Available online at www.sciencedirect.com
ScienceDirect Procedia Technology 23 (2016) 344 – 351
3rd International Conference on Innovations in Automation and Mechatronics Engineering, ICIAME 2016
Experimental Investigation and Prediction of Flatness and Surface Roughness during Face Milling Operation of WCB Material Saurin Sheth a*, P M George b a*
Associate professor, Mechatronics Engineering Department, G.H. Patel College of Engineering & Tech, VV Nagar, Gujarat, India b Professor & Head, Mechanical Engineering Department, B V M Engineering College, VV Nagar, Gujarat, India.
Abstract In this work, effect of machining parameters spindle speed, feed and depth of cut were investigated during Face Milling of Wrought Cast Steed grade B (WCB). WCB is widely used in manufacturing valves due to its less cost. 23 full factorial design with four centre points is selected to perform the reliable experiments. Here the response parameters selected are surface roughness and flatness, a form control of Geometric Dimensioning & Tolerancing (GD&T). The values of flatness and surface roughness affect a lot during leakage testing of dual plate check valve. To achieve the desire value of flatness and surface roughness machining parameters need to be controlled. The right selection of process parameters can be achieved through a predictive model. ANOVA has been carried out to know the significance of input parameters. The values predicted from the model and experimental values are very close to each other.
© 2016 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license © 2016 The Authors. Published by Elsevier Ltd. (http://creativecommons.org/licenses/by-nc-nd/4.0/). Selection and/or peer-review under responsibility of the Organizing Committee of ICIAME 2016 Peer-review under responsibility of the organizing committee of ICIAME 2016 Keywords: Speed. Feed; Depth of Cut; Flatness; Surface Roughness; Face Milling; Modeling; GD & T
1. Introduction GD & T is a symbolic language used to specify the size, form, orientation and location of part features. It is based on the standard, Dimensioning and Tolerancing ASME Y14.5M-1994 which is later on updated as ASME Y14.5-2009. Drawings with properly applied geometric tolerancing provide the best opportunity for uniform interpretation and cost-effective assembly. GD&T was created to insure the proper assembly of mating parts, to improve quality, and to reduce the rework and its associated cost [1, 24]. With GD & T, variations of a part from its specified size and form are controlled to ensure part functionality and interchangeability. Flatness control is commonly used on planar surface capable of resting on matting planar surface without any significant rocking. The mechanism behind the formation of flatness as geometric tolerance is very dynamic, complicated, and process dependent. The Flatness specification may be verified with a dial indicator, CMM or by other methods [5, 6, 19]. Vishal Francis et al discussed the use of Taguchi method and Response surface methodology for optimization of surface roughness in machining of Gun metal with a
* Saurin Sheth. Tel.: +91 9428076097 E-mail address:
[email protected]
2212-0173 © 2016 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). Peer-review under responsibility of the organizing committee of ICIAME 2016 doi:10.1016/j.protcy.2016.03.036
Saurin Sheth and P.M. George / Procedia Technology 23 (2016) 344 – 351
345
HSS tool [2]. Amit Joshi et al use Taguchi methodology to investigate the effects of various parameters during end milling process by varying spindle speed, depth of cut, feed rate on surface finish of aluminum cast heat treatable alloy. The results of analysis of variance indicate that the feed rate is most influencing factor for modeling surface finish [3]. D. Baijic et al studied the effect of cutting speed, feed rate and depth of cut on the surface roughness during face milling. Regression analysis and neural networks had been applied on the experimentally determined data to predict surface roughness [15, 17]. Sheth et al [8] measured the vibration and analyzed its effect on surface roughness by varying machining parameters. Patel et al [9, 10] analyzed the effect of MRR during flashing operation of precision steel ball manufacturing using 2 3 replicate experimental design. They have developed a fuzzy logic based model also to predict MRR [13]. Schmitz et al have presented a case study on comparison of error sources in high speed milling [22]. Woo et al studied high speed cutting characteristics using design of experiments [21]. Abdelilah et al presented choice of cutting tool during milling operation [20]. The flatness on the milling components is of prime importance in dual plate check valve as shown in fig 1 [16]. Here two matting planar surfaces together create metal to metal seal, for a dual plate check valve. A perfect metal to metal seal without significant rocking can be achieved by desired flatness and proper surface roughness. Kovac et al developed a fuzzy logic & regression based model to predict surface roughness during face milling operation [18]. It seems that the investigation of geometrical features along with the surface roughness needs to be addressed to meet the functional requirement of such kind of products produced with face milling operation. So, surface roughness and flatness control helps in reduction of leakage and subsequently it helps in meeting the functional requirement of an assembly. So, the investigation of flatness and surface roughness of WCB material creates new scientific domain in the machining. The selection of proper cutting parameters for milling process becomes a vital requirement for reduction of rework and to increase the productivity. Few main parameters like Spindle speed, Feed rate, Depth of Cut are considered for the present investigation. Their right selection may optimize the flatness error and surface roughness during machining.
Fig. 1. Model of valve body and door [5,6,7]
2. Experimentation 2.1. Work piece material, Machine tool and cutting tool The test work piece material is WCB, wrought cast steel with grade B. Its chemical composition along with its few mechanical properties are shown in table 1. It is the most widely used material in manufacturing of valves due to its less cost [11, 14]. Machinability tests were carried out on the 3-axis CHIRON FZ 16 L/CNC Milling having spindle motor power of 5.7kw. Miracle coated VP15TF insert with specifications as shown in table 2 is used for this investigation. Table 1. Chemical composition and mechanical properties of WCB material Steel type Metal code, Standard C Si Mn P S Cr Ni Mo Others Tensile strength (N/mm2) Yield strength (N/mm2)
Carbon steel WCB, ASTM A216 ≤0.30 ≤0.60 ≤1.00 ≤0.040 ≤0.045 ≤0.50 ≤0.50 ≤0.20 Cu : ≤0.30 V : ≤0.03 485-655 ≥250
Re (mm)
Table 2. Insert Specification Geometry
F1 (mm)
Insert
S1 (mm)
Saurin Sheth and P.M. George / Procedia Technology 23 (2016) 344 – 351
D1 (mm)
346
12.7
3.97
1.4
0.8
SOMT12T308PEER-JM
2.2. Experimental Procedure Experiments are carried out on blocks having size of 60 mm x 60mm x 50 mm of WCB material. Spindle speed, Feed and Depth of cut are selected as input variables to perform experiments according to 2^3 full factorial with four center points experimental design. The levels of input variables are shown in Table 3. Block is clamped by using hydraulic vice as shown in Figure 2 (a). All six sides of the block is initially machined and then after face milling process is carried out on each face of block to create 50 mm wide slot throughout length [16]. Here no. of passes and coolant flow rate are constant. The machined work pieces are shown in Figure 2 (b). Table 3. Factors and Levels Factors Spindle speed (rpm) Feed (mm/min.) Depth of cut (mm)
Coded factors A B C
Fig. 2. (a) Machining of work-piece
Low level (-) 500 150 0.1
High level (+) 1200 300 0.5
Center points 850 225 0.3
(b) machined work-pieces [7]
2.3. Measuring Techniques Flatness measurement is carried out on hexagon make CNC co-ordinate measuring machine as shown in below fig. 3 (a). For measuring flatness, rectangle grid extraction strategy is used to extract points from the surfaces [23]. Points are extracted from the surfaces having 35 mm x 35 mm cross section with 5 mm grid size. The sample reading of flatness i.e. PC-Dmis report for treatment combination 1 is shown in figure 4. Surface roughness measurement is carried out on Surf test SV-2100 as shown in fig. 3 (b). A sample reading for surface roughness for treatment combination 1 is shown in the fig. 5. Table 4 shows the responses according to coded factors and treatment combinations using 23 full factorial with the four center point experimental design [4].
Fig. 3. (a) Flatness measurement on CMM
Fig. 3. (b) Surface Roughness Measurement
347
Saurin Sheth and P.M. George / Procedia Technology 23 (2016) 344 – 351
Fig. 4. Sample reading of flatness for treatment combination 1
Fig. 5. Sample reading of surface roughness for treatment combination 1
3. Result, Analysis and Regression modeling Analysis of variance (ANOVA) is performed to know the significant factor, individual factor effect on response and percentage contribution of each factor. Table 4. Result of surface roughness and flatness for various treatment combinations Treatment combination Coded factors
Responses
A
B
C
Flatness (mm)
Surface roughness (µm)
1
-
-
-
0.027
2.1195
a
+
-
-
0.019
2.1035
b
-
+
-
0.038
5.7466
ab
+
+
-
0.028
4.4532
c
-
-
+
0.026
2.5713
ac
+
-
+
0.023
1.7697
bc
-
+
+
0.040
6.5157
abc
+
+
+
0.028
2.0633
0
0
0
0.021
2.8351
0
0
0
0.018
2.8004
0
0
0
0.020
3.0775
0
0
0
0.018
2.4305
center points
3.1. ANOVA for Flatness Table 5. ANOVA table for Flatness Source of Variation
Degree of Freedom
Sum of Square
Mean Square
F
p-value
% Contribution
Main Effects
3
0.00032937
0.00010979
48.80
0.005
A
1
0.00013612
0.00013612
60.50
0.004
22.95
B
1
0.00019013
0.00019013
84.50
0.003
32.06
C
1
0.00000313
0.00000313
1.39
0.324
0.52
348
Saurin Sheth and P.M. George / Procedia Technology 23 (2016) 344 – 351
2-Way Interactions
3
0.00001638
0.00000546
2.43
0.243
AB
1
0.00001513
0.00001513
6.72
0.081
AC
1
0.00000112
0.00000112
0.50
0.530
BC
1
0.00000012
0.00000012
0.06
0.829
3-Way Interactions
1
0.00000613
0.00000613
2.72
0.198
Curvature
1
0.00023438
0.00023438
104.17
0.002
Pure Error
3
0.00000675
0.00000225
Total
11
0.00059300
From Table 5 it is clearly understood that spindle speed and feed rate are the most significant parameters, while depth of cut is the least significant one. 3.2 Main Effects plots and Interaction plot: Fig. 6 (a) shows main effect plot of Spindle speed, Feed and Depth of cut vs. Flatness. It is concluded that the Flatness is minimum at center points of each input cutting parameters. Fig. 6 (b) shows the Interaction plot of Spindle speed, Feed and Depth of cut vs. Flatness. It shows that the interaction is not present between factors for Flatness as response.
Fig. 6 (a) Main Effect plot of Spindle speed, Feed and Depth of cut vs. Flatness
Fig. 6 (b) Interaction plot of Spindle speed, Feed and Depth of cut vs. Flatness
3.2. ANOVA for Surface Roughness Table 6. ANOVA table for Surface Roughness of WCB Material Source of Variation
Degree of Freedom
Sum of Square
Mean Square
F
p-value
% Contribution
Main Effects
3
18.7097
6.2366
87.44
0.002
A
1
5.3853
5.3853
75.50
0.003
20.67
B
1
13.0420
13.0420
182.86
0.001
50.06
C
1
0.2824
0.2824
3.96
0.141
1.08
2-Way Interactions
3
5.3596
1.7865
25.05
0.013
AB
1
3.0363
3.0363
42.57
0.007
AC
1
1.9453
1.9453
27.27
0.014
BC
1
0.3781
0.3781
5.30
0.105
Saurin Sheth and P.M. George / Procedia Technology 23 (2016) 344 – 351
3-Way Interactions
1
0.7043
0.7043
9.87
0.052
14.93
0.031
Curvature
1
1.0649
1.0649
Pure Error
3
0.2140
0.0713
Total
11
26.0525
349
From Table 6 it is clearly understood that, Spindle speed and Feed Rate are significant. while Depth of cut has very less contribution on Surface Roughness. 3.3. Main Effects plots and Interaction plot:
Fig. 7 (a) Main Effect plot of Spindle speed, Feed and Depth of cut vs. Surface Roughness
Fig. 7 (b) Interaction plot of Spindle speed, Feed and Depth of cut vs. Surface Roughness
Figure 7 (a) shows main effect plot of Spindle speed, Feed and Depth of cut vs. Surface Roughness. It can be concluded that the Surface Roughness is minimum at center points of Spindle speed and depth of cut while at low level of Feed rate. Fig 7 (b) shows, Interaction plot of Spindle speed, Feed and Depth of cut vs. Surface Roughness. So it shows that the small interaction is present between Spindle speed and Feed. There is also significant amount of interaction is present between the Feed and Depth of cut for Surface Roughness as response. There is no interaction between the Feed and Depth of cut. 3.4. Regression model The regression model widely used to predict the responses is an algebraic representation of the regression line and is used to describe the relationship between the responses and predictor variables [4, 12]. Response = constant + coefficient (predictor) + . . . . + Coefficient (predictor) ݕൌ ߚ ߚଵ ݔଵ ߚଶ ݔଶ ߚଵଶ ݔଵ ݔଶ Ǥ Ǥ Ǥ Ǥ ߚ ݔ ݔ Where constant (ߚ ) is the value of the response variable when the predictor variables is zero. Hereݔଵ , ݔଶ ǡ ݔଷ are the predictor variables associates with the spindle speed, feed rate and Depth of cut. Coefficients (ߚ ǡ ߚଵ ǡ ߚଶ ǡ ߚଵଶ ǡ Ǥ Ǥ Ǥ Ǥ ǡ ߚ ) represents the estimated change in response for each unit change in predictor value. Flatness can be predicted using Equation 1. But the presence of curvature and less R2 value suggests higher order equation as shown in equation 2. Still it can be improved by performing more no of experiments. Same way equation 3 gives prediction of surface roughness with 1 st order and equation 4 with 2nd order. Flatness = 0.0255 - 0.004125 A + 0.004875 B + 0.000625 C - 0.001375 A*B + 0.000375 A*C - 0.000125 B*C - 0.000875 A*B*C (1) (R2 = 59.34%) With curvature Effect or higher order terms, Flatness = 0.01925 - 0.004125 A + 0.004875 B + 0.000625 C + 0.009375 A*A - 0.001375 A*B + 0.000375 A*C - 0.000125 B*C (2) 0.000875 A*B*C (R2 = 98.86%) Surface Roughness = 3.20719 - 0.820425 A + 1.27685 B - 0.18785 C - 0.616025 A*B - 0.493075 A*C - 0.21735 B*C - 0.296675 (3) A*B*C (R2 = 95.09%) With curvature Effect or higher order terms, Surface Roughness = 2.78587 - 0.820425 A + 1.27685 B - 0.18785 C + 0.631975 A*A - 0.616025 A*B - 0.493075 A*C - 0.21735 (4) B*C - 0.296675 A*B*C (R2 = 99.18%)
350
Saurin Sheth and P.M. George / Procedia Technology 23 (2016) 344 – 351
4. Validation of developed model In order to validate this models four new experiments are conducted at different levels with different combinations other than that used to develop the model. Table 7 shows the measured values, predicted values and error while prediction using developed models.
No. of Experiment
Table 7. Error of flatness and surface roughness Parameters (Inputs)
Measured Flatness (mm)
Measured Surface Roughness
Predicted Flatness (mm)
Predicted Surface Roughness (mm)
%
%
Error
Error
Flatness
Surface Roughness
(mm) A
B
1
-1
0
C 1
0.032
4.112
0.033
4.5434
3.125
10.49
2
1
0
-1
0.022
3.683
0.0235
3.2783
6.81
10.98
3
0
1
1
0.023
3.496
0.024625
3.6575
7.06
4.62
4
1
1
0
0.027
3.303
0.028
3.2582
3.70
1.35
5. Conclusion & Future Scope A second order regression model to predict flatness and surface roughness is developed in context of input parameters speed, feed and depth of cut for WCB material. Table 7 shows the prediction error associated with the validation experiments. It shows that the max error by regression model to predict flatness is 7.06 % and average error is 5.17 %. While predicting the surface roughness the maximum error is 10.98% and average error is 6.86 %. The controversy in depth of cut leads to multi objective optimization. But the influence of Depth of Cut is very less on both the responses. So, to achieve the desire quality more focus should be made in the right selection of spindle speed and feed. The developed predictive model is very useful to the practicing engineers to reduce the scrape and rework. Even it may be helpful in optimizing the machining parameters to obtain the desire value of surface roughness and flatness. The presence of curvature effect in ANOVA implies that still a higher order of regression model may lead to better results. Though the equation is of 2nd order but still all the terms are not included due to inadequate number of experiments. To overcome this more no of experiments at 5 levels using rotatable Central Composite Design are recommended. Then the developed model may be more accurate in prediction. The same data is modeled using fuzzy logic where max error in predicting flatness and surface roughness are 3.12% and 8.56% respectively [16]. Such kind of modelling can be done for various other materials as well as other geometrical errors also. Acknowledgement The authors would like to thank Mr. Paresh Thakar and Vivek Thakar of Flovel valves Pvt Ltd. for their support during experimentation. References [1] Drake. P. J, 2009. Dimensioning and Tolerancing Handbook, McGraw-Hill. [2] Francis. V and Dubey. A, 2013. Application of Taguchi and response surface methodologies for surface roughness in face milling operation. International Journal of Mechanical and Production Engineering Research and Development, 3 (2), pp. 213-220. [3] Joshi. A and Kothiyal. P, 2012. Investigating effect of machining parameters of CNC milling on surface finish by Taguchi method. International Journal on Theoretical and Applied Research in Mechanical Engineering, 1(2), pp. 60-65. [4] Montgomery D. C, 2009. Design and Analysis of Experiments, 7th Edition, John Wiley & Sons. [5] Mistry. B, George. P. M and Sheth. S, 2013. Study and scope of DFMA and GD&T in manufacturing process: A case study on dual plate check valve. Proceeding of 7th International Conference on Advanced Computing and Communication Technologies (ICACCT™-2013). DOI: 10.13140/2.1.2294.8642. [6] Modi. B. S, George. P. M and Sheth. S, 2013. Study and investigate effect of cutting parameters on flatness for dual plate check valve. Proceeding of 7th International Conference on Advanced Computing and Communication Technologies (ICACCT™-2013). DOI: 10.13140/2.1.3867.7286 [7] Modi. B, 2014. Effect of Cutting Parameters on flatness of dual plate check valve, PG Thesis, BVM Engineering College, Gujarat, INDIA. [8] Sheth. S, Modi. B. S, Patel. D and Chaudhari. A. B, 2015. Modeling and prediction using regression, ANN and Fuzzy logic of real time vibration monitoring on lathe machine in context of machining parameters. Bonfring International Journal of man machine interface, 3 (3), pp. 30-35. [9] Patel. P. J and Sheth. S, 2013. Effect of Various Parameters on Material Removal Rate in Flashing Operation of Precision Steel Ball Manufacturing Process. 1st International and 16th National Conference on Machines and Mechanisms (iNaCoMM2013), pp. 332-338. [10] Patel. P. J, Sheth. S and Chauhan. P, 2014. Effect of Various Parameters on Spread in Flashing Operation of Precision Steel Ball Manufacturing Process. Procedia Materials Science, 5, pp. 2224-2232. [11] Patel. T, Sheth. S, Modi. B. S and Patel. P, 2015. Experimental investigation and Forecast of Weld penetration in MIG Welding Process on WCB Material. Proceeding of the international conference on Advances in Production and industrial Engineering, pp. 186-191. DOI: 10.13140/RG.2.1.3550.4163 [12] Patel. P, Modi. B. S, Sheth. S and Patel. T, 2015. Experimental Investigation, Modelling and Comparison of kerf width in laser cutting of GFRP. Bonfring International Journal of Industrial Engineering and Management Science, 5 (2), pp. 55-62.
Saurin Sheth and P.M. George / Procedia Technology 23 (2016) 344 – 351 [13] Sheth. S, Modi. B. S, George. P. M and Patel. P, 2014. A Fuzzy Logic based Model to Predict MRR in Flashing Operation of Precision Steel Ball Manufacturing Process. Procedia Materials Science. 5, pp. 1837-1845. [14] Sheth. S, Modi. B. S, Patel. T and George. P. M, 2014. A Fuzzy Logic Based Model to Predict Weld Width–A Case Study of Hard Facing Process Using MIG Welding on Dual Plate Check Valve. Applied Mechanics and Materials, 592-594, pp. 8-12. DOI : 10.4028/www.scientific.net/AMM.592-594.8 [15] D. Bajic, B. Lela, D. Zivkovic, “Modeling of machined surface roughness and optimization of cutting parameters in face milling,” Metalurgija 47 (2008) 4, pp. 331-334. [16] Sheth, S., George, P. M. (2016). Experimental Investigation and Fuzzy Modelling of Flatness and Surface Roughness for WCB Material Using Face Milling Operation. In CAD/CAM, Robotics and Factories of the Future, Springer India. pp. 769-777. DOI: 10.1007/978-81-322-2740-3_74 [17] B. Lela, D. Bajic & S. Jozic, 2009. Regression analysis, support vector machines, and Bayesian neural network approaches to modeling surface roughness in face milling. International Journal of Advanced Manufacturing Technology. 42, pp. 1082-1088. [18] Kovac. P, Rodic. D, Pucovsky. V, Savkovic. B, Gostimirovic. M. 2013, Application of fuzzy logic and regression analysis for modeling surface roughness in face milling. Journal of intelligent manufacturing. 24, 755-762. [19] Frechette. S. P, Jones. A. T and Fischer. B. R, 2013. Strategy for testing conformance to geometric dimensioning & tolerancing standards. Procedia CIRP. 10, pp. 211-215. [20] Elmesbahi. A, Rechia. A and Jaider. O, 2014. Optimized – automated choice of cutting tool machining manufacturing features in milling process. 11th World congress on computational mechanics (WCCM XI), Spain. [21] Woo kang kim, Hong T. K. Geun. M. R., Geon H.K. 2013. Study on High-speed cutting characteristics using design of experiments. International journal of precision engineering and manufacturing. 14 (10), 1869-1872. [22] Schmitz, T. L., Ziegert, J. C., Canning, S., and Zapata, R. 2008. Case study: A comparison of error sources in high-speed milling. Precision Engineering, 32,126133. [23] Moroni, Giovanni; Petro, Stefano. Inspection strategies and multiple geometric tolerances. Procedia CIRP. 2013, 10, 54-60. [24] Sheth S, George PM, Experimental Investigation, Prediction and Optimization of Cylindricity and Perpendicularity during Drilling of WCB Material Using Grey Relational Analysis, Precision Engineering (2016), http://dx.doi.org/10.1016/j.precisioneng.2016.01.002
351
