Available online at www.sciencedirect.com
ScienceDirect Procedia CIRP 61 (2017) 517 – 522
The 24th CIRP Conference on Life Cycle Engineering
Optimization of Machining Parameters for Improving Energy Efficiency using Integrated Response Surface Methodology and Genetic Algorithm Approach Kuldip Singh Sangwana*, Girish Kanta a Department of Mechanical Engineering, Birla Institute of Technology and Science, Pilani 333031, India * Corresponding author. Tel.: +911596515205; fax: +91-1596-242-183. E-mail address:
[email protected]
Abstract Machine tools consume enormous amount of energy during machining, build-up to machining, post machining and idling condition to drive motors and auxiliary equipments in the manufacturing system. Reduction of energy consumption during the machining phase is extremely important to improve the environmental performance over the entire life cycle. This paper presents a predictive and optimization model based on integrated response surface methodology and genetic algorithm approach to predict the energy consumption and the corresponding machining parameters during the turning of AISI 1045 steel with a tungsten carbide tool. Experiments using Taguchi design are performed to develop the predictive model. The developed predictive model is used to formulate the objective function for genetic algorithm. The confirmation experiments are performed to validate the developed model and the results are found within 4% error. The statistical significance of the developed model has been tested by the analysis of variance test. This research will be beneficial for a number of manufacturing industries for selection of machine tools on the basis of energy consumption. The reduction of peak load through optimization will results in lowering the energy consumption of the machine tools during non-cutting time. © Published by Elsevier B.V. This © 2017 2017The TheAuthors. Authors. Published by Elsevier B.V.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 scientific committee of the 24th CIRP Conference on Life Cycle Engineering. Peer-review under responsibility of the scientific committee of the 24th CIRP Conference on Life Cycle Engineering Keywords:Optimization; Sustainability; Response Surface Methodology; Energy Efficiency; Machining; Genetic Algorithms
1. Introduction Optimizing the energy efficiency of processes has become a priority in the manufacturing sector; driven by soaring energy costs and the environmental impact caused by high energy consumption levels [1]. More than 99% of the environmental impacts are due to the consumption of electrical energy used by the machine tools in discrete part machining processes like turning and milling [2]. The cost of energy used over a ten-year period is about 100 times higher than the initial purchase cost of the machine tools used to manufacture products, and therefore, if energy consumption is reduced, the operating cost and the environment impact generated from energy production are diminished [3]. With the implementation of sustainability principles in machining technologies, end-users have the potential to save money and improve their environmental performance even if their production stays in the same range or decreases [4]. In
machining processes, money saving and sustainability performance can be improved by reducing energy consumption [3]. Lot of research has been done on machining processes but energy consumption aspect of machine tools has not given significant attention. In the past, metal cutting operations have been mainly optimized on the basis of economical and technological considerations without the environmental dimension [5]. Reduction in energy consumption will improve the environmental impact of machine tools and manufacturing processes. The optimization of machining parameters for minimum energy requirement is expected to lead not only to the application of lower rated motors, drives and auxiliary equipment, but also energy saving during machining, build-up to machining, post machining and idling condition [6]. The first step towards reducing the energy consumption in machining is to analyze the impact of machining parameters on energy consumption. Recently, the researchers have started to analyze and optimize the energy
2212-8271 © 2017 The Authors. Published by Elsevier B.V. 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 scientific committee of the 24th CIRP Conference on Life Cycle Engineering doi:10.1016/j.procir.2016.11.162
518
Kuldip Singh Sangwan and Girish Kant / Procedia CIRP 61 (2017) 517 – 522
consumption in machining [7-9]. The literature review reveals that researchers have focused on various predictive modelling and optimization techniques to determine optimal or near-optimal machining parameters [10-13]. This work aims at developing a predictive and optimization model to obtain minimum power consumption (P) as a function of machining parameters; cutting speed (v), feed rate (f) and depth of cut (d) using Response Surface Methodology (RSM) and Genetic Algorithm (GA). The work makes an important contribution to the machining science by reducing the power consumption through optimal selection of machining parameters and resulting in minimizing harmful emissions. This research will be beneficial for a number of manufacturing industries for selection of machine tools on the basis of power consumption. It can be further extended during process planning for various advanced materials, cutting conditions and cutting tools.
widely used in infrastructure, agriculture and automotive industries. The workpieces are machined from a solid cylindrical bar to a final dimension of 47 mm diameter and 250 mm cutting length. The power measurement method and experimental details are shown in Kant and Sangwan [6].Taguchi’s L27 orthogonal array is used to design the experimental plan. Experiments are carried out for all combinations using full factorial array. It is the most suitable array which has 27 runs and 26 degrees of freedom. The three main factors (cutting speed, feed rate and depth of cut) take 6 degree of freedom and the remaining are taken by interactions.
2. Research Methodology Fig. 1 shows the research methodology used for the study. The research methodology can be broadly classified into four phases – experimental planning; predictive modeling by using RSM; optimization using GA; and confirmation experimental studies. In the first phase machine tool, cutting tool, workpiece material, machining parameters (cutting speed, feed rate and depth of cut) and their levels and corresponding response variable (power consumption) were selected to develop the experimental plan for the study. Taguchi’s L27 orthogonal array was used to design the experiments. Next, machining experiments were conducted for the 27 combinations to get the power consumption values. In the next phase, RSM was used to develop the second order polynomial response surface mathematical model for the prediction of the power consumption. This predictive model provides the relationship between machining parameters and power consumption using regression analysis. The statistical significance of the developed model was analyzed using Analysis of Variance (ANOVA). The adequacy of developed model was checked using normal probability plot and residual plot. The parametric influence of machining parameters on power consumption was analyzed using main effect and 3D surface plot. In the last phase, GA was used to find the optimum values of the machining parameters for minimization of power consumption. Next, experimental tests were conducted at the optimum values of the machining parameters predicted by integrated RSM and GA approach to confirm the results. 3. Experimental work A centre lathe having a maximum spindle speed of 2300 rpm and a spindle power of 5.5 kW was utilized to perform experiments in dry conditions. The TNMG 16 04 04 tungsten carbide inserts from Sandvik were mounted on the tool holder PTGNR 2020 K16 having a rake angle of 7º, clearance angle of 6º and 0.4 mm nose radius. The workpiece material used for the experiments is AISI 1045 steel due to its unique combination of strength, formability and versatility; and
Fig. 1. Research methodology used in this study.
519
Kuldip Singh Sangwan and Girish Kant / Procedia CIRP 61 (2017) 517 – 522
4. Results and Discussion The experimental machining parameters and the corresponding measured power consumption values are given in Table1. Next, RSM is used to develop mathematical model for prediction of power consumption.
are shown in Figure 2. The comparison of predicted and experimental values shows that the predicted values of the power consumption are close to the experimental values. The mean relative error between the experimental and predicted values is 5.1%.
Table 1 Experimental design and measured power consumption S. No
v (m/min)
f (mm/rev.)
d (mm)
P (kW)
1
103.31
0.12
0.50
0.497
2
103.31
0.12
1.00
0.684
3
103.31
0.12
1.50
1.021
4
103.31
0.16
0.50
0.543
5
103.31
0.16
1.00
1.013
6
103.31
0.16
1.50
1.269
7
103.31
0.20
0.50
0.586
8
103.31
0.20
1.00
0.980
9
103.31
0.20
1.50
1.512
10
134.30
0.12
0.50
0.574
11
134.30
0.12
1.00
0.824
12
134.30
0.12
1.50
1.285
13
134.30
0.16
0.50
0.604
14
134.30
0.16
1.00
1.115
15
134.30
0.16
1.50
1.547
16
134.30
0.20
0.50
0.721
17
134.30
0.20
1.00
1.309
18
134.30
0.20
1.50
1.794
19
174.14
0.12
0.50
0.690
20
174.14
0.12
1.00
0.982
21
174.14
0.12
1.50
1.631
Regression 9
6.4714 6.4714 0.7191 80.08 0.000
97.69
22
174.14
0.16
0.50
0.815
Linear
3
6.0254 6.1153 2.0384 227.0 0.000
90.96
23
174.14
0.16
1.00
1.173
v
1
1.0860 1.0702 1.0702 119.1 0.000
16.40
24
174.14
0.16
1.50
2.002
f
1
0.8515 0.8777 0.8777 97.74 0.000
12.85
25
174.14
0.20
0.50
0.883
d
1
4.0879 4.1674 4.1674 464.1 0.000
61.71
26
174.14
0.20
1.00
1.888
Square
3
0.0091 0.0091 0.0030 0.34
0.800
0.14
27
174.14
0.20
1.50
2.430
v*v
1
0.0047 0.0047 0.0047 0.52
0.480
0.07
f*f
1
0.0003 0.0003 0.0003 0.03
0.855
0.00
d*d
1
0.0041 0.0041 0.0041 0.45
0.511
0.06
Interaction 3
0.4369 0.4369 0.1456 16.22 0.000
6.60
v*f
1
0.0901 0.0901 0.0901 10.03 0.006
1.36
v*d
1
0.1904 0.1904 0.1904 21.21 0.000
2.87
f*d
1
0.1564 0.1564 0.1564 17.42 0.001
2.36
Residual Error
17
0.1527 0.1527 0.0090
2.31
Total
26
6.6241
4.1. Development of mathematical model RSM is a collection of mathematical and statistical techniques that are useful for the modelling and analysis of problems in which a response of interest is influenced by several variables [14]. A second-order polynomial response surface mathematical equation is used to develop the mathematic model for power consumption as a function of machining parameters. The developed mathematical model to predict power consumption (P) is: ܲ ൌ ʹǤͳͻ െ ͲǤͲͳʹͲ ݒെ ͳͲǤͲ݂ െ ͳǤͳͶͳͻͶ݀ ͲǤͲͲͲͲʹʹͻͲͶ ݒଶ ͶǤͶͻͳ݂ ଶ ͲǤͳͲͶ݀ ଶ ͲǤͲͳͲͲͳͶ ݂ݒ ͲǤͲͲͲͻͶͷͲ ݀ݒ ͷǤͲͺ͵͵݂݀ (1) Predicted values of power consumption from the developed mathematical model and the experimental values
Fig. 2. Experimentally measured and predicted values of power consumption.
4.2. Analysis of Variance The Analysis of Variance (ANOVA) was applied to analyze the significance of the developed model and the effect of machining parameters on the power consumption. The last column of the Table 2 shows the percentage contribution of each source to the total variation indicating the degree of influence on the results. Table 2 Analysis of variance (ANOVA) results Source
R2 = 0.9770
DOF
Seq.
Adj.
Adj.
SS
SS
MS
R2 (Pred.) = 0.9439
F
p
Contribution %
R2 (Adj.) = 0.9648
Regression F-value of 80.08 indicates that the model is significant. The percentage contribution of each term is also shown in Figure 3. Depth of cut (d) was found to be the most significant factor on the power consumption which explains 61.71% contribution of total variation. The next contributions
520
Kuldip Singh Sangwan and Girish Kant / Procedia CIRP 61 (2017) 517 – 522
on the power consumption come from the cutting speed and feed rate having contribution of 16.40% and 12.85% respectively. The quadratic terms [v2, f2, d2] do not have statistical significance on power consumption because they have much lower level of contribution and their p-value is also more than the confidence level. Interactions [(v×f), (v×d) and (f×d)] have 1.36%, 2.87% and 2.36% contributions respectively. It can also be seen that residual error has only 2.31% contribution. The value of R2 is 0.9770 which indicates that 97.70% of the total variations are explained by the model. The value of the R2 (Adj.) = 0.9648 indicating 96.48% of the total variability is explained by the model after considering the significant factors. R2 (Pred.) = 0.9439 which shows that the model is expected to explain 94.39% of the variability in new data.
Fig. 5. Plot of residual versus fitted power consumption values
4.4. Parametric influence on power consumption The main effects of machining parameters on mean power consumption are shown in Figure 6.
Fig. 3. Percentage contribution of machining parameters on power consumption
4.3. Model fitness check The residuals are examined to investigate the adequacy of the developed modal. The residuals are the differences between the experimental measured value and the predicted value. These are examined using normal probability plots of the residuals and the plots of the residuals versus the predicted response. Figure 4 reveals that the residuals are not showing any particular trend and the errors are distributed normally. The residual versus the power consumption plot in Figure 5 also shows that there is no obvious pattern and unusual structure.
Fig. 4. Normal probability plot of residual for power consumption
Fig. 6. Main effect plot of power consumption
It reveals that power consumption increases with increase in cutting parameters. The slope of depth of cut is maximum as compare to cutting speed and feed rate, which shows that depth of cut, has more impact on power consumption. This trend is also supported by the ANOVA results shown in Table 2. However, the results of Camposeco-Negrete [15] while working on AISI 6061 T6 material demonstrated that feed rate is the most significant factor and cutting speed is the least significant parameter for minimizing the total power consumption. Abhang and Hameedullah [16] demonstrated that the feed rate has the most significant effect on power consumption, followed by depth of cut, tool nose radius and cutting speed during turning of EN-31 steel with tungsten carbide tool. Bhattacharya et al. [17] reported that cutting speed is the most significant factor followed by depth of cut to reduce the power consumption and feed rate has no significant effect on power consumption during turning of AISI 1045 steel with coated carbide tools under high speed machining. More power is consumed due to increase in material removal rate, cutting forces and temperature. Further the power is also consumed to provide the relative movement to the cutting tool with respect to workpiece (feed rate) and rotation of spindle (cutting speed). Main effect plot clearly shows that the mean
Kuldip Singh Sangwan and Girish Kant / Procedia CIRP 61 (2017) 517 – 522
power consumption was minimum, when the smaller values of cutting speed, feed rate and depth of cut was selected. The 3D merged surface and contour plots for the power consumption are shown in Figure 7. Figure 7(a) shows the surface and contour plots for power consumption at 0.16 mm/rev. feed rate. It shows that at lower values of cutting speed and depth of cut, the power consumption is minimum and power consumption increases drastically with increase in depth of cut even at lower value of cutting speed. It again supports this observation that depth of cut has more influence on power consumption as compared to cutting speed.
tool designer to estimate the suitable range of parameters during the design of machine tools.
Fig. 7(c). Surface and contour plot for power consumption for varying depth of cut and feed rate at cutting speed 134.30 mm/rev.
4.5. Optimization using genetic algorithm
Fig. 7(a). Surface and contour plot for power consumption for varying cutting speed and depth of cut at 0.16 mm/rev. feed rate
Figure 7(b) reveals that at the lower values of cutting speed and feed rate, the power consumption is minimum. Increase in cutting speed at lower values of feed rate has negligible effect on the power consumption but power consumption increases with increase in cutting speed at higher values of feed rate.
The selection of optimum operating parameters has always been a difficult task in the machining. In practice, the machining parameters are generally selected on the basis of human judgment and experience and referring the available catalogues and handbooks which leads to non-optimal parameters and hence loss of power consumption. The optimum machining parameters can be achieved efficiently by varying cutting conditions by using an appropriate optimization method. Therefore, the machining parameters are defined in the standard optimization format and solved using genetic algorithm. The minimization of power consumption problem formulated in the standard mathematical format is as below: Minimize ʹǤͳͻ െ ͲǤͲͳʹͲ ݒെ ͳͲǤͲ݂ െ ͳǤͳͶͳͻͶ݀ ͲǤͲͲͲͲʹʹͻͲͶ ݒଶ ͶǤͶͻͳ݂ ଶ ͲǤͳͲͶ݀ ଶ ͲǤͲͳͲͲͳͶ݂ݒ ͲǤͲͲͲͻͶͷͲ ݀ݒ ͷǤͲͺ͵͵݂݀ Subjected to constraints: 103.31 m/min ≤ v ≤ 174.14 m/min 0.12 mm/rev. ≤ f ≤ 0.16 mm/rev. 0.5 mm ≤ d ≤ 1.5 mm
Fig. 7(b). Surface and contour plot for power consumption for varying cutting speed and feed rate at 1 mm depth of cut
Figure 7(c) indicates that power consumption is minimum at smaller values of depth of cut and feed rate and maximum at larger values of depth of cut and feed rate. These plots are useful to find the optimum values of cutting speed, feed rate and depth of cut for a particular value of power consumption. These 3D surface plots can be used for estimating the power consumption values for any suitable combination of the machining parameters namely cutting speed, feed rate and depth of cut. This is useful in practice for the operator or the part programmer. These plots can also be used by the machine
A genetic algorithm was used to solve the above objective function. In this work optimization toolbox of MATLAB was used for implementing GA. A population size of 200 and initial population range covering the entire range of values for v, f and d has been used to avoid getting local minimum. The cross over rate used was 0.8 and mutation function was uniform. The scaling function and selection function were rank and uniform respectively. The optimum machining parameters and corresponding minimum power consumption obtained by GA are shown in Figure 8. The optimal solution was obtained after 34 generations.
521
522
Kuldip Singh Sangwan and Girish Kant / Procedia CIRP 61 (2017) 517 – 522
for the cutting process has been estimated based on cutting force prediction equations, which is limited to the energy consumption of the tool tip [1]. This work can be extended to measure, predict and optimize the total energy consumption of the selected machine tool. More reliable prediction of unit process energy consumption will enable industry to develop potential energy saving strategies during product design and process planning stages [18]. References Fig. 8 variation of best fitness with generations and the optimum machining parameters
5. Experimental confirmation The confirmation experiments were performed to facilitate the verification of the obtained feasible optimal machining parameters (v = 119.05 m/min, f = 0.12 mm/rev and d = 0.5 mm) for the power consumption. The results of the confirmation runs for the power consumption are listed in Table 3. The validation error between the predicted results and the confirmation results is 3.29%. Table 3 Confirmation results for power consumption Optimum machining parameters v f d m/min mm/rev mm 119.05 0.12 0.50
Power consumption (P) kW Experimental
GA
Validation Error (%)
0.547
0.529
3.29
6. Conclusions This paper presents prediction and optimization of machining parameters leading to minimum power consumption during turning of AISI 1045 steel. The predictive power consumption values determined using response surface methodology is close to the experimental values. The mean relative error is 5.1% which shows that the developed model has good accuracy in predicting the power consumption. The significance of the developed model has been tested by using ANOVA and examination of residuals. The study of parametric influence on power consumption has given valuable data showing the optimal and near optimal values of machining parameters. It is expected that these predicted information will be useful for practitioners during part programming and also for the machine tool manufacturers to provide the required range of machining parameters for the required range of power consumption. The optimization results have been confirmed by experiments results. The optimal values will be useful for the industry during selection of machine tools to get the desired power consumption. It has been shown that the integrated RSM and GA approach can be used an effective and alternative approach for costly and time consuming experimental studies and can contribute to economic optimization of machining. The results have also proved that the depth of cut is the main influencing machining parameter for the power consumption followed by the cutting speed and feed rate. However, the power required
[1] Li, W, Kara, S. An empirical model for predicting energy consumption of manufacturing processes: a case of turning process. Proc. of the Institution of Mechanical Engineers, Part B: J. of Engineering Manufacture. 2011; 225: 1636–1646. [2] Li, W, Zein, A, Kara, S, Herrmann, C. An investigation into fixed energy consumption of machine tools. In: Glocalized Solutions for Sustainability in Manufacturing: Proceedings of the 18th CIRP International Conference on Life Cycle Engineering. Braunschweig: 2011. 268–275. [3] Pusavec, F, Kramar, D, Krajnik, P, Kopac, J. Transitioning to sustainable production – part II: evaluation of sustainable machining technologies. J. Clean. Prod. 2010b;18:1211–1221. [4] Pusavec, F, Krajnik, P, Kopac, J, Transitioning to sustainable production – Part I: application on machining technologies. J. Clean. Prod. 2010a;18:174–184. [5] Yan, J, Li, L. Multi-objective optimization of milling parameters – the trade-offs between energy, production rate and cutting quality. J. Clean. Prod. 2013;52:462–471. [6] Kant, G, Sangwan, KS. Prediction and optimization of machining parameters for minimizing power consumption and surface roughness in machining. J. Clean. Prod. 2014;83:151–164. [7] Zhang, C, Li, W, Jiang, P, Gu, P. Experimental investigation and multiobjective optimization approach for low-carbon milling operation of aluminum. Proc. Inst. Mech. Eng. Part C J. Mech. Eng. Sci. 2016;1–20. [8] Kant, G, Sangwan, K.S. Predictive modelling for power consumption in machining using artificial intelligence techniques. Procedia CIRP. 2015;26:403–407. [9] Kant, G, Sangwan, KS. Predictive modelling for energy consumption in machining using artificial neural network. Procedia CIRP. 2015;37:205– 210. [10]Garg, A, Lam JSL, Gao L. Power consumption and tool life models for the production process. J. Clean. Prod. 2016;131:754-764. [11]Kant, G, Sangwan, KS. Predictive modelling and optimization of machining parameters to minimize surface roughness using artificial neural network coupled with genetic algorithm. Procedia CIRP. 2015;31:448-453. [12]Sangwan, KS, Saxena, S, Kant, G. Optimization of machining parameters to minimize surface roughness using integrated ANN-GA approach. Procedia CIRP. 2015;29:305-310. [13]Kant, G, Rao V, V, Sangwan, KS. Predictive modelling of turning operations using response surface methodology. Applied Mechanics and Materials. 2013;307:170–173 [14]Myers RH, Montgomery DC, Christine M. Anderson-Cook CM. Response surface methodology: process and product optimization using designed experiments. John Wiley & Sons; 2016. [15]Camposeco-Negrete, C. Optimization of cutting parameters for minimizing energy consumption in turning of AISI 6061 T6 using Taguchi methodology and ANOVA. J. Clean. Prod. 2013;53:195–203. [16]Abhang, LB, Hameedullah, M. Power Prediction Model for Turning EN31 Steel Using Response Surface Methodology. J. Eng. Sci. Technol. Rev. 2010;3:116–122. [17]Bhattacharya, A, Das, S, Majumder, P, Batish, A Estimating the effect of cutting parameters on surface finish and power consumption during high speed machining of AISI 1045 steel using Taguchi design and ANOVA. Prod. Eng. 2009;3:31–40. [18]Kara, S, Li, W. Unit process energy consumption models for material removal processes. CIRP Annals-Manufacturing Technology. 2011;60: 37-40.
