SURAPOL RAADNUI
Department of Production Engineering, Faculty of Engineering
King Mongkut’s Institute of Technology North Bangkok
1518 Pibulsongkram Road, Dusit, Bangkok, 10800, THAILAND
ABSTRACT
Surface roughness parameters can be measured by variety of typical surface roughness testers available in the
market. Parameters provide from the measuring devices are plethora, and, in turn, are used to represent the
characteristic of the surfaces under investigation. In this particular research, the author seeks his own initiative to
utilize a statistical - based mean to classify the differences between major machining processes. Several plotting
techniques are employed in the surface classification process. In short, the technique may be employed to
identify/differentiate some certain characteristics between typical machining surface. The techniques used are based
mainly on the multivariate statistical approach. The relationship between conventional surface roughness parameters
was studied utilizing the more fundamental exploratory statistical data analysis in hyperspace i.e. Draftsman Plot,
Principal Component Analysis and Andrews’ Plot. Preliminary results for classification of machining surfaces
commonly used in industry, namely planing, milling, grinding and turning processes both within and amongst the
process are presented. It was found out that these techniques were the most useful for a machined surface profile
roughness description in comparison with currently applied parameters, with respect to differentiating between
machining processes.
Frequency
with a TKK 300 stylus (5 µm radius of spherical tip) was Turning
.
surface. R q (µm)
Planing Turning
0.5 GRINDING
PC1 0
-0.6 -0.4 -0.2 -0.5 0 0.2 0.4 0.6
-1
PLANING -1.5
-2
-2.5
-3
8 PC 2
·Fine (R =6.3micron)
z , Medium (Rz=10micron) ° coarse (R =20micron)
z
ƒ(θ)
θ
- - - - -
Fig.6 Two-dimensional plot (Rt vs Wt) for milling process.
Fig.8 Andrews’ plot of the most four influential parameters of
planing process.
4. CONCLUSIONS
Several techniques for obtaining a visual representation of
surface roughness multivariate data are presented. The
preliminary examination of most data, in this work, is
facilitated by the use of diagrams. Diagrams prove nothing,
but bring outstanding features readily to the eye. Such
techniques should be useful for a preliminary low level
Fig.7 Two-dimensional plot (Rt vs Wt) for turning process. investigation of surface roughness multivariate data and in
several cases, shown here, may also be used in assisting
In addition, Andrews’ plot of Wt, Rm, Rp and Rq of planing communicating results to the investigator involved.
and cylindrical grinding processes can be seen in Figures 8 From the above research work, it can be concluded
and 9. that:
- Surface roughness parameters are plethora, it has
been demonstrated that these parameters are
significantly useful for identification of
machining processes.
- Utilization of multivariate statistical analysis is
proved to be important in machining surface
roughness classification processes. Parameter Formulae and/or description
- Multiple surface roughness parameters should be
implemented as part of Quality Control (QC) l
1
l ∫0
process and/or productivity improvement of Ra Ra = y ( x) dx
machining operation.
1
5. REFERENCES R z = ( R1 + R3 + R5 + R7 + R9
[1] W.P. Dong, P.J. Sullivan and K.J. Stout,, RZ 5
“Comprehensive study of parameters for characterizing − ( R2 + R4 + R6 + R8 + R10 ))
three-dimensional surface topography - I: Some inherent
properties of parameter variation”, Wear, 159 (1992), pp. Rm Rm = 6 Ra
161-171
1/ 2
[2] W.P. Dong, P.J. Sullivan and K.J. Stout, “Comprehensive 1 L
Rq = ∫ ( y ( x )) dx
2
study of parameters for characterizing three-dimensional Rq
surface topography - II: Statistical properties of L 0
parameter variation”, Wear, 167 (1993), pp. 9-21
[3] W.P. Dong, P.J. Sullivan and K.J. Stout, “Comprehensive Rp Rp =
1
(R p1 + R p 2 + ... + R p 5 )
study of parameters for characterizing three-dimensional 5
surface topography - III: Parameters for characterizing
amplitude and some functional properties”, Wear, 178 Rv Rv =
1
(Rv1 + Rv 2 + ... + Rv 5 )
(1994), pp. 29-43 5
[4] B. Nowicki, “Multiparameter representation of surface
roughness”, Wear, 102 (1985), pp. 161-176 R R1 = R p1 + Rv1
[5] T. Klimczak, “Origins, magnitude and statistical
significance of differences between roughness parameters
of two- and three-dimensional characteristics”, Wear, 156 Rtm Rtm =
1
(Rm1 + Rm 2 + ... + Rm5 )
(19920, pp. 19-31 5
[6] J. Michalski and P. Pawlus, “Characterization of the 1 i=n
shape of the roughness profile ordinate distribution of Rpm R pm = ∑ R pi
n i =1
honed cylinder surfaces”, Wear, 161 (1993), pp. 135-143
[7] B.F.J. Manly, Multivariate statistical methods, 2nd
edition, Chapman & Hall, London, England, 1994
Rt Rt=Rm+Rp
[8] D.F. Morrison, Multivariate statistical methods, 3rd
edition, McGraw-Hill Inc., Singapore, 1990
Wt Maximum profile depth
[9] B.S. Everitt and P. Nicholls, “Visual techniques for Pt Maximum waviness depth
representing multivariate data”, The Statistician, 24
(1975), Vol.24, pp. 37-49 R3z Arithmetic mean third peak-to-valley height
CC.
1 2 3 4
Process
CC.
1 2 3 4
Process
Wt vs Ra R3z vs Rzi Ra vs Rzi Rpm vs Rp
Milling
-0.4% 1.3% -1.9% 5.1%