当前位置:首页 >> 机械/仪表 >>

Energy consumption model and energy ef


Journal of Cleaner Production 104 (2015) 264e272

Contents lists available at ScienceDirect

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

clepro

A hybrid approach to energy consumption modelling based on cutting power: a milling case
N. Liu, Y.F. Zhang*, W.F. Lu
Department of Mechanical Engineering, National University of Singapore, 9 Engineering Drive 1, Singapore 117576

a r t i c l e i n f o
Article history: Received 11 December 2014 Received in revised form 16 April 2015 Accepted 13 May 2015 Available online 22 May 2015 Keywords: Energy ef?ciency Energy consumption modelling Cutting power Hybrid approach Slot milling process

a b s t r a c t
Soaring energy price, increasing environmental concerns, and stringent legislations have forced enterprises to reduce energy consumption in discrete parts manufacturing. In order to achieve this reduction, accurate estimation of energy consumption at the manufacturing planning stage is needed. In this paper, a new model for energy consumption prediction in machining processes is proposed and validated in experiments. Using the established cutting force model, cutting power at the tool tip is obtained analytically. Further, the relationship between total power consumed by the machine tool and cutting power at the tool tip is empirically characterized. To demonstrate the new model, this hybrid approach has been applied for machine power prediction in a slot milling process. A comparison study has also been conducted between the proposed model and existing models under various cutting conditions. The results show that the proposed model could achieve better prediction accuracy than other models. Moreover, a major limitation of the existing models is successfully addressed. By considering cutting power at the tool tip, the proposed hybrid model could re?ect the nature of material removal and is able to provide valuable information regarding the impact of speci?c cutting parameters on power consumption. This knowledge is important in evaluating operation parameters, process plans, and even production schedule in terms of energy ef?ciency. ? 2015 Elsevier Ltd. All rights reserved.

1. Introduction and state of the art With the depletion of natural resources of the planet earth, affordable energy sources have been one of the major interests from industry. Environmental concerns and stringent legislations have imposed pressure on manufacturing companies nowadays. It was reported that a large share of energy was consumed by the manufacturing industries (Du?ou et al., 2012; Park et al., 2009). Improvement in energy ef?ciency will not only bene?t manufacturing enterprises, but also help alleviate the worldwide pollution problem in the long run. As a consequence, much research has been conducted to improve the energy ef?ciency in manufacturing. In general, energy ef?ciency in discrete parts manufacturing can be considered at three levels, i.e., operation, process planning, and scheduling. In order to achieve energy ef?cient manufacturing,

* Corresponding author. Tel.: ?65 65162868; fax: ?65 67791459. E-mail addresses: liuning@u.nus.edu (N. Liu), mpezyf@nus.edu.sg (Y.F. Zhang), mpelwf@nus.edu.sg (W.F. Lu). http://dx.doi.org/10.1016/j.jclepro.2015.05.049 0959-6526/? 2015 Elsevier Ltd. All rights reserved.

a good understanding of the energy consumption is required. Previous research in this area generally follows two directions: (1) identi?cation or optimization of process parameters, and (2) energy consumption modelling of machine tools. In the ?rst category, process parameters are identi?ed and optimized to improve the energy ef?ciency in a certain unit process. Draganescu et al. (2003) investigated the ef?ciency surface response of various process parameters during a milling process based on which a statistic model was presented. Rajemi et al. (2010) derived an economic tool-life model for turning process meeting minimum energy requirements. Based on this model, several critical parameters were identi?ed to minimize the total energy consumption. Oda et al. (2012) tried to optimize the tool angles and cutting speed to minimize the energy consumption in a 5-axis milling process. Hana? et al. (2012) applied grey relational theory and Taguchi method to optimize the cutting parameters in the process of turning PEEK-C30 using TiN coated tools. Similarly, Taguchi methodology and ANOVA were employed by CamposecoNegrete (2013) to analyze cutting parameters in turning of AISI 6061 T6 for minimizing energy consumption. In CNC turning of Al alloy SiC particle composites, Bhushan (2013) optimized the

N. Liu et al. / Journal of Cleaner Production 104 (2015) 264e272

265

cutting parameters by integrating energy consumption and tool life. Recently, Campatelli et al. (2014) optimized the process parameters with response surface method in order to minimize the power consumption in a milling process. A systematic approach of process planning and scheduling for sustainable machining was presented by Wang et al. (2015). In their work, the process parameters in the milling process were optimized using several optimization algorithms to improve the energy ef?ciency. Research work in this category mainly focused on identifying critical parameters to reduce the total energy consumption. However, an accurate estimation method for power consumption of a speci?c process is lacking. In the second category, research aims at modelling the power consumption of a certain process on a particular machine. These models can be further classi?ed into two types: machine statebased models and process parameter-based models. The machine state-based models deal with the energy consumption by considering various machine states, such as idle, air cut, and spindle acceleration/deceleration. Avram and Xirouchakis (2011) estimated the use phase energy requirements of a machine tool system as the sum of spindle energy requirements for acceleration, rotating without cutting, cutting, and deceleration, respectively. Similarly, Mori et al. (2011) modelled the energy consumption as a sum of different power demands corresponding to basic operation, cutting, and accelerating/decelerating the spindle, respectively. He et al. (2012) further divided total energy consumption for NC machining into energy consumed by spindle, feed, tool, coolant pump, and components in ?x state, respectively. An improved model for direct electrical energy requirements in machining was presented by Balogun and Mativenga (2013). Recently, Aramcharoen and Mativenga (2014) linked the energy demand of a milling process to toolpaths by modelling key energy states of the milling machine tool. For these models to work accurately, an assumption is required, i.e., the power consumed by the common components in one stage would remain the same in other stages (e.g., the power consumption due to auxiliary components in idle stage would not change in cutting stage). However, since the working conditions are not necessarily the same in different stages, this assumption may not hold. Besides, implementation of these models in machining processes is not so straightforward. In the parameter-based models, a power consumption model is established as a function of certain process parameters. These models are promising due to their good accuracy and easy implementation in machining processes. The ?rst power consumption model in this subcategory was introduced by Gutowski et al. (2006, 2007). The power consumption (P) in a unit process is broken down into a constant idle part (Pidle) and a variable part that is proportional to the material removal rate (MRR) and can be expressed as

where C0 and C1 are the coef?cients related to the speci?c machine tool. This model has since been validated in milling process (Diaz et al., 2011; Kara and Li, 2011) and grinding process (Li et al., 2012). Noting that the spindle speed contributes signi?cantly to the total power for a machine tool with relatively low standby power, Li et al. (2013) presented an improved model as

SEC ? k0 ? k1

n 1 ? k2 MRR MRR

(3)

P ? Pidle ? k$MRR

(1)

where k is the coef?cient related to the speci?c process. However, this model lacks experimental validation. For a speci?c process on a particular machine tool, it fails to give a clear de?nition of each factor, which limits its practical application. The ?rst model that can be practically used for energy consumption prediction was presented by Li and Kara (2011). Using an empirical method, they found an inverse relationship between the speci?c energy consumption (SEC) and MRR for a CNC turning machine as shown in Equation (2).

where n is the spindle speed; k0, k1, and k2 are the coef?cients related to the machine tool. Both of the above models have been proved to have good accuracy level in machining experiments (Li et al., 2013; Li and Kara, 2011). Although those existing models have decent accuracy level, there are some puzzling issues that need further study. These models take MRR (Li and Kara, 2011) or MRR and spindle speed (Li et al., 2013) as the only factors. However, the same MRR may come from different combinations of parameter and different tooleworkpiece couples, even at the same spindle speed. Machining experiments show that SEC may change under different cutting conditions at the same MRR and spindle speed. Further, the prediction of cutting power consumed by the cutting forces has not been thoroughly studied. In previous research (Li et al., 2013; Li and Kara, 2011), it was generally believed that such cutting power was negligible because it only provided the minimal energy to remove the material. However, this was not validated by experiments. As the milling force model includes both the cutting component due to material removal and the edge component caused by rubbing or ploughing (Budak et al., 1996; Lee and Altintas, 1996), cutting power at the tool tip may contain not only the minimal energy to remove materials, but also other unproductive energy such as heat and sound. As such, the role of cutting power in milling processes may not be negligible as previously perceived. Besides, due to the empirical nature of these studies, it is dif?cult to answer directly the energy decomposition of the machine tools. The work presented in this paper was motivated by the following observations: (1) power consumption of a machine tool at the same MRR and spindle speed changes with combinations of cutting parameters (e.g., depth-of-cut and feedrate), (2) power consumption of a machine tool does not change signi?cantly even when MRR changes sharply, and (3) power consumption using the same machine tool changes with tooleworkpiece couples. Observation (1) contradicts with major existing models described in Eqs. (1)e(3), according to which the predicted power and SEC should remain the same. Meanwhile, observations (2) and (3) cannot be explained directly by these existing models. Such observations indicate that there could be some hidden factors that affect the power consumption. In order to reveal these factors, energy consumption modelling is carried out from a new viewpoint of cutting power that re?ects the nature of the material removal process. Based on cutting power, a hybrid energy consumption model is developed. The model can make full use of cutting power (analytical method) and measured data (empirical method) in a slot milling process. The novel power/SEC prediction model is validated in experiments and can address most of the problems mentioned above. It could also help to achieve sustainable manufacturing at different levels (operation, process planning, and scheduling).

2. Cutting power in slot milling process In the past decades, cutting force prediction models in milling process have been established. In this section, the adopted cutting

SEC ? C0 ?

C1 MRR

(2)

266

N. Liu et al. / Journal of Cleaner Production 104 (2015) 264e272

force model is brie?y introduced, followed by the derivation of cutting power in terms of process parameters. 2.1. The cutting force model Fig. 1 shows a ?at-end mill in a slot milling process. The differential cutting forces (N) on an in?nitesimal cutting edge in terea coordinates are expressed as (Budak et al., 1996; Lee and Altintas, 1996).

cutting power at the tool tip, the two motions are considered separately. Since both dFr and dFa are perpendicular to cutting speed in rotation motion, no power will be consumed by them. At the same time, dFt is in line with cutting speed. Therefore, the differential powers (W) consumed at each in?nitesimal cutting edge in tangential, radial, and axial directions are

dPn;t ? dFt $V dPn;r ? 0 dPn;a ? 0

(6a) (6b) (6c)

dFt ?j; z? ? Kte dS ? Ktc st sin j dz dFr ?j; z? ? Kre dS ? Krc st sin j dz dFa ?j; z? ? Kae dS ? Kac st sin j dz

(4a) (4b) (4c)

where Ktc, Krc, Kac (N/mm2) and Kte, Kre, Kae (N/mm) are the speci?c cutting and edge force coef?cients; st (mm/rev) is the feed per tooth; j (rad) is the radial immersion angle of the in?nitesimal cutting edge; dS (mm) is the length of the in?nitesimal cutting edge; z (mm) is the position in axial direction; dz (mm) is the differential length in axial direction. To obtain the resultant cutting forces in Cartesian (xeyez) coordinates, the cutting forces in Eq. (4) are resolved as

where V (m/s) is the cutting speed and V ? 2pnr/60. The total power consumed by the in?nitesimal cutting edge due to rotation motion can then be obtained as

dPn ? dFt $V
Similarly, in the feed motion, the differential powers are

(7)

dPf ;t ? dFt cos j$f =60; 000 dPf ;r ? dFr sin j$f =60; 000 dPf ;a ? 0

(8a) (8b) (8c)

2

?cos j dFx 4 dFy 5 ? 4 sin j dFz 0

3

2

?sin j ?cos j 0

0 dFt 0 54 dFr 5 dFa ?1

32

3 (5)

The total force can then be obtained by integrating all the in?nitesimal cutting edges. 2.2. Cutting power calculation In a slot milling process shown in Fig. 1, power is consumed at each in?nitesimal cutting edge to overcome the resistance due to material removal. There are mainly two kinds of motions related to a ?at-end mill. One is rotation driven by the spindle motor and the other is feed driven by the feed motors. In order to calculate

where f (mm/min) is the feedrate. The total power consumed by the in?nitesimal cutting edge due to feed motion is

dPf ? dFt cos j$f =60; 000 ? dFr sin j$f =60; 000
Considering Eq. (5), Eq. (9) can be represented as

(9)

dPf ? ?dFx $f =60; 000

(10)

With dPn and dPf, Pn and Pf (W) can be obtained by integrating all the in?nitesimal cutting edges engaged in cutting. Finally, cutting power (W) at the tool tip can be given by

Z Z Pcutting ? Pn ? Pf ? dPn ? dPf Z Z ? V $ dFt ? f =60; 000$ ??dFx ?

(11)

This is an instantaneous power at a certain rotation angle. In this paper, however, it is the average power rather than the instantaneous power that is of concern because the rotation period is too short for the measurement device to capture. The average cutting power (W) during one period is

Pcutting

1 ? fp

Zfp Pcutting ? Pn ? Pf
0

(12)

where fp (rad) is the pitch angle de?ned by fp ? 2p/N with N denoting the number of ?utes of the ?at-end mill; Pn , Pf ?W? are the average powers due to rotation and feed, respectively.

3. Experiment details The experiment was performed on a 3-axis CNC vertical milling centre under various cutting conditions. Sections 3.1 and 3.2 describe the experimental design and measurement details,

Fig. 1. A ?at-end mill in a slot milling process.

N. Liu et al. / Journal of Cleaner Production 104 (2015) 264e272

267

Fig. 2. Speci?cations of SSM 2000 ?at-end mill (Sumi Tool, 2014).

respectively. Section 3.3 shows how the hybrid energy consumption model is obtained. 3.1. Design of experiment The experiments were carried out on a 3-axis CNC vertical milling centre, Makino-V55, with an 18.5 kW spindle motor and a maximum spindle speed of 14,000 rpm. As shown in Fig. 2, a solid carbide ?at-end mill (SSM 2000) was used in this experiment for the slot milling under dry cutting condition. The 2-?ute cutter has a helix angle of 30 and a radius of 5 mm. The workpiece was a block of ASSAB 760 steel with a dimension of 273 mm ? 50 mm ? 90 mm. ASSAB 760 steel has a hardness of about 200 HB, tensile strength of

640 MPa, and a 0.2% proof stress of 340 MPa. Its nearest equivalents are AISI 1148, SIS 1650, and En43. In the ?rst set of experiments (Experiment-I), depth-of-cut (ap), feedrate (f), and spindle speed (n) were varied in three levels (see Table 1). There are 27 combinations of parameter in total. Based on Taguchi L9 orthogonal array, a total of nine parameter settings were selected for Experiment-I as shown in Table 2. In the second set of experiments (Experiment-II), depth-of-cut and feedrate were varied under the same MRR (20 mm3/s) and spindle speed (1500 rpm). As a result, a total of four parameter settings were selected. In all experiments, the force pro?le of the cutter and the power pro?le of the machine tool were collected. 3.2. Measurement

Table 1 Three levels of cutting parameters used in Experiment-I. Variables Depth-of-cut [mm] Spindle speed [rpm] Feedrate [mm/min] Level 1 2.0 1000 50 Level 2 3.0 1500 75 Level 3 4.0 2000 100

Fig. 3(a) shows the experiment setup in which the workpiece was mounted on the dynamometer table (see Fig. 3(b) for an enlarged view), where slot milling was conducted along x-axis. The forces in xeyez coordinates were measured with a 3-channel dynamometer (Kistler 9265B) at a sampling frequency of 48 kHz. The signals captured by the dynamometer were ?ltered at a low

Table 2 Cutting parameters and power readings in Experiment-I. No. 1 2 3 4 5 6 7 8 9 ap [mm] 2.0 2.0 2.0 3.0 3.0 3.0 4.0 4.0 4.0 n [rpm] 1000 1500 2000 1000 1500 2000 1000 1500 2000 f [mm/min] 50 75 100 75 100 50 100 50 75 MRR [mm3/s] 16.67 25.00 33.33 37.50 50.00 25.00 66.67 33.33 50.00 Pair [W] 1600 1598 1600 1601 1601 1603 1599 1600 1602 P [W] 1620 1630 1630 1633 1640 1630 1652 1633 1645 SEC [J/mm3] 97.20 65.20 48.90 43.55 32.80 65.20 24.78 48.99 32.90

268

N. Liu et al. / Journal of Cleaner Production 104 (2015) 264e272

Fig. 3. The experiment setup. (a): the workpiece mounted on the dynamometer table. (b): a closer view of the workpiece and the dynamometer is shown. (c): an example of the measured three-component force data (Fx, Fy, Fz) is shown.

Fig. 4. The installation block diagram of Eco-Power Meter KW9M.

pass frequency of 100 Hz and then processed in a PC with an A/D board connected to the dynamometer. Fig. 3(c) shows an example of the measured force pro?les within a period of time. The power consumption of Makino-V55 was measured with a Panasonic Eco-Power Meter KW9M. The installation block diagram is illustrated in Fig. 4. At block A, it was installed directly to the 3P4W (three phases, four wires) mains while at block B it was connected via three CTs (current transformers). The CTs can produce reduced currents (secondary currents) proportional to the original currents (primary currents) accurately. They are widely used when the original current in a circuit is too high to measure directly. In this experiment, the CTs used are MICRO 19 split core CTs (200:5 Type) from HOBUT. With such installation, good measurement accuracy was guaranteed. The power pro?le was recorded at a frequency of 10 Hz. In Table 2, all the power readings were averaged values over 10 s during steady states (excluding cutter entry and exit); Pair is the power reading during air cut mode and P is the power reading during cutting mode.

3.3. Energy consumption modelling For the prediction of cutting power, according to Eq. (11), both dFt and dFx are required. It is therefore essential to ?nd the cutting force coef?cients. This can be achieved by using the measured force data. The measured force pro?le consists of a large number of data points, with each corresponding to a force reading at a time instant. In this paper, the standard linear least squares (LLS) method was applied to determine the cutting coef?cients by using m selected points as

K ? ? Kte

Ktc

Kre

 ? 1 Krc ?T ? X T X XT d

(13)

P P P where X ?i? ? ??dS cos f st dz cos f sin f dS sin f P st dz sin f sin f? with i ? 1, 2, …, m; d is an m ? 1 vector with d(i) ? Fx(f); f is the angular position of the cutter. Kae and Kac are not considered because the axial force does not contribute to Ft. For

N. Liu et al. / Journal of Cleaner Production 104 (2015) 264e272

269

Fig. 5. Comparison between the predicted and measured pro?les of Fx. (a): comparison is conducted under setting 3 (ap ? 2 mm; n ? 2000 rpm; f ? 100 mm/min). (b): comparison is conducted under setting 8 (ap ? 4 mm; n ? 1500 rpm; f ? 50 mm/min).

implementation, 3000 data points of Fx from the collected data set under setting-1 in Experiment-I (ap ? 2 mm; n ? 1000 rpm; f ? 50 mm/min) were used to calculate the cutting force coef?cients. The result is K ? [10.11 3020.52 8.91 1493.53]T. To validate the obtained K, the predicted pro?les and measured pro?les of Fx under two different cutting conditions (setting-3 and setting-8 in Experiment-I) were compared and the results are shown in Fig. 5(a) and (b), respectively. It is worth noting that data collected from these two settings were not used in the calculation of K. Clearly, the predicted pro?les agree very well with the measured pro?les. Using Eq. (5), Ft can be calculated based on the obtained K, which will have the same level of accuracy. With Ft and Fx, cutting power can be obtained using Eqs. (11) and (12). The overall procedure is described as follows:

2011). This indicates the power consumed by the unproductive components in air cutting stage will decrease when the machine is switched to cutting mode. Hence, machine power consumption under cutting mode cannot be simply treated as the sum of Pair and Pcutting . Further, the power due to feed motion is very small and nearly all the power consumption at the tool tip is  caused by rotation motion. In addition, cutting power ef?ciency Pcutting =P ? increases from 3.85% to 13.67% as cutting power gets larger. In other words, when cutting power increases, cutting ef?ciency also increases. The authors are still in the process to ?nd explanations to these two phenomena. On the other hand, there seems a strong linear relationship existing between P and Pcutting (see Fig. 6) from Table 3. Based on this observation, a new energy consumption model is proposed as

For the nine parameter settings in Experiment-I, the calculated cutting forces and powers are shown in Table 3, in which Ft and Fx are the average cutting forces in tangential and feed direction; Pn andPf are the average powers due to rotation and feed;Pcutting is the average cutting power at the tool tip; Pgap is the power gap between normal cut and air cut; Pcutting =P is the cutting power ef?ciency. From Table 3, two interesting phenomena are observed. First, Pgap is smaller than Pcutting in all settings, which contradicts with the perceptions in the existing literature (Li et al., 2013; Li and Kara,

P * ? C0 ? C1 Pcutting Pcutting C0 ? C1 MRR MRR

(14a)

SEC * ?

(14b)

There are two coef?cients in the proposed power consumption model, C0 and C1. C0 is closely related to the air cut power; and C1 is the proportionality coef?cient of cutting power. For this particular slot milling process, C0 ? 1609.56 and C1 ? 0.18.

270

N. Liu et al. / Journal of Cleaner Production 104 (2015) 264e272 Table 3 Calculated cutting forces and power for the settings in Experiment-I. No. 1 2 3 4 5 6 7 8 9 Ft [N] 119.18 119.18 119.18 250.88 226.85 106.72 430.65 174.34 190.36 Pn [W] 62.40 93.60 124.80 131.36 178.17 111.76 225.49 136.93 199.34 Fx [N] ?50.42 ?50.42 ?50.42 ?103.63 ?94.29 ?47.61 ?175.51 ?75.93 ?82.16 Pf [W] 0.04 0.06 0.08 0.13 0.15 0.04 0.29 0.06 0.10 Pcutting [W] 62.44 93.66 124.88 131.49 178.32 111.80 225.78 136.99 199.44 Pgap [W] 20 32 30 32 39 27 53 33 43 P [W] 1620 1630 1630 1633 1640 1630 1652 1633 1645 Pcutting =P 3.85% 5.75% 7.66% 8.05% 10.87% 6.86% 13.67% 8.39% 12.12%

Fig. 6. A linear relationship between P and Pcutting is observed.

Table 4 Cutting parameters and power readings in Experiment-II. No. 1 2 3 4 ap [mm] 1.0 2.0 3.0 4.0 n [rpm] 1500 1500 1500 1500 f [mm/min] 120 60 40 30 MRR [mm3/s] 20 20 20 20 Pair [W] 1601 1600 1600 1601 P [W] 1621 1625 1625 1632 SEC [J/mm3] 81.05 81.25 81.25 81.60

Table 5 Comparison between predicted powers and actual power readings in Experiment-II. No. 1 2 3 4 Ft [N] 88.38 99.96 111.53 123.08 Pn [W] 69.41 78.51 87.59 96.67 Fx [N] ?36.43 ?42.95 ?49.48 ?56.02 Pf [W] 0.07 0.04 0.03 0.02 Pcutting [W] 69.48 78.55 87.62 96.69 P* [W] 1622.07 1623.70 1625.33 1627.00 P [W] 1621 1625 1625 1632 Accuracy 99.94% 99.92% 99.98% 99.70%

For other slot milling processes with a new workpieceetool couple, the proposed model is expected to be also applicable. It can be implemented with only a few runs of experiments. The cutting force coef?cients are determined with cutting force readings, and C0 and C1 can then be obtained with the calculated cutting power and the corresponding measured machine power readings. Compared to those existing empirical models, cutting power is considered directly for the ?rst time in the proposed model. In order to obtainPcutting , the speci?c parameter combination and workpieceetool couple in the process must be considered, which differentiates the proposed model from other empirical models.

4. Results and discussions For effectiveness testing, the proposed model was ?rstly validated using the data from Experiment-II. Subsequently, a comparison between the proposed model and well-known existing models was conducted using data from both Experiment-I and Experiment-II. 4.1. Model validation To validate the proposed model, a second set of experiments (Experiment-II) were carried out. All four parameter settings are shown in Table 4, together with the power readings of Pair and P

N. Liu et al. / Journal of Cleaner Production 104 (2015) 264e272 Table 6 Model-A and model-B based on data in Experiment-I. Model Power Model-A (Li and Kara, 2011) PA ? 0.4571MRR ? 1614.97 Model-B (Li et al., 2013) PB ? 0.4365MRR ? 0.0017n ? 1613.18

271

Table 7 Model comparison using data from Experiment-I. No. 1 2 3 4 5 6 7 8 9 Mean error P* [W] 1620.80 1626.42 1632.04 1633.23 1641.66 1629.68 1650.20 1634.22 1645.46 PA [W] 1622.59 1626.40 1630.21 1632.11 1637.83 1626.40 1645.44 1630.21 1637.83 PB [W] 1622.16 1626.64 1631.13 1631.25 1637.56 1627.49 1643.98 1630.28 1638.41 P [W] 1620 1630 1630 1633 1640 1630 1652 1633 1645 Error* 0.049% 0.21% 0.12% 0.01% 0.10% 0.01% 0.10% 0.07% 0.02% 0.08% ErrorA 0.16% 0.22% 0.01% 0.05% 0.13% 0.22% 0.40% 0.17% 0.44% 0.20% ErrorB 0.13% 0.21% 0.07% 0.11% 0.15% 0.15% 0.49% 0.17% 0.40% 0.21%

Table 8 Model comparison using data from Experiment-II. No. 1 2 3 4 Mean error P* [W] 1622.07 1623.70 1625.33 1627.00 PA [W] 1624.11 1624.11 1624.11 1624.11 PB [W] 1624.46 1624.46 1624.46 1624.46 P [W] 1621 1625 1625 1632 Error* 0.06% 0.08% 0.02% 0.30% 0.11% ErrorA 0.20% 0.05% 0.05% 0.48% 0.20% ErrorB 0.21% 0.04% 0.03% 0.46% 0.19%

Fig. 7. The prediction errors of the three models are compared using data from Experiment-II.

under each setting. It can be seen that, among all the settings, both MRR and spindle speed were kept unchanged while depth-of-cut and feedrate were varied in four levels. One can see that with the same MRR and spindle speed, P changes with different parameter settings. On the other hand, for the same four settings, the calculated cutting power ?Pcutting ? and total machine power (P*) using the proposed model are shown in Table 5. It can be seen that a high prediction accuracy of over 99% is achieved for all four cases, indicating the effectiveness of the proposed energy consumption model.

4.2. Model comparison To further test the effectiveness of the proposed model, a comparative study was carried out against two well-known existing empirical models: model-A (Li and Kara, 2011) and model-B

(Li et al., 2013), with data from both Experiment-I and Experiment-II. The coef?cients of the two empirical models (A and B) were ?rst obtained based on the experimental data in Experiment-I. The resultant formulae are shown in Table 6. The proposed model is then compared with model-A and model-B using the collected data in Experiment-I and ExperimentII, respectively. The comparison results, based on relative prediction error and mean error, are shown in Tables 7 and 8, respectively. In Table 7, where data from Experiment-I are used, it can be seen that the proposed model achieved lower mean error of 0.08% compared to model-A (mean error ? 0.20%) and model-B (mean error ? 0.21%). Using the data from Experiment-II (see Table 8), the proposed model (mean error ? 0.11%) also outperformed model-A (mean error ? 0.20%) and model-B (mean error ? 0.19%). Therefore, based on the two sets of experiments, the proposed model achieved smaller power prediction error than model-A and model-B. Nevertheless, it is also noted that model-A and model-B are fairly accurate, possibly because Makino V55 has a very large idle power (around 1600 W). Another interesting observation from Table 8 is that the power predicted by model-A and model-B remains the same for every parameter setting (due to the same MRR and n), while the actual power readings showed variation with different combinations of parameter (ap and f). Fig. 7 shows the relative prediction errors of these three models for the four settings in Experiment-II. It can be seen that model-A and model-B are equally accurate as the proposed model when ap and f are in moderate level (see setting-2 and setting-3 in Fig. 7). However, when ap and f are either too high or too low, model-A and model-B tend to be much less accurate. This reveals that MRR and n alone are not suf?cient to predict the power effectively. In contrast, by making use of cutting power at the tool tip, the proposed model effectively takes ap and f (as well as n) into

272

N. Liu et al. / Journal of Cleaner Production 104 (2015) 264e272 milling of carbon steel. J. Clean. Prod. 66, 309e316. http://dx.doi.org/10.1016/ j.jclepro.2013.10.025. Camposeco-Negrete, C., 2013. Optimization of cutting parameters for minimizing energy consumption in turning of AISI 6061 T6 using Taguchi methodology and ANOVA. J. Clean. Prod. 53, 195e203. http://dx.doi.org/10.1016/j.jclepro.2013. 03.049. Diaz, N., Redelsheimer, E., Dornfeld, D., 2011. Energy consumption characterization and reduction strategies for milling machine tool use. Glocalized Solut. Sustain. Manuf. 263e267. http://dx.doi.org/10.1007/978-3-642-19692-8_46. Draganescu, F., Gheorghe, M., Doicin, C.V., 2003. Models of machine tool ef?ciency and speci?c consumed energy. J. Mater. Process. Technol. 141, 9e15. http:// dx.doi.org/10.1016/S0924-0136(02)00930-5. Du?ou, J.R., Sutherland, J.W., Dornfeld, D., Herrmann, C., Jeswiet, J., Kara, S., Hauschild, M., Kellens, K., 2012. Towards energy and resource ef?cient manufacturing: a processes and systems approach. CIRP Ann. Manuf. Technol. 61, 587e609. http://dx.doi.org/10.1016/j.cirp.2012.05.002. Gutowski, T., Dahmus, J., 2007. A thermodynamic characterization of manufacturing processes. In: Electronics & the Environment, Proceedings of the 2007 IEEE International Symposium on, pp. 137e142. http://dx.doi.org/10.1109/ISEE.2007. 369382. Gutowski, T., Dahmus, J., Thiriez, A., 2006. Electrical energy requirements for manufacturing processes. In: 13th CIRP International Conference on Life Cycle Engineering, vol. 31, pp. 623e638. Hana?, I., Khamlichi, A., Cabrera, F.M., Almansa, E., Jabbouri, A., 2012. Optimization of cutting conditions for sustainable machining of PEEK-CF30 using TiN tools. J. Clean. Prod. 33, 1e9. http://dx.doi.org/10.1016/j.jclepro.2012.05.005. He, Y., Liu, F., Wu, T., Zhong, F.-P., Peng, B., 2012. Analysis and estimation of energy consumption for numerical control machining. Proc. Inst. Mech. Eng. Part B: J. Eng. Manuf. http://dx.doi.org/10.1177/0954405411417673. Kara, S., Li, W., 2011. Unit process energy consumption models for material removal processes. CIRP Ann. Manuf. Technol. 60, 37e40. http://dx.doi.org/10.1016/j.cirp. 2011.03.018. Lee, P., Altintas, Y., 1996. Prediction of ball-end milling forces from orthogonal cutting data. Int. J. Mach. Tools Manuf. 36, 1059e1072. http://dx.doi.org/10.10 16/0890-6955(95)00081-X. Li, L., Yan, J., Xing, Z., 2013. Energy requirements evaluation of milling machines based on thermal equilibrium and empirical modelling. J. Clean. Prod. 52, 113e121. http://dx.doi.org/10.1016/j.jclepro.2013.02.039. Li, W., Kara, S., 2011. An empirical model for predicting energy consumption of manufacturing processes: a case of turning process. Proc. Inst. Mech. Eng. Part B: J. Eng. Manuf. 225, 1636e1646. http://dx.doi.org/10.1177/20412975113 98541. Li, W., Winter, M., Kara, S., Herrmann, C., 2012. Eco-ef?ciency of manufacturing processes: a grinding case. CIRP Ann. Manuf. Technol. 61, 59e62. http:// dx.doi.org/10.1016/j.cirp.2012.03.029. Mori, M., Fujishima, M., Inamasu, Y., Oda, Y., 2011. A study on energy ef?ciency improvement for machine tools. CIRP Ann. Manuf. Technol. 60, 145e148. http:// dx.doi.org/10.1016/j.cirp.2011.03.099. Oda, Y., Mori, M., Ogawa, K., Nishida, S., Fujishima, M., Kawamura, T., 2012. Study of optimal cutting condition for energy ef?ciency improvement in ball end milling with tooleworkpiece inclination. CIRP Ann. Manuf. Technol. 61, 119e122. http:// dx.doi.org/10.1016/j.cirp.2012.03.034. Park, C., Kwon, K., Kim, W., Min, B., 2009. Energy consumption reduction technology in manufacturingda selective review of policies, standards, and research. Int. J. Precis. Eng. Manuf. 10, 151e173. http://dx.doi.org/10.1007/ s12541-009-0107-z. Rajemi, M., Mativenga, P., Aramcharoen, A., 2010. Sustainable machining: selection of optimum turning conditions based on minimum energy considerations. J. Clean. Prod. 18, 1059e1065. http://dx.doi.org/10.1016/j.jclepro.2010.01.025. Sumi Tool, 2014. Solid Carbide Spiral Endmills SSM 2000 Type. Wang, S., Lu, X., Li, X.X., Li, W.D., 2015. A systematic approach of process planning and scheduling optimization for sustainable machining. J. Clean. Prod. 87, 914e929. http://dx.doi.org/10.1016/j.jclepro.2014.10.008.

consideration and is able to predict power consumption more accurately. 5. Conclusions This paper presents a new machining energy consumption model as a function of cutting power at the tool tip. While cutting power can be obtained analytically through calculation of cutting forces, the relationship between total power and cutting power is obtained empirically using experiment data. This proposed model has been implemented and validated in a slot milling process (under dry cutting condition) on a 3-axis CNC vertical milling centre (Makino V55). A comparison study showed that the proposed model could achieve better prediction accuracy than existing models considering either MRR only or both MRR and spindle speed. Moreover, the proposed model successfully addresses one major limitation of the existing models, which cannot explain the phenomenon that machining under the same MRR and spindle speed may result in different power consumption. By considering cutting power at the tool tip, the proposed hybrid model could re?ect the nature of material removal and is able to provide valuable information regarding the impact of speci?c cutting parameters on power consumption. It therefore offers great potential in developing optimization strategies to improve energy ef?ciency at various levels of manufacturing planning (i.e., operation, process planning, and scheduling). The next stage of this study will extend the hybrid modelling method to other processes performed on the same and different machine tools. Study is underway to look for a comprehensive explanation to the phenomena observed in Section 3.3. Besides, energy consumption with non-uniform machining on complex geometry, such as mould cavity, will also be studied. References
Aramcharoen, A., Mativenga, P.T., 2014. Critical factors in energy demand modelling for CNC milling and impact of toolpath strategy. J. Clean. Prod. 78, 63e74. http:// dx.doi.org/10.1016/j.jclepro.2014.04.065. Avram, O.I., Xirouchakis, P., 2011. Evaluating the use phase energy requirements of a machine tool system. J. Clean. Prod. 19, 699e711. http://dx.doi.org/10.1016/ j.jclepro.2010.10.010. Balogun, V.A., Mativenga, P.T., 2013. Modelling of direct energy requirements in mechanical machining processes. J. Clean. Prod. 41, 179e186. http://dx.doi.org/ 10.1016/j.jclepro.2012.10.015. Bhushan, R.K., 2013. Optimization of cutting parameters for minimizing power consumption and maximizing tool life during machining of Al alloy SiC particle composites. J. Clean. Prod. 39, 242e254. http://dx.doi.org/10.1016/j.jclepro. 2012.08.008. Budak, E., Altintas ? , Y., Armarego, E.J.A., 1996. Prediction of milling force coef?cients from orthogonal cutting data. J. Manuf. Sci. Eng. 118, 216e224. http://dx.doi.org/ 10.1115/1.2831014. Campatelli, G., Lorenzini, L., Scippa, A., 2014. Optimization of process parameters using a response surface method for minimizing power consumption in the


相关文章:
1.US Energy Consumption by Fuel(1980-2030)
1.US Energy Consumption by Fuel(1980-2030)_英语考试_外语学习_教育专区。英语雅思小作文The graph shows energy consumption in the US from 1980 to 2012,...
U.S Fuel energy consumption overview
U.S Fuel energy consumption overview The multiple-line graph display information from a 2008report about consumption of energy in the USA since 1980 with ...
数控机床专业英语翻译题
Without this calibration, models can be inaccurate. Keywords ? ? ? ? Machine tool; Energy model; Energy consumption; Ecodesign 1. Introduction The issue...
Source of imperfect information Source of energy consumption
Vector space model 2.2. k-Partial set cover problem 3. Problem statement 3.1. Source of imperfect information 3.2. Source of energy consumption 4. ...
基于无尺度网络理论的无线传感器网络模型设计毕业论文外文翻译
the maximum number of links for each node is introduced into the algorithm, which can make energy consumption more balanced than the previous model (...
无线传感器网络模型设计_英文文献翻译.doc
the maximum number of links for each node is introduced into the algorithm, which can make energy consumption more balanced than the previous model (...
NORM INEN 2659
3.1.21 Energy Factor (EF). Overall measure at ...Page 4 of 52 3.1.34 Representative model. All ... of water consumption and energy consumption. ...
生态足迹概述
and energy and environment indexes.By means of EF model we Call transform the consumption of energy and resources and the Oatlalt of green house gases ...
The Primary Energy Consumption in the United States and in China in 2011.
The Primary Energy Consumption in the United States and in China in 2011._高等教育_教育专区。全国大学生写作竞赛。图表题The Primary Energy Consumption in ...
OP015 Rules for efficient use of energy如何有效地使用能源
如何有效地使用能源 Rules for Efficiency Use of Energy 如何有效地使用能源 Rules for Efficient use of Energy Refrigeration System 冷藏系统 ? Verify that ...
更多相关标签:
energy consumption | consumption model | energy based model | energy fuel acs ef | ef model first | ef 由数据库生成model | ef model | ef 修改一个model |