Available online at www.sciencedirect.com
ScienceDirect Materials Today: Proceedings 5 (2018) 24313–24322
www.materialstoday.com/proceedings
IConAMMA_2017
Parametric Optimization of Surface Roughness of Deburring Process on Aluminum Alloy Using SCARA Manipulator Dr. PVS Subhashinia*, N.Amulyab, Y P Deepthic a
Assistant professor,Department of Mechanical Engineering,Vasavi College of Engineering, Hyderabad, 500031, India b ME student,Department of Mechanical Engineering, ,Vasavi College of Engineering, Hyderabad , 500031, India c Assistant professor,Department of Mechanical Engineering,Amrita School of Engineering, Bengaluru, Amrita Vishwa Vidyapeetham, India
Abstract This paper presents an attempt to analyze deburring operation using a SCARA (Selective Compliance Articulated Robot Arm) manipulator. There are various deburring techniques available in the literature but this work aims at deburring of small sized components. In this paper parametric optimization of surface roughness using SCARA Manipulator using various deburring parameters has been performed with Genetic and TLBO optimization algorithms. A rectangular work piece of Aluminum is considered and the end effector used in this case is a grinding wheel of aluminum oxide. The machining parameters considered are Spindle speed, Feed rate and Depth of cut. An orthogonal array L9 was modeled using Taguchi Method to conduct the experiments. The surface roughness of the workpiece is measured using an instrument called Talysurf. Later the end effector of SCARA is used for deburring on the same workpiece to check the efficiency of SCARA for removal of burrs. Regression analysis is employed to analyze the effect of the deburring parameters on material. Optimization of parameters was done using Genetic and TLBO algorithms in MATLAB environment. An optimum deburring parameters for minimizing the surface roughness was obtained at the Spindle speed 5250 (rpm), feed rate 0.859 (mm/sec) and depth of cut 0.3 (mm) and Spindle speed was found to be the most significant factor followed by feed rate for Surface roughness. © 2018 Elsevier Ltd. All rights reserved. Selection and/or Peer-review under responsibility of International Conference on Advances in Materials and Manufacturing Applications [IConAMMA 2017]. Keywords:SCARA,Genetic Algorithm, TLBO algorthm, Parametric Optimization, Deburring, Spindle speed, Feed rate.
* Corresponding author. Tel.: +91 9866802894. E-mail address:
[email protected] 2214-7853© 2018 Elsevier Ltd. All rights reserved. Selection and/or Peer-review under responsibility of International Conference on Advances in Materials and Manufacturing Applications [IConAMMA 2017].
24314
PVS Subhashini/ Materials Today: Proceedings 5 (2018) 24313–24322
1. Introduction To overwhelm the problems encountered with the manual involvement in the production operations, it is required to incorporate automation. Recent sector where manufacturer is focusing is on programmable robots. SCARA is a notable manipulator exclusively used for faster applications. It consists of 4-DOF where 3 are Revolute joints and one is prismatic joint. The end effector of SCARA robot designed to handle rapid operations like pick & place work, circuit board assembly, mechanical assembly with high accuracy. They are less priced compared to other robots and able to adapt to different functions. SCARA robots provide automated solution for electronic assembly to control the problems like extensive production runs and resolve the problem of low productivity etc. Deburring is a method of removing unwanted burrs using deburring tool. Burrs are most commonly generated by machining operations like grinding, drilling, milling, turning. A small burr left out on machined part can cause delay in production due to injury of the worker. To reduce the time delay in production such as industrial accidents, one has to make sure that burrs are removed. In view of the inexpensive nature and flexibility of SCARA robots, it is felt that small and medium sized parts can be deburred with ease. As products of automobiles, small machinery, mechanisms are of small to medium size, it is required to analyze this aspect from the point of view of productivity as happen in the case of Printed Circuit Board assembly. Nomenclature SCARA DOC BD AD SR RPM mm mm/sec µm X1 X2 X3 Ra
Selective Compliance Articulated Robot Arm Depth of Cut Before Deburring After Deburring Surface Roughness Revolution per Minute Millimetre Millimetre per Sec Microns Cutting speed Depth of cut Feed Rate Surface Roughness
2. Literature review To achieve deburring accurately tool-path modifications have been proposed using a CAD (computer-aided design) model. The position errors of the component are also considered. Impedance control is also assumed by applying contact force in excess manner and finally force control is improved. Matching teaching point with the obtained CAD model of tool path, deburring path is optimized. Finally, performance is optimized using ICP generated algorithm [1]. Changhoon Kim et al. [2] proposed model robotic manipulator is controlled with the deburring tool. Double acting system with two pneumatic actuators is considered for the work. Arm and the tool for deburring are considered as two subsystems. Coordination between them are developed by controllers. Rigid, integrated pneumatic and single pneumatic tools were considered for simulation purpose. In the conclusion, the accuracy of operation i.e. deburring process is improved using arm selected for deburring using different tools. Nirosh Jayaweera et al. [3] work was based on deburring path for robot on real time basis a deburring method has been introduced. A sensor to decide the exact location of workpiece prior to the operation i.e. deburring process is used. In this process, there are group of algorithms to generate the required path for robot for the deburring
PVS Subhashini/ Materials Today: Proceedings 5 (2018) 24313–24322
24315
operation. This paper describes the development of operation of deburring for a simple straight path and their further application to complicated paths. The algorithm generated were evaluated using workpieces made from RR1000, Titanium and super CMV alloys by considering a spindle and attaching that to the robot. Wang et al. [4] predicted the contact force considering burr variations a model is presented. To reduce the burrs effect on machined surface am adaptive algorithm is used. Mohammad et al. [5] investigations are related to off-line simulation of finished edges using DELMIA tool were presented. From this process part finishing steps are featured. Impact with the environment were detected and removed through the simulation of cutting path before the path is programmed to the actual hardware. Naoki et al. [6] investigated the deburring application in press working process using a material handling robot. Stepien et al. [7] designed a microprocessor. This microprocessor controls the forces between workpiece and robot. Force control was implemented in deburring experiments. Carlos Valente et al. [8] proposed trajectory programming for burr removal, contact evaluation and control parameters. Catalin G Dumitras et al. [9] presented a relation between cutting parameters and surface finish, forces involved in cutting process and metal removal for deburring process in a mathematical way. L B Abhang et al. [10] carried out Experimental works on EN31 steel alloy to optimize the parameters involved during machining process. D. Philip Selvaraj et al. [11] applied Taguchi optimization technique, to optimize the process parameters, which minimizes the surface roughness of the dry turning of AISI 304(Austenitic Stainless Steel). The present paper aims at parametric optimization of machining parameters in deburring process using SCARA Manipulator by Taguchi Method. Regression analysis is employed to determine a fittest mathematical model to represent our experimental analysis and to find the effect of the machining parameters on material and was done by using statistical software MINITAB-17. 3. Experimental Setup The work material used is rectangular workpiece of aluminum of 25mm square cross section is considered and is made to fix to the table. Table 1 shows the chemical composition of the material used in the present paper. Aluminum is a ductile material which is light in weight, strong and light silver in color. The chemical composition of Aluminium 6061 is based on ASTM standards. Table 1. Chemical Composition of Aluminum 6061 ( ASTM Standards) Material
Al (percentage)
Si
0.4-0.8
Fe
Up to 0.7
Cu
0.15-0.40
Mn
Up to 0.15
Mg
0.8-1.2
Cr
0.04-0.35
Zn
Up to 0.25
Ti
Up to 0.15
Al
Balance (95.85-98.56)
Table 2. Specifications of the Cutting Tools. Tool
Grade
Aluminium Oxide
A60S
A - Abrasives Grain - Aluminum oxide 60- grit size – Medium S - Grade / hardness- Medium And another entry
24316
PVS Subhashini/ Materials Today: Proceedings 5 (2018) 24313–24322
The end effector used in this case is a grinding wheel of aluminum oxide. Path considered is linear. The specifications and geometry of cutting tools are illustrated in Table 2. The SCARA manipulator has two revolute and a prismatic joint. A rectangular workpiece of 25mm square cross section is considered and fixed to the table. The SCARA manipulator is operated with the help of python program. The SCARA manipulator is shown in Fig. 1. (a). To examine the influence of machining parameters on process parameters, experiments are carried out on all the three materials listed above (of square cross section 25mm) using an end effector made of aluminum oxide. a
b
Fig. 1. (a) SCARA Manipulator( Customized) (b) Surface Roughness tester (Mitutoyo SJ-210)
The surface roughness (Ra) was calculated using an advanced hand-held surface roughness tester called Mitutoyo SJ-210, which had an accuracy of about 0.001 microns. The instrument that was used is shown in the Fig. 1. (b). 4. Experimental Design and Analysis In the present paper, Taguchi Orthogonal array [12-13] has been used in conducting the experiments and analyzing the results. The various steps for experimentation and results are shown in the succeeding article. Literature survey was carried out to fix the limits to cutting speed(rpm), feed rate(mm/sec) and depth of cut(mm) for machining of material aluminum from various journals discussed in literature survey. Various levels of the 3 factors for material aluminum 6061 are shown in the Table 3. Table 3. Experimental Cutting Conditions for material Aluminum. Work Material
Cutting Speed
Aluminum
3250, 4250, 5250
Feed (mm/sec)
DOC (mm)
Cutting medium
0.859, 1.8, 2.5
0.3, 0.5, 0.7
Dry
(rpm)
4.1. Selection of orthogonal array Deciding a particular orthogonal array from all the standard orthogonal arrays depends on the number of factors, and levels of each factor. Then orthogonal array was selected using Taguchi technique.
PVS Subhashini/ Materials Today: Proceedings 5 (2018) 24313–24322
i) ii) iii)
24317
Number of control factors = 3 (X, Y, Z) Number of levels for each control factors: 3(1, 2, 3) Number of experiments to be conducted = 9 Table 4. Taguchi Design of Experiments. Exp. No.
X
Y
Z
1
1
1
1
2
1
2
2
3
1
3
3
4
2
1
2
5
2
2
3
6
2
3
1
7
3
1
3
8
3
2
1
9
3
3
2
(ADAPTED :Taguchi methods: DOE page 77) L9 orthogonal array was selected and it is shown in the table 4. 4.2. Data collection After conducting the experiments, data has been collected using measurement technique for Surface Roughness of the machined samples. A sample of work piece after machining is shown in Fig.2.
Fig. 2. Work piece after machining on a SCARA manipulator.
24318
PVS Subhashini/ Materials Today: Proceedings 5 (2018) 24313–24322
5. Results and Discussion Data collected from the experiments pertaining to output responses, SR from training data set table are used to implement the proposed methodology. The necessity in developing the mathematical relationships is to produce the machining responses to the cutting parameters in continuation facilitating the optimization of the deburring process. Regression analysis is a simple statistical tool used to find the mathematical relation among the control variables and their responses. MINITAB software was used in which, a module called Multiple Regression was used to find the empirical formula. The regression analysis for Aluminum was performed and discussed below. The experiments are conducted and the results for Aluminum 6061 are tabulated in. Table 5 for surface roughness (Ra).
Table 5. Experimental Results of Surface Roughness of Material Aluminum S. No.
Cutting Speed
DOC
Feed Rate
Sur. Fin. (BD)
Sur. Fin. (AD)
(rpm)
(mm)
(mm/sec)
(µm)
(µm)
1
3250
0.3
0.859
2.162
2.276
2
3250
0.5
1.8
3.891
4.005
3
3250
0.7
2.5
4.130
4.244
4
4250
0.3
1.8
2.244
2.358
5
4250
0.5
2.5
3.055
3.169
6
4250
0.7
0.859
2.420
2.534
7
5250
0.3
2.5
1.980
2.094
8
5250
0.5
0.859
1.389
1.503
9
5250
0.7
1.8
2.066
2.180
Multiple Regression for OUTPUT Summary Report Comments
Is there a relationship between Y and the X variables? 0
0.1
> 0.5
Yes
The following terms are in the fitted equation that models the relationship between Y and the X variables: X1: cuttigspeed X2: depth of cut X3: feedrate
No
P = 0.001
The relationship between Y and the X variables in the model is statistically significant (p < 0.10).
If the model fits the data well, this equation can be used to predict OUTPUT for specific values of the X variables, or find the settings for the X variables that correspond to a desired value or range of values for OUTPUT.
% of variation explained by the model 0%
100%
Low
High R-sq = 94.62%
94.62% of the variation in Y can be explained by the regression model.
OUTPUT vs X Variables cuttigspeed
depth of cut
feedrate
4
3
A gray background represents an X variable not in the model.
2 00 30
00 40
00 50
30 0.
45 0.
60 0.
8 0.
1.6
2.4
Fig.3. Summary report for material Aluminum. ( Obtained from Minitab)
PVS Subhashini/ Materials Today: Proceedings 5 (2018) 24313–24322
24319
The fitness measures for SR is good with a value of 94.62 %. This presents that the model can explain the variation in the Ra up to the extent of 94.62 %.Based on high values of R2 it presents that the model is adequate in representing the relation. Fig.4 gives expressions for Ra (SR). X1: cuttigspeed X2: depth of cut X3: feedrate
Final Model Equation OUTPUT = 4.132 - 0.000791 X1 + 1.859 X2 + 0.657 X3
Fig. 4. Regression equation for Material Aluminum, ( Obtained from Minitab)
Multiple Regression reports generated through MINITAB for above model for SR shown in Fig. 3. The report in Fig. 3 shows the fitness model equation along with R-sq. percentage of each factor and its interactions for Ra. From the above report, it is clear that cutting speed has a maximum percentage contribution to the surface roughness with 57 %, followed by feed rate with 28 % and depth of cut has less contribution. Hence for material Aluminum cutting speed was proved to be most significant factors. The generated regression equation as shown in Fig.4 was used to predict the values of SR for various settings and is compared with the experimental results. A graph was plot using x-axis as experiment number and y-axis for respective responses. According to the graph no high differences or errors are seen which indicate the developed models are satisfactorily. The comparison graphs are shown and presented in Fig.5 and Fig.6.
Fig. 5.Comparison of predicted values and experimental values of validation data set for SR.
Fig.5. shows the comparison between SR results of experimental data and data predicted through fitness model, obtained through Regression analysis. Fig.6. presents graph between experiment number and surface roughness. It was noticed that the surface roughness is minimized after deburring operation on the SCARA manipulator. Surface roughness is minimized over a range of 0.496 µm to 0.12 µm.
24320
PVS Subhashini/ Materials Today: Proceedings 5 (2018) 24313–24322
Fig.6. Surface Roughness(µm) Before and After Deburring.
In this process of optimization, the aim is to minimize the surface roughness. So, in the present work, the deburring problem is framed as a single objective optimization problem in which Ra is minimized, subject to the feasible bounds of process control variables. Table 6 presents the feasible bounds of the variables and the mathematical model for surface roughness is given by Equation shown in Fig. 4 and are used for formulation of the objective function and bounds. The objective function and bounds are represented as follows
Minimum Surface roughness for material Aluminum The bounds for the parameters for material Brass are given as: 3250 rpm ≤ X1 ≤ 5250 rpm 0.3 mm ≤ X2 ≤ 0.7 mm 0.859 mm/sec ≤ X3 ≤ 2.5 mm/sec Fig.7 shows the Optimization tool in MATLAB for material Aluminum. The solver used was “GA – Genetic Algorithm” As the optimization problem is formulated, then it is solved using an evolutionary technique called Genetic technique in the MATLAB environment. The code or algorithm is written using the MATLAB editor and saved with file extension “.m”. The GUI based GA module is opened through MATLAB apps and the required fitness function is called. Evolutionary techniques are generally strong towards variations of control parameters. Table 6. GA control parameters ( Data used in MATLAB) Population size
50
Number of generations (Maximum)
100
Number of independent runs
51
Crossover probability (%)
80
Crossover fraction (%)
80
Selection method
Stochastic uniform
Fitness measure
R2
Maximum depth of tree
6
PVS Subhashini/ Materials Today: Proceedings 5 (2018) 24313–24322
24321
Fig. 7.Optimization Tool in MATLAB for material Aluminum 6061
However, the same mathematical models given by the Equation in Fig. 4 is now attempted by the TLBO algorithm to check the result. Initially both the models are attempted individually as a single objective function. The population size is used randomly starting with low value. The Fig.8 shows the GUI of TLBO in MATLAB for material Aluminum.
Fig. 8.TLBO in MATLAB for material Aluminum Table 7. TLBO control parameters (Programmed in MATLAB) Population size:
10
Number of generations (Maximum):
50
Number of independent runs:
25
Elite count
3
24322
PVS Subhashini/ Materials Today: Proceedings 5 (2018) 24313–24322
The TLBO control parameters are shown in table 7. The TLBO algorithm has given a surface finish of value 1.28 µm. the optimized parameters obtained or this result are given in Table 8 along with comparison with the other results. Table 8. Single Objective Optimization Results for material Aluminum Parameters and Objective Function
Genetic Algorithm Result
TLBO Algorithm Result
Surface roughness
1.3 µm
1.28 µm
Speed
5250 rpm
5250 rpm
Depth of Cut
0.3 mm
0.3 mm
Feed rate
2.5 mm/sec
2.5 mm/sec
Population size
50
10
6. Conclusion In this paper, parametric optimization of surface roughness using SCARA Manipulator with various deburring parameters was performed using Taguchi Method. An optimum deburring parameters for minimizing the Surface roughness was obtained for Aluminum material. Deburring parameters optimization was done using a TLBO Algorithm. Validation was carried out using Genetic Algorithm. In all the cases, the TLBO algorithm has given similar results to that of Genetic Algorithm. For material Aluminum, the optimum deburring parameters obtained are the Spindle speed 5250 (rpm), Feed rate 0.859 (mm/sec) and Depth of cut 0.3 (mm) and Spindle speed was found to be the most significant factor followed by feed rate for Surface roughness. Surface roughness is minimized over a range of 0.496 µm to 0.12 µm. References [1] Song, Hee-Chan, and Jae-Bok Song. "Precision robotic deburring based on force control for arbitrarily shaped workpiece using CAD model matching." International Journal of Precision Engineering and Manufacturing 14.1 (2013): 85-91. [2] Hong, Deukjo, et al. "HIGHT: A new block cipher suitable for low-resource device." International Workshop on Cryptographic Hardware and Embedded Systems. Springer Berlin Heidelberg, 2006. [3] Jayaweera, Nirosh, Phil Webb, and Craig Johnson. "Measurement assisted robotic assembly of fabricated aero-engine components." Assembly Automation 30.1 (2010): 56-65. [4] Wang, Jue, Maneesh Agrawala, and Michael F. Cohen. "Soft scissors: an interactive tool for realtime high quality matting." ACM Transactions on Graphics (TOG). Vol. 26. No. 3. ACM, 2007. [5] Bousquet, Jean, et al. "Allergic rhinitis and its impact on asthma (ARIA) 2008." Allergy 63.86 (2008): 8-160. [6] Yazawa, Koji, et al. "Molecular dynamics of regioregular poly (3-hexylthiophene) investigated by NMR relaxation and an interpretation of temperature dependent optical absorption." The Journal of Physical Chemistry B 114.3 (2010): 1241-1248. [7] Stepien, Tomasz, et al. "Control of tool/workpiece contact force with application to robotic deburring." IEEE Journal on Robotics and Automation 3.1 (1987): 7-18. [8] Dumitras, Catalin G. "An Approach on Micro-Cutting (Deburring) Process." World Congress on Engineering. 2007. [9] Abhang, L. B., and M. Hameedullah. "Chip-tool interface temperature prediction model for turning process." International Journal of Engineering Science and Technology 2.4 (2010): 382-393. [10] Howard, Andrew W., et al. "The California planet survey. I. Four new giant exoplanets." The Astrophysical Journal 721.2 (2010): 1467. [11] Selvaraj, D. Philip, and P. Chandramohan. "Optimization of surface roughness of AISI 304 austenitic stainless steel in dry turning operation using Taguchi design method." Journal of engineering science and technology 5.3 (2010): 293-301. [12] Radhika N, “Fabrication of LM25/SiO2metal matrix composite and optimization of wear process parameters using design of experiment”.Tribology in Industry, vol. 39, pp. 1-8, 2017. [13] S. Vivek and Amrith, V., “Spirituality and productivity - A relationship perspective”, Purusharta, vol. 10, pp. 60-69, 2017.