ABSTRACT
Fused Deposition Modeling is most popular rapid prototyping process because of its
faster, economical and clean technology, however it suffers from low surface finish quality.
To improve its surface finish quality, various attempts had been made by several researchers
by controlling various process parameters. The main objective of this research is to apply
chemical treatment processes through Design of Experiments using different chemicals with
variant conditions like different levels of concentration, time of exposure, temperatures and
initial roughness, interaction effects of the process parameters have also been analyzed.
ANOVA technique is used to find out the significant factors affecting the surface finish.
Results show satisfactory improvement in surface finish of FDM parts (ABS) with simple
inexpensive and harmless chemical treatment processes.
I. INTRODUCTION
Rapid prototyping (RP) technologies provide the ability to fabricate initial prototypes
from various model materials. Stratasyss Fused Deposition Modeling (FDM) is a typical RP
process that can fabricate prototypes out of ABS plastic [1]. FDM rapid prototyping process
is quite popular in industry for various reasons such as: different raw materials (thermo
plastics) can be used as long as the appropriate hot head is available; FDM parts are very
strong and hence can work as functional parts; it does not employ lasers, hence is less
531
International Journal of Mechanical Engineering and Technology (IJMET), ISSN 0976
6340(Print), ISSN 0976 6359(Online) Volume 3, Issue 3, Sep- Dec (2012) IAEME
expensive and there are no safety related issues; It does not use liquid/powder raw materials
and hence is a clean process; It can be kept in an office environment as a 3D printer; very
easy to remove the support material; this is probably the easiest of all RP processes; this is
the cheapest technology; etc.
Parts produced by FDM are, however, less accurate than those produced by other
rapid prototyping processes such as Stereo lithography Apparatus (SLA), Solid Ground
Curing (SGC). Besides, FDM process is very slow as every point of the volume is addressed
by a mechanical device. The key issue with FDM process is surface roughness because of its
staircase effect (the angle between the vertical axis and surface tangents) [2]. The poor
surface finish affects the functioning of RP parts, depending on the geometry of the enclosing
surface, the building strategy, layer thickness and orientation of the part; this drawback may
outweigh the advantages of RP parts [3].
In literature, several researchers have proposed various methods to reduce the surface
roughness of the FDM parts of ABS material. One of the prominent methods is to control the
process parameters like layer thickness, build orientation, raster width, raster angle, air gap
etc.. In this method process parameters were optimized using statistical techniques like design
of experiments and gray relational analysis have been integrated for obtaining the optimum
process parameter values [3]. The process parameters influence the responses in a highly non-
linear manner; therefore, prediction of overall dimensional accuracy is made based on
artificial neural network (ANN) [4]. Several algorithms were also developed to obtain
optimum part deposition orientation for fused deposition modeling process for enhancing part
surface finish and reducing build time [5, 6].
Another method is adaptive slicing scheme in which slices of different thicknesses in
different zones are produced while building the part [7-10]. Daekeon Ahn et al investigated
the relation between surface roughness and overlap interval [11], they also analyzed and
discussed the effects of surface angle and filament section shape to the surface roughness.
Debapriya Chakraborty et al introduced a new kind of deposition method called Curved layer
FDM or CLFDM which offers solution to the issues of surface roughness and strength for
thin curved shell-type parts, this process proposes an entirely new building paradigm for
FDM, the filaments would be deposited along curved (essentially non horizontal) paths
instead of planar (horizontal) paths [12]. A mathematical technique has been developed by
W. Rattanawong et al to determine best part orientation based on minimum volume error
(VE) in the part due to staircase effect [13]. Noshir A. Langrana et al have developed a
method to fabricate the highest quality of multi-material parts. In this method, a virtual
simulation system and experimental real time video microscopy have been developed. In this
virtual simulation, one can check or test a variety of the layered manufacturing process
parameters, and make the best selection of tool path and other parameters to obtain high
quality parts [14]
One more method for improving surface finish is chemical treatment method which has
been proposed by L.M. Galantucci et al [2]. In this chemical treatment method
,Dimethylketone (Acetone) with 90% concentrated solution and 10% water was used and
parts were immersed in diluted solution for 5 minutes and also suggested that further studies
need to be conducted on freeform products, also using other dimethylketone solvents such as
ethylene and using designed experiments to optimize the process in terms of the solution
concentration and process time. To the best of the authors knowledge, no investigations of
chemical treatment method have been reported since the work of L.M. Galantucci et al. and
hence the present study has been undertaken by the authors to investigate the optimum
conditions for obtaining best surface finish from the chemical treatment process.
532
International Journal of Mechanical Engineering and Technology (IJMET), ISSN 0976
6340(Print), ISSN 0976 6359(Online) Volume 3, Issue 3, Sep- Dec (2012) IAEME
III.METHODOLOGY
In this paper we will be optimizing the chemical treatment process using Design of
Experiments (DOE). The factors affecting the chemical treatment process were identified by
performing numerous trials, based on these trials concentration, temperature, time of
exposure and initial roughness were identified as possible main factors. These are analyzed
using Design of Experiments (DOE). DOE is done for two different chemicals i.e.
Dimethylketone (Acetone) and Methylethylketone (MEK), test specimen selected are shown
in Fig.1(a) to Fig.1(e)... The optimization method is based on Design of Experiments (DOE)
and Analysis of Variance (ANOVA). It identifies significant parameters affecting the surface
finish, to which more attention must be paid in order to attain best possible results.
Figure1(a) Test Specimen with 0.2540 mm Figure1(b) Test specimen with 0.3302
mm layer thickness layer thickness of Sample1/A1
Figure 1(c) Sample 2/A2 Figure 1(d) Sample 3 Figure 1(e) Sample 4
In total, L16 has 15 degrees of freedom. The remaining (15-13) two degrees of freedom are used for
error. The design of experiments based on the L16 array for the present case is shown in Table 2.
Inability to distinguish effect of factors and interactions is called confounding [16]. As it is expected
that factors C, Tm, Tp to interact, no factors are assigned to columns 5, 6 and 7. This is done to avoid
confounding. The results of surface roughness value of various FDM samples with combinations of
parameters for chemical 1 are shown in Table 2. Similar results for chemical 2 is shown in Table 3
534
International Journal of Mechanical Engineering and Technology (IJMET), ISSN 0976
6340(Print), ISSN 0976 6359(Online) Volume 3, Issue 3, Sep- Dec (2012) IAEME
C(1) Tp(2) Ri(3) Tm(4) CxTp(5) CxTm(6) TpxTm(7) A11 A12 A13 A2
535
International Journal of Mechanical Engineering and Technology (IJMET), ISSN 0976
6340(Print), ISSN 0976 6359(Online) Volume 3, Issue 3, Sep- Dec (2012) IAEME
Trials were carried out according to the various combinations of parameters displayed
in the Orthogonal Array and the results for surfaces roughness values were recorded. Three
readings were taken on sample1, A11 correspondence to the top surface, A12 correspondence
to the reading along the build direction, A13 correspondence to the reading perpendicular to
the build direction. These were then analyzed to obtain the optimum condition using
MINITAB software. The Data Means plot for main effects and interaction plots for Chemical
1 is shown in Fig.2.The S-N ratio plot for main effects and interaction plots for Chemical 1 is
shown in Fig 3. The experimental data was solved using both Data Means and S-N Ratio. The
condition was S-N Ratio taken was Smaller is Better hence we will be accepting the higher
value as preferred value from the graph where as in means graph lower value will be taken as
preferred value. Results from both Data Means and S-N Ratio give the same optimized
condition. Data Means plot for main effects and interaction plots for Chemical 2 is shown in
Fig.4 and S-N ratio plot for main effects and interactions is shown in Fig.5.
Main Effects Plot (data means) for Means Interaction Plot (data means) for Means
concent rat ion temp 25 55
3 concentration
2.5
70
2.0 2 80
concentr ation 85
1.5 90
1
1.0
Mean of Means
0.5 3 temp
25
70 80 85 90 25 55 2 55
roughness t ime temp
1
2.5
2.0 3 time
5
1.5
2 10
1.0 time
1
0.5
0.2540 0.3302 5 10 70 80 85 90 5 10
Main Effects Plot (data means) for SN ratios Interaction Plot (data means) for SN ratios
25 55
concent ration temp
10 10 concentration
70
5 80
0
concentr ation 85
0
90
Mean of SN ratios
-5 -10
10 temp
-10 25
70 80 85 90 25 55 55
0
temp
roughness time
10
-10
5 10 time
5
0
10
0
time
-5
-10 -10
0.2540 0.3302 5 10 70 80 85 90 5 10
536
International Journal of Mechanical Engineering and Technology (IJMET), ISSN 0976
6340(Print), ISSN 0976 6359(Online) Volume 3, Issue 3, Sep- Dec (2012) IAEME
Main Effects Plot (data means) for Means Interaction Plot (data means) for Means
25 55
concentration temp
2.5 4
concentration
20
2.0
25
2 30
1.5 concentr ation
35
1.0
Mean of Means
0 4
0.5 temp
25
20 25 30 35 25 55 55
2
temp
roughness time
2.5
2.0 4 0
time
1.5 3
6
1.0 2
time
0.5
0
0.2540 0.3302 3 6 20 25 30 35 3 6
Main Effects Plot (data means) for SN ratios Interaction Plot (data means) for SN ratios
concent ration temp 25 55
10 concentration
10 20
5 25
0 concentr ation 30
0 35
Mean of SN ratios
-10
-5
temp
-10 10 25
20 25 30 35 25 55 55
roughness time temp 0
10
-10
5 time
10 3
0
6
0 time
-5
-10
-10
0.2540 0.3302 3 6 20 25 30 35 3 6
The results obtained for surface roughness data from the white light interferometer (WLI)
data are analyzed by using the statistical tool ANOVA. It determines the relative effect of the
individual factors and their interactions on the surface roughness of parts. The analysis by
using ANOVA technique is done analytically. An equation for total variation may be written
as
SS = SS C + SS + SS + SS + SS + SS + SS (1)
T Tp Ri Tm TmxC TpxC TmxTp
where SST is total sum of squares, SSC, SSTp, SSRi, SSTm, are sum of squares for Concentration
C, Temperature Tp, Initial roughness Ri, Time Tm. SSTmxc, SSTpxc, , SSTmXTp are sum of
squares of Concentration-Temperature, Concentration -Time and Time-Temperature
interactions respectively and SSE is sum of square of the error. If T is the sum of all (N)
Surface roughness values, the total sum of squares is given by
N
T 2 (2)
SS T = Fi 2
N
i = 1
N Ci 2
T 2
SS C = (3)
i=1 N Ci N
where, N is the number of levels of Concentration factor, Ci and NCi are the sum and number
of observations respectively under ith level. Similarly, sum of squares of other five factors can
also be calculated. Sum of squares of interaction of C and Tm is given by
n ( CXTm ) T 2 2
= i
SS SS SS (4)
C Tm
i =1
CXTm
N N
( CXTm )i
where (C xTm)i and N(CXTm)i are the sum and number of observations (surface roughness)
respectively under ith condition of the combinations of factors C and Tm and n is the number
of possible combinations of the interacting factors C and Tm. Similarly, the sum of squares
for other two interactions can also be found out. The results obtained from ANOVA for
chemical 1 and chemical 2 are given in Table 4 and Table 5 respectively.
538
International Journal of Mechanical Engineering and Technology (IJMET), ISSN 0976
6340(Print), ISSN 0976 6359(Online) Volume 3, Issue 3, Sep- Dec (2012) IAEME
539
International Journal of Mechanical Engineering and Technology (IJMET), ISSN 0976
6340(Print), ISSN 0976 6359(Online)
(Online) Volume 3, Issue 3, Sep-
Sep Dec (2012) IAEME
IV. RESULTS
All the test sample from group 1 to 16 having average minimum and maximum
roughness 5.56 micron and 6.67 micron respectively had experienced reduced roughness
value after the chemical treatment.
treatment The minimum average roughness observed is equal to
0.1755 micron and maximum average roughness equal to 3.47 micron for Chemical 1 and
average minimum 0.134 micron and maximum 3.58 micron for chemical 2. Fig.6 shows
effect of chemical treatment on average roughness value for chemical 1 and chemical 2.
The roughness
ness values are analyzed on the basis on DOE and ANOVA for both the
chemicals. Following are the detailed explanation of results:
results:-
4.1 CHEMICAL 1
From the ANOVA table (Table 4) we find that for chemical 1 the most important
factor is concentration contributing 65.9%, concentration-temperature
temperature interaction is the
second most important factor contributing 11.16% followed by initial roughness 7.3% time of
exposure is the least significant factor. Fig.2 and Fig.3 also display the similar results.
7
6
5
4
3
2
1
0
15 16
12 13 14
7 8 9 10 11
Ra(initial)
5 6
Ra(acetone)
4
Ra(MEK)
1 2 3
Ra(initial)
Ra(acetone)
Ra(MEK)
540
International Journal of Mechanical Engineering and Technology (IJMET), ISSN 0976
6340(Print), ISSN 0976 6359(Online) Volume 3, Issue 3, Sep- Dec (2012) IAEME
4.2 CHEMICAL 2
Table 6 Results for optimum condition and roughness values at optimum levels.
Tp(C) 55 25 55 55
Tm (Time of exposure 5 10 3 6
in Min.)
Further a comparison is made between results obtained with chemical 1 and chemical 2 on
four different samples which were manufactured on Stratasyss Dimension SST 1200 FDM
machine. Sample 1 is a cube as shown in Fig.1(a) & Fig.1(b), sample 2 is shown in Fig.1(c),
sample 3 is shown in Fig.1(d) and sample 4 is shown in Fig.1(e).The results with
comparisons for chemical 1 and chemical 2 is shown in Table 7. Fig.7 shows the original
part where as Fig.8 and Fig.9 show the parts treated with Chemical 1 and Chemical 2
respectively at their optimum conditions.
541
International Journal of Mechanical Engineering and Technology (IJMET), ISSN 0976
6340(Print), ISSN 0976 6359(Online) Volume 3, Issue 3, Sep- Dec (2012) IAEME
Sample 4 3.3 Very 1.5 Less than 2.12 Glossy 3 Less than
smooth 0.5% 1%
The original part as shown in Fig.7 is made of ABS material in white & blue color, but the
chemically treated parts as shown in Fig.8 and Fig.9 are made of ABS material in white
color.
The size of the specimen was measured before and after the chemical treatment process in
order to account for the variation in dimensions due to chemical treatment process. Base
lengths were taken as l1 and l2, height of the specimen was taken as H. Readings were taken
by ACCURATE SPECTRA co-ordinate measuring machine. The results show less than 1%
deviation. Detailed results are shown in Table 8.
542
International Journal of Mechanical Engineering and Technology (IJMET), ISSN 0976
6340(Print), ISSN 0976 6359(Online) Volume 3, Issue 3, Sep- Dec (2012) IAEME
The cost of the chemical treatment process is compared with the commercially available
systems and the same is listed in the Table 9. It is observed from the Table 9 that the
proposed system is economical to use and has very small setup cost as compared to the
commercial system available in the market.
Table 9 Cost comparison of proposed chemical treatment process with other available
commercial system.
Sr. Capital Cost, Depreciation, Raw Power Labour
No INR(Approx.) INR Material Consumption Cost
cost per cost per per
part*, hour, INR hour,
INR INR
1 Acetone
10000 2.78 per day 32 6 50
process
2 MEK
10000 2.78 per day 21 6 50
process
3 Commercial
35, 00000 959 per day 42 15 50
system
* for part size 50x50x25 mm
V.CONCLUSION
In this paper the surface roughness of FDM prototype parts is addressed, the parameters that
have significant effect on the surface roughness (Ra) value in the chemical treatment process have
been identified. The chemical treatment process is optimized in terms of solution concentration, time
of exposure, initial roughness and temperature of the chemical bath using Design of Experiments and
ANOVA. Two different chemical were taken, i.e. Dimethyl ketone (Acetone) and Methyl ethyl ketone
(MEK), in case of Acetone it was observed that the solution concentration, concentration-temperature
interaction and the initial roughness are the most significant factors. For Methyl ethyl ketone chemical
treatment process, it was observed that the concentration, concentration - temperature interaction and
concentration-time interaction are the most important factors, surprisingly for MEK the initial
roughness and time of exposure have negligible effect on the process. The process was applied for
simple parts to complex free form parts. The optimum levels for the parameters for chemical
treatment process are found out which shows drastic improvement in surface finish. The appearance
of the finished parts is comparable to plastic moulded parts, the parts have glossy finish and the
maximum curing time is about 2 to 4 hours. The process is very much economical compared to other
commercial systems available in the market. Further studies can be carried out to commercialize this
process to make it available in the market at an affordable price.
543
International Journal of Mechanical Engineering and Technology (IJMET), ISSN 0976
6340(Print), ISSN 0976 6359(Online) Volume 3, Issue 3, Sep- Dec (2012) IAEME
REFERENCES
544