You are on page 1of 5

Surface roughness model for

turning
M. Hasegawa* A. Seireg** and R. A. Lindberg**

The effect of the cutting parameters on the surface roughness in a turning


operation has been investigated using response surface methodology. The
mathematical prediction models for surface roughness have been obtained for
a common mild steel. The equation is used for determining the optimal cut-
ting conditions for a required surface roughness.

In general, surface roughness has an important role in all The workpiece used was mild carbon steel (SAE 1020)
areas of tribology. This work was aimed at investigating the ranging in size from 63.5 mm to 152.4 mm diameter. These
effect of cutting parameters on surface f'mish, A large num- diameters made it possible to obtain surface speeds up to
ber of analytical and experimental surface roughness studies 1000 ft/min.
on turning operations have been made, as shown by a review
The cutting toolhad tungsten carbide disposable inserts.
of the literature.
The inserts used had a 5° positive back rake. The tool hol-
Some investigators have reported that the surface roughness der had the following geometry;
improves with increasing cutting speed and tool nose radius Back rake angle: - 5 °
and with decreasing feed ~'4,s. In particular, it has been Side rake angle: - 5 °
observed that the surface roughness 3 has maximum values End relief angle: + 5°
for the speeds below 200 ft/min. Of the factors influenc- Side relief angle: + 5 °
ing the surface roughness, the depth of cut was found to End cutting edge angle: +15 °
have the least effect. This was found to corroborate the Side cutting edge angle: +15 °
findings of Sata 6. In order to simulate the production turning conditions a
Other researchers have found that the tool wear causes the preliminary rough cut was taken on the surfaces under the
surface finish to deteriorate rapidly and has a direct effect following conditions:
on the maximum roughness s. Similarly,"the principal cut- Depth of cut: 2.54 mm
ting edge and Brinell hardness number of the workpiece Feed rate: 0.254 mm/rev
meterial also affect the surface roughness2. In general, Cutting speed: 121.92 m/min
theoretical values of surface finish derived from geo- Tool nose radius: 0.0468 mm.
metric analysis show considerable difference in comparison The cutting conditions for the finishing cuts were:
with the experimental values. Recently, statistical tech- Cutting speed: 60.96 rn/min
niques, developed by Box and Wilson, have been applied to 182.88 m/min
establishing a prediction equation for the relationship be- 304.80 m/rain
tween tool life or surface finish and cutting parameters 7,9 . Feed rate: 0.127 mm/rev
However, we know little of the relationship between the 0.508 mm/rev
surface finish and cutting parameters, including the effect 0.889 mm/rev
of the tool nose radius. Depth of cut: 0.3048 mm
0.508 mm
These turning investigations were aimed at developing a
0.7112 mm
similar equation to that used in the prediction of surface Tool nose radius: 0.79375 mm
finish in terms of the four variables: speed, V(m/min); 1.58750 mm
feed, f(mm/rev); depth of cut, d(mm); and tool nose radius, 2.38125 mm
R(mm), for a common mild steel No coolant was used. The experiments were carried out
with sharp tools.
Experimental conditions
The equipment used for turning was 1 L saddle type turret
lathe equipped with a 10 HP variable speed motor and a Test procedure
modified variable feed.
To obtain the prediction equation of the surface roughness,
* Department o[Mechanicai Engineering, National Defence Aca- 81 experiments were carried out at three levels for each in-
demy, 1-10-20, Hashirimuza, Yokosuka, Japan dependent variable to develop a second-order model. A
** Department o f Mechanical Engineering, Faculty o f Engineering,
series of bars (Fig l) were prepared from 63.5-152.4 mm
University o f Wisconsin-Madison, 513 University Auenue, Madison diameter materials. Nine bars were used, each being divided
Wisconsin, 53706, USA into ten blanks. Number 0 in Fig 1, which was used to estimate

TRIBOLOGY international December 1976 285


• 330'2 mm 0'254 mm
Y =OloX0 +alX1 +°t2X 2 +c~3x 3 +~4X4 + 6 (4)

It should be developed so that the model more adequately


represents the data. A second order model in four dimen-
tions is as follows:
Chucked 9 8 7 6 5 4 3 2 I No.O
end Y - e = Y = box 0 + blX 1 + b2x 2 + b3x 3 + b4x 4 + b12XlX 2
Fig I Schematic diagram o f test section
+ b13XlX 3 + bl4XlX 4 + b23x2x 3

+ b24x2x 4 +b34x3x 4 + b123XlX2X 3


the precutting surface, was always cut by a tungsten carbide
disposable insert with a nose radius of 1.19063 mm. Under + b124XlX2X 4 + b134XlX3X 4 + b234x2x3x 4
given cutting speed and depth of cut, numbers 1,2 and 3
were cut by a tool with a nose radius of 0.79375 mm. Num- + bllX2 + b22x 22 + b33x 23 + b44x2 + bll2XlX2
2
bers 4, 5 and 6 were cut by a tool with a nose radius of 2 2
1.5875 mm and numbers 7, 8 and 9 by a tool with nose + bll3XlX3 + b l l 4 X l X 4 + b221X2Xl
radius of 2.38125 mm. Each of the bars was cut in turn
with a feed rate of 0.127 mm/rev, 0.508 mm/rev, and + b223x2x3 + b224x2x4 + b331X2Xl
0.889 mm/rev, respectively. Each of the test bars was cut
with a new part of the tool cutting edge. The workpiece 2 + b334x2x4 + b441X4Xl
+ b332x3x2
was removed from the machine and the surface measured
2 2
parallel and perpendicular to the tool grooves by a Talysurf + b442x4x2 + b443x4x 3 (5)
surface roughness measuring instrument (Model IV). A sur-
face roughness cut off value of 2.54 mm was used. Values where Y is the logarithmic value of the measured response,
used in the results are the average of five readings taken at e is an experimental error and the b's are estimates of the/3
equal intervals around the workpiece. parameters. Y is the estimated response on a logarithmic
scale.
A pen recording was also taken of the surface profile along
the feed marks. The 'Peak to Valley' height was measured The b's were estimated by the method of least squares
from the recorded profile curves of the surface finish. where the basic formula is given by:

Model postulation for response b = (x'x)-lx'y (6)


The relationship between response (surface roughness) of where b is the matrix of parameter estimates, x is the cal-
the cutting operation and independent variables can be rep- culation matrix, x' is the transpose of x, x ' x is the variance
resented by the following equation: matrix, (x'x) -1 is the covariance matrix which is the inverse
o f x ' x and y is the matrix of measured response on a logari-
Rma x = koVklfk;Idk ~ k4 (1) thmic scale.

where Rrnax is the response (that is, 'Peak to Valley' height), If an attempt is made to portray the nature of the obser-
while V, f, d andR are the cutting speed, feed rate, depth of vations with another of the most elemental mathematical
cut and tool nose radius, respectively. However, k 0, k 1, k 2, form possible, we can also consider the prediction equation
k 3 and k 4 are constants to be determined. This equation on a nonlogarithm scale:
can be written in the linear form as:
rl = f(V, f, d, R ) (7)
lnRma x = Ink0 + k l l n V + k21nf+ k31nd+ k41nR (2)
The proposed first-order model is:
which may represent the following mathematical model as
a continuous function of four independently controllled Y =/30 +/31 V+/32.(+/33 d +/34R + e (8)
variables:
Further, equation (8) could be written in the second-order
r/= f(Xl, X2, .7(3, X4) (3) model such as equation (5):

where Y - e = l3"=/30 +/31 V+/32/+/33 d +/34R

X 1 = In V +/312 V f + . . . +/3443R2d (9)

X 2 = Inf By using equation (6),/30,/31, - - •/3443 could also be deter-


mined by the same method.
X3 =lnd

X4 = lnR
Models for turned surafce roughness
As in the four dimensional case, the first step in exploring The experimental results are listed in Table 1. The relation-
this function is fit the first-order model: ship between the independent variables and code for calcu-

286 TRIBOLOGY intemational December 1976


Table 1 Experimental conditions and experimental results Table 3 Computed simple correlation and t-values on the
related to 1Beak to Valley' height (/~m) constants of Equation (14)

Depth Tool Correlation x vs y Computed t value


of Nose Feed, mm/rev
Velocity, cut, Radius, x1 -0.32779 -5.4093
m/min mm mm 0.127 0.508 0.889 x2 0.74605 12.311
x3 0.01311 0.21634
(0.79375 28.96 39.16 94.56 x4 0.23815 -3.9300
0.3048 41.5875 17.53 39.42 49.07
t2.38125 13.70 64.30 44.38
(0.79375 55.60 52.16 110.04 four main variables, we obtain the following prediction
60.96 0.508 [1.5875 31.68 25.40 67.77 equation for the surface roughness of the 81 experiments:
t2.38125 47.20 30.72 57.20
(0.79375 18.10 43.20 100.87 Y= 3.38985 - 0.47778 x 1 + 1.1269 x 2 + 0.017134 x 3 -
0.7112 41.5875 33.21 28.40 76.15 0.32439x 4 (14)
t2.38125 17.84 22.57 24.43
(0.79375 4.62 32.76 101.54 This prediction equation was transformed using equations
0.3048 ~1.5875 10.50 22.94 64.44 (10), (11), (12) and (13) to represent the 'Peak to Valley'
t2.38125 21.66 16.48 34.28 height surface roughness, Rmax, as a function of the cutting
(0.79375 14.60 43.44 110.24 speed, feed, depth of cut and tool nose radius:
182.88 0.508 41.5875 9.96 25.08 71.08
t2.38125 17.48 19.16 49.32 Rma x = 624.2432 V-0"4349f 0"8129 d 0"0335 R -0"4680 (15)
(0.79375 9.00 35.85 99.72
0.7112 [1.5875 8.58 17.70 61.07 For the purpose of checking the significance of the indepen-
t2.38125 10.58 17.07 44.53 dent variables, the computed simple correlation and t-values
' (0.79375 6.20 34.50 91.12 on the constants of equation (14) are ~hown in Table 3.
0.3048 41.5875 3.90 9.61 54.48 From the information given in Table 3, it is assumed that
t2.38125 5.95 10.46 34.89 the depth of cut has little significance on the surface rough-
(0.79375 6.32 42,80 113.00 ness. Thus, we can obtain the following first-order predict-
304.8 0.508 41.5875 5.04 22.32 77.40 ion equation by computer:
t2.38125 7.20 13.92 51.60
(0.79375 5.53 37.63 101.77 I~= 3.40452 - 0.51355x 1 + 1.1065x 2 - 0.31602x 4 (16)
0.7112 [1.5875 4.96 17.84 62.16
t2.38125 5.81 11.20 38.88 The above equation (16) also was transformed:
Measuring lenth for roughness is absed upon diS B 0601-1970
Rma x = 722.49685 V-0"46745f 0"79817R -0"45592 (17)

lation are shown in Table 2. In Table 2, the relationship be- On the other hand, from the physical view point based on
tween the code and independent variables are as follows: the computed results of Table 3, the mathematical model,
that is the second-order model of equation (5) for the sur-
In V - In 182.88 face roughness within the experimental conditions, can be
(10) developed to obtain reliability for practical use. The most
Xl = In 182.88 - In 60.96
suitable computed results in regard to equation (5) shows
the mean values of experimental results, without consider-
l n F - in 0.508 ing the depth of cut:
x2 = (11)
~0.508 -~0.127
lnRma x = 3 . 1 4 3 8 4 - 0 . 3 8 0 8 1 x 1 + 1.9438x 2 - 0 . 4 2 5 7 5 x 4
In D - In 0.508
x3 = (12) + 0.93562 X l X 2 - 0.12552 X l X 4 - 0.66543 x2x 4
In 0.508 - In 0.3048
+ 1.0726x~+O.3457x~x 2 -0.39173x~x 2
lnR - In 1.5875
x4 = (13)
In 1.5875 -- In 0.79375 - 0.21737 X l X 2 X 4 (18)

Using Table 1 ,Table 2, and least squares method, we can Moreover, the most suitable form of equation (9) on the
calculate the Y about equation (4). That is, if we limit the nonlogarithmic scale for the surface roughness is:

Table 2 Experimental conditions and codes for calculations

V, m/min Code f, m/rev Code d, mm Code R, mm Code

60.96 - 1 0.127 - 1 0.781 - 1 0.79375 - 1


182.88 0 0.508 0 1.563 0 1.58750 0
304.80 0.46498 0.889 0.40368 2.344 0.65753 2.38125 0.58495

TRIBOLOGY international December 1976 287


sidered the most useful for predicting surface roughness.
R=O-78mm o f-O.127mmlrev
By using equation (19), the relationship between the surface
. . . . . . Eq.(18) • f-O'508mm/rev
roughness and cutting parameters can be shown clearly in
Eq.(I9) • f-OB89mrn/rev
Figs 5(a), (b) and (c).

Conclusion
I00
The main results of this study are:
(1) The cutting parameter which affects roughness most
E
::L is the feed rate
<
(2) The depth of cut has little effect on the surface
roughness.
(3) As cutting speed and tool nose radius increase, sur-
50 face roughness also decreases.
(4) Experimental results did not bear out previous re-
Q
search that

R max = f2/SR

but found that

o ' ,60 ' 26o ' 360 Rmax tx f-0.798 R -0A56


V, m/rain
(5) As the feed rate increases surface roughness increases.
Fig 2 Comparison of the experimental results with the (6) Using response surface methodology, a mathmatical
second order model[or turning operation [R = O.78 ram) model for the surface roughness was developed in
terms of the cutting speed, feed and tool nose radius,
iO0
by the use of minor variations.
R=l.56mrn o f-O-127rnm/rev (7) By use of any one of the four surface roughness pre-
. . . . . . Eq.(18) = f-O.5OSmm/rev diction equations, the surface roughness can be selec-
Eq. (19) • f-O-889mrn/rev ted for given cutting conditions.

E Acknowledgements

,( 50 The authors wish to thank the Computer Center and Machine


~r Shop of the Department of Mechanical Engineering, Uni-
oc
versity of Wisconsin-Madison, for their assistance. Thanks
are also due to Wisconsin Alumni Research Foundation for
financial support of this study.

o ' ,6o ' 26o ' 36o


V, m/min R=2"34 mrn o f = O127mm/rev
...... Eq. (18) • f = 0-508rnm/rev
Fig 3 Comparison of the experimental results with the Eq. (19) • f =O889mrn/rev
second order model for turning operation (R = 1.56 mm)

I00

Rma x = 22.6215 - 6.6133 V - 29.277 f - 8.6483 R


E
zl_
+ 5.9375 V f - 14.988 f R + 3.0989 V2
~E

+ 13.951 f2 + 6.0822 R 2 _ 3.3042 V2f


50
- 6.4242 f2R (19)

To calculate equation (19), the code of independent vari-


ables at the lower, centre, and higher levels were selected as
- 1 , 0 and 1. The standard error of estimates on equation
(19) was 3.92512 (/an). Figs 2 - 4 show the comparison
between the computed results from equations (18) and (19) 0 I00 200 300
and experimental results. As a natural consequence, equa-
v, mlmm
tion (18) and (19) show better agreement with experimen-
tal results than the first-order models., Further, we can see Fig 4 Compairson of the experimental results with the
equation (19) is much better than equation (18) and is con- second order model for turning operation {R = 2.34 mm)

288 TRIBOLOGY international December 1976


0 0 0 0 0 0 "79375 10(3"00
0 0 0 0 0 95"00
o
O~79375/
oJ -- 90'00
85'00
500 80'00
75"00
70"OO
1.5875(: 65'00
E
60"00
E
1"5875C 55OO
~ 50'OO
45"OO
2'38125
6096 182.88 304"8O
30"OC V, m/rain
2'38125
60.96 182.88 304.80 Fig 5 Relationship between the response surface and cutt-
V, m/rain ing parameters f o r turning. Top left: t = 0.127 mm/rev.
Above: t = 0.508 mm/rev. B o t t o m left: t = 0.889 mm/rev

0'79375 ~45.00 4000


~ 35"00

30.00

and Research, Proceeding of the 3rd lnternational MTDR Con-


25.00
ference, Pergamon Press, New York, 1962, pp 255-243
Bhnttaclmnya A., Faria-Gonzalex R., and Inyon$ Ham Re-
gression Analyds for Predicting Surface Finish and Its Appli-
E 2000 cation in the Determination of Optimum Machining Con-
E 1"58750 ditions, Trans. ASME, Vo192, No 3, Series B (1970),
pp 711-714
Cktnd/ranumi K. L. and Cook N. H. Investigations on the
Nature of Surface Finish and Its Variation with Cutting Speed.
Trans. A SME, Yo186, No 2, Series B, (1964), pp 134-140
Olwn K. V. Surface Roughness in Turned Steel Components
and the Relevant Mathematical Analysis, The Production
Engineer, December 1968, pp 593-606
2.38125[ Sara T. Surface Finish in Metal Cutting, C.LR.P. Ann Alen
60'96 182.88 304'80
Band XII, Heir, 4. (1964/, pp 190-197
V, m/rain Tataman K. Multi Machining Output-Multi Independent Vari-
able Turning Research by Response Surface Methodology,
Int, J. Prod. Res., Vo112, No 2 (1974) pp 233-245
References Solaja Vlsdim/r. Wear of Carbide Tools and Surface Finish
1 Albrech A. B. How to Secure Surface Finish in Turning Generated in Finish Turning of Steel, Wear, Voi 2, (1958/59),
Operations, American Machinist, Vo1100, 1956, pp 133-136 pp 40-58
2 Ansell C. T. and Taytor J., The Surface Finish Properties of a Wu S. M. Tool Life Testing by Response Surface Methodology
Carbide and Ceramic Tool, Advances in Machine Tool Design Parts I and H, Trans, Am. Soc. Mech. Engrs. 86 (1964) p 105

TRIBOLOGYinternationalDecember1976 289

You might also like