0% found this document useful (0 votes)
18 views11 pages

Measurement: Sensors: Mohamed Zakaria Zahaf, Mohieddine Benghersallah

Uploaded by

MAKKI
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
18 views11 pages

Measurement: Sensors: Mohamed Zakaria Zahaf, Mohieddine Benghersallah

Uploaded by

MAKKI
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 11

Measurement: Sensors 13 (2021) 100035

Contents lists available at ScienceDirect

Measurement: Sensors
journal homepage: www.journals.elsevier.com/measurement

Surface roughness and vibration analysis in end milling of annealed and


hardened bearing steel
Mohamed Zakaria Zahaf, Mohieddine Benghersallah *
Laboratory of Advanced Technologies in Mechanical Production (LRTAPM), Badji Mokhtar Annaba University, BP12 23000, Algeria

A R T I C L E I N F O A B S T R A C T

Keywords: This experimental study involves statistical analysis, mathematical modeling and optimization of machined
AISI 52100 steel surface roughness (Ra) well as the level of workpiece displacements (D) recorded during the shoulder and contour
Hardness milling of AISI 52100 bearing steel. Dry end milling of annealed and quenched/tempered condition with TiAlN
Dry end milling
coated carbide inserts tool. A Taguchi experimental plan of 27 tests was adopted to determine the influence of the
Surface roughness
Throbs/displacements
cutting parameters (Vc, fz, ae) considered as input factors, surface roughness and level of workpiece throbs/
Modeling displacements considered as response factors recorded during machining. To show the degree of influence of each
Optimization cutting parameter on the milled surface roughness and displacements, analysis of variance and mathematical
regression models were adopted. The objective of this study is to see the influence of the radial depth of cut ae
during shoulder, grooves and contour milling when we have a strong ap and a weak ae. Usually when face milling,
we adopt a strong ae and a low ap. The experimental results obtained shown that for two states of steel tested that
the cutting speed and feed per tooth is the most influencing factor on the milled surface roughness Ra evolution.
Radial depth of cut influences considerably the level of displacements. Multi-objective optimization gave the
desired cutting conditions with a desirability function DF ¼ 0.897.

Nomenclatures DA/DH Level displacement Annealed/Hardened


RSM Response surface methodology
Vc Cutting speed (m/min) PVD Physical vapour deposed
fz Feed per tooth (mm/t) CVD Chemical vapour deposed
ap Axial depth of cut (mm) CNC Computer numerical control
ae Radial depth of cut (mm) AISI American Iron and Steel Institute
T Tool life (min) ANOVA Analysis of variance
HRc Hardness value by Rockwell MS Mean of squares
Ra Arithmetic Surface roughness (μm) %C Percentage of contribution
TiN Titanium nitride df Degree of freedom
Disp Displacements (Peak to peak) (mm) SS Sum of squares
Ra Machined Surface roughness (μm) S/N Ratio signal-to-noise ratio
RaHS Horizontal Surface roughness (μm) AISI American Iron and Steel Institute
RaVS Vertical Surface roughness (μm)

* Corresponding author.
E-mail address: [email protected] (M. Benghersallah).

https://doi.org/10.1016/j.measen.2020.100035
Received 27 April 2020; Received in revised form September 2020; Accepted 11 December 2020
Available online 18 December 2020
2665-9174/© 2020 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-
nc-nd/4.0/).
M.Z. Zahaf, M. Benghersallah Measurement: Sensors 13 (2021) 100035

1. Introduction manganese-chromium steel. The ANOVA analysis of the taguchi plan L18
adopted has shown that the feed per tooth is the determining parameter
AISI 52100 steel is often used to manufacture mechanical parts such on the evolution of the roughness and the cutting speed considerably
as extrusion tools and bearings. It is recommended for high mechanical affects the wear of the tool. A quadratic linear regression model is
stresses, friction wear and corrosion. Recent studies have shown that the applied. The test results showed that the Taguchi method was very suc-
mechanical properties of this steel can be improved by bainitic quench- cessful in the optimization of machining parameters. CVD TiCN/A-
ing heat treatment. Dry cutting in the material removal industries has l2O3-coated carbide inserts exhibited better performance and could be
received much attention due to high cost of cutting fluids. The treatment recommended for use in the milling of tested steel. V. Sathyamoorthy
of cutting fluids is expensive. They are also dangerous for the health of [18], machined a metal block (240  120  60) in magnesium alloy
machine tool operators. Dry cutting is desirable because not only does it using a TiN coated carbide end mill with 10 mm diameter. The developed
reduce manufacturing costs, but it eliminates negative effects on the RSM model was tested through ANOVA. Based on the response surface
environment. method (RSM), a quadratic polynomial regression model has been
[1–3]. The feasibility of dry cutting in the material removal industries developed and tested by ANOVA. The results of the analysis revealed that
has received much attention due to high cost of cutting fluids. R.B. Da the feed rate is the most significant factor affecting the roughness of the
Silva et al., [4] declared that the use of coatings for cutting tools becomes milled area. The contribution percentages are respectively 76.18% for
essential, today more than 40% of the tools used in machining are coated. feed rate, 2.94% for the depth of cut and 1.99% for the cutting speed. . He
M. Alauddin et al. [9], El Houssein et al. [10], Confirmed in their study showed that the RSM method is an efficient method to optimize the
that carbide cutting tools are widely used in metal cutting industry for milling parameters to obtain a minimum surface roughness. A good
cutting various hard materials. In recent years, various statistical and surface finish is often required for applications since magnesium (Mg) is
experimental studies have been carried out on the basis of design and replacing aluminum (Al) and finding applications in the automotive and
analysis experimental methods for determining the effects of cutting aerospace industries. Mohamed Walid Azizi et al., [19] used Response
parameters on machinability of different received and hard materials Surface Methodology (RSM) to optimize the machining parameters
[5–8]. allowing to give the best results of the surface roughness, the radial vi-
Several researchers working on material machining by turning, brations of the tool and the volume of the chip removed. Hard turning of
milling or drilling processes use response surface methodology (RSM), EN19 alloy steel with coated carbide (GC3015) cutting tools was studied.
analysis of variance (ANOVA) and mathematical modeling tools to The development of a mathematical model combined with a central
determine the influence of each cutting parameter and its contribution on composite design centered on the face (FCCD) was used for statistical
the responses [9–11]. analysis. A second-order regression model were developed to correlate
Aouissi H et al. [12], examined the effect of cutting speed, feed per the cutting parameters with surface roughness, tool vibration and ma-
revolution and depth of cut on surface roughness using a mathematical terial removal rate. In conclusion it has been shown that feed rate has an
model. Turning experiments on hardened AISI H11steel (HRc 50) with influence of the order (72.19%) on Surface roughness, followed by the
CBN insert tool according to a plan based on Box–Behnken designs were quadratic effect of the cutting speed (10.69%) and finally by the
carried out. Additional work was done to assess the influence of two quadratic effect of the feed rate (1.86%). cutting speed, feed rate and
layers of coating on the surface roughness. Choudhury et al., [5] depth of cut are the three factors that increase the material of chip
formulated a mathematical model for flank wear, and concluded that the removal. The variation in surface roughness is directly proportional to
cutting velocity (speed) and the index of diffusion coefficient were the the radial vibration of the tool.
most significant factors, followed by the feed and depth of cut. CM, Cakir In recent year Benghersallah et al., [20] have presented a contribution
et al. [13] developed a mathematical model to evaluate the roughness of in characterizing end milling process of annealed AISI 52100 ball bearing
the machined surface depending on the type of coating material and the steel through statistical analyses of variance (ANOVA). They have re-
cutting conditions. Danielle Martins Duarte Costa et al., [14] tested in ported that Regression analyses have conducted to the development of
end milling rectangular pieces (100  100  300) in AISI1045 steel. The simplified empirical models that can be effectively used to predict surface
tool material used in dry milling is a cemented carbide insert coated with roughness and tool wear in face milling process. cutting speed and feed
TiCN and TiN by PVD process. Four input parameters and six response per tooth are the most important factors affecting tool life and machined
variables were considered. The normal boundary intersection (NBI) is a surface roughness. In this year S.Wojciechowski et al. [21], presented an
multi-objective optimization method mainly developed. The optimiza- analysis of the relationships between the instantaneous displacements of
tion results found by the NBI-MMSE approach compared to the weighted ball milling cutter and machined surface roughness along an inclined
traditional sum were able to reach a maximum high chip removable with plane. They have shown that the value of overhang tool significantly
minimum roughness. Bhuvnesh Bhardwaj et al. [15], In their study, they affects the mechanisms for generating surface roughness during milling.
developed a model of predicting the best surface roughness when fin- A good correlation of the kinematic-geometric model is obtained for a
ishing milling of EN 353 steel using carbide insert. A quadratic model that rigid overhanging tool (l ¼ 35 mm) which is not the case when (l ¼ 85
predicted best surface roughness with 92% accuracy. Another improved mm). Danil Yurievich Pimenov et al. [22] Have studied the variations of
quadratic model using Box-Cox transformation with RSM has been the radial immersion factor ao from the milling cutter to the workpiece.
developed to give more precision. The value of the ao factor directly characterizes the kinematics of the
Muhammmad Yasir et al. [16], studied the effects of cutting param- surfacing process. In the case where a0 ¼ 0, the forces are weakest (in the
eters on surface topography during milling of AISI 316L stainless steel. directions X and Z) and the vibrations (in the direction Y). During
Using a metal carbide cutter for 5 mm diameter, they machined rectan- surfacing with higher a0 values the thickness of the uncut chips is rela-
gular plates according to the cutting conditions (Feed rates ranges from tively high. Therefore, the dynamic contact between the tooth and the
0.10 mm/rev to 0.14 mm/rev while the cutting speeds ranges from 80 workpiece causes forced vibrations, which revealed the tendency for
m/min to 120 m/min). A mathematical model defining a relationship growth and acceleration of vibrations in the Y direction. The researchers
between cutting parameters and surface roughness has been established. proposed a multi-criteria optimization, based on the determination of the
From the results, it was found that the feed rate was the main factor total desirability function, which made it possible to obtain the optimal
affecting the surface roughness while the cutting speed had a negligible a0 factor.
effect. Analysis of variance revealed that the percentage contribution of The major concern of manufacturers is compliance with the technical
the feed rate was 10.38% and the cutting speed 2.1%. Turgay Kıvak [17], requirements imposed by the designers. The wrong choice of cutting
used the Taguchi method to optimize surface roughness and skin wear conditions, particularly in the milling operations of the shoulders,
insert coated of PVD TiAlN and CVD TiCN/Al2O3 during dry milling of grooves and profiles leads to dispersions affecting the precision of the

2
M.Z. Zahaf, M. Benghersallah Measurement: Sensors 13 (2021) 100035

mechanical parts. Compared to research carried out on end milling, the  Drilling and tapping (2xØM8x1.25)
weak radial depth of cut ae and strong axial depth of cut ap strategy used  Heat treatment (Quenching and tempering) as 50HRc
in hard milling is less studied. in our study we propose another method  Finishing by grinding the profile shoulder (70  50  10)
for measuring surface roughness Ra based on the exploration of the front
face and the vertical face constituting the geometry of the shoulders, 2.1.2. Machining by CNC hard milling
grooves and profiles. The displacements acting on the parts can also
guide researchers on the irregularities obtained on machined surface and  Face milling of the as received block (80  50  25)
the geometric integrity.  shoulder milling profile (28.5  8.5)
This paper presents a statistical analysis based on Taguchi (L27)  Chamfer milling (5  45  2)
experimental design. A mathematical model of regression is proposed  Milling grooves (5  4  50)
allowing the determination of the cutting parameters according new  Drilling and tapping (2xØM8x1.25)
strategy of end milling of shoulders, grooves and contours using low  Heat treatment (Quenching and tempering) as 50HRc
radial depth of cut ae, strong axial depth of cut ap. The objective is to  Finishing by hard milling the profile shoulder (30  10  80). Ra ¼
present for the company the results based on the optimization of cutting 0.8 μm
conditions during end milling of AISI 52100 steel in the annealed and
treated state. 2.2. Developing of experimental strategy

2. Experimental procedure The major challenge of hard milling to obtain the required shapes on
dies and molds is to ensure dimensional and surface finish integrity.
Two mechanical manufacturing companies "ETRAG and FERROVIAL" Therefore this strategy is developed to overcome this challenge. Based on
want to adopt hard milling of shapes such as shoulders and contouring on the guidelines of the manufacturers of cutting tools and the recommen-
dies and molds using CNC machine tool. dations of the literature, the proposed strategy is as follows:
So research investigation on hard milling is further more increasingly
required but a huge amount of research work is expected because of two  Performance of CNC machine tool
main issues: in one hand each type of milling operation requires specific  Selecting cutting parameters
cutting tool and cutting conditions and in the other hand the range of  Defining methodology of measurement output parameters (Wear,
hardness values of hardened steel is wide. The experimental procedure is surface roughness, cutting forces, vibrations level)
established according to the following strategy.  Defining an experimental plan
The drawing below Fig. 1 represents a model of dies among others
similar manufactured by companies. Compliance with technical specifi-
cations and manufacturers’ requirements. 2.2.1. CNC machine tool
Machining tests have been carried on a vertical EMCO Mill E350 CNC
milling Machine with a (50–10000 rpm) speed range, spindle motor
2.1. Comparison between two manufacturing methods power (6,8 Kw) and spindle torque (34Nm).

2.1.1. Machining by the conventional method 2.2.2. Workpiece materials


The shoulder milling is carried out on blocks (80  80  100) made of
 Face milling of the as received block (80  50  25) annealed and tempered AISI 52100 steels. The corresponding chemical
 shoulder milling profile (28.5  8.5) composition is given in Table 1. The heat treatments and the corre-
 Chamfer milling (5  45  2) sponding hardnesses of the parts are shown in Table 2.
 Milling grooves (5  4  50)
2.3. Selecting cutting parameters

2.3.1. Cutting inserts and tool holder geometry


According to Sandvik Coromant [28], the Sandvick type inserts made
of PVD (Ti,ALN) coated GC1030 cemented carbide is designed to ma-
chine hardened steel up to 40HRc with Vc of 90 m/min and fz in a range
of (0.01 to 0.15). A 25 mm diameter Sandvik Coromant Coromill
R390-020A20-11M tool holder has been used to handle 3 "R390-11T3
08M-PL1030". Sandvick type inserts. [28].

2.3.2. Selecting cutting parameters


According to the recommendations of tool manufacturers, the cutting
parameters selected for dry milling the shoulders on annealed and
treated steel are presented in the Table 3.

Table 1
Chemical composition of AISI 52100 steel (in weight percent).
C Si Mn P S

0.95 0.35 0.4 0.03 0.025


Cr Ni Mo V Fe
1.50 0.1 0.1 0.05 Balance
Fig. 1. Drawing of an example of parts manufactured by the company.

3
M.Z. Zahaf, M. Benghersallah Measurement: Sensors 13 (2021) 100035

Table 2 radial of cut. Table 4 shows the Taguchi 27 layout. Experimental results
Heat treatments. have been treated using the analysis of variance, a mathematical
States Heat treatments Temperature  c Hardness HRC modelling between roughness and displacements according to the cutting
parameters (Vc, fz, ap) during end milling of AISI 52100 steel.
1 Annealed 950 28
2 Quenched 850 52
3 Tempered 200 45 3. Result and discussion

3.1. Experimental design Taguchi table L27


Table 3
Cutting parameters.
Parameters Values
3.2. Evolution of output responses according the input cutting parameters
Vc (m/min) 75-95-115
fz mm/t) 0,01-0,05-0,1
ae (mm) 0,5-1-1,5 Fig. 4 shows the evolution of end milling machined surface roughness
ap (mm) 9 Ra for the different combinations of the experimental design. The graph
analysis shows that the surface roughness values are mostly higher for
annealed steel. The difference is greater for the cutting speed range (Vc ¼
2.4. Methodology of measurement output parameters
75 m/min). During the cutting speed interval (95-115 m/min), stability is
observed and the Ra values are fairly close.
2.4.1. Surface roughness measurement
Surface roughness has been measured using Mitutoyo Surftest SJ-
3.3. Evolution of workpiece displacements
201M tester adjusted on the shoulder of the machined workpiece along 9
mm deep as shown in Fig. 3. Three measurements have been conducted
Fig. 5 shows the evolution of level displacements in end milling
along the length of the face surface RaHS and three measurements on
process for the different combinations of experimental design. Level
profile surface RaVS workpiece.
displacement values recorded during milling of hardened steel are much
Surface roughness Ra of the shoulder is measured along the hori-
higher than the values recorded during milling of annealed steel. Milling
zontal face and the vertical face. The value shown in the Table 4 is the
of annealed steel is more stable. The combination of higher feed per tooth
average of six measurements according to the following equation.
and the radial depth of cut gives the peaks of displacements.
ðRaVs  3Þ þ ðRaHs  3Þ
Ra ¼ (1)
6 3.4. Taguchi L27 analysis

2.4.2. Vibrations level measurement 3.4.1. Main effect plot of


The vibration signal generated during the machining operation is Fig. 6 gives the main factor plot of means for milling surface rough-
detected by piezoelectric vibration acceleration sensors of VM-6360 se- ness. During the cutting speed (75–95) interval, the surface roughness Ra
ries type accelerometer with 10 Hz to 1 kHz frequency resolution, for annealed steel appears to be an almost-linear decreasing function of
displacement (Peak to peak, 0.001–2.0 mm) with RS232C interface. The Vc. This result is in perfect adequacy with the literature. During (95–115)
sensor has been attached by magnet on the back of the workpiece in order interval, the surface roughness appears to be an almost-linear horizontal
to record vibrations in the transverse (Y) direction see Fig. 2 Vibrations function of Vc. This result shows that during this interval the machining
measurements have been made in terms of displacements while the is more stable. During (0,01-0,1) interval appears to be an almost-linear
cutter is engaged in a cycle of milling operation cycle allowing then to increasing function of feed per tooth fz. During (0,01-1) interval for radial
monitor the changes during the machining such as wear on the inserts, depth of cut ae, average machined surface roughness decreases. In (1-1,5)
defect in the removable material, and chip adhesion. interval surface roughness increases. In hardened steel state in the first
interval surface roughness decreasing function of Vc in (75–95) interval,
2.4.3. Defining an experimental plan after it increases. During (0,01-0,1) fz interval and (0,5-1,5) ae interval,
Design experiments have been achieved using Taguchi’s technique surface roughness Ra appears to be an almost-linear increasing function.
which consists in analysing the collected data from surface roughness The best value of Ra is obtained with medium cutting speed value (Vc ¼
measurements and instantaneous displacement throbs on work piece. 95 m/min). The influence of other factors fz and ae is not significant. The
The input parameters are the cutting speed, the feed rate and the depth best surface roughness Ra value is obtained for (fz ¼ 0.02 mm/t and ap ¼

Fig. 2. Measurement Displacement/throbs on Workpiece during milling shoulder cycle.

4
M.Z. Zahaf, M. Benghersallah Measurement: Sensors 13 (2021) 100035

Fig. 3. Experimental setup.

Table 4
Taguchi table L27 and experimental results.
Input parameters Output parameters

Trial No. Vc m/min Fz mm/t Ae mm Ra A (μm) Annealed steel Ra H (μm) Hardened steel D A (mm)/Annealed steel D H (mm)/Quenched steel

1 75 0,01 0,5 0,46 0,29 0,015 0038


2 75 0,01 1 0,42 0,39 0,036 0114
3 75 0,01 1,5 0,43 0,44 0,045 0191
4 75 0,05 0,5 0,75 0,45 0,063 0137
5 75 0,05 1 0,69 0,49 0,072 0247
6 75 0,05 1,5 0,74 0,58 0,084 0378
7 75 0,1 0,5 0,96 0,48 0,044 0230
8 75 0,1 1 0,89 0,52 0,065 0225
9 75 0,1 1,5 0,92 0,79 0,079 0355
10 95 0,01 0,5 0,36 0,21 0,031 0151
11 95 0,01 1 0,31 0,25 0,050 0206
12 95 0,01 1,5 0,41 0,35 0,059 0241
13 95 0,05 0,5 0,43 0,39 0,038 0171
14 95 0,05 1 0,39 0,32 0,053 0175
15 95 0,05 1,5 0,45 0,36 0,062 0267
16 95 0,1 0,5 0,77 0,44 0,039 0206
17 95 0,1 1 0,71 0,37 0,054 0229
18 95 0,1 1,5 0,96 0,39 0,071 0373
19 115 0,01 0,5 0,31 0,27 0,045 0078
20 115 0,01 1 0,28 0,31 0,101 0127
21 115 0,01 1,5 0,46 0,36 0,110 0163
22 115 0,05 0,5 0,39 0,41 0,047 0145
23 115 0,05 1 0,34 0,33 0,071 0165
24 115 0,05 1,5 0,47 0,41 0,089 0188
25 115 0,1 0,5 0,81 0,44 0,096 0147
26 115 0,1 1 0,75 0,53 0,112 0237
27 115 0,1 1,5 0,95 0,62 0,117 0379

0.5 mm).
Fig. 7 gives the main factor plots for level displacements, a) for
annealed steel and b) for quenched and tempered steel. During the cut-
ting speed (75–95) interval for machined annealed steel, the level of
displacements decreases. In the second interval the displacements in-
creases. During machining hardened steel in (75–95) interval level of
displacements increases. In the second interval the displacements de-
creases. For the two states of milled steel the level of displacements in-
creases with increasing feed per tooth fz and the depth of cut ap.

3.4.2. Analysis of variance


Fig. 4. Evolution of Machined surface roughness Ra for both states.
Tables 5–8 show these results of ANOVA, respectively, for Ra and D
for annealed and hardened steel. This analysis was out for a 5%

5
M.Z. Zahaf, M. Benghersallah Measurement: Sensors 13 (2021) 100035

relationship between the input factors and the measured quantities was
modeled by the following quadratic regression equations.
Equation (2) given relation of milled surface roughness (Ra A), the
coefficient of determination R2 ¼ 86%

RaðAÞ ¼ 8; 03  0; 1698 Vc þ 48; 1 fz  0; 642 ae


(2)
þ 0; 000888 Vc2  767 fz2 þ 0; 560 ae2
Equation (3) given relation of milled surface roughness (Ra H), the
coefficient of determination R2 ¼ 79%

RaðHÞ ¼ 4; 119  0; 0822 Vc þ 10; 2 fz  0; 544 ae


(3)
þ 0; 000431 Vc2  44 fz2 þ 0; 756 ae2
Equation (4) given relation of level displacement D(A), the coefficient
of determination R2 ¼ 84%
Fig. 5. Evolution of level displacements DA and DH during experiments.
DðAÞ ¼ 0; 0486  0; 00243 Vc þ 8; 04 fz þ 0; 0978 ae
(4)
þ 0; 000013 Vc2  172; 2 fz2  0; 0409 ae2
significance level, i.e., for a 95% confidence level.
Equation (5) given relation of level displacement D(H), the coefficient
3.4.3. Regression equations of determination R2 ¼ 81%

3.4.3.1. Surface roughness and level displacements quadratic models. The

Fig. 6. Main effect plot for mean on Machined surface roughness Ra.

Fig. 7. Main effect plot for mean on Displacements levels D.

Table 5
Analysis of variance for Ra (A) (Annealed steel).
Source DL SS CM F P %C Remak

Vc 2 0,76,549 0,382,744 340,50 0,000 48% Significant


fz 2 0,58,302 0,291,511 259,33 0,000 36,46% Significant
ae 2 0,01496 0,007478 6,65 0,011 0,90% Insignificant
Vc*fz 4 0,20,716 0,051,789 46,07 0,000 13% Significant
Vc*ae 4 0,01469 0,003672 3,27 0,050 0,92% Insignificant
Erreur 12 0,01349 0,001124 0,8% Admissible
Total 26 1,59,880

S ¼ 0,335,272 R square ¼ 99,1% R square (adjust) ¼ 98,17%.

6
M.Z. Zahaf, M. Benghersallah Measurement: Sensors 13 (2021) 100035

Table 6
Analysis of variance for Ra (H) (Hardened steel).
Source DL SS CM F P %C Remak

Vc 2 0,17,905 0,089,526 22,76 0,000 38% Significant


fz 2 0,12,847 0,064,237 16,33 0,000 21,1% Significant
ae 2 0,06352 0,031,759 8,07 0,006 13,5% Significant
Vc*fz 4 0,01126 0,002815 0,72 0,597 2,38% Insignificant
Vc*ae 4 0,04361 0,010,904 2,77 0,077 9,2% Insignificant
Erreur 12 0,04720 0,003933 10% Admissible
Total 26 0,47,312

S ¼ 0,0627,163 R square ¼ 90,02% R square (adjust) ¼ 78,38%.

Table 7
Analysis of variance for level displacements D (A) (Annealed steel).
Source DL SS CM F P %C Remak

Vc 2 0,000157 0,000078 5,60 0,019 1.65% Significant


fz 2 0,004183 0,002092 149,40 0,000 43.76% Significant
ae 2 0,003680 0,001840 131,43 0,000 38.58% Significant
Vc*fz 4 0,001369 0,000342 24,46 0,000 14.32% Significant
Vc*ae 4 0,000001 0,000000 0,01 1000 0.4% Insignificant
Erreur 12 0,000168 0,000014 1.75% Admissible
Total 26 0,009558

S ¼ 0.0087695 R square ¼ 84% R square (adjust) ¼ 71%%.

Table 8
Analysis of variance for level displacements D (H) (Hardened steel).
Source DL SS CM F P %C Remak

Vc 2 0,000801 0,000401 0,24 0,790 0.3% InSignificant


fz 2 0,078,557 0,039,278 23,61 0,000 34.5% Significant
ae 2 0,105,085 0,052,543 31,58 0,000 41.6% Significant
Vc*fz 4 0,018,583 0,004646 2,79 0,075 8.2% InSignificant
Vc*ae 4 0,004686 0,001172 0,70 0,604 2% Insignificant
Erreur 12 0,019,966 0,001664 8.77% Admissible
Total 26 0,227,678

S ¼ 0.0087695 R square ¼ 81% R square (adjut) ¼ 75%.

DðHÞ ¼ 0; 345 þ 0; 00549 Vc þ 18; 84 fz  0; 052 ae  Table 10 shows the RSM results when optimizing (surface roughness
(5)
0; 000029 Vc2  312 fz2 þ 0; 354 ae2 (Ra) and level displacements (D)) in order of decreasing desirability
level. The optimized surface roughness (Ra A, Ra H) and level displace-
3.4.4. 3D surface analysis of output factors during end milling process ments (Disp A, Disp H) are (0.31, 0,21)μm and (0,015, 0,038) mm,
Figs. 8 and 9 illustrate 3D surface response results relative to the respectively. According this results, summary recommended parameters
impact of each input cutting parameters. For each plot, the non presented for company’s manufacturers are proposed in Table 11.
variables are held at a constant mean value. Analysis of the surface dia-
gram of Ra (A) shows that milled surface roughness increases with 3.5.2. Desirability function
increasing feed per tooth. The best combination for low roughness is (Vc To improve the manufacturing process in turning or milling or other
¼ 90 m/min, fz ¼ 0,01 mm/tooth) for machined surface when machining process, we always seek the optimal cutting conditions. Various re-
annealed steel. For machining hardened steel, feed per tooth has the searchers have applied the desirability function (DF). Kahhal and al [23],
greatest influence on machined surface roughness. The best combination Montgomery and al [24], are the first to use this method which is the
to obtain low roughness is (Vc ¼ 100 m/min, fz ¼ 0,01 mm/tooth). combination of several response factors in a single desired function in an
With regards to level displacements in radial depth of cut ae has great interval of (0, less desirable to 1, more desirable). During optimization, it
effect. So confirm the nodes observed during the principal effects plots is necessary to reduce the roughness of the surface and the level of dis-
analysis. The combination of radial depth of cut and feed per tooth, has a placements. Optimizing the response surface is an optimal procedure for
significant effect. determining the best cutting parameters combined in end milling. To
solve this type of parameter design problem, M. Mourabet and al [25],
3.5. Optimization has presented an objective function defined as follows:
The individual desirability for the predicted values for each depen-
3.5.1. Optimization of cutting conditions dent variable are then combined as a function of the overall desirability
Considering that the optimal manufacturing conditions for bearing DF, by calculating the geometric mean of the different values di.
steel (AISI 52100) are those minimizing the values of machined surface DF ¼ (d1  d2  d3  ⋅⋅ ⋅dn) 1/n (6)
roughness component (Ra) and throbs workpiece level displacements
(D) during the end milling process. The goals and the parameter ranges where di indicates the desirability of the response and n is the number of
defined for the optimization process are summarized in Table 9. responses. If one of the answers is greater than di, then the global func-
Figs.10 and 11 show the surface response plot for optimizing output tion becomes zero. It can be extended to
technological factors milling surface roughness (Ra) and level displace-
ments (D) according to the input parameters (Vc, fz, ae).

7
M.Z. Zahaf, M. Benghersallah Measurement: Sensors 13 (2021) 100035

Fig. 8. 3D surface plot of Ra (A) and Ra (H) versus (Vc; fz).

Fig. 9. 3D surface plot of level displacements D (A) and D (H) versus (fz, ae).

“target is the best” kind, then its individual desirability function is:
Table 9
Parameter ranges for optimization.  P
YI  Li
di ¼ if Li  Yi  Ti
Name Goal Lower Upper Lower Upper Importance Ti  Li
Limit Limit Weight Weight
 q
a:Vc is in 75 115 1 1 3 Y I  Ui
range di ¼ if Li  Yi  Ti (8)
Ti  Ui
b:fz is in 0,01 0,1 1 1 3
range
C:ae is in 0,5 1,5 1 1 3 di ¼ 1 if Yi ¼ Ti
range
Ra minimize 0,31 0,96 1 1 3
(A)
di ¼ 0 if Yi < Li or Yi ¼ Li
Ra minimize 0,21 0,79 1 1 3
(H) with the exponents p and q determining how important it is to hit the
D (A) minimize 0,015 0117 1 1 3 target value. If a response is to be maximized instead, the individual
D (H) minimize 0,038 0379 1 1 3 desirability is defined as shown in Eq. (6)

 1=n di ¼ 0 if Yi  Li
DF ¼ d1α1  d2α1  …  dn1=n (7)
 P
YI  Li
0  ∝i  1; ði ¼ 1; 2; 3; ::::::nÞ; α1 þ α2 þ :::: þ αn ¼ 1 di ¼ if Li  Yi  Ti (9)
Ti  Li

where di indicates the desirability of the different responses Yi (i ¼ 1, 2,


di ¼ 1 if  Yi  Ti
3, …, n) and i represents the importance of responses. So, maximum over
all desirability function DF depends on the, i (importance) value. Finally, if a response is to be minimized, the individual desirability
Depending on whether a particular response Yi is to be maximized, (di) is calculated according to Eq. (7)
minimized or assigned a target value, the different desirability functions
di (Yi) used are those proposed by Derringer and Suich [26], Walid and al di ¼ 1 if Yi  Ti
[27], Let Li, Ui and Ti be the lower, upper and target values, respectively,
that are desired for response Yi, with Li  Yi  Ui. If a response is of the

8
M.Z. Zahaf, M. Benghersallah Measurement: Sensors 13 (2021) 100035

Fig. 10. Response surface plot of Ra (A) and Ra (H).

Fig. 11. Response surface plot of D (A) and D (H).

Table 10
Optimization results (surface roughness (Ra) and level displacements (D)) in order of decreasing desirability.
N Vc fz ae Ra (A) Ra (H) D (A) D (H) Desirability
1 103,795 0,010 0500 0,317 0210 0,043 0071 0,897 Selected
7 105,278 0,010 0502 0,312 0206 0,044 0069 0,896
12 105,151 0,010 0500 0,315 0207 0,044 0070 0,895
16 107,447 0,010 0500 0,304 0200 0,046 0066 0,894
21 106,093 0,010 0517 0,310 0207 0,045 0070 0,893

Table 11
Summary table of recommended parameters for the company’s manufacturers.
Annealed steel Parameters Vc(m/min) fz(mm/t) Ae(mm) Ra A(μm) DA(mm)

Draft 75 0,01 1,5 0,43 0,045

Finishing 115 0,01 0,5 0,31 0,045

Quenched and tempered steel Parameters Vc(m/min) fz(mm/t) Ae(mm) RaH(μm) DH(mm)
Draft 75 0,05 0,5 0,45 0,132
Finishing 115 0,01 0,5 0,27 0,078

 q
Y I  Ui The Figs. 12 and 13 show the graphic optimization. The overlay plot
di ¼ if Li  Yi  Ti (10)
Ti  Ui highlights the required response regions. The contours are plotted at the
limits specified by the criteria (0,31  Ra A  0,95, 0,25  Ra H  0,79
di ¼ 0 if Yi  Ui and 0,015  D A  0,117, and 0,038  D H  0,379). From this analysis it
is found that the cutting speed, 90,55 m/min; feed rate, 0.016 mm/tooth

9
M.Z. Zahaf, M. Benghersallah Measurement: Sensors 13 (2021) 100035

Fig. 12. Graphics for optimizing Ra (A) annealed steel, Ra (H) hardened steel.

Fig. 13. Graphics for optimizing level displacements D (A)annealed steel, D (H) hardened steel.

value of function DF(x) ¼ 0.897 located in the factor space shaded blue
approximately the optimal solution, indicating that small variations in
the region where DF increases of: cutting speed, 95 m/min; feed rate,
0.07 mm/tooth and radial depth of cut, 0,51 mm are predicted to not
change the overall desirability drastically. However, the importance of
performing confirmatory runs at the estimated optimal operating con-
ditions should be emphasized.

4. Conclusion

This study allows us to give the following conclusions:


The machinability of hardened and tempered steel is more difficult
than that of annealed steel. This conclusion is given by the majority of
researchers. The breaking strength of hardened steel increases. Chip
formation requires greater energy.
Multilayer Coated TiN, TiCN, TiAlN. inserts 1030 [Sandvik coromant
2019], can give good machined surface roughness, larger tool life,. These
insert can be recommended for industrial applications.
Fig. 14. Graphic of optimizing desirability function. For milling shoulders, grooves and profile contours, down milling is
recommended because it gives good results for surface roughness Ra and
and radial depth of cut, 0,51 mm are the optimum values of machining not sticking chip, especially for annealed steel.
parameters while the optimum value of surface roughness, level dis- For annealed steel, the depth of cut affects vibration more than the
placements are Ra A ¼ 0.455 μm, Ra H ¼ 0,368 μm, D A ¼ 0,038 mm, D cutting speed and the feed rate. On the other hand, in the case of hard-
H ¼ 0,134 mm, respectively. ened steel, the cutting speed is the most influential parameter. Increasing
Fig. 14 shows a 2D contour plot of the overall desirability function the hardness of the workpiece decreases the life of the tool and increases
DF(x) for the (Vc, f) plane when ae is fixed at 0,51 mm. The maximum the level of vibrations generated during milling.
The established mathematical models allow to predict the response.

10
M.Z. Zahaf, M. Benghersallah Measurement: Sensors 13 (2021) 100035

The graphical representation of the outputs facilitates the interpretation [10] K.A. Abou-El-Hossein, K. Kadirgama, M. Hamdi, K.Y. Benyounis, Prediction of
cutting force in end-milling operation of modified AISI P20 tool steel, J. Mater.
of the results and clearly illustrates the effect of the input factors on the
Process. Technol. 182 (2007) 241–247, https://doi.org/10.1016/j.jmat
responses. protec.2006.07.037.
The machining of hardened AISI 52100 steel (55HRC) by coated [11] C. Kompan, J. Somkiat, Optimization parameters of tool life model using the
carbide inserts tools presents an economic alternative compared with taguchi approach and response surface methodology, IJCSI Int. J. Comput., Sci.
Issues 9 (Issue 1) (2012). No 3.
grinding. [12] AouiciH, BouchelaghemH, Yallese M. A., ElbahM, FnidesB (2014), Machinability
The mathematical models obtained are of significant industrial in- investigation in hard turning of AISI D3 cold work steel with ceramic tool using
terest for mechanical manufacturers. response surface methodology, Int. J. Adv. Manuf. Technol., doi: 10.1007/s00170-
014-5950-0.
The multi-objective optimization of the technological parameters (Ra [13] C.M, Cakir, C. Ensarioglu, Ir Demirayak,. Mathematical modeling of surface
(A), Ra (H), and D (A), D (H)) using the desirability function allowed us roughness for evaluating the effects of cutting parameters and coating material. J.
to find an optimal regime, which are (Vc) ¼ 95 mm/min, (fz) ¼ 0.07 Mater. Process. Technol.. https://doi.org/10.1016/j.j mat protec. 2008.01.050.
[14] D, Martins D, Costa , T, Gonçalves B, A Paulo de Paiva, R. C Leme , P. P Balestrassi.
mm/tooth and (ae) ¼ 0.51 min. This op regime corresponds to the A normal boundary intersection with multivariate mean square error approach for
following technological parameters: Ra(A) ¼ 0.31 μm, Ra (H) ¼ 0.303 dry end milling process optimization of the AISI 1045 steel. J. Clean. https://doi.
μm and D (A) ¼ 041 mm, D (H) ¼ 0.126 mm with desirability DF ¼ org/10.1016/j.jclepro.2016.01.062.
[15] Bhuvnesh Bhardwaj, Rajesh Kumar, K. Singh Pradeep, An improved surface
0.813. roughness prediction model using Box-Cox transformation with RSM in end milling
of EN 353, J. Mech. Sci. Technol. 28 (12) (2014) 5149–5157, https://doi.org/
Declaration of competing interest 10.1007/s12206-014-0837-4.
[16] Muhammmad Yasir, Turnad Lenggo Ginta, Bambang Ariwahjoedi, Adam
Umar Alkali, Mohd Danish, Effect of cutting speed and feed rate on surface
The authors declare that they have no known competing financial roughness of aisi 316l ss using end-milling, ARPN J. Eng. Appl. Sci. 11 (February
interests or personal relationships that could have appeared to influence 2016). N . 4.
the work reported in this paper. [17] Turgay Kıvak. Optimization of surface roughness and flank wear using the Taguchi
method in milling of Hadfield steel with PVD and CVD coated inserts. Measurement
50 doi.org/10.1016/j.measurement.2013.12.017.
Acknowledgments [18] V. Sathyamoorthy*, S. Deepan, S. P. Sathya Prasanth and L. Prabhu. Optimization of
machining parameters for surface roughness in end milling of magnesium AM60
alloy. Indian J. Sci. Technol., Vol 10(32),doi: 10.17485/ijst/2017/v10i32/104651.
We thank the companies ETRAG and FERROVIAL for funding this [19] R. Singh, J.S. Dureja, Comparing Taguchi method and RSM for optimizing flank
research. wear and surface roughness during hard turning of AISI D3 steel, in: Proceedings of
the International Conference on Research and Innovations in Mechanical
Engineering, Springer, New Delhi, 2014, pp. 139–152, https://doi.org/10.1007/
References 978-81-322-1859-3.
[20] M. Benghersallah, S. Benchiheuband, A. Amirat, Statistical characterisation of end
[1] J. Chakraborty, I. Manna, Development of Ultrafine Ferritic Sheaves/plates in SAE milling of AISI 52100 annealed bearing steel, Adv. Mater. Res. 7 (2) (2018) 83–95,
52100 Steel for Enhancement of Strength by Controlled Thermo Mechanical. https://doi.org/10.12989/amr.2018.7.2.137.
doi.org/10.1016/j.msea.2012.03.056. [21] S. Wojciechowski, M. Wiackiewicz, G.M. Krolczyk, Study on metrological relations
[2] Yoon-Uk Heo, You-Young Song, Seong-Jun ark, H. K. D. H. Bhadeshia & Dong-Woo between instant tool displacements and surface roughness during precise ball end
Suh. Influence of silicon in low density Fe-C-Mn-Al steel. Metall. Mater. Trans., milling, Measurement 129 (2018) 686–694, https://doi.org/10.1016/
https://doi:10.1007/s11661-012-1149-x. j.measurement.2018.07.058.
[3] A. Basu, J, S.M. Chakraborty, G. Shariff, S.V. Padmanabham, G. Joshi, Dutta J. [22] Danil Yurievich Pimenov, Amauri Hassui, Szymon Wojciechowski, Mozammel Mia,
Majumdara Sundararajan, I. Manna, Laser surface hardening of austempered Aristides Magri, Daniel I. Suyama, Andres Bustillo, Grzegorz Krolczyk, Munish
bainitic ball bearing steel, Materialia (2007), https://doi.org/10.1016/ Kumar Gupta, Effect of the relative position of the face milling tool towards the
j.scriptamat.2007.01.029. workpiece on machined surface roughness and milling dynamics, Appl. Sci. 9
[4] R.B. Da Silva, J.M. Vieira, R.N. Cardoso, H.C. Carvalho, E.S. Costa, A.R. Machado, (2019) 842, https://doi.org/10.3390/app9050842.

R.F. De Avila, Tool wear analysis in milling of medium carbon steel with coated [23] P. Kahhal, Y.A. Brooghani, H.D. Azodi, Multi-objective optimization of sheet metal
cemented carbide inserts using different machining lubrication/cooling systems, forming die using genetic algorithm coupled with RSM and FEA, J. Fail. Anal. Prev.
Wear 271 (2011) 2459–2465, https://doi.org/10.1016/j.wear. 2010.12.046. 13 (6) (2013) 771–778, https://doi.org/10.1007/s11668-013-9747-5.
[5] I.A. Choudhury, M.A. El Baradie, Tool life prediction model by design of [24] R. H. Myers and D. C. Montgomery, Response Surface Methodology: Process and
experiments for turning high strength steel, J. Mater. Process. Technol. 77 (1998) Product Optimization Using Designed Experiments, https://doi.org/10.1080/
319–326, https://doi.org/10.1016/S0924-0136 (97) 00435-4. 00224065.2004.11980252.
[6] M. Cemal, Cakir, Cihat Ensarioglu, Ilker Demirayak, Mathematical modeling of [25] M. Mourabet, A. El Rhilassi, H. El Boujaady, M. Bennani-Ziatni, R. El Hamri,
surface roughness for evaluating the effects of cutting parameters and coating A. Taitai, Removal of fluoride from aqueous solution by adsorption on Apatitic
material, J. Mater. Process. Technol. 209 (2009) 102–109, https://doi.org/ tricalcium phosphate using Box–Behnken design and desirability function, Appl.
10.1016/j.jmat.protec.2008.01.050. Surf. Sci. 258 (2012) 4402–4410, https://doi.org/10.1016/j.apsusc.2011.12.125.
[7] H. Aouissi, M.A. Yalles, K. Chaoui, T. Mabrouki, J.-F. Rigal, Analysis of Surface [26] G. Derringer, R. Suich, Simultaneous optimization of several response variables, J.
Roughness and Cutting Force Components in Hard Turning with CBN Tool: Qual. Technol. https://doi.org/10.1080/00224065.1980.11980968.
Prediction Model and Cutting Conditions Optimization, vol. 45, 2012, pp. 344–353, [27] M. Walid Azizi, O. Keblouti, L. Boulanouar, M. Athmane Yallese, Design
https://doi.org/10.1016/j.Measurement.2011.11.011. optimization in hard turning of E19 alloy steel by analysing surface roughness, tool
[8] A. Slaimia, M.A. Yalles, H. Bensouilah, I. Meddour, I. khettabi, Modeling and vibration and productivity, Struct. Eng. Mech. 73 (No. 5) (2020), https://doi.org/
optimization in dry face milling of X2CrNi18-9 austenitic stainless steel using RMS 10.12989/sem.2020.73.5.501.
and desirability approach, Measurement 107 (2017) 53–67, https://doi.org/ [28] Sandvik Coromant, Milling, Main Catalogue, AB Sandvik Coromant, Sandviken,
10.1016/j. measurement.2017.05.012. September, 2017. Sweden, 2019.
[9] M. Alauddin, M.A. Ei Bardie, M.S.J. Hasmi, Prediction of tool life in end milling by
response surface methodology, J. Mater. Process. Technol. 71 (1997) 456–465,
https://doi.org/10.1016/S0924-0136 (97) 00111-8.

11

You might also like