You are on page 1of 11

Journal of Cleaner Production 148 (2017) 174e184

Contents lists available at ScienceDirect

Journal of Cleaner Production


journal homepage: www.elsevier.com/locate/jclepro

A process parameters optimization method of multi-pass dry milling


for high efficiency, low energy and low carbon emissions
Hua Zhang a, b, Zhaohui Deng a, b, *, Yahui Fu a, b, Lishu Lv a, b, Can Yan a, b
a
Hunan Province Key Laboratory of High Efficiency and Precision Machining of Difficult-to-Cut Materials, Hunan University of Science and Technology,
Xiangtan, 411201, China
b
College of Mechanical and Electrical Engineering, Hunan University of Science and Technology, Xiangtan, 411201, China

a r t i c l e i n f o a b s t r a c t

Article history: Responding to the current urgent need for low carbon and high efficiency manufacturing, the re-
Received 23 September 2016 lationships between the processing time, power energy consumption and the carbon emissions with the
Received in revised form milling process parameters were researched during dry milling processes. The characteristics of the
13 January 2017
power energy consumption and carbon emissions were also analyzed. The lowest energy, high efficiency
Accepted 14 January 2017
Available online 20 January 2017
(the minimum processing time) and lowest carbon emission functions were separately constructed. Next,
multi-objective optimization model to achieve high efficiency, low energy consumption and low carbon
emissions were constructed. The multi-objective optimization model was converted into a single goal
Keywords:
High efficiency
with weight coefficients introduced. The procedure to identify the empirical model function coefficients
Low energy consumption was built by principal component analysis and regression analysis based on experimental data.
Low carbon emission Considering the processing constraints from machine equipment performance and machining quality,
Multiple objectives optimization the operational flow chart was given to solve the optimization model using the genetic algorithm. The
Multi-pass feasibility of the process parameter optimization method to trade off three responses (low processing
time, low energy consumption and low carbon emissions) was validated using a practical example. From
the results, large feed rates and large cut width can benefit to three responses if the constrains can be
met with. The optimized solution and program of cutting parameters was useful to the optimum per-
formance of the machine tool with the cutter, and benefited to reach sustainable manufacturing. The
balance of the efficiency, energy consumption and carbon emissions can be helpful for the enterprises to
develop the cutting parameters and to relieve the impact to the environment.
© 2017 Elsevier Ltd. All rights reserved.

1. Introduction turning and milling (Li et al., 2011). But the average energy effi-
ciency of machining process is less than 30% (Liu et al., 2014a).
Due to the urgent need for more environmentally friendly Therefore, it is of great importance to research how to decrease the
manufacturing, it is essential to consider sustainable energy consumption and carbon emissions produced during
manufacturing to reduce energy consumption and carbon emis- machining processes in order to relive the impact to the
sions. As an important part of national industry, manufacturing environment.
consumes a large quantity of energy and resources, leading to large Energy savings up to 6e40% can be realized by optimizing cut-
amounts of emissions (Du et al., 2015). Manufacturing operations ting parameters, tools and optimum tool path design (Newman
account for as much as 19% of greenhouse gas (GHG) emissions et al., 2012). The cutting parameters directly affect the production
(Herzog, 2009). More than 99% of the environmental impact asso- efficiency, the production cost and the quality of the product. In
ciated with manufacturing is due to the consumption of electricity addition, the material removal power and tool life are also
used by machine tools in discrete part machining processes such as impacted by the cutting process parameters. The relationships
between the processing time, power energy consumption and the
carbon emissions with the milling process parameters were
* Corresponding author. Hunan Province Key Laboratory of High Efficiency and researched during dry milling processes performed on a modern
Precision Machining of Difficult-to-Cut Materials, Hunan University of Science and CNC machining center in this paper. The optimized solution and
Technology, Xiangtan, 411201, China.
E-mail address: edeng0080@vip.sina.com (Z. Deng).
program of cutting parameters was proposed to trade off three

http://dx.doi.org/10.1016/j.jclepro.2017.01.077
0959-6526/© 2017 Elsevier Ltd. All rights reserved.
H. Zhang et al. / Journal of Cleaner Production 148 (2017) 174e184 175

responses (the processing time, the power energy consumption tool life (Camposeco-Negrete, 2013). Bhushan (2013) minimized
and the carbon emissions), so that it can relive the impact to the power consumption and maximized tool life during CNC turning
environment and do not significantly affect the production machining operations. Ahilan et al. (2010) presented a multi-
efficiency. response optimization using the grey-based Taguchi method with
the responses being surface roughness and power consumption.
2. Literature review Yan and Li (2013) presented a multi-objective optimization method
to model the relationships between material removal rate, cutting
Sustainable machining processes can be achieved by reducing energy and surface roughness. However, the afore-mentioned
the power consumption (Camposeco-Negrete, 2013). If the energy optimization for machining processes were mainly based on
consumption is reduced, the environmental impact generated from machining science and economic considerations and did not
power production is correspondingly diminished (Pusavec et al., consider carbon emissions.
2010a). Therefore, most studies focused on the evaluation of As low carbon and high efficiency manufacturing is becoming
manufacturing processes based on operating state and components more sophisticated, more researchers are focusing on using ma-
of the machine tools, thermal equilibrium theoretical approaches, chine tools for sustainable manufacturing. Heddeghem et al. (2012)
and empirical modeling of machine tools. Draganescu et al. (2003) provided a mathematical model for calculating carbon footprints
attempted to establish relationships between the machine tool and proposed a manufacturing footprint and geographical region
efficiency, specific energy consumption, energy consumption and parameter. Jeswiet and Nava (2009) proposed a carbon emission
different machining parameters. Liu et al. (2014a) proposed a signature method to calculate the total carbon emissions from
practical method to predict the energy consumption of the main electricity consumption in a manufacturing process. Branker et al.
drive system during a machining process. A theoretical power (2011) presented a microeconomic machining model to optimize
consumption equation based on a thermal equilibrium approach the machining parameters with respect to carbon emissions and
was introduced by Gutowski et al. (2006). An empirical approach to cost sensitivity. Cao et al. (2012) presented a carbon efficiency
develop unit process energy consumption models for material approach to characterize the life-cycle carbon emissions of
removal processes was presented by Kara and Li (2011). For all machining process, and proposed strategies to reduce carbon
these models, the ultimate goal was energy savings. More work is emissions by improving energy efficiency and matching production
needed to determine environmental impacts. tasks to appropriate equipment. Liu et al. (2014b) presented a
Traditional machine process optimization goals included time, multi-objective optimization model to simultaneously minimize
cost and quantity (Anderberg et al., 2010). The used commonly cutting energy and surface roughness and maximum material
optimization criteria were material removal rate, surface rough- removal rate. Li et al. (2015) presented a carbon emission objective
ness, cutting force, tool life and power consumption (Yan and Li, function model that included electric power, cutting tools and
2013). Rangarajan and Dornfeld (2004) argued that there was an cutting fluid, and proposed an optimization model to minimize the
optimal orientation that minimized cycle time for machining pro- carbon emissions associated with drill machining. Li et al. (2013)
cess. Mori et al. (2011) measured the power consumption of a presented a multi-optimization model to minimize processing
certain machine center in various conditions. They summarized time and carbon emissions. Li and Liu (2015) established an opti-
that modifications to the cutting conditions could decrease energy mization model that reduced low carbon emissions and chatter
consumption, but excessive cutting speeds and feed rates would stability to couple high unit milling profits. Those studies analyzed
exacerbate tool wear. Mativenga and Rajemi (2011) presented a the relationships between the targets and the cutting parameters.
methodology for selecting the optimum cutting parameters for the With sustainable manufacturing requirements being adopted,
minimum energy footprint for a machined component. An empir- reducing carbon emissions is becoming an active research topic.
ical model was presented by Li and Kara (2011) to characterize the The operational range of cutting parameters must be carefully
relationships of the specific energy consumption with processing selected to prevent unacceptable decreases in the quality of
parameters. The specific energy consumption could vary consid- machined parts and excessive material removal rate, as these could
erably for a given material and was affected by the cutting speed, lower productivity (Pusavec et al., 2010b). Liu et al. (2015) devel-
feed, tool rake angle, etc. (Boothroyd and Knight, 1989). A oped an algorithm based on the non-dominated sorting genetic
machining cost model was proposed by Anderberg et al. (2010), algorithm II to solve the mathematic model in order to minimize
which compared machining costs and energy consumption with the makespan and the total carbon dioxide emission. Campatelli
cutting parameters in a CNC machining environment. A mathe- et al. (2014) utilized the response surface method to analyze the
matical model was proposed by Shroufa et al. (2014) to minimize effect of cutting parameters on energy consumption during dry
energy consumption costs for single machine production sched- milling process. Kant and Sangwan (2014) utilized grey relational
uling during production processes. Significant works have been analysis coupled with principal component analysis and response
performed to optimize cutting parameters based on machining surface methodology to obtain the optimum machining parame-
science and economic considerations. However, single objective ters. It is more reasonable to optimize cutting parameters based on
approaches have limited value with regards to fixing the optimal both environment sustainable and economic objectives (Yan and Li,
cutting parameters. For this, several different and contradictory 2013). However, the complex relationships between economic
objectives must be simultaneously optimized. Hence, multi- objectives (such as processing time and energy consumption) and
objective approaches to consider several different and contradic- carbon emission with cutting process parameters for dry cutting
tory objectives were applied for cutting parameter optimization. A are less researched. Therefore, in this paper, the complex re-
multi-objective mathematical programming model was proposed lationships between three responses with milling parameters by
by Mouzon et al. (2007) to minimize the energy consumption and principal component analysis and regression analysis based on
total completion time using operation dispatching rules. A math- experimental data were investigated, and a machining parameter
ematical model was proposed by Yildirim and Mouzon (2011) to optimization method for multi-pass dry milling processes to eval-
reduce total completion time and minimize the energy consump- uate the balance of three responses was proposed. Experiments
tion in a single machine. Machining parameters such as cutting were performed to verify the feasibility of the optimization
speed, feed rate, depth of cut and nose radius were optimized by method, and the results were analyzed to provide detailed process
multi-response considerations, namely, power consumption and parameters selection strategies.
176 H. Zhang et al. / Journal of Cleaner Production 148 (2017) 174e184

3. Optimization models for multi-pass dry milling during waiting periods, the operational phase energy consumption
Eui before the tool will be contact with the workpiece, the cutting
The power consumed during milling processes was mainly phase energy consumption Eci, and the tool changing energy con-
affected by the cutting parameters (Peng and Xu, 2014). The same sumption Etool. These are summed as follows:
may be said for the tool life. Therefore, the milling process pa-
rameters (cutting speed yc, feed rate f, cutting depth ap, cutting X
N X
N X
N

width ae) were taken as optimized variables, although the influence


Ein ¼ E0 þ Ewi þ Eui þ Eci þ Etool ; (3)
i¼1 i¼1 i¼1
weights of each variable were different.
where i is the pass number of the milling process and N is the total
3.1. Optimization of objective function number of passes.
The variable tttc is the tool changing time, which is treated as a
The lowest energy, high efficiency (the minimum processing fixed value, Ti is the tool life during the i-th pass of the milling
time) and lowest carbon emission functions were separately con- process, in min, tdi is the time spent idle during the i-th pass milling
structed while the characteristics of the power energy consumption process, twi is the time spent in the air cutting phase prior to contact
and carbon emissions were analyzed. Then a optimization model with the workpiece during the i-th pass milling process, and tci is
was proposed for multi-pass dry milling processes to evaluate the the time spent milling process during the i-th pass milling process.
balance of three responses. To simplify the calculation, the changing tool power is considered
to be equal to idle power, thus, Equation (3) becomes:
3.1.1. High efficient and low energy objective functions
X
N X
N N 
X
The power consumed during machine tool milling is shown in
Fig. 1. It is clear that the machining power is a non-constant value.
Ein ¼ E0 þ P0 tdi þ ðP0 þ kni þ bÞtwi þ P0 þ kni þ b
i¼1 i¼1 i¼1
To establish the functional relationships between processing time
and energy consumption with milling process parameters, the
 X
N
tci
þ k1 nxc 1 f x2 axp3 axe4 tci þ tttc P0
input power Pin for the machine tool system is decomposed into 60Ti
i¼1
three parts: idle power P0, which has nothing to do with the milling
(4)
process parameters, operational power Pu, which is related mainly
to the cutting speed and the main driving system, and the cutting As the time for the machine startup and stop phases is very
power Pc, which is closely related to the milling process short compared to the process time, it is neglected. The variables tdi
parameters: and twi have nothing to do with the milling process parameters and
are approximated as a constant:
Pin ¼ Pu þ Pc ¼ P0 þ kn þ b þ Pc ; (1)
Qi
where Pu ¼ P0 þ kn þ b, b and k are coefficients related to the tci ¼ ; (5)
MRRi
machine tool (Balogun and Mativenga, 2013). The P0 is treated as a
fixed value (as it is only related to the machine tool, in W). The where Qi is the material removal volume during the i-th pass
variable n is the spindle speed in rpm. The cutting power Pc can be milling process in mm3 and MRRi is the material removal rate
estimated using the same form as the milling force in the empirical during the i-th pass milling process in mm3/s:
model (Lu and Sun, 2006):
yf ap ae 1000yc fap ae
Pc ¼ k1 yxc 1 f x2 axp3 axe4 ; MRRi ¼ ¼ ; (6)
(2) 60 60pD

where yc is the cutting speed in m/min, f is the feed rate in mm/r, ap where yf is the feed speed in mm/min and D is the diameter of the
is the cutting depth in mm, ae is the cutting width in mm and k1, x1, cutter head in mm. The total processing time tTotal is determined
x2, x3 and x4 are coefficients that can be determined based on the from the following expression:
experimental data.
Thus, the total energy consumption for a single milling process X
N XN
tci
tTotal ¼ N  td þ N  tw þ tci þ tttc : (7)
(Fig. 1) consists of the machine startup phase and stopping phase 60T i
i¼1 i¼1
energy consumption E0, the idle phase energy consumption Ewi

Fig. 1. Energy consumption composition of multi-pass milling process.


H. Zhang et al. / Journal of Cleaner Production 148 (2017) 174e184 177

From Equations (4)e(7), simultaneous high efficiency and low (3). The chips are cleaned from the machine tool and recycle. The
energy objective functions can be expressed as follows: environmental load caused by the recycling of raw materials
is 50% of the burden of the system and the receiving system

8 " #
>   XN
Qi XN y1
Qi yc f y2 ap
y3
>
> tTotal yc ; f ; ap ; ae ¼ min Nðtd þ tw Þ þ A þ Btttc
>
>
>
< y fa a
i¼1 c p e i¼1
yc f ap ae
2   3 (8)
>
> X XN Qi P0 þ kni þ b þ k1 yx1 f x2 ax3 ax4 X
>
>   N
c p e
N
Qi yy1 y2 y3
c f ap 5
> 4
: Ein yc ; f ; ap ; ae ¼ min C þ ktw
> ni þ A
yc f ap ae
þ BP0 tttc
yc f ap ae
i¼1 i¼1 i¼1

where A ¼ 60 pD pDCT for recycling materials (Yang et al., 2002). The carbon emis-
1000 , B ¼ 1000VB , and C ¼ E0 þ NP0 td þ NP0 tw þ Nbtw . The
variables A, B, E0, P0, D, tw and tttc are constants that are related to sion from chip generation may be given as:
the machine tool, the cutter and the workpiece. The coefficients in
CW ¼ Cpw  Q  r; (12)
the empirical models can be obtained from the metal cutting
manual or be identified from experimental data. where Cpw is the coefficient of the carbon emissions for the chips in
kg CO2/kg. For the work piece is steel, the Cpw is taken to be 10.2
3.1.2. Low carbon objective function kgCO2/kg (Yang et al., 2002; Yin, 2014), and r is the density of the
The carbon emissions associated with dry cutting processes work piece in kg/mm3 (taken as 7.95E-6 kg/mm3 here).
include mainly the carbon emission Cp from the electricity Combining Equations (9)e(12), the low carbon emission objec-
consumed by the machine tool system, the carbon emissions CTool tive function can be expressed as:
from tool wear, and the carbon emission Cw from the raw material  
minCO2Total yc ; f ; ap ; ae ¼ Cpe  minEin þ Cpt  W  B
consumption (which includes emissions associated with its fabri-
cation and transport). X yc f ap
N y1 y2 y3
 min þ Cpw  Q  r:
i¼1
yc fap ae
(1). In this paper, the energy consumed by the machine tools is
electric energy. The carbon emission coefficient Cpe is taken (13)
to be 0.7538 kg CO2/kWh from the China carbon trading
network (2015). Thus, the total carbon emissions generated
during the milling process can be expressed as: 3.2. Multi-objective model transforming
Cp ¼ Cpe  Ein (9)
Considering of the total processing time tTotal, the total energy
consumption of the machine tools Ein and the total carbon emis-
(2). Cutter tool wear during the milling process is inevitable. The sions CO2Total associated with the milling process, a multi-object
embodied energy in the cutters is consumed during wear of optimization model was built and expressed from the milling
the cutter as part of the cutting process. The tool life is process parameters:
shortened due to the non-use of lubricating fluid as part of  
minFun yc ; f ; ap ; ae ¼ ðmintTotal ; minEin ; minCO2 Total Þ: (14)
the dry milling process (Wang, 2000), and the tool life is
estimated using the extended Taylor formula: To facilitate solving the multi-objective optimization model, the
model could be transformed using the following equation in which
VB
Ti ¼ y y ; (10) the processing time and the total energy consumption and carbon
CT yc 1 f y2 ap3 emissions are weighted different value.

where VB is fixed flank wear (Astakhov, 2004; Li, 2012). CT is a   tmax  tTotal Emax  Ein
maxj yc ; f ; ap ; ae ¼ max  u1 þ  u2
constant which depends on the tool-workpiece combination tmax  tmin Emax  Emin
(Karandikar et al., 2014; Li, 2012), and CT, y1, y2 and y3 can deter- 
CO2max  CO2Total
mined via the manual (Lu and Sun, 2006) or identified from þ  u3 :
CO2max  CO2min
experimental data (Alauddin et al., 1997; Astakhov, 2004). Thus, the
carbon emissions from tool wear during the milling process may be (15)
expressed as:
where J is comprehensive optimization index, 0 < J  1. A greater
X X y y value of J, represents that the time, energy consumption and
N
tci N
Qi yc 1 f y2 ap3
CTool ¼ Cpt  W  ¼ Cpt  W  B  ; carbon emissions are closer to the minimum values and indicates
i¼1
60T i¼1
yc fap ae better performance in terms of high efficiency, low energy con-
(11) sumption and low carbon emissions. The variables tmin, Emin and
CO2min are the minimum processing time, the minimum energy
where Cpt is the carbon emission coefficient of the cutter and W is consumed and the lowest carbon emissions, respectively, that
the weight of the cutter. The cutter used in this paper is a Sandvik satisfy the constraints during the milling process. Correspondingly,
R290-12t308M. The value of Cpt is taken to be 29.6 kgCO2/kg tmax, Emax and CO2max are the maximum values for the same con-
(Dahmus and Gutowski, 2004; Li et al., 2015) and W is 20 g. ditions. The variables u1, u2 and u3 are the weight of the total
178 H. Zhang et al. / Journal of Cleaner Production 148 (2017) 174e184

processing time tTotal, the total energy consumption of the machine (2). Cutter constraints
tools Ein and the total carbon emissions CO2Total, respectively.
Considering high efficiency and low energy consumption, the If the cutter is changed too frequently, it affects the continuity
weight of process time and energy consumption are equal and precision of the machining process. Trail milling should be
(u1 ¼ u2). Considering the sustainable manufacturing, the weight of performed to select adaptive tool life and the cutting speed range
carbon emissions should be equal to the total weight of the process scope:
P
time and energy consumption (u3 ¼ (u1 þ u2)). So, the weight of
process time, energy consumption and carbon emissions are sup- Tmin < T < Tmax ; (17)
posed as 0.25, 0.25 and 0.5 respectively.
where Tmin and Tmax are the minimum and maximum values of the
tool life recommended by the tool manufacturers and T is the
3.3. Constraints of the milling process predicted tool life.

The milling process parameters should be constrained by the (3). Constraints associated with machining quality
performances of the machine tool and the cutters. They should also
satisfy the requirements of machining quality. Therefore, the mill- The feed rate has a great influence on the surface roughness of
ing process parameters should be within a limited region of the the milled surface (Benardos and Vosniakos, 2003). According to
machine tool range. predicted surface roughness Ra, a feed rate range was determined
(Alauddin et al., 1995; White house, 1994):
(1). Machine tool constraints
ðf =ZÞ2
Ra ¼  Ramax ; (18)
During the milling process, the milling power cannot exceed the 32r
rated power of the designed machine tool. In addition, the milling
where Ra max is the maximum required finished surface roughness
speed should be in the range of the machine tool, and the milling
in mm, r is the radius of the cutter, f is the feed per revolution in mm/
force should meet the designed machine tool torque:
rad, and Z is the number of teeth in the milling cutters.
8
>
> Pc ¼ Fc yc  Pe h
<
Fc D=2 < Te
(16) 4. Identify the coefficients of empirical model functions
>
> yf min < yf < yf max
:
nmin < yc =ðpDÞ < nmax
The relationships between operational power, cutting power
and tool life with the milling process parameters were analyzed
where Pe is the rated power of the designed machine tool, h is the
based on the experiments conducted in this work. The operational
power efficiency of the machine tool, Te is the designed machine
power empirical model Pu ¼ P0 þ kyc/(pD) þ b, cutting power
tool torque, D is the diameter of the milling cutter head, nmin and
nmax are the minimum and maximum values of the spindle speed, empirical model Pc ¼ k1 yxc 1 f x2 axp3 axe4 and predicted tool life model
respectively, and yfmin and yfmax are the minimum and maximum T ¼ VB =CT yyc 1 f y2 ayp3 were identified by the objective functions. The
feed speeds, respectively. procedure to confirm the models is shown in Fig. 2. The identified

Fig. 2. The procedure used to identify the empirical model function coefficients.
H. Zhang et al. / Journal of Cleaner Production 148 (2017) 174e184 179

models were adapted for the next optimization example.

4.1. Experiment introduction

The milling experiments were carried out using a KVC800ma-


chining center (Fig. 3). The rated power of the spindle motor was
9 kW, the spindle speed could be varied from 20 to 6000 rpm, and
the feed speed could be varied from 0 to 10 m/min according to the
documents that came with the machining center. The workpieces
were medium-carbon steel blanks (C45, Fig. 4). Sandvik R290-
12t308M cutters (hard alloy coating cutter) were installed on the
cutter head. The tools had a diameter of 100 mm (Fig. 4). The power
demand of the milling process was measured using three AWS2103
sensors AWS2103 (Fig. 5) connected to a computer. The force
dynamometer used to measure the cutting forces was a Kistler
9257B (Fig. 6). Scanning electron microscopy (Fig. 7) was performed
using a JSM6700 to observe the worn length of the cutters.

4.2. Identification of the air cutting empirical model coefficients


Fig. 5. AWS2103 power sensor.
Fifteen fetch experiments were designed using cutting speeds in
the range from 31.4 to 417.2 m/min. Each fetch experiment was

Fig. 6. Kistler 9257B force dynamometer.

Fig. 3. KVC800 machining center.

Fig. 7. JSM6700 scanning electron microscopy.


Fig. 4. Tool and clamped workpiece.
180 H. Zhang et al. / Journal of Cleaner Production 148 (2017) 174e184

tested three times. Fig. 8 shows that the operational power in- Table 1
creases with the cutting speed. The relationship between the Factors and levels of the orthogonal experiment.

operational power and the cutting speed was approximately linear. Factors Level1 Level2 Level3 Level4 Level5
The operational power empirical model was identified via regres- yc,m/min 125.66 188.50 251.33 314.16 376.99
sion analysis, as shown in Fig. 2. The fitting is shown in Fig. 9 and yf,mm/min 400 500 600 700 800
the empirical model function was expressed as: ap,mm 0.1 0.15 0.2 0.25 0.3
Pu ¼ 1880 þ 1.268yc þ 183 (Goodness of fit: R ¼ 0.976), ae,mm 10 20 30 40 50

yc ¼ 31.4e471.2 m/min, and k ¼ 0.395.

could be predicted (Kwon and Fischer, 2003) based on the tool


4.3. Empirical function coefficients for cutting power and tool life
worn observed using JSM6700 scanning electron microscopy. The
The effects of the milling parameters on the cutting power and fitting parameters T ¼ 168:89ðyc =60Þ1:847 f 0:163 a0:379
p (Goodness
on the tool life were analyzed through an orthogonal design with of fit: R2 ¼ 0.971) are shown in Fig. 11.
four factors and five levels (Table 1). Each set was tested three
times, and the average value was used to analyze the main effect of 5. Optimization example
the milling process parameters by ANOVA. Based on the experi-
mental data (Table 2), cutting power and predicted tool life 5.1. Detail of workpiece to mill and experiments
empirical models were identified by regression analysis, as
described in Fig. 2. One seat of a certain machine tool was multi-pass dry milled on
The fitting parameters Pc ¼ 22:71ðyc =60Þ1:089 f 0:627 a0:526
p a0:753
e the same KVC800 machining center. The optimization model was
(Goodness of fit: R2 ¼ 0.981) are shown in Fig. 10. Assuming that the applied to select a superior set of milling process parameters. The
fixed flank wear VB was equal to 0.3 mm (Li, 2012), the tool life validity of the identified models was then assessed using trail

Fig. 8. Measured operational power at different cutting speeds.

Fig. 9. Fitting figure for operational power function.


H. Zhang et al. / Journal of Cleaner Production 148 (2017) 174e184 181

Table 2
Orthogonal experiment design and their responses.

No yc, m/min f, mm/r ap, mm ae, mm Cutting power Pc,W Tool life T, min

1 125.66 1.000 0.1 10 91.96 86.4


2 125.66 1.250 0.3 30 425.00 56.2
3 125.66 1.500 0.25 50 637.31 65.1
4 125.66 1.750 0.2 20 282.33 68.3
5 125.66 2.000 0.15 40 478.42 72.3
6 188.50 0.667 0.3 50 551.72 30.0
7 188.50 0.833 0.25 20 302.75 37.1
8 188.50 1.000 0.2 40 493.87 40.1
9 188.50 1.167 0.15 10 165.18 34.4
10 188.50 1.333 0.1 30 381.40 44.1
11 251.33 0.500 0.25 40 576.46 19.2
12 251.33 0.625 0.2 10 221.43 23.8
13 251.33 0.750 0.15 30 398.88 22.6
14 251.33 0.875 0.1 50 517.64 20.0
15 251.33 1.000 0.3 20 593.32 14.0
16 314.16 0.400 0.2 30 429.78 15.7
17 314.16 0.500 0.15 50 676.31 16.1
18 314.16 0.600 0.1 20 310.73 20.5
19 314.16 0.700 0.3 40 1010.67 9.8
20 314.16 0.800 0.25 10 319.42 14.6
21 376.99 0.333 0.15 20 300.30 9.9
22 376.99 0.417 0.1 40 434.56 13.4
23 376.99 0.500 0.3 10 291.02 8.5
24 376.99 0.583 0.25 30 811.70 7.0
25 376.99 0.667 0.2 50 1081.09 10.3

Fig. 10. Fitting figure for the cutting power.

Fig. 11. Fitting figure for tool life.


182 H. Zhang et al. / Journal of Cleaner Production 148 (2017) 174e184

milling experiments.
The seat was welded using C45 medium-carbon steel blanks and
was rough machined prior to semi finish and finish milling. The
residual amount to be machined was 3 mm and the roughness Ra of
the finished surface was 1.6 mm. Due to the constraints, the finish
milling process parameters were initially determined as
yc ¼ 188.50e314.16 m/min, f ¼ 0.25e0.75mm/r, ae ¼ 10e40 mm,
semi-milling ap ¼ 0.25e0.5 mm, and fine milling
ap ¼ 0.15e0.25 mm. The cutter was Sandvik R290-12t308M (hard
alloy coating cutter).
The measured data from the trail milling experiments were
shown in Table 3. The finished surface roughness from the trail
experiments satisfied the quality requirement. The errors in cutting
power were small (less than 10%) between the measured cutting
power and that predicted from the empirical model. The deviation
of the predicted tool lives from the measured tool lives was also
small (less than 10%). Therefore, the results proved the feasibility of
the identified models.
The tool path for milling the seat was shown in Fig. 12. Each pass
milling processing started from point 1 and proceeded sequentially
to point 16 before stopping. The total milling process consisted of
multi-pass semi finish milling and only one finish milling pass. The
depth of the milling should satisfy the relation
PN1
i¼1 ða Þ
psemi­finish i þ apfinish ¼ 3.
The range of milling parameters were defined according to the
milling experiments when the machine tool and the cutter were
chose. The genetic algorithm was adopted to solve the optimized
problem for the good global search ability and the operation flow Fig. 13. Operation flow chart for the genetic algorithm.

Table 3
Trail milling experiments data.

No yc m/min ap mm f mm/r ae mm Cutting power Pc, W Tool life T, min Roughness Ra, mm Cutting power predicted, W Tool life predicted, min

1 188.50 0.15 0.25 20 122 48.9 1.327 117 52.5


2 188.50 0.20 0.50 20 226 39.5 1.593 209 42.0
3 188.50 0.25 0.38 20 205 37.8 1.472 197 40.5
4 251.33 0.15 0.38 20 203 26.7 0.382 206 28.9
5 251.33 0.20 0.25 20 178 25.6 0.362 185 27.6
6 251.33 0.25 0.50 20 347 21.6 1.192 322 22.7
7 314.16 0.15 0.50 20 334 18.2 0.714 314 18.2
8 314.16 0.20 0.38 20 327 15.6 0.693 305 17.1
9 314.16 0.25 0.25 20 290 15.4 0.234 266 16.8

Fig. 12. Workpiece size and tool path.


H. Zhang et al. / Journal of Cleaner Production 148 (2017) 174e184 183

Table 4
Constants of the optimization models.

Variables Q, mm3 E0, kW h P0, W td, s tw, s tttc, s D, mm k VB, mm CT r


Values 1.0E5 0.033 1880 180 20 90 100 0.395 0.3 1.78  103 7.95 E-6 kg/mm3

Table 5
Optimized objectives and the milling process parameters corresponding.

Objective functions Optimized values yc,m/min f,mm/rad ap,mm ae,mm Notes

Comprehensive index: J 0.992 314.16 0.75 0.472 40 1ste6th pass semi-finish


Processing time: tTotal 2845 s 308.19 0.75 0.168 40 Fine milling
Energy consumption: Ein 2.27 kW h
Carbon emissions:CO2Total 27.78 kgCO2

chart for the genetic algorithm was shown in Fig. 13. Then the actual values. When the machining area of the workpiece was
milling parameters were optimized using MATLAB software to larger, the optimization objective values were closer to the actual
trade off three responses after the optimized values of single re- values.
sponses were got. The constant coefficients and variables are
shown in Table 4. The rotational speed of the spindle should be an 6. Conclusions
integer.
The relationships between high efficiency, low energy con-
sumption and low carbon emissions with the milling process pa-
5.2. Results and discussions
rameters were analyzed experimentally in this paper. And an
optimization model to trade off process time, energy consumption
The contribution and effects of the milling parameters on the
and carbon emissions was proposed to relieve the impact to the
three responses were investigated via the measured results. In
environment. While the coefficients were identified by principal
general, large feed rates f and large cut widths ae can shorten the
component analysis and regression analysis, an optimization
processing time and can reduce energy consumption and carbon
example was presented to demonstrate the feasibility of the pro-
emissions if the requirements of machining quality can be properly
cess parameter optimization method to trade off three responses.
met.
Process time, energy consumption and carbon emissions could
When the optimization goals were high efficiency and low en-
directly be affected by the cutting parameters. Large feed rates and
ergy consumption, the cutting speed yc should be high and the
large cut width could benefit to three responses if the constrains of
cutting depth ap should be large to shorten the processing time and
milling process were met with. Large cutting speed and large cut-
reduce energy consumption if the finished surface roughness can
ting depth could reduce the processing time and energy con-
meet with the requirements.
sumption, but could shorten the tool life and increase carbon
When the optimization goal was low carbon emissions, the
emissions. The balance of the efficiency, energy consumption and
choice of cutting speed yc should be based on the combined effects
carbon emissions could be useful to the optimum performance of
of tool wear and energy consumption. As the cutting speed
the machine tool with the cutter, and be helpful to reach green high
increased, the cutter wear became more severe and the carbon
efficiency milling.
emissions from tool wear increased, whereas overall energy con-
Because of the complexity of the milling process, the impact of
sumption decreased at the same time. Thus, the total carbon
the tool path was ignored. However, the time required for the tool
emissions did not necessarily decreased with the increase to the
to transverse the workpiece cannot be ignored if the shape of the
cutting speed.
workpiece was complex. Therefore, future work should focus on
When the optimization goal was to trade off three responses,
constructing an optimization model for complex shape workpieces.
the selecting of the cutting speed and the cutting depth should
consider the complex effect if the constraints are to be met. From
Acknowledgment
the results, processing time, energy consumption and carbon
emissions increased slightly compared to their respective mini-
This research was supported in part by the project
mum values (tmin ¼ 2800s, Emin ¼ 2.24 kW h, CO2min ¼ 27.56 kg)
2014AA041504 which was funded by the National High Technology
when the comprehensive optimization index was maximized,
R&D Program (863 Program) of China and was also supported by
which indicated the best simultaneous performance of the three
the project CX2015B430 which was funded by Hunan Provincial
responses. The increases were 1.60%, 1.01% and 0.78%, separately,
Innovation Foundation for Postgraduate. This research was sup-
where tTotal ¼ 2845s, Ein ¼ 2.27 kW and CO2Total ¼ 27.78 kg. This
ported in part by Graduate Student Innovation Talent Cultivation
indicated that the effect of the milling parameters was significant.
Fund Project from Advanced Mining and New Energy Equipment of
The milling process parameters were shown in Table 5.
Hunan Province of 2011 Collaborative Innovation Center too.
The milling process parameters optimized in Table 5 were used
to test. The deviations between theoretical value and measured
value of cutting power Pc were 1.2% and 0.63%, respectively, and References
the deviation of energy consumption was 0.8%. The feasibility of the
Ahilan, C., Kumanan, C., Sivakumaran, N., 2010. Application of grey based Taguchi
proposed optimization model was verified by experimentation. method in multi-response optimization of turning process. Int. J. Adv. Prod.
One solution to optimize the milling process parameters was Manag. 5, 171e180.
described. Due to neglecting the effect of the milling tool path, the Alauddin, M., Baradie, M.A.E1, Hashmi, M.S.J., 1995. Computer-aided analysis of a
surface-roughness model for end milling. J. Mater. Process. Technol. 55,
minimum values for the processing time tmin, machine tool energy 123e127.
consumption Emin, and carbon emissions CO2min were less than the Alauddin, M., Baradie, M.A.E.1., Hashmi, M.S.J., 1997. Prediction of tool life in end
184 H. Zhang et al. / Journal of Cleaner Production 148 (2017) 174e184

milling by response surface methodology. J. Mater. Process. Teehnology 71, numerical simulation technologies. Int. J. Refactory Metals Hard Mater. 35,
456e465. 143e151.
Anderberg, S.E., Kara, S., Beno, T., 2010. Impact of energy efficiency on computer Li, C.B., Cui, L.G., Liu, F., Li, L., 2013. Multi-objective NC machining parameters
numerically controlled machining. Proc. Inst. Mech. Eng. Part B J. Eng. Manuf. optimization model for high efficiency and low carbon. J. Mech. Eng. China 49
224, 531e541. (9), 87e91.
Astakhov, V.P., 2004. The assessment of cutting tool wear. Int. J. Mach. Manuf. 44, Li, Y., Liu, Q., 2015. Service-oriented research on multi-pass milling parameters
637e647. optimization for green and high efficiency. J. Mech. Eng. China 51 (11), 89e98.
Balogun, V.A., Mativenga, P.T., 2013. Modeling of direct energy requirements in Li, W., Zein, A., Kara, S., Herrmann, C., 2011. Glocalized solutions for sustainability in
mechanical machining processes. J. Clean. Prod. 41, 179e186. manufacturing. In: Proceedings of the 18th CIRP International Conference on
Benardos, P.G., Vosniakos, G.C., 2003. Predicting surface roughness in machining: a Life Cycle Engineering. Braunschweig, Germany, pp. 268e273.
review. Int. J. Mach. Tools Manuf. 43, 833e844. Li, W., Kara, S., 2011. An empirical model for predicting energy consumption of
Bhushan, R.K., 2013. Optimization of cutting parameters for minimizing power manufacturing processes: a case of turning. Proc. Inst. Mech. Eng. Part B J. Eng.
consumption and maximizing tool life during machining of Al alloy SiC particle Manuf. 225 (9), 1636e1646.
composites. J. Clean. Prod. 39 (1), 242e254. Liu, F., Xie, J., Liu, S., 2014a. A method for predicting the energy consumption of the
Boothroyd, G., Knight, W.A., 1989. Fundamentals of Machining and Machine Tools, main driving system of a machine tool in a machining process. J. Clean. Prod.
second ed. Marcel Dekker, Inc., New York, USA. 105, 171e177.
Branker, K., Jeswiet, J., Kim, I.J., 2011. Green house gases emitted in manufacturing a Liu, C.G., Yang, J., Lian, J., Li, W.J., Evans, S., Yin, Y., 2014b. Sustainable performance
productea new economic model. CIRP Ann. Manuf. Technol. 60 (1), 53e56. oriented operational decision-making of a single machine systems with
Campatelli, G., Lorenzini, L., Scippa, A., 2014. Optimization of process parameters deterministic product arrival times. J. Clean. Prod. 85, 318e330.
using a response surface method for minimizing power consumption in the Liu, C., Dang, F., Li, W., et al., 2015. Prediction and optimization of machining pa-
milling of carbon steel. J. Clean. Prod. 66, 309e316. rameters for minimizing power consumption and surface roughness in
Camposeco-Negrete, C., 2013. Optimization of cutting parameters for minimizing machining. J. Clean. Prod. 105, 285e299.
energy consumption in turning of AISI 6061 T6 using Taguchi methodology and Lu, J., Sun, J., 2006. Metal Cutting Theory and Cutting Tool, fourth ed. China Machine
ANOVA. J. Clean. Prod. 53, 195e203. Press, Beijing, China.
Cao, H., Li, H., Cheng, H., Luo, Y., Yin, R., Chen, Y., 2012. A carbon efficiency approach Mativenga, P.T., Rajemi, M.F., 2011. Calculation of optimum cutting parameters
for life-cycle carbon emission characteristics of machine tools. J. Clean. Prod. 37, based on minimum energy footprint. CIRP Ann. Manuf. Technol. 60 (1),
19e28. 149e152.
China carbon trading network, 2015. China low Carbon Technology Fossil Fuel Po- Mori, M., Fujisima, M., Inamasu, Y., Oda, Y., 2011. A study on energy efficiency
wer Generation Voluntary Emission Reduction Project Regional Power Grid improvement for machine tools. CIRP Ann. Manuf. Technol. 60 (1), 145e148.
Benchmark Line Emission Factor [EB/OL]. http://www.tanjiaoyi.com/article- Mouzon, G., Yildirim, M.B., Twomey, J., 2007. Operational methods for minimization
14961-1.html. of energy consumption of manufacturing equipment. Int. J. Prod. Res. 45,
Dahmus, J.B., Gutowski, T.G., 2004. An environmental analysis of machining. ASME 4247e4271.
International Mechanical Engineering Congress & Exposition vol. 15, 643e652. Newman, S.T., Nassehi, A., Imani-Asrai, R., Dhokia, V., 2012. Energy efficient process
Draganescu, F., Gheorghe, M., Doicin, C.V., 2003. Models of machine tool efficiency planning for CNC machining. CIRP J. Manuf. Sci. Technol. 5, 127e136.
and specific consumed energy. J. Mater. Process. Technol. 141, 9e15. Peng, T., Xu, X., 2014. Energy-efficient machining systems: a critical review. Int. J.
Du, Y.B., Yi, Q., Li, C.B., Liao, L., 2015. Life cycle oriented low-carbon operation Adv. Manuf. Technol. 72, 1389e1406.
models of machinery manufacturing industry. J. Clean. Prod. 91, 145e157. Pusavec, F., Krajnik, P., Kopac, J., 2010a. Transitioning to sustainable pro-
Gutowski, T., Dahmus, J., Thiriez, A., 2006. Electrical energy requirements for ductionepart I: application on machining technologies. J. Clean. Prod. 18,
manufacturing processes. In: Proceedings of 13th CIRP International Conference 174e184.
on Life Cycle Engineering. Leuven, Belgium, pp. 623e628. Pusavec, F., Kramar, D., Krajnik, P., Kopac, J., 2010b. Transitioning to sustainable
Heddeghem, W.V., Vereecken, W., Colle, D., Pickavet, M., Demeester, P., 2012. productionepart II: evaluation of sustainable machining technologies. J. Clean.
Distributed computing for carbon footprint reduction by exploiting low- Prod. 18, 1211e1221.
footprint energy availability. Future Gener. Comput. Syst. 28 (2), 405e414. Rangarajan, A., Dornfeld, D., 2004. Efficient tool paths and part orientation for face
Herzog, T., 2009. World Greenhouse Gas Emissions: 2005. World Resources Insti- milling. CIRP Ann. Manuf. Technol. 53 (1), 73e76.
tute, Washington, D.C. Shroufa, F., Ordieres-Mere , J., García-Sa
nchez, A., Ortega-Mier, M., 2014. Optimizing
Jeswiet, J., Nava, P., 2009. Applying CES to assembly and comparing carbon foot- the production scheduling of a single machine to minimize total energy con-
print. Int. J. Sustain. Eng. 2 (4), 232e240. sumption costs. J. Clean. Prod. 67, 197e207.
Kant, G., Sangwan, K.S., 2014. Prediction and optimization of machining parameter Wang, X., 2000. Green machining technology in metal cutting. Chin. J. Mech. Eng.
for minimizing power consumption and surface roughness in machining. 36 (8), 6e14.
J. Clean. Prod. 83, 151e164. White house, D.J., 1994. Handbook of Surface Metrology. CRC Press.
Kara, S., Li, W., 2011. Unit process energy consumption models for material removal Yan, J., Li, L., 2013. Multi-objective optimization of milling parametersethe trade-
processes. CIRP Ann. Manuf. Technol. 60, 37e40. offs between energy, production rate, and cutting quantity. J. Clean. Prod. 52,
Karandikar, J.M., Abbas, A.E., Schmitz, T.L., 2014. Tool life prediction using Bayesian 462e471.
updating. Part 1: milling tool life model using a discrete grid method. Precis. Yang, J., Xu, Ch, Wang, R., 2002. Product Life Cycle Assessment Method and its
Eng. 38, 9e17. Application. Meteorological Press, China, Beijing.
Kwon, Y., Fischer, G.W., 2003. A novel approach to quantifying tool wear and tool Yildirim, M.B., Mouzon, G., 2011. Single-machine sustainable production planning to
life measurements for optimal tool management. Int. J. Mach. Tools Manuf. 43, minimize total energy consumption and total completion time using a multiple
359e368. objective genetic algorithm. In: IEEE Transactions on Engineering Management,
Li, A.P., Gu, Z.Y., Zhu, J., et al., 2015. Optimization of cutting parameters for multi- pp. 1e13.
pass hole machining based on low carbon manufacturing. Comput. Integr. Yin, R.X., 2014. The Study and Application of Process Planning Decision Model for
Manuf. Syst. 12 (6), 1515e1522. Low Carbon Manufacturing Based on Carbon Emissions Evaluation. Chongqing
Li, B., 2012. A review of tool wear estimation using theoretical analysis and University.

You might also like