Professional Documents
Culture Documents
net/publication/235275194
CITATIONS READS
36 144
1 author:
John Kechagias
University of Thessaly
94 PUBLICATIONS 596 CITATIONS
SEE PROFILE
Some of the authors of this publication are also working on these related projects:
Special Issue "Sustainable Manufacturing Processes and Machine Tool Technology" View project
Time prediction algorithm for raster Rapid Prototyping systems View project
All content following this page was uploaded by John Kechagias on 24 March 2020.
Abstract
Purpose – To investigate laminated object manufacturing (LOM) process quality, using a design of experiments approach.
Design/methodology/approach – The quality characteristics measured were in-plane dimensional accuracy, actual layer thickness (ALT), and mean time
per layer. The process parameters tested were nominal layer thickness (LT), heater temperature (HT), platform retract (PR), heater speed (HS), laser speed (LS),
feeder speed (FS) and platform speed (PS). A typical test part has been used, and matrix experiments were carried out based on Taguchi design. Optimal process
parameter values were identified and finally, additive and regression models were applied to the experimental results and tested using evaluation experiments.
Findings – The statistical analysis of the experimental results shows that error in X direction was higher than error in Y direction. Dimensional accuracy
in X direction depends mainly on the HS (89 percent) and HT (5 percent), and in Y direction on HS (50 percent), LT (31 percent), LS (9 percent), PS
(6 percent), and HT (3 percent). On the other hand, ALT depends mainly on the nominal ALT (96 percent), HS (2 percent), HT (1 percent), and PR
(1 percent). Finally, mean time per layer depends mainly on HS (59 percent), LS (17 percent), FS (17 percent), and PS (4 percent).
Research limitations/implications – Future work should involve extensive matrix experiments using parameters such as dimensions of test part
(Xmax, Ymax, Zmax), hatch spacing in X and Y directions, and delay time between sequential layers.
Practical implications – Using the extracted models, the quality of LOM parts can be predicted and appropriate process parameter values selected. This
means minimization of post processing time, easier disengagement between supporting frame and part, easier decubing, process optimization, less finishing
and satisfactory final LOM parts or tools. Also, ALT prediction and mean time per layer analysis could be used to improve LOM build time predictions.
Originality/value – The above analysis is useful for LOM users when predictions of part quality, paper consumption, and build time are needed. This
methodology could be easily applied to different materials and initial conditions for optimisation of other LOM-type processes.
316
Investigation of LOM process quality Rapid Prototyping Journal
John Kechagias Volume 13 · Number 5 · 2007 · 316 –323
Figure 1 LOM process The process parameters tested were nominal layer thickness
(LT), HT, platform retract (PR), HS, LS, feeder speed (FS)
Laser Speed (mm/sec) and platform speed (PS). Two values of each process
Heater parameter (levels) were used according to the L8 (27)
(Speed mm/sec) standard orthogonal matrix (Phadke, 1989). Taguchi design
(Temperature °C) demands each one of the matrix experiments to be executed
in order to have proper analysis of the results. Thus,
Platform preliminary experiments were executed to define the final
Retract (mm) experimental area. After that, a typical test part was used, and
Part experiments were carried out. Then an analysis of means and
an analysis of variances (ANOVA) were carried out in order to
characterize the LOM process parameters weightings with
Platform respect to the dimensional accuracy, actual layer thickness
(ALT) and mean time per layer. In addition, additive and
Layer regression models were extracted using the experimental
Thickness (mm) Feeder Speed
(mm/sec) results. It was expected that using the extracted regression
Platform Speed models, the quality of LOM parts as well as the ALT could be
(mm/sec) predicted and appropriate process parameter values selected.
load caused by the heated roller during bonding, materials Design of experiments
properties (paper and glue), laser beam compensation, and
moisture exchange. Pham and Gault (1998) noted in-plane Conditions of experiments
dimensional accuracy of LOM parts of about ^ 0.127 mm. The rapid prototyping machine used was a LOM 1015 from
A number of researchers have tried to investigate and improve Helisys. The paper used was LPH042 and LPH080 with a LT
of 0.1 and 0.2 mm, respectively. The machine uses a 25 W
the quality of LOM parts proposing mathematical models
CO2 laser. The percentage of laser power was 5.5 and
(analytical or statistical) for good bonding of laminates (Pak and
11 percent for LPH042 and LPH080, respectively. Moreover,
Nisnevich, 1994; Reece and Styger, 1995; Sonmez and Hahn,
at the PR value of 1.27 mm, the gap between platform and
1998; Flach et al., 1998), or mathematical models for predicting heater was adjusted to 0.508 mm.
quality characteristics when varying process or geometry Additionally, in each experiment, 20 layers of paper were
parameters (Reeves and Cobb, 1996; Chryssolouris et al., set on the platform before starting the process. This resulted
1999; Paul and Voorakarnam, 2001; Chryssolouris et al., 2003; in obtaining the desirable temperature of the surface. In
Kechagias, 2007). Besides, many researchers have proposed addition, chamber temperature was kept constant. In each
modifications to the LOM process for improving quality experiment, two dog bones were built as is described in
characteristics (Cho et al., 2000; Ahn et al., 2002; Yu et al., previous work (Chryssolouris et al., 2003). Cubes produced
2003; Liao et al., 2006), as well as optimisation of part build by hatching (20 £ 20 £ 7 mm3) were used as test parts for
orientation (Campbell et al., 2002). in-plane measurements (Figure 2). In each experiment, a
Wang et al. (1996) investigated the in-plane processing mean value of five cubes measurements were used for error in
accuracy of the slicing solid manufacturing process, which is the X and Y directions correspondingly.
very similar WTO the LOM process. They introduced an Also, for each experiment total build time (Ttot), actual height
integrated method that combined orthogonal experimental (Zmax), and total number of layers (Nact) were measured. Then,
design and analysis, and neural networks analysis to determine ALTand mean time per layer (Tlayer) were calculated as follows:
the optimal processing conditions. Park et al. (2000) studied the Z max
precision and accuracy of the LOM process and the stability of ALT ¼ ð1Þ
N act
LOM parts. Cross hatching size, left and right heater margin,
heater speed (HS), heater temperature (HT), laser power, and and:
T tot
laser speed (LS) were found to affect cutting accuracy and T layer ¼ ð2Þ
N act
bonding efficiency. The experiments conducted showed that
mean values and standard deviations of each dimension were
Figure 2 Cubes produced by hatching
different. The most affected dimension was Z, followed by X
and then Y. Additionally, the Z dimension displayed the largest
Hx=20mm
instability due mainly the moisture absorption.
Preliminary experiments (Kechagias et al., 1997) showed
differences in quality characteristic measures as well as total
build times and total number of final layers for the same part
when different sets of process parameters were used. Having
taken into account the above literature and the preliminary
work, it was concluded that the LOM process needs further
analysis and investigation in order to improve the final part Hy=20mm
quality. Thus, an orthogonal experiment was designed to
determine the quality characteristics of LOM parts according to
the process parameters used.
317
Investigation of LOM process quality Rapid Prototyping Journal
John Kechagias Volume 13 · Number 5 · 2007 · 316 –323
318
Investigation of LOM process quality Rapid Prototyping Journal
John Kechagias Volume 13 · Number 5 · 2007 · 316 –323
Results analysis and discussion Thus, the width of the two-standard-deviation confidence
interval, which is an approximately 95 percent confidence
Effects of process parameters interval for each estimated effect, is:
The mean values of each parameter (S/N ratios of seven
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
parameters according to each level) are reported in Table VI 1
for each quality characteristic. For example: ex ¼ ^2 · · 0:0004 ¼ ^28 mm ð5Þ
2
1
mLTðLevel 1Þ Ex ¼ ðh1 þ h2 þ h3 þ h4 Þ and:
4 rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1 1
¼ ð20:17 2 0:225 2 0:188 2 0:23Þ ey ¼ ^2 · · 0:00026 ¼ ^23 mm ð6Þ
4 2
319
Investigation of LOM process quality Rapid Prototyping Journal
John Kechagias Volume 13 · Number 5 · 2007 · 316 –323
–170 LT
HT
Ex (µm) –190 PR
HS
–210 LS
FS
–230 PS
–250
1 2 1 2 1 2 1 2 1 2 1 2 1 2
Levels
30
LT
15 HT
PR
Ey (µm)
0
HS
–15 LS
–30 FS
PS
–45
–60
1 2 1 2 1 2 1 2 1 2 1 2 1 2
Levels
PR
160 HS
145 LS
FS
130
PS
115
100
1 2 1 2 1 2 1 2 1 2 1 2 1 2
Levels
E x ¼ 20:2 þ mHTj þ 0:2 þ ðmHSl þ 0:2Þ ^ 0:028 ð8Þ ALT ¼ 0:166 þ ðmLTi 2 0:166Þ þ mHTj 2 0:166
ð10Þ
þ ðmPRk þ 0:166Þ þ ðmHSl þ 0:166Þ ^ 0:017
E y ¼ 20:008 þ ðmLTi þ 0:008Þ þ mHTj þ 0:008 where: Ex, Ey, are X and Y error predictions; ALT is ALT
prediction; 2 0.2 is the overall mean error in X; 20.008 is the
þ ðmHSl þ 0:008Þ þ ðmLSm þ 0:008Þ ð9Þ
overall mean error in Y, 0.166 is the overall mean of ALT;
þ ðmPSk þ 0:008Þ ^ 0:023 and, mLTi , mHTj , mPRk , mHSl , mLSm , mPSk are the mean values
320
Investigation of LOM process quality Rapid Prototyping Journal
John Kechagias Volume 13 · Number 5 · 2007 · 316 –323
60
LT
58
HT
Tlayer (secs) PR
56 HS
LS
FS
54
PS
52
1 2 1 2 1 2 1 2 1 2 1 2 1 2
Levels
Table VII ANOVA table for dimensional accuracy in X direction having the levels 1 or 2 represented in Tables VI, according to
its quality characteristic, respectively.
DOF Sum of squares Mean square F Percentage
LT 1 0.0001 0.0001 1.29 1 Regression model
HT 1 0.0004 0.0004 5.13 5 Engineering experimenters often wish to find the conditions
PR 1 0.0000 0.0000 0.24 0 under which a certain process attains the optimal results. That
HS 1 0.0065 0.0065 84.94 89 is, they want to determine the levels of the design parameters
LS 1 0.0001 0.0001 1.68 2 at which the response reaches its optimum. The optimum
FS 1 0.0000 0.0000 0.11 0 could be either a maximum or a minimum of a function of the
PS 1 0.0001 0.0001 1.68 2 design parameters. One of the methodologies for obtaining
Error 0 0.0000 the optimum is the response surface technique. Response
Total 7 0.0073 surface methodology is a collection of statistical and
(Error) (5) (0.0004) (0.00008) mathematical methods that are useful for the modeling and
analyzing of engineering problems. In this technique, the
main objective is to optimize the response surface that is
Table VIII ANOVA table for dimensional accuracy in Y direction influenced by various process parameters. Response surface
methodology also quantifies the relationship between the
DOF Sum of squares Mean square F Percentage
process parameters and the obtained response surface.
LT 1 0.0088 0.0088 68.51 31 Assuming that the process parameters are continuous and
HT 1 0.0009 0.0009 7.049 3 controllable in the experiments, the response can be expressed
PR 1 0.0000 0.0000 0.024 0 as follows:
HS 1 0.0140 0.0140 109.5 50
LS 1 0.0026 0.0026 20.51 9 Y ¼ b1 þ b2 LT þ b3 HT þ b4 PR þ b5 HS þ b6 LS
FS 1 0.0003 0.0003 1.976 1
PS 1 0.0017 0.0017 12.9 6 þ b7 FS þ b8 PS ^ e ð11Þ
Error 0 0.0000
Total 7 0.0282 where, Y is the response of each quality characteristic; and bi,
(Error) (2) (0.00026) (0.00013) coefficients, which should be determined. In general,
equation (11) can be written in a matrix form:
Table IX ANOVA table for actual layer thickness Table X ANOVA table for mean time per layer
DOF Sum of squares Mean square F Percentage DOF Sum of squares Mean square F percentage
LT 1 0.0167 0.0167 352.52 96 LT 1 2.4865 2.4865 2.97 2
HT 1 0.0002 0.0002 3.41 1 HT 1 0.0145 0.0145 0.02 0
PR 1 0.0001 0.0001 2.37 1 PR 1 0.0084 0.0084 0.01 0
HS 1 0.0003 0.0003 6.58 2 HS 1 68.0945 68.0945 81.41 59
LS 1 0.0000 0.0000 0.67 0 LS 1 19.5313 19.5313 23.35 17
FS 1 0.0001 0.0001 1.27 0 FS 1 19.2821 19.2821 23.05 17
PS 1 0.0001 0.0001 1.05 0 PS 1 5.1520 5.1520 6.16 4
Error 0 0.0000 Error 0 0.0000
Total 7 0.0175 Total 7 114.5692
(Error) (3) (0.0001) (0.00005) (Error) (3) (2.5094) (0.83645)
321
Investigation of LOM process quality Rapid Prototyping Journal
John Kechagias Volume 13 · Number 5 · 2007 · 316 –323
322
Investigation of LOM process quality Rapid Prototyping Journal
John Kechagias Volume 13 · Number 5 · 2007 · 316 –323
Campbell, R.I., Martorelli, M. and Lee, H.S. (2002), “Surface Park, J., Tari, M.J. and Hahn, H.T. (2000), “Characterization
roughness visualisation for rapid prototyping models”, of the laminated object manufacturing (LOM) process”,
Computer Aided Design, Vol. 34 No. 10, pp. 717-25. Rapid Prototyping Journal, Vol. 6 No. 1, pp. 36-49.
Cho, I., Lee, K., Choi, W. and Song, Y. (2000), Paul, B. and Voorakarnam, V. (2001), “Effect of layer
“Development of a new type rapid prototyping system”, thickness and orientation angle on surface roughness in
International Journal of Machine Tools & Manufacture, laminated object manufacturing”, SME Journal of
Vol. 40, pp. 1813-29. Manufacturing Processes, Vol. 3 No. 2, pp. 94-101.
Chryssolouris, G., Kechagias, J., Kotselis, J., Mourtzis, D. Phadke, M.S. (1989), Quality Engineering Using Robust Design,
and Zannis, S. (1999), “Surface roughness modeling of the Prentice-Hall, Englewood Cliffs, NJ.
laminated object manufacturing parts”, in Campbell, R.I. Pham, D.T. and Gault, R.S. (1998), “A comparison of rapid
(Ed.), Proceedings of the 8th European Conference on Rapid prototyping technologies”, International Journal of Machine
Prototyping & Manufacturing, University of Nottingham, Tools & Manufacture, Vol. 38, pp. 1257-87.
Nottingham, pp. 141-52. Reece, S.R. and Styger, L.J. (1995), “Laminated object
Chryssolouris, G., Kechagias, J., Moustakas, P. and Koutras, manufacturing: process practice and research experience”,
E. (2003), “An experimental investigation of the tensile paper presented at First National Conference on Rapid
strength of parts produced by laminated object Prototyping & Tooling Research, Buckinghamshire College,
manufacturing (LOM) process”, CIRP Journal of UK, pp. 101-10.
Manufacturing Systems, Vol. 32 No. 5, pp. 319-22. Reeves, P.E. and Cobb, R.C. (1996), “Surface deviation
Flach, L., Jacobs, M., Klosterman, D. and Chartoff, R. modelling of LMT processes: a comparative analysis”, in
(1998), “Simulation of laminated object manufacturing Dickens, P.M. (Ed.), Proceedings of the 5th European
with variation of process parameters”, Solid Freeform Conference on Rapid Prototyping & Manufacturing, Helsinki,
Fabrication Symposium Proceedings, The University of Finland, pp. 59-76.
Texas at Austin, Texas, pp. 407-16. Sonmez, F. and Hahn, T. (1998), “Thermomechanical
Jacobs, P.F. (1996), Stereolithography and Other RP&M analysis of the laminated object manufacturing process”,
Technologies: From Rapid Prototyping to Rapid Tooling, Rapid Prototyping Journal, Vol. 4 No. 1, pp. 26-36.
Society of Manufacturing Engineers in Co-operation with the Wang, W., Feng, W., Yan, Y. and Fuh, J. (1996),
Rapid Prototyping Association of SME, ASME Press, “Experimental design and analysis of in-plane processing
New York, NY. accuracy for SSM process”, Materials & Design, Vol. 17
Kechagias, J. (2007), “An experimental investigation of the No. 3, pp. 159-66.
surface roughness of parts produced by LOM process”, Wring, T. (1994), “Review of the applications possible with a
Rapid Prototyping Journal, Vol. 13 No. 1, pp. 17-22. Helisys LOM model”, in Dickens, P.M. (Ed.) paper
Kechagias, J., Anagnostopoulos, V., Zervos, S. and presented at 3rd European Conference on Rapid
Chryssolouris, G. (1997), “Estimation of build times in Prototyping & Manufacturing, University of Nottingham,
rapid prototyping processes”, in Dickens, P.M. (Ed.),
Nottingham, pp. 87-102.
Proceedings of the 6th European Conference on Rapid
Yu, G., Ding, Y., Li, D. and Tang, Y. (2003), “A low cost
Prototyping & Manufacturing, University of Nottingham,
cutter-based paper lamination rapid prototyping system”,
Nottingham, pp. 137-48.
International Journal of Machine Tools & Manufacture,
Kechagias, J., Maropoulos, S. and Karagiannis, S. (2004),
Vol. 43, pp. 1079-86.
“Process build-time estimator algorithm for laminated
object manufacturing”, Rapid Prototyping Journal, Vol. 10
No. 5, pp. 297-304. Further reading
Kruth, J.P. (1991), “Material incress manufacturing by rapid
prototyping techniques”, CIRP Annals, Vol. 40 No. 2, Box, G.E.P., Hunter, W.G. and Hunter, J.S. (1978), Statistics
pp. 603-14. for Experimenters, Wiley, New York, NY.
Liao, Y.S., Li, H.C. and Chiu, Y.Y. (2006), “Study of D3500-76 (1986), Annual Book of ASTM Standards,
laminated object manufacturing with separately applied Standards Methods of Testing Plywood in Tension.
heating and pressing”, International Journal of Advanced Peace, G.S. (1993), Taguchi Methods: A Hands-on Approach,
Manufacturing Technology, Vol. 27, pp. 703-7. Addison-Wesley Publishing Company, Reading, MA.
Pak, S.S. and Nisnevich, G. (1994), “Interlaminate strength
and processing efficiency improvements in laminated object
Corresponding author
manufacturing”, Proceedings of the 5th International
Conference on Rapid Prototyping, Dayton, Ohio, pp. 171-80. John Kechagias can be contacted at: jkechag@teilar.gr
323