 Open Access
 Total Downloads : 503
 Authors : M. Maruthi Prasad, O. Y. Venkatasubbareddy, N. Jaya Krishna
 Paper ID : IJERTV3IS120483
 Volume & Issue : Volume 03, Issue 12 (December 2014)
 Published (First Online): 22122014
 ISSN (Online) : 22780181
 Publisher Name : IJERT
 License: This work is licensed under a Creative Commons Attribution 4.0 International License
Improving the Surface Roughness of FDM Parts By using Hybrid Methods
M. Maruthi Prasad
Associate Professor, Department of Mechanical Engineering,
Annamacharya Institute of Technology and Sciences, Rajampet516126
India

Y. Venkatasubbareddy
Assistant Professor, Department of Mechanical Engineering,
Annamacharya Institute of Technology and Sciences, Rajampet516126
India
N. Jaya Krishna
Assistant Professor, Department of Mechanical Engineering,
Annamacharya Institute of Technology and Sciences, Rajampet516126 India
AbstractFused deposition modeling (FDM) is one of the rapid prototyping process that uses the plastics materials such as ABS (Acrylonitrile butadiene styrene) in the semi molten state to produce the products directly from CAD model. FDM is an additive manufacturing process and the prototypes are made by layer by layer addition of semimolten plastic material onto the platform from bottom to top. . The design investigates the effect of the process parameters layer thickness, raster width, raster angle and air gap that influences the surface roughness of the part produced by the process of Fused Deposition Modeling. Hence, the Optimization of these process parameters of FDM is able to make the system more specific and repeatable and such progression can guide to use of FDM in rapid manufacturing applications rather than only producing prototypes. The novel ABS M30 biomedical material was used in this research work to build parts. The effect of process parameters on response is studied via Response surface methodology (RSM). It is used to calculate the regression coefficients and the function is made with the significant factors. Then optimization of process parameters to be done through Genetic algorithm in order to minimize the surface roughness of the part.
KeywordsRapid prototyping, fused deposition modelling, Response surface Methodology, Genetic algorithm

INTRODUCTION
The competition in the world market is growing tremendously and it is the vital need to make sure that the new products reach the market as soon as possible. Rapid Prototyping (RP) is an additive manufacturing technology that automatically builds functional assemblies using CAD model of the part. These "3D printers" allow designers to quickly create tangible prototypes of their designs, rather than just twodimensional pictures. In general, RP process includes five basic steps to build a part model automatically:
(a) Create a CAD model of the design (b) Convert the CAD model to STL format (c) Slice the STL file into thin cross sectional layers (d) Construct the model one layer atop
another (e) Clean and finish the model. Surface roughness is the key property of RP build parts. Surface finish is considered as a vital feature and parts must be prepared in line with the product finishing specifications. However, there are certain materials for which it is difficult to meet the specifications, thus an optimum and achievable choice of material and application conditions is essential. Consequently, the operating conditions that optimally suit a material must be employed and their characteristics have to be taken into account.
The surface finish of parts obtained through these manufacturing processes is important, especially in cases where the components are in contact with other elements or materials in their service life. For example building moulds to produce components by means of Solid Free Form Manufacturing Processes, or cases of other functional components where their surface characteristics will have a considerable effect on their mechanical properties such as fatigue, wear, and corrosion. Therefore, it is important to have prior knowledge, by means of conceptual models, of the manufacturing process parameters that allow the user to predict the surface finish of manufactured prototypes.
Fused Deposition Modeling (FDM) is a leading RP technology that is used for fabricating solid prototypes in various materials directly from a computeraided design (CAD) data. The quality and the strength of the FDM build parts are dependent essentially on the process parameters. In order to understand the performance and the behavior of FDM build parts, the influence of the process parameters on outcome quality of the build parts must be studied. Earlier studies (Mahapatra, et al, 2009), (Ahn, et al, 2002) have reported that FDM parameters such as layer thickness, air gap, raster width, and raster orientation were significantly impacting the quality characteristics of build parts. The FDM systems available in the market are different in their build speed, build volume, range of parameter settings and build materials (Masood, et al, 2010). In relevant empirical
studies, parametric optimization was used to develop the quality characteristics of FDM parts or the process performance where the number of FDM process parameters were studied and optimized. For instance, (Lee, et al, 2005) and (Laeng, et al, 2006) investigated the elasticity performance of ABS material. Similarly, (Anitha, et al, 2001) optimized the FDM process parameters improving the surface roughness of build parts, while (Gregorian, et al., 2001), (Sood, et al., 2010) have looked into the dimensional accuracy of FDM parts

MATERIALS AND METHODS
The material used for the present investigation is ABS M30 plastic. The chemical composition of the material is given in the Table 1.
Table 1: chemical composition of the material
S.NO
PROPERTIES
SPECIFICATIONS
1.
STRUCTURE
AMORPHOUS
2.
SPECIFIC DENSITY
1.05
3.
WATER ABSORPTION RATE(%)
0.27
4.
ELONGATION (%)
20
5.
TENSILE STRENGTH (MPA)
29.64
6.
COMPRESSION STRENGTH (MPA)
62.05
7.
FLEXURAL STRENGTH (MPA)
63.43
8.
FLEXURAL MODULUS (MPA)
2068.48
9.
IMPACT (JOULES)
8.94
10.
HARDNESS
R110
11.
ULTRASONIC WELDING
EXCELLENT
12.
MACHINING
GOOD
13.
MIN. UTILIZATION TEMP (DEG.C)
40
14.
MAX. UTILIZATION TEMP (DEG.C)
90
15.
MELTING POINT (DEG.C)
105
16.
COEFFICIENT OF EXPANSION
0.000053
17.
ARC RESISTANCE
80
18.
DIELECTRIC STRENGTH (KV/MM)
16
19.
TRANSPARENCY
TRANSLUCENT
20.
UV RESISTANCE
POOR
21.
CHEMICAL RESISTANCE
GOOD
22.
BONDING
EXCELLENT

Response Surface Methodology
RSM is a combination of experimental designs and statistical techniques for empirical model building and optimization. RSM was originally dveloped for the model fitting of physical experiments by Box and Draper and later extended to other fields. RSM is very useful for modeling and analysis where a response of interest is influenced by several variables and the objective is to optimize this response. By conducting experiments and the posterior application of regression analysis a model of the response variable of interest is obtained. The real relationship between the response and the independent variables is unknown. For that reason, the first step in RSM is to find an approximation of the true functional relationship between the response and the independent variables.
In general, the relationship between the response y and independent variables 1, 2 k is
Y= f (1, 2k) + (1)
Where includes effects such as measurement error on the response, background noise, the effect of other variables, and so on. Usually is treated as a statistical error, often assuming it to have a normal distribution with mean zero and variance 2. Then,
E(y) = =E [f (1, 2k)] + E () =f (1, 2..k) (2)
The variables 1, 1., k in equation (2) are usually called the natural variables, because they are expressed in the natural units of measurement, such as degrees Celsius, pounds per square inch, etc. in much RSM work it is convenient to transform the natural variables to coded variables x1, x2, , xn, which are usually defined to be dimensionless with mean zero and the same standard deviation. In terms of the coded variables the response function will be written as f (x1, x2 xn); is called response surface. In most of the RSM problems the form of relationship between the response and the independent variable is unknown. Thus the first step in RSM is to find a suitable approximation for the true functional relationship between Y and set of independent variables employed. Usually a second order model is utilized in RSM.
= 0+kj=1 jXj+kj=1 jjXj2 +kj=2 ijXiXj (3)
The coefficients, used in the above model can be calculated by means of using least squares technique. The second order model is normally used when the response function is not known or nonlinear.

Optimization by Genetic Algorithm
Genetic algorithms are search procedures that emulate the process of evolution in nature, as Darwins theory explains. They were proposed by Holland in the early 70s, and developed later on by Goldberg Based on survival of the fittest and reproduction, they search for new solutions in each stage or generation. A solution is coded in a chromosome. The most frequently used way of encoding is binary codification, although other systems like real value numbers have been used. The solution of the optimization
problem with GA begins with a set of potential solutions or chromosomes that are randomly generated and selected the entire set of chromosomes comprises a population. The chromosomes evolve during several iterations or generations. New generations are generated using a crossover and mutation technique. Crossover involves splitting two chromosomes and then combining one half of each chromosome with other pair. Mutation involves flipping a single bit of a chromosome. The chromosomes are then evaluated using a certain fitness criteria and the best ones are kept while the others are discarded. This process is repeated until one chromosome has the best fitness and thus is taken as the best solution to the problem.


EXPERIMENTAL PLAN
A trial run was performed in which a series of samples were built on the FDM machine using ABS M30 material. The machine is equipped with Insight software that assists the user to adjust the variable parameters in building part specification. Principally, the FDM variables are considered as four groups of operating parameters, as follows; FDM build specification, FDM environment/machine, and material specification. The full factor experiment was obtained to develop the experimentation plan for five parameters and three levels, considering the highest number of experimentation runs for the specified number of runs and levels in order to optimize the maximum parameters combinations. In this study, Full factor experiment, Box Behnken design (three levelsfive factors) has been selected initially according to the number of FDM variable parameters and number of settings or levels. The dimensions of the samples were selected according to specimen as shown in figure 1.
Figure 1: specimen
Figure 2: measuring of surface roughness by using talysuf
With the help of CATIA V5 software 3D solid model of prototype is modeled and are the converted to STL file. STL file is imported to FDM software (Insight). Now, control factors listed in Table3 are set as per shown experiment plan (Table 2). Four parts per experiment are fabricated by the use of FDM Vantage SE machine. ABS m30 is the material used for fabricating the designed part. The surface roughness is taken to be the representative value respectively. Mitutoyo Talysurf is used to measure the surface roughness
Table 2: Fixed factors for FDM machine
FIXED FACTORS
Factor
Value
Part fill style
Perimeter Raster
Counter width (mm)
0.464
Part interior style
Solid Normal
Visible surface
Normal raster
XY&Z shrink factor
1.0038
Perimeter to raster air gap (mm)
0
Table 3: Control factors for FDM machine
CONTROL FACTORS
Factor
Symbol
Levels
1
0
1
Layer thickness
(mm)
A
0.127
0.178*
0.254
Orientation
(Â°)
B
0
15
30
Raster angle
(Â°)
C
0
30
60
Raster width
(Â°)
D
0.4064
0.4564
0.5064
Air gap (mm)
E
0
0.004
0.008
*modified centre level value
S.NO
A
B
C
D
E
Surface
roughne ss
1
1
0
0
0
1
4.86
2
0
0
1
0
1
4.417
3
1
1
0
0
0
9.178
4
1
0
1
0
0
9.883
5
1
0
0
0
1
4.9418
6
0
0
1
1
0
4.9932
7
1
0
0
1
0
4.2356
8
0
1
0
0
1
4.8067
9
1
0
1
0
0
4.1415
10
0
0
0
1
1
4.9056
11
1
1
0
0
0
8.8538
12
0
0
1
1
0
4.6988
13
0
0
0
0
0
5.005
14
1
1
0
0
0
4.1372
15
1
0
1
0
0
9.419
16
0
1
0
1
0
6.8732
17
1
0
0
0
1
5.5376
18
1
0
1
0
0
4.0672
19
0
1
1
0
0
4.4592
20
0
0
0
1
1
4.076
21
0
0
1
1
0
5.5603
Table 4: Experimental plan based on RSM
22
0
0
0
0
0
5.0454
23
0
1
0
0
1
5.0544
24
0
1
0
0
1
340648
25
0
1
0
1
0
4.173
26
0
0
1
0
1
4.5032
27
0
0
0
0
0
4.2956
28
1
0
0
0
1
5.1026
29
0
1
1
0
0
4.68
30
0
0
1
1
0
4.8993
31
1
0
0
1
0
4.4412
32
0
1
0
1
0
5.331
33
0
1
1
0
0
4.554
34
0
0
0
0
0
5.1948
35
1
0
0
1
0
11.046
36
0
0
1
0
1
8.728
37
0
0
0
1
1
6.217
38
0
1
0
1
0
5.6863
39
1
0
0
1
0
5.7563
40
0
0
0
0
0
5.1333
41
0
1
0
0
1
4.195
42
1
1
0
0
0
4.5153
43
0
1
1
0
0
9.7465
44
0
0
1
0
1
6.4857
45
0
0
0
1
1
5.4303
46
0
0
0
0
0
6.5212
Term Coef SE Coef T P
Constant 0.098288 0.01866 15.268 0.000
A 0.048334 0.00497 4.075 0.043
B 0.024167 0.004497 3.982 0.035
C 0.0051843 0.00497 0.486 0.031
D 0.055528 0.00497 1.457 0.157
E 0.013461 0.00497 0.299 0.767
A*A 0.002969 0.01608 0.049 0.961
B*B 0.037414 0.01608 3.614 0.544
C*C 0.069676 0.01608 4.144 0.263
D*D 0.044359 0.01608 0.729 0.473
E*E 0.037692 0.01608 0.619 0.541
A*B 0.023335 0.01809 1.037 0.441
A*C 0.004164 0.01809 0.463 0.647
A*D 0.066667 0.01809 1.741 0.465
A*E 0.031666 0.01809 0.352 0.728
B*C 0.0233333 0.01809 1.483 0.151
B*D 0.070000 0.01809 0.778 0.044
B*E 0.033335 0.01809 0.371 0.014
C*D 0.058530 0.01809 0.651 0.521
C*E 0.193016 0.01809 2.146 0.052
D*E 0.102698 0.01809 1.142 0.264
Table5.estimated regression coefficients for % change in surface roughness

RESULTS AND DISCUSSIONS
Analysis of the experimental data obtained from Box Behnken design runs is done on minitab16 software by the use of full quadratic response surface model which is given by
Where xi is ith factor and Y is the response.
In the ANOVA table the value of F is checked. Probability of F value is greater than calculated F value due to noise is indicated by P value. The significance of corresponding term is established, if P value is less than
0.05. The value of P must be greater the 0.05 for the lack of fit. An insignificant lack of fit is desired because it is the indication that anything left out of the model is not important and the developed model fits.
Based on analysis of variance (ANOVA) test full quadratic model was found to be suitable for surface roughness with regression pvalue less than 0.05 and lack of fit more than 0.05.
Response Surface Regression: Surface Roughness Vs A, B, C, D, E. The analysis was done using coded units
In the analysis, the factor A, B,C and interaction B*E, B*D are important because their P value is less than 0.05. The coefficient of determination (RSq) which indicates the goodness of fit for the model so the value of RSq = 85.32%, which indicate the high significance of the model. F (surface roughness) = 0.098288 – 0.048334*A + 0.024167*B + 0.0051843*C 0.07*(B*D) +
0.033335*(B*E)
Source
DF
Seq SS
ADj SS
ADjMS
F
P
A
1
249.7
249.7
249.7
2.88
0.109
B
1
427.7
427.7
427.7
4.93
0.041
C
1
1096
1096
1096
12.6
0.003
D
1
3.35
3.35
3.35
0.04
0.847
E
1
39.6
39.6
39.6
0.46
0.508
A*B
1
17.49
17.49
17.49
0.2
0.659
A*C
1
0.37
0.37
0.37
0
0.949
A*D
1
60.78
60.78
60.78
0.7
0.415
A*E
1
90.12
90.12
90.12
1.04
0.323
B*C
1
221.6
221.6
221.6
2.55
0.13
B*D
1
0.88
0.88
0.88
0.01
0.921
B*E
1
5.04
5.04
5.04
0.06
0.813
C*D
1
49.15
49.15
49.15
0.57
0.463
C*E
1
1.61
1.61
1.61
0.02
0.893
D*E
1
11.21
11.21
11.21
0.13
0.724
Error
16
1388.2
1388.2
86.77
Total
31
3663.2
Table6.Analysis of variance for % change in surface roughness
Optimization of Parameters using Genetic Algorithm
Fitness function is given by,
F (surface roughness) = 0.098288 – 0.048334*A + 0.024167*B + 0.0051843*C 0.07*(B*D) +
0.033335*(B*E)
Genetic algorithm tool in MATLAB 2010 is used to optimize the process parameters in coded form.The fitness function is saves in Matlab and is called in the column named fitness function by @fitness_function
FIGURE 2 : FITNESS GRAPH
The optimized results obtained in coded form are:
A= 0.967, B = 0.935, C = 0.309, D = 0.976, E = 0.067
The optimized values of the process parameters in un coded form are:
A= 0.25, B = 290, C = 380, D = 0.5063, E = 0.004

CONCLUSION
In the present work, influence of five process parameters namely, layer thickness, part build orientation, raster angle, raster width and air gap each taken at three different levels are studied for the surface finish of the FDM processed part. The surface roughness value at initial stage is 9.12 microns. The effect of process parameters on response is studied via Response surface methodology (RSM). With the help of RSM significant factors and their interaction are identified. In order to improve surface roughness of the build part it is required that the parts are manufactured with optimum process variables through a structured method. The method of genetic algorithm is used to get the optimum process parameters so that surface roughness is increased. Genetic algorithm shows that layer thickness, part build orientation, raster angle of will fabricate the part with overall improvement in accuracy of dimensions. Percentage deviation is observed in surface roughness with the optimum values. Small percentage error establishes the fitness of the present model. The obtained optimum values are surface roughness is 3.046 microns.

REFERENCES


Hua, K., Hong, H., and Ho, S., (1999), Rapid tooling technologies, part 1, A comparative study, The International Journal of Advanced Manufacturing Technologies.

Levy G., Schindel, R. and Kruth , J., (2003), Rapid manufacturing and rapid tooling with layer manufacturing (LM) technologies, state of the art and future perspectives CIRP Annals – Manufacturing Technology

Ahn, S., Lee, C., and Jeong, W. (2004), Development of translucent FDM parts by postprocessing, Rapid prototyping Journal. 10 (4), pp. 218224,

Reeves, P. and Cobb, R., (1997), Reducing surface roughness deviation of Steriolithography using in process techniques, Rapid prototyping journal. 3 (1), pp. 20 31.

Masood, S. Mau, K., Hong, J., and Song, W., (2010), Tensile properties of processed FDM polycarbonate material, Materials Science Forum. 654 656, pp. 25562559.

Mahapatra, S., Sood, A., Patel, S., and Sahu, S. (2009) Optimization of Process Parameters in Fused Deposition Modeling using Weighted Principal Component Analysis, Administrative Staff College of India (ASCI), Hyderabad, India.

Ahn, S., Montero, M., Odell, D., Roundy, S., and Wright, P., (2002), Anisotropic material properties of fused deposition modeling ABS, Rapid prototyping Journal. 8 (4), pp. 248257.

Lee, B., Abdullah, J., and Khan, Z., (2005), Optimization of rapid prototyping parameters for production of flexible ABS object, Journal of Materials Processing Technology. 169 (2005), pp. 5461.

Laeng, J., Zahid, A., Khu, S., and Chandrasekaran, M., (2006), Optimization flexible behavior of bow prototype using Taguchi approach, Journal of Applied Science. 6(3), pp.622633.

Ang, K., Leong, K., Chua, C., and Chandrasekaran, M., (2006), Investigation of the mechanical properties and porosity relationships in fused deposition modelingfabricated porous structures, Rapid Prototyping Journal. 12(2), PP.100105.

Essaid, S., Foyos, J., Noorani, R., Mendelson, M., Marloth, R., and Pregger, A., (2000), Effect of layer orientation on mechanical properties of rapid prototyping samples, Material and manufacturing processes, Vol. 15, pp 107122.

Anitha, R., Arunachalam, S., Radhakrishnan, P., (2001), Critical parameters influencing the quality of prototypes in fused deposition modeling Journal of Materials Processing Technology. 118, pp. 385388.

Gregorian, A., Elliot, B., Navarro, R., Ochoa, F., Singh, H., Monge, E., Foyos, J., Noorani, R., Fritz, B., and Jayanthi, S., (2001), Accuracy improvement in rapid prototyping machine (FDM1650), Solid Freeform Fabrication Proceedings, pp. 7784.

Sood, A., Ohdar, S., and Mahapatra S.,(2010), Parametric appraisal of fused deposition modeling process using the grey Taguchi method, Proceeding of the Institution of Mechanical Engineers, part B: Journal of Engineering Manufacture, 224(1), pp. 135145.

Wang, C., Lin, T., and Hu, S., (2007), Optimizing the rapid prototyping process by integrating the Taguchi method with the Gray relational analysis, Rapid Prototyping Journal, 13(5), pp.304 315.