Optimization of machining parameters considering ...

Journal of Cleaner Production xxx (2015) 1e9

Contents lists available at ScienceDirect

Journal of Cleaner Production journal homepage: www.elsevier.com/locate/jclepro

Optimization of machining parameters considering minimum cutting fluid consumption Zhigang Jiang a, *, Fan Zhou a, Hua Zhang a, Yan Wang b, John W. Sutherland c a

College of Machinery and Automation, Wuhan University of Science & Technology, Wuhan, 430081, China Department of Computing, Engineering and Mathematics, University of Brighton, Brighton, BN2 4GJ, United Kingdom c Division of Environmental and Ecological Engineering, Purdue University, West Lafayette, IN, 47907, USA b

a r t i c l e i n f o

a b s t r a c t

Article history: Received 15 April 2014 Received in revised form 8 May 2015 Accepted 2 June 2015 Available online xxx

Dry or near dry machining is often regarded as an effective strategy for reducing ecological impacts of the cutting processes. However, due to the application limitations of dry or near dry machining, reduction of cutting fluid supply through machining parameter optimization offers a cost effective alternative. To this end, an optimization model of machining parameters considering minimum cutting fluid consumption and cost is proposed. Process cost and cutting fluid consumption are treated as the two objectives in the optimization model, which are affected by four variables, namely cutting depth, feed rate, cutting speed, and cutting fluid flow. In the model, process cost includes production operation cost and cutting tool cost, whilst cutting fluid consumption by a machining process, which consists of reusable cutting fluid and non-reusable cutting fluid, ie., the remaining cutting fluid deposited on the workpiece and chips as well as that diffused into the environment. The multi-objective optimization problem is solved by a hybrid genetic algorithm programmed in Matlab 7. An illustrative case study was implemented to verify the effectiveness of the multi-objective optimization model, and the simulation results showed 17% reduction of fluid consumption compared to that without optimization. This indicates that the proposed optimization is effective and has great potential to be adopted by industry. © 2015 Elsevier Ltd. All rights reserved.

Keywords: Machining parameters Cutting fluids Multi-objective optimization Genetic algorithm (GA)

1. Introduction Machining is a widely used class of industrial manufacturing processes in which cutting fluids often play a crucial role in terms of machining quality and efficiency due to their lubricant, cooling, and chip removal functions (Cetin et al., 2011). Cutting fluids can significantly improve the productivity of machining operations, tool life, and workpiece quality. Also, they prevent the cutters and machine tools from overheating (Tazehkandi et al., 2014). Despite the benefits of cutting fluid, negative impacts on the environment have attracted much attention (Thepsonthi et al., 2009). Disposal of cutting fluids can be very costly and subject to stringent government regulations. Treatment and improper disposal of cutting fluids may present serious environmental problems (e.g., water and soil pollution). Use of cutting fluids has also been linked to worker health problems (e.g., lung cancer, respiratory diseases, dermatological and genetic diseases) (Wagabayashi et al., 1998). Moreover,

* Corresponding author. E-mail addresses: [email protected], [email protected] (Z. Jiang).

cutting fluids usage also increases the total process cost (Kuram et al., 2013). A survey carried out in the German automotive industry showed that workpiece related manufacturing costs incurred in connection with the deployment of cutting fluids was at a level of 7e17%, which were several times higher than tool costs that accounted for only approximately 2e4% (Klocke and Einesblatter, 1997). In order to reduce the high cost and mitigate the environmental burden associated with the use, treatment and disposal of cutting fluids, some new technologies have been sought to minimize or even avoid the use of cutting fluids in machining operations. These technologies include dry machining or machining with minimum quantity lubrication (MQL). Dry machining operations are now of great interest as they can eliminate environmental impacts associated with cutting fluids (Dhar et al., 2006). Devillez et al. (2011) utilized dry machining to make the machining processes more environmental friendly by employing a coated carbide tool and increasing the cutting speed. Bahi et al. (2012) presented a hybrid analytical numerical approach to model the chip formation under dry machining and tool wear. Canteli et al. (2010) focused on the heat treatment of inserts which permitted the selection of the

http://dx.doi.org/10.1016/j.jclepro.2015.06.007 0959-6526/© 2015 Elsevier Ltd. All rights reserved.

Please cite this article in press as: Jiang, Z., et al., Optimization of machining parameters considering minimum cutting fluid consumption, Journal of Cleaner Production (2015), http://dx.doi.org/10.1016/j.jclepro.2015.06.007

2

Z. Jiang et al. / Journal of Cleaner Production xxx (2015) 1e9

Nomenclature ap b Cm cp Ct Cv Cw D f Fc fp g h H KF L La lc lw Mu

Cutting depth (mm) Width of the chip being cut (mm) Machining cost rate (dollars/min) Specific heat of the cutting fluid (J/(kg K)) Tooling cost per cutting edge (in dollars) A fixed value related to cutting conditions Machining cost (in dollars) Diameters of the workpiece (mm) Feed rate (mm/revolution) Cutting force (N) The component of power vector along the x-direction (kw) Gravity constant (m/s2) Thickness of the chip (mm) Latent heat of vaporization (J/kg) The correction coefficient of cutting force Cutting fluid flow (L/min) Reusable cutting fluid flow (L/min) The cutting length of the workpiece (mm) The processing length (mm) Cutting fluid consumption of machining (L)

optimal processing parameters in dry machining. Huang et al. (2014) used a variable pitch tool to dry mill a thin-walled titanium alloy Tie6AIe4V in order to reduce cutting vibration and decrease cutting temperature. Dry machining is considered a very promising technology since it can eliminate the problems associated with cutting fluids (Davoodi and Tazehkandi, 2014). Often, however, dry machining is limited to operating at lower cutting speeds, this leads to a low production rate. Therefore, cutting fluids are still required for machining when higher machining efficiency, better surface finish quality required and severe cutting conditions exist. A cost effective alternative would be to reduce the consumption of cutting fluid whist maintaining the same productivity and quality. Techniques to reduce the quantity of cutting fluids have been investigated during the last decade focused on Minimum Quantity Lubrication (MQL) and near dry machining. For instance, Fratila (2009) presented some requirements for a successful MQL application in gear milling to ensure process safety and product quality. Thepsonthi et al. (2009) proposed a minimal-cutting-fluid technique to reduce the use of cutting fluids in the high-speed milling of hardened steel using a coated carbide ball end mill. Similarly, Yan et al. (2009) utilized an MQL method to reduce the use of cutting fluids and conducted a comparison between dry and wet cutting using cemented carbide tools to mill high strength steel, the result of which indicated that MQL could reduce tool wear and improve the quality of the machined surface. This prior research suggests that MQL is an emerging technology for minimizing the use of cutting fluids during machining and offer both economic and environmental benefits. However, the application of MQL in machining requires special machines and tools that impose significant manufacturing costs. In view of the existing processes and equipment, choosing an appropriate values for the machining parameters to reduce the amount of cutting fluids without sacrificing productivity plays an important role in improving overall machining process performance in terms of efficiency, costs and environmental impact, enabling the transition to dry or near dry machining. Yi et al. (2013) investigated process parameter optimization for a CNC machine to

Ma Reusable cutting fluid consumption (L) Mn Non reusable cutting fluid consumption (L) tm Process cutting time (min) tr Process auxiliary time (min) tc Tool change time (min) T Cutting tool life (min) Tvap Vaporization temperature (
C) T∞ Ambient temperature (
C) vc Cutting speed (m/min) 1/m, 1/n, 1/q The coefficient of tool durability xF, yF, zF The coefficients value related to cutting conditions and tool material D Machining allowance (The planned deviation between the dimension after rough machining and the intended final dimension, mm) b Friction angle (
) h Machine power effective coefficient l0 Rake angle (
) r Density of the cutting fluid (g/cm3) s Surface tension (N/m) ts Workpiece plastic flow stress 4 Shear angle (
)

minimize both carbon emissions and processing time. Fratila and Caizar (2011) applied the Taguchi optimization method to optimize the machining parameters (cutting speed, the depth of cut, the feed rate and cutting fluid flow) during the machining of AlMg3 with a HSS tool to obtain optimal surface roughness and minimum power consumption. These studies have achieved good progress in investigating the optimal combination of machining parameters such as cutting speed, depth of cut and feed to realize optimal objective functions, but less attention has been paid to the determination of the optimal cutting fluid consumption in terms of machining parameters on cutting fluid supply (Kuram et al., 2010). Traditionally, the most commonly used criteria for machining process optimization are material removal rate (MRR), surface roughness, processing time, and cost (Gopalsamy et al., 2009). However, in order to address the challenges of environment burden associated with machining processes and process cost, this paper specifically deals with two objectives optimization including process cost and fluid consumption with cutting speed, feed rate, cutting depth and fluid flow rate as the decision variables. The goal is to reduce process cost and the cutting fluid consumption through optimization of cutting parameters and flow rate. There are many conventional and non-conventional optimization techniques that could be employed for this type of problem. One such approach is the Genetic Algorithm (GA), a technique that mimics the process of natural evolution to optimize multi-dimensional nonlinear problems, is the current trends for the machining process optimization (Zain et al., 2010). Multi-objective problems optimization using GAs has been reported by many researchers. For instance, Yu and Wei (2012) developed a hybrid model that combines a GA and System Dynamics (SD) to predict China's coal production-environmental pollution. Liu et al. (2014) developed an algorithm based on the non-dominated sorting GA II to solve the mathematical model with two objectives including minimizing carbon dioxide emission and making span for processing all types of products in seru production systems. Good progress has been achieved to solve optimization problems using GAs. However, traditional GAs are prone to premature and local convergence; with in mind, this paper explores a hybrid genetic algorithm combined with self-adaptive penalty

Please cite this article in press as: Jiang, Z., et al., Optimization of machining parameters considering minimum cutting fluid consumption, Journal of Cleaner Production (2015), http://dx.doi.org/10.1016/j.jclepro.2015.06.007

Z. Jiang et al. / Journal of Cleaner Production xxx (2015) 1e9

function and crossover to conduct the optimization calculation and simulation. This allows each generation population to remain part of the non-feasible solution in order to obtain the global optimal solution on both sides of feasible and non-feasible solutions. In comparison with prior work, the novelty of this paper lies in the following three points: (1) the cutting fluid consumption is treated as an objective in addition to process cost for the multiobjective optimization problem; (2) a hybrid method integrating penalty function and GA is developed, allowing global optimization that could not be achieved using a GA alone; and (3) a novel constructive model linking decision variables and objectives is established.

2. Multi-objective optimization model for machining process parameters 2.1. Optimization variables and objective function The most important parameters affecting the machining process are cutting speed, feed rate and cutting depth (Rajemi et al., 2010). In practice, the amount of cutting fluid is usually determined by process requirements; cutting fluid has a direct impact on tool life, process costs and environmental pollution. Therefore, cutting fluid flow rate is considered as a decision variable in addition to the other decision variables: cutting speed, feed rate and cutting depth, for the multi-objective optimization model. The vector expression of the optimization variable X is as follows:


T X ¼ vc ; f ; ap ; L

(1)

Where, vc is cutting speed, f is feed rate, ap is cutting depth, and L is cutting fluid flow rate. (1) Process cost function The objective function of process cost is expressed as below (Jiang et al., 2002):

  tm tm þ Ct Cw ¼ Cm tm þ tr þ tc T T tm ¼

(2)

pDlw D 1000vc fap

1

1

The total amount of cutting fluid is divided into two parts for machining processes: i) reusable cutting fluid consumption Ma (cutting fluid that is be reused for the next machining operation), and ii) non-reusable cutting fluid consumption Mn (cutting fluid that is lost owing to deposition on the workpiece and chips, and then diffused into the surrounding environment). Thus, cutting fluid consumption Mu can be expressed as follows:

Mu ¼ Ltm ¼ Ma þ Mn

(6)

In operation, the operating status of the machine tool is unsteady because of the dynamic nature of the cutting force, voltage fluctuations, and random factors. Therefore, in reality the operation of the machine tool is a dynamic process, leading to varying cutting condition, thus, the reusable cutting fluid consumption Ma can be calculated as follows:

Ztm La ðtÞdt ¼ La

Ma ¼ 0

pDlw D 1000vc fap

(7)

Where, La represents the reusable cutting fluid flow rate. The non-reusable cutting fluid consumption Mn is further divided into three paths for machining process: (1) a liquid waste stream created through cutting fluid sticking on the chips generated during the machining process, (2) a liquid waste stream resulting from remaining cutting-fluid on the workpiece, and (3) a vapor waste stream generated through cutting-fluid diffused into the surrounding environment. The total consumption of the additional cutting fluid can be expressed as (Munoz and Sheng, 1995):

Mn ¼ MChip þ MWorkpiece þ MEvap

(8)

To find the fluid on chip Mchip, a chip geometry with height h and thickness b can be defined. Assuming that the dominant mechanism for the remaining fluid on chips is surface tension, the geometry of the liquid can be modeled as cylindrical, with curvature in the direction of the smallest length scale of the chip surface. The total volume of cutting fluid sticking on the chips during a machining process is dependent on the aggregated length of the chip (or chips) formed (Munoz and Sheng, 1995):

MChip ¼

Cv vm f n ap

(2) Cutting fluid consumption

(3)

Where, Cm is the machine cost rate, Ct is the tooling cost per cutting edge, tm is the process cutting time, tr is the process auxiliary time, tc is the tool change time, T is the tool durability time, lw is the cutting length of the work-piece, D is the diameter of the workpiece, and D is the machining allowance. According to the generalized Taylor's Formula (Ai and Xiao, 1994), the tool durability time equation is:

T¼

3

(4)

1 q

Where, Cv is a constant, 1/m, 1/n, and 1/q are the coefficients of tool durability related to cutting conditions. Therefore, the process cost Eq. (2) can be rewritten as:

8s lw rg

(9)

Where, s is surface tension, r is density of the cutting fluid, g is earth gravity, and lw is processing length. The total volume of cutting fluid sticking to the workpiece can be determined in a similar way by using the major length scale of the workpiece as lc as follow (Munoz and Sheng, 1995):

MWorkpiece ¼

8s lc rg

(10)

Where, lc is the length of the workpiece. The fluid lost through evaporation is calculated using an energy balance method. The volume of cutting fluid mass loss due to evaporation can be estimated as (Munoz and Sheng, 1995):

1q

Cw ¼ Cm t r þ

pDlw DCm pDlw Dap q ðCm tc þ Ct Þ 1m 1n þ vc m f n 1000vc fap 1000Cv

(5)

MEvap ¼

fp vc
 cp Tvap  T∞ þ H

(11)

Please cite this article in press as: Jiang, Z., et al., Optimization of machining parameters considering minimum cutting fluid consumption, Journal of Cleaner Production (2015), http://dx.doi.org/10.1016/j.jclepro.2015.06.007

4

Z. Jiang et al. / Journal of Cleaner Production xxx (2015) 1e9

 fp ¼

cosðb  l0 Þ cosðb þ 4  l0 Þ



ts bh sin 4

 (12)

Where, fp is the component of power in the x-direction, cp is the specific heat of the cutting fluid, Tvap is the vaporization temperature, T∞ is the ambient temperature, H is the latent heat of vaporization, ts is the workpiece flow stress, b is the friction angle, l0 is the rake angle, 4 is the shear angle, b is width of the chip being cut, and h is thickness of the chip. The cutting fluid flow is treated as a constant value since it contains only negligible fluctuations, when a machine tool runs at a constant speed. Therefore, cutting fluid consumption in formula (6) can be rewritten as:

Mu ¼ L ¼

pDlw D 1000vc fap

fp vc 8s 8s pDlw DLa  lc þ lw þ
þ rg rg cp Tvap  T∞ þ H 1000vc fap

(13)

2.2. Optimization model

f ðXÞ ¼ ðCw ; Mu ÞT
T X ¼ vc ; f ; ap ; L

(14)

Subject to:

8 pDnmin pDnmax > > >  vc  > > 1000 1000 > > > > > f  f  f > max > < min CF ap xF f yF vzF KF  Fmax > > > > Fc vc > > > > 1000h  Pmax > > > > : Lmin  L  Lmax

FðxÞ ¼

2 X

li Fi ðxÞ

(16)

i¼1

Where, li is the weighted coefficient of the ith objective function. To make a reasonable compromise between each objective, the li can be determined as li ¼ 1/Fi(x)*, where Fi(x)* is the ith single-objective optimization value. With the vector of optimization variables in Eq. (1), the evaluation function is:

FðxÞ ¼

Cw Cw

*

þ

Mu Mu *

(17)

Where, Cw * and Mu * are the single-objective optimization values of process cost and cutting fluid consumption respectively. Each single-objective value deviates from the optimization value by assuming that f(X) ¼ (Cw)T for Cw * and f(X) ¼ (Mu)T for Mu * in Eq. (14). 3.2. Hybrid genetic algorithm for machining parameter optimization

The multi-objective parameter optimization problem aiming to minimize process cost and cutting fluid consumption is defined as follows:

(

minimum in order to obtain the model solution (Chikkerur et al., 2011).

(15)

Where, nmin and nmax are the minimum and maximum spindle speeds; fmin and fmax are the minimum and maximum feed rates; Fmax is the maximum cutting force; Pmax is the maximum cutting effective power; Fc is the cutting force; Lmin and Lmax are the minimum and maximum cutting fluid flow rates achievable with the nozzle pump system. In the above mathematical expressions, Eq. (14) sets the optimization objectives to minimize both process cost and cutting fluid consumption. The operation constraints are then presented in Eq. (15), which typically sets the boundaries for the cutting parameters (cutting depth, feed rate, cutting speed, and cutting fluid flow). 3. Hybrid genetic algorithm for parameter optimization 3.1. Multi-objective function transformation In solving this multi-objective optimization problem, the two objective functions cannot be simply summed to compose the evaluation function, because each objective function has a different physical meaning and should be given a corresponding weight. In this paper, a linear-weighted summation method is used as shown in Eq. (16). For the purpose of optimization, each objective function should be as close as possible to their

The optimization model is complicated as it contains both linear and nonlinear constraints. The penalty function method is commonly used to transform a constrained problem into an unconstrained problem which has the disadvantage of low accuracy and computational complexity. Genetic algorithm (GA) is an adaptive method that has been widely used to solve complex search and optimization problems (Yi et al., 2013), but the traditional GA is prone to premature and local convergence. Therefore, a hybrid GA based on both penalty function method and GA is established, then each generation of a population retains part of the non-feasible solutions, and the global optimal solution can be obtained on both sides of feasible and nonfeasible solutions. As shown in Fig. 1, the hybrid genetic algorithm starts with a random generated population size, where individual chromosomes are encoded, and an initial population of strings is created. The process then iteratively selects individuals from the population that undergo some form of transformation to create a new population. The new population is then tested to see if it fulfills some stopping criterion. If it does, then the process halts, otherwise, iterations continue. 3.2.1. Population size and coding choices The initial population is randomly selected in the variable optimization space, and it should be set to a reasonable integer N, whose value size plays an important role in the convergence of the algorithm. If N is too small, it would be easy to fall into a local optimum. If too large it would affect the efficiency of evolution. According to literature (Wang et al., 2014), the population size generally ranges from dozens to hundreds. Note that there is a variety of encoding schemas whose choice is contingent on application features. In this case, a 10-bit binary string is used to denote a single chromosome representing the four machining parameters: vc, f, ap, and L. 3.2.2. Fitness function with self-adaptive penalty term The determination of the fitness function, which is used to reflect the closeness of the individual corresponding target number and design goals is a critical step in the process of GA.

Please cite this article in press as: Jiang, Z., et al., Optimization of machining parameters considering minimum cutting fluid consumption, Journal of Cleaner Production (2015), http://dx.doi.org/10.1016/j.jclepro.2015.06.007

Z. Jiang et al. / Journal of Cleaner Production xxx (2015) 1e9

5

solutions and non-feasible solutions, then the penalty factor remains the same.

3.2.3. Genetic algorithm operations (1) Selection The selection operator aims at either taking the optimization individual to the next generation directly or producing a new individual gene in the next generation by cross-matching based on the individual fitness degree evaluation. The probability for an individual to be selected can be calculated as follows:

, pi ¼ Fi

N X

FN

(21)

k¼1

Where, N is the population size, and Fi is the individual fitness value. (2) Adaptive crossover The parent individuals are selected randomly by a crossover operator to form its recombinant offspring individual. The number of offspring individuals is determined by the crossover probability Pc. Therefore, a crossover probability with a self-adaptive penalty function is proposed to improve search ability, thus, improving the global optimality of the algorithm. Its expression can be written as follows:

8 > < Pc ðtÞ 
ðFmax  FÞ ; F  Favg Fmax  Favg Pc ðt þ 1Þ ¼ > : Pc ðtÞ; F < Favg

Where, Fmax is the maximum fitness value in the population, Favg is the average fitness value in each generation, F is the larger of the

Fig. 1. Flowchart of the hybrid genetic algorithm.

A fitness function with self-adaptive penalty term is established as follows:

SðxÞ ¼ FðxÞ þ HðxÞ

(18)

Where, x is the chromosome, F(x) is the evaluation function. H(x) is the penalty function, which is expressed as follows:

8 0; > > > 2 > P > > > ui ðxÞ; rðtÞ > > > i¼1 < 2 P HðxÞ ¼ rðtÞ ðui ðxÞ  abi Þ2 ; > > > > i¼1 > > 2 > P > > > ðui ðxÞ  cbi Þ2 ; : rðtÞ

Table 1 Chemical composition of C45. Composition

C

Si

Mn

Cr

Ni

Cu

Fe

Wt%

0.42e0.50

0.17e0.37

0.50e0.80

0.25

0.30

0.25

Bal

Table 2 Physical and mechanical properties of C45.

When x is feasible abi  ui ðxÞ  cbi ui ðxÞ < abi

(22)

(19)

Poisson Density Yield Tensile Heat Trademark Rockwell Elastic (g/cm3) strength strength treatment hardness modulus ratio (
C) (GPa) (MPa) (HRC) (MPa) C45

55

210

0.31

7.85

355

600

850

ui ðxÞ > cbi

i¼1

Where, abi and cbi are the upper and lower bounds of the ith constraint that is not satisfied, t is the evolution number, r(t) is the penalty factor. In order to improve the performance of the algorithm, a self-adaptive penalty factor is constructed:

8 < d  rðtÞ; d < 1 rðt þ 1Þ ¼ rðtÞ : j  rðtÞ; j > 1

Table 3 Specifications of the CNC lathe machine. nmin

nmax

vmin

vmax

fmin

fmax

Lmin

Lmax

Fmax

Pmax

h

100

2500

45

200

0.1

3.0

0

50

5000

15

0.8

(20)

In Eq. (20), if the tth generation has feasible solutions, then in the (t þ 1)th generation the penalty factor is increased, otherwise, the penalty factor is reduced. If the tth generation has both feasible

Table 4 Tool life and cutting force coefficient. Cv

1/m

1/n

1/q

CF

xF

yF

zF

KF

b

4

l0

64,136

5

1.75

0.75

1600

1.0

0.6

0.2

1.0

17


28


20


Please cite this article in press as: Jiang, Z., et al., Optimization of machining parameters considering minimum cutting fluid consumption, Journal of Cleaner Production (2015), http://dx.doi.org/10.1016/j.jclepro.2015.06.007

6

Z. Jiang et al. / Journal of Cleaner Production xxx (2015) 1e9

Fig. 2. Schematic diagram of the cutting fluid supply setup.

fitness values of its two parents, Pc(t) is the crossover rate of tth generation. The previous generation Pc(t) could be reduced to make excellent gene individuals well into the next generation when the fitness value is higher than the average fitness of population, otherwise, Pc(t) could be maintained when the fitness value is lower than the average fitness of population. Crossover is performed iteratively until a new population is created. Then the iteration starts again with selection and repeats until the stopping criteria are met. (3) Mutation Whereas a crossover operator is used to combine existing genes in order to obtain new chromosomes, a mutation operator creates new chromosomes by causing small perturbations in genes. Therefore, it helps to prevent the population from stagnating and increase the diversity of the population to extend the solution space. Mutation can be performed either during selection or crossover (though crossover is more usual). For each string element in the mating pool, the GA checks to see if it should perform a mutation. If it should, it randomly changes the element value to a new one. In our binary strings, 1s are changed to 0s, and 0se1s.

properties. The workpiece was machined on a CNC lathe, whose specifications are listed in Table 3. The cutting tool used in this optimization was a YT15 carbide tool, whose parameters are showed in Table 4. The cutting fluid supply setup is shown in Fig. 2. The setup consists of a thermal sensing unit, an ultrasonic atomization unit, and a nozzle atomization unit. The ultrasonic atomization unit consists of a blower, a water supply component and two ultrasonic atomizers. The nozzle atomization unit contains an air compressor, a water pump, and two atomization nozzles. The nozzle unit consists of two nozzles and an atomizing nozzle. In this setup, water and oil are mixed in the ultrasonic atomization unit before the mixture of oil and water is injected on the chipetool interface at a high velocity through the nozzle unit. In order to implement the optimization, some parameters and data should be determined in advance. In practice, the cutting depth is usually determined by machining allowance and accuracy requirement and has relatively small influence on tool wear. Hence,

4. Case study 4.1. Background of the optimization problem To verify the proposed model, machining trials on a CNC lathe were undertaken as an example to demonstrate the impact of adjusting parameter settings on achieving the objectives of minimizing both cutting fluid consumption and process cost. The process was carried on a 45# carbon steel cylindrical bar with a 200 mm length and 50 mm diameter. The chemical composition and physical and mechanical properties of C45 are given in Tables 1 and 2. C45 is widely used in machinery manufacturing due to its good comprehensive mechanical

Table 5 Values of coefficients in the optimization model.

D

s

Ct

tr

tc

Cm

g

ts

cp

r

100

15,000

0.057

3.85

0.1

7.55

10

390

4074

7.85

Fig. 3. Non-dominated front considering cutting fluid consumption and process cost objectives.

Please cite this article in press as: Jiang, Z., et al., Optimization of machining parameters considering minimum cutting fluid consumption, Journal of Cleaner Production (2015), http://dx.doi.org/10.1016/j.jclepro.2015.06.007

Z. Jiang et al. / Journal of Cleaner Production xxx (2015) 1e9

the cutting depth is set as a constant ap ¼ 1 mm; processing length lw ¼ 150 mm; and it was decided to use a cutting fluid. Other coefficients in the optimization model are determined by referring to mechanical engineering manual (Zhang, 1992), includingCt, D, s, tr, tc, Cm, ts, g, cp and r, whose values are listed in Table 5. Production cost is estimated to be 3.42 dollars/min, including direct cost from depreciation of machine tool, cutting fluid, and salary for worker, and indirect cost from management. The purchase price for the hard alloy cutting tool is 7.55 dollars/piece. These data are used for the determination of the production and cutting tool cost. Each of the four variables in the optimization model has an initial value interval. In detail, due to the limitation of rotating velocity of the machine tool, the cutting speed ranges from 45 to 120 m/min, the intervals of feed and cutting fluid flow for machining are 0.1e3.0 mm/revolution and 0e50 L/min. Based on these, the ranges of optimization limitation variables are established as follows.

7

8 45  vc  200 > > > > 0:1  f  3:0 < s:t: 0  L  50 > >  3:125 f 0:6  v0:2 > c > : 0:6  v0:8 f c  7:5

4.2. Simulation for the optimization model Hybrid GA was used to obtain the optimal solution for the optimization of machining parameters. There is no standardized method to determine the population size and generations for the GA (Wang et al., 2014). The parameters for hybrid genetic algorithm were set as follows: iteration number is 100, initial population size N is 100, initial penalty factor r (1) ¼ 1. In addition, crossover probability is 0.95, and mutation probability is 1/s, where s is the number of decision variables, d is 0.5, and 4 is 2. Matlab 7

Fig. 4. Non-dominated front for the machining parameter optimization problem.

Please cite this article in press as: Jiang, Z., et al., Optimization of machining parameters considering minimum cutting fluid consumption, Journal of Cleaner Production (2015), http://dx.doi.org/10.1016/j.jclepro.2015.06.007

8

Z. Jiang et al. / Journal of Cleaner Production xxx (2015) 1e9 Table 6 Comparison of scenarios optimization. Scenario

tm (min)

vc (m/min)

f (mm/r)

ap (mm)

L (L/min)

Surface roughness (mm)

Process cost ($)

Cutting fluid consumption (L)

1 2 3 4 5

3.27 3.25 3.35 3.38 3.12

129 135 117 112 143

0.31 0.37 0.35 0.33 0.28

1 1 1 1 1

9.85 9.96 12.15 11.95 9.32

9.55 9.55 9.12 9.25 9.30

2.55 3.65 2.94 2.35 3.45

36.7 37.1 44.1 45.7 34.6

(MathWorks, USA) software with GA toolbox was used to simulate the optimization model, and the trade-off considering cutting fluid consumption (F1) and process cost (F2) objectives can be obtained as shown in Fig. 3. As shown in Fig. 3 the optimization model with two objectives converges to a solution. With the simulation, the optimal machining parameters are achieved with cutting speed, feed rate, and cutting fluid flow as 129 m/min, 0.31 mm/revolution and 9.85 L/min, respectively, and the optimal objective values of process cost and cutting fluid consumption are $2.55 and 36.7 L, respectively. 5. Discussion To reveal the performance of the optimization model, a set of scheme optimizations are listed in Table 6. Scenario 1 is the optimal result of the optimization model considering both cost and cutting fluid consumption objectives. Scenario 2 is the result when three objectives are considered cost, cutting fluid consumption, and surface roughness. Scenario 2 showed very similar results to scheme 1. This is the case, because in a machining operation, the surface roughness of the workpiece, a main indicator of process quality (Yang et al., 2013), is very strong correlated with cost. This correlation makes it difficult for the three objectives optimization to identify an optimal condition. A nondominated front as shown in Fig. 5 is conducted for the three objective optimization model. The obtained 3D front in Fig. 4(a) is a curve, which indicated that some kind of redundancy exists in the three objectives. As shown in Fig. 4(b), the cutting fluid consumption objective and cost objective possess very good tradeoff between them, so do cost objective and quality objective in Fig. 4(d). However, the cutting fluid consumption objective and quality objective do not possess too much of a tradeoff between them as shown in Fig. 4(c). It is because the machining parameters have little effect on surface roughness as the parameters have already been restricted in a reasonable range by cost as argued by Chaudhari et al. (2011). Therefore, for the computation efficiency, the two-objective optimization instead of three-objective optimization is chosen for the identified problem.

The variable settings for scenario 3 in Table 6 were selected by an experienced workshop technician. Scenario 1 showed a 17% reduction of cutting fluid consumption in comparison with scenario 3. Scenario 4 and Scenario 5 in Table 6 are the optimal results of single-objective optimization considering process cost and cutting fluid consumption respectively. Scenario 4 which is the cost oriented optimization results in lower cutting speed, more processing time and more cutting fluid consumption in comparison to other cases. The increase in Ma is more significant than the decrease in Mn, leading to the increase in the total cutting fluid consumption as depicted in Fig. 5. On the other hand, the low cutting fluid consumption oriented optimization leads to faster cutting speed, less processing time and increased efficiency, but the faster the speed the quicker the tool wear leading to the higher process cost. As a result, our proposed 2 objective model is certainly a balanced solution allowing low-cutting fluid consumption and process cost without sacrificing quality performance due to the strong correlation between quality and cost. 6. Conclusions A multi-objective optimization of machining parameters considering cutting fluid consumption and process cost based on hybrid GA algorithm is proposed in this paper. A cylindrical turning process case study is used to demonstrate the application of the model. Based on the results in the paper, the following conclusions may be drawn: (1) The relationship between cutting fluid consumption and machining parameters is established and the two objective optimisation model is constructed, which has not been seen in prior art; (2) Hybrid GA algorithm is adapted as opposed to the traditional GA method to allow computation efficiency; (3) 17% reduction of fluid consumption is achieved as a result of the optimization in comparison to the existing one without optimization. In addition, future study may be focused on machining parameter optimisation considering fluid consumption, energy consumption and tool life based on the proposed method in this paper. Acknowledgements The work described in this paper was supported by the National Natural Science Foundation of China (51205295), National Science and Technology supporting program (2014AA041504), Wuhan Youth Chenguang Program of Science and Technology (2014070404010214), and A1202 supported by Science Foundation of Wuhan University of Science and Technology. These financial contributions are gratefully acknowledged. References

Fig. 5. Composition of cutting fluid consumption under different optimization goals.

Ai, X., Xiao, S.G., 1994. A Concise Manual of Machining Parameters. China Machine Press, Beijing.

Please cite this article in press as: Jiang, Z., et al., Optimization of machining parameters considering minimum cutting fluid consumption, Journal of Cleaner Production (2015), http://dx.doi.org/10.1016/j.jclepro.2015.06.007

Z. Jiang et al. / Journal of Cleaner Production xxx (2015) 1e9 Bahi, S., Nouari, M., Moufki, A., Mansori, M.E., Molinari, A., 2012. Hybrid modelling of slidingesticking zones at the toolechip interface under dry machining and tool wear analysis. Wear 286, 45e54. mez, B., Gordo, E., Migue lez, M.H., 2010. Canteli, J.A., Cantero, J.L., Marín, N.C., Go Cutting performance of TiCNeHSS cermet in dry machining. J. Mater. Process. Technol. 210 (1), 122e128. Cetin, M.H., Ozcelik, B., Kuram, E., Demirbas, E., 2011. Evaluation of vegetable based cutting fluids with extreme pressure and cutting parameters in turning of AISI 304L by Taguchi method. J. Clean. Prod. 19 (17), 2049e2056. Chaudhari, D.S.S., Khedkar, S.S., Borkar, N.B., 2011. Optimization of process parameters using Taguchi approach with minimum quantity lubrication for turning. Int. J. Eng. Res. Appl. (IJERA) 1 (4), 1268e1273. Chikkerur, S., Sundaram, V., Reisslein, M., Karam, L.J., 2011. Objective video quality assessment methods: a classification, review, and performance comparison. Broadcast. IEEE Trans. 57 (2), 165e182. Davoodi, B., Tazehkandi, A.H., 2014. Experimental investigation and optimization of cutting parameters in dry and wet machining of aluminum alloy 5083 in order to remove cutting fluid. J. Clean. Prod. 68, 234e242. Devillez, A., Le Coz, G., Dominiak, S., Dudzinski, D., 2011. Dry machining of Inconel 718, workpiece surface integrity. J. Mater. Process. Technol. 211 (10), 1590e1598. Dhar, N.R., Islam, M.W., Islam, S., Mithu, M.A.H., 2006. The influence of minimum quantity of lubrication (MQL) on cutting temperature, chip and dimensional accuracy in turning AISI-1040 steel. J. Mater. Process. Technol. 171 (1), 93e99. Fratila, D., 2009. Evaluation of near-dry machining effects on gear milling process efficiency. J. Clean. Prod. 17 (9), 839e845. Fratila, D., Caizar, C., 2011. Application of Taguchi method to selection of optimal lubrication and cutting conditions in face milling of AlMg3. J. Clean. Prod. 19 (6), 640e645. Gopalsamy, B.M., Mondal, B., Ghosh, S., 2009. Taguchi method and ANOVA: an approach for process parameters optimization of hard machining while machining hardened steel. J. Sci. Ind. Res. 68 (8), 686e695. Huang, P.L., Li, J.F., Sun, J., Zhou, J., 2014. Study on performance in dry milling aeronautical titanium alloy thin-wall components with two types of tools. J. Clean. Prod. 67, 258e264. Klocke, F., Eisenbl€ atter, G., 1997. Dry cutting. Ann. CIRP 46 (2), 519e526. Jiang, B., Yang, S.C., Zheng, M.L., 2002. Multi-goal optimization of NC cutting parameters. Tool Eng 36 (7), 22e25. Kuram, E., Simsek, B.T., Ozcelik, B., Demirbas, E., Askin, S., 2010. Optimization of the cutting fluids and parameters using Taguchi and ANOVA in milling. In: Proceedings of the World Congress on Engineering, vol. 2.

9

Kuram, E., Ozcelik, B., Bayramoglu, M., Demirbas, E., Simsek, B.T., 2013. Optimization of cutting fluids and cutting parameters during end milling by using D-optimal design of experiments. J. Clean. Prod. 42, 159e166. Liu, C., Dang, F., Li, W., Lian, J., Evans, S., Yin, Y., 2014. Production planning of multistage multi-option seru production systems with sustainable measures. J. Clean. Prod. http://dx.doi.org/10.1016/j.jclepro.2014.03.033 (in press). Munoz, A.A., Sheng, P., 1995. An analytical approach for determining the environmental impact of machining processes. J. Mater. Process. Technol. 53 (3), 736e758. Rajemi, M.F., Mativenga, P.T., Aramcharoen, A., 2010. Sustainable machining: selection of optimum turning conditions based on minimum energy considerations. J. Clean. Prod. 18 (10), 1059e1065. Tazehkandi, A.H., Pilehvarian, F., Davoodi, B., 2014. Experimental investigation on removing cutting fluid from turning of Inconel 725 with coated carbide tools. J. Clean. Prod. 80, 271e281. Thepsonthi, T., Hamdi, M., Mitsui, K., 2009. Investigation into minimal-cutting-fluid application in high-speed milling of hardened steel using carbide mills. Int. J. Mach. Tools Manuf. 49 (2), 156e162. Wagabayashi, T., Sato, H., Inasaki, I., 1998. Turning using extremely small amount of cutting fluid. JSME Int. J. Ser. C 41 (1), 143e148. Wang, Q., Liu, F., Wang, X., 2014. Multi-objective optimization of machining parameters considering energy consumption. Int. J. Adv. Manuf. Technol. 71 (5e8), 1133e1142. Yan, L.T., Yuan, S.M., Liu, Q., 2009. Effect of cutting parameters on minimum quantity lubrication machining of high strength steel. Mater. Sci. Forum 626, 387e392. Yang, Y., Li, X.Y., Gao, L., Shao, X.,Y., 2013. Modeling and impact factors analyzing of energy consumption in CNC face milling using GRASP gene expression programming. Int. J. Adv. Manuf. Technol. 2013, 1e17. Yi, Q., Tang, Y., Li, C., Li, P., 2013. Optimization of CNC machine processing parameters for low carbon manufacturing. In: Automation Science and Engineering (CASE), 2013 IEEE International Conference on. IEEE, pp. 498e503. Yu, S., Wei, Y.M., 2012. Prediction of China's coal production-environmental pollution based on a hybrid genetic algorithm-system dynamics model. Energy policy 42, 521e529. Zain, A.M., Haron, H., Sharif, S., 2010. Application of GA to optimize cutting conditions for minimizing surface roughness in end milling machining process. Expert Syst. Appl. 37 (6), 4650e4659. Zhang, Q.R., 1992. Mechanical Engineering Practical Manual. Tianjin University Press.

Please cite this article in press as: Jiang, Z., et al., Optimization of machining parameters considering minimum cutting fluid consumption, Journal of Cleaner Production (2015), http://dx.doi.org/10.1016/j.jclepro.2015.06.007