Int J Adv Manuf Technol 2008. doi:10.1007/s00170-008-16823.
[20] Hur SM, Choi KH, Lee SK, Chang PK. Determination of fabricating orientation
and packing in SLS process. J Mater Process Technol 2001;112:23643.
[21] Senthilkumaran K, Pandey PM, Rao PVM. Shrinkage compensation along single
direction dexel space for improving accuracy of SLS process. In: Proceedings of
the fourth annual IEEE conference on automation science and engineering;
2008. p. 82732.