Professional Documents
Culture Documents
Cost Optimization of Slab Milling Operation Using Genetic Algorithms
Cost Optimization of Slab Milling Operation Using Genetic Algorithms
net/publication/265731282
CITATIONS READS
0 207
1 author:
SEE PROFILE
Some of the authors of this publication are also working on these related projects:
Experimental investigation to studythe effect of micro coated tool on the surface topology of hard to machine materials View project
All content following this page was uploaded by Dr. Sanket N. Bhavsar on 18 September 2014.
Abstract
In order to optimize the objective function for machining, one needs to determine cutting
parameters, such as speed, feed and depth of cut. Here, it is presented the use of Genetic Algorithms
(GAs) to minimize the cost for slab milling operation on horizontal milling machine. The depth of cut
and feed values are set to suit the machine condition and surface finish requirements, while cutting
velocity is optimized for minimum cost of production. A correction factor is incorporated to GAs for
correcting the results of optimum cutting speed and optimum cost of production. At the end the cost
optimum cutting speed and optimum cost generated by GAs are compared with the results generated
by traditional optimization method of using Taylor’s equation of tool life, for its validation. Comparison
of results evolves the efficiency of GAs for finding the cost optimum cutting parameters.
1. INTRODUCTION
Of vital interest to the manufacturing engineer are production cost and production rates. Although, in
practice a high production rate would probably mean low production costs, it should be pointed out
that these two factors must be considered separately and that the manufacturing conditions giving
maximum production rate will not be identical to those conditions giving minimum cost of production.
The machining economics problem (shaw, M.C.; Lissaman, A.J. and Martin, E.J.; Chapman, P.C.)
consists in determining the process parameters, usually the cutting speed, feed rate and the depth of
cut, in order to optimize the objective function. Moreover, in determining these parameters, special
attention is usually given to the restrictions or the constraints imposed on the particular operation by
the machine tool, the cutting tool and the work piece.
For the purposes of the discussion it is necessary to explain that feed is the distance moved by the
tool relative to the work piece in the feed direction for each revolution of the tool or the work piece or
each stroke of the tool or work piece. Confusion may arise in certain multipoint tool operations, such
as milling, where the feed setting on the machine refer to the relative speed between the tool axis and
the work piece in the feed direction (the feed speed). Thus, if the feed speed in the milling operation is
vf and the rotational frequency of the tool is nt, the work piece feed during each revolution of the cutter
is given by vf / nt, and the maximum cutting speed v in a milling operation is given by πdtnt, where dt is
the tool diameter. It now follows that if it is required to double the cutting speed in a milling operation
while keeping the feed constant, it would be necessary to double both the rotational frequency of the
cutter nt and the feed speed vf (Boothroyd, G.).
A number of machining optimization approaches have been reported, some of these approaches were
limited to single pass operations (Wang, J. et al.; Kirov, K.P. and Hristov, H.I.). The parameters
usually optimized in a single pass are the cutting speed and the feed rate. The other approaches
considered the multi pass operations (Sonmet, A.I. et al.; Onwubolu, G.C. and Kumalo, T.;
Amiolemhen, P.E. and Ibhadode, A.O.A.). In the multi pass operations, the parameters to be
optimized are the depth of cut or the number of pass, the cutting speed and the feed rate. Onwubolu
et al. shows that Genetic Algorithms (GAs) is most suitable for the optimization of machining
parameters. Simulated Annealing is also used to find out the optimum machining parameters in multi
pass operations (Saravanan, R. et al.). In this work slab milling operation is selected for single pass
operation. GAs were applied to find out the cutting velocity, which gives minimum cost of production
for given feed rate and depth of cut. At the same time the results of optimization generated by GA are
compared with the results obtained by routine analytical method of optimization which uses Taylor’s
equation of optimum tool life.
1
It is known that the correct feed to use in roughing operations is the highest the machine tool can
withstand in terms of tool forces and power consumption. When a finishing cut is to be taken, the
appropriate feed will be that which will give an acceptable surface finish.
2.2 Choice of Cutting Speed
2
Fig. 1. Approximate values of the cutting speed v r when tool life t r = 60 sec (Boothroyd, G.)
where v is the cutting speed and K is a constant for the particular operation. In slab milling
operation, the value of K will be given by π d l / f , where l is the length to be traveled by cutter, d
is the diameter of the cutter and f is the feed.
The relationship between the production cost and the cutting speed can now be obtained by
substitution of equations ( 6 ) and ( 7 ) in equation ( 2 ):
K K (1− n )
C pr = M t l + M + 1 (M t ct + Ct ) v n (8)
v v nt
r r
To find the cutting speed vc for minimum cost, equation ( 8 ) must now be differentiated with respect
to v and equated to zero. Thus
n
n M tr
vc = v r (9)
1 − n M t ct + C t
2.2.2 Tool Life Determination
In analyzing practical machining operations it is convenient to employ expression for the optimum tool
life for minimum cost t c . This expression can be obtained by substitution of equation ( 9 ) in Taylor’s
tool life equation ( 4 ). Thus,
1− n C
tc = t ct + t ( 10 )
n M
Finally the corresponding optimum cutting speeds can be found from
3
n
t
vc = v r r ( 11 )
Q tc
The value of Q for slab milling can be obtained from geometries shown in figure 2. Thus for slab
milling,
θ 1 1 2a
Q= = + arcsin e − 1 ( 12 )
2π 4 2π dt
where a e is the working engagement and d t is the tool diameter.
4. Implementation
The procedure explained in section 3 for GAs is implemented in C++ language for the following
cutting conditions on a Horizontal Milling machine for slab milling operation.
2
Work piece Material = Cast Iron ( Tensile Strength = 425 MN/m )
Cutting Tool Material = High Speed Steel
Cutter Diameter, d = 0.050 m
Length of Work piece = 0.1 m
Length to be Traveled by Cutter, l = 0.15 m
Feed Rate, f = 0.01 mm/rev
M = Rs. 4/- per min
t l = 180 sec
t r = 60 sec (figure 1)
t ct = 300 sec
v r = 3.5 m/s (figure 1)
C t = Rs. 100/- per cutting tool
n = 0.125
4
Fig. 2. Slab milling operation
According to the machine tolerable limit and surface roughness requirements, values of depth of cut
and feed rate have been selected for machining. Here cutting speed is required to be optimized for
the operation mentioned above. The Genetic Algorithms is planned to take initial population of ten
different speed values from the range of 1 rpm to 4095 rpm. For this purpose a total of 12 digit binary
11
number is generated for each speed selected in a generation, i.e. 2 . After getting the values of
cutting speeds, we need to calculate the total cost of production for individuals ( fx (i ) ) with the help of
equation (8). As GAs are used to obtain the maximum value, our problem is converted to
maximization by subtracting each cost value from the maximum cost value and gives difference of
costs for ten speed values ( f 1(i ) ) . Now to find out probability of individual cutting speeds, each
( f 1(i ) ) is divided by ∑ f 1(i) . The cumulative probability is determined for reproduction from the
initial population. Two point cross over operator is applied on this reproduction pool for swapping the
strings. Afterwards the strings are mutated with probability of 0.05. This completes one generation.
Table 1 shows the result of sixth generation.
Table2 shows the optimum result of each generation. The values of velocity in table2 does not
consider the effect of parameter Q = 0.14 (Boothroyd, G.), which must be considered for slab milling
operation according to equation (11). Hence the corrected optimum velocity from table2 is,
1.8003 1.8003
= = 2.3019 m/s
Qn 0.14 0.125
Now the corrected minimum cost can be calculated by putting this value of cutting velocity in equation
(8), which is 151.9033 Rs.
Sr. Binary Values Speed Cost Cost Prob. Cum. Random String No. of
No. of Speed RPM fx (i) difference Prob. No. No. Strings
Rs f 1(i )
Rs
1 011010111110 1726 8067.647 1394.431 0.02 0.02 0.48 5 0
2 000001110110 118 520.733 8941.346 0.13 0.15 0.17 3 0
3 000010111110 190 727.951 9134.127 0.13 0.28 0.54 5 1
4 001011100110 742 114.669 9347.409 0.13 0.41 0.49 5 1
5 001011100110 742 114.669 9347.409 0.13 0.55 0.87 8 4
6 001011100110 742 114.669 9347.409 0.13 0.68 0.68 6 2
7 001010111110 702 112.281 9349.798 0.13 0.82 0.34 4 0
8 001010111101 701 112.256 9349.822 0.13 0.95 0.51 5 2
9 011001111110 1662 6204.885 3257.193 0.05 1.00 0.66 6 0
10 011011100110 1766 9462.078 0.000 0.00 1.00 0.85 8 0
Sr. Population after String 1st site 2nd site Result of Result of
No. Reproduction no. for of of Cross over Mutation
Cross Cross Cross
5
over over over
1 000010111110 8 2 6 001011111110 001011111110
2 001011100110 4 5 9 001011100110 011011000110
3 001011100110 9 1 3 001011100110 001001100110
4 001011100110 2 5 9 001011100110 001011100110
5 001011100110 6 8 9 001011100110 001011100110
6 001011100110 5 8 9 001011100110 001011100110
7 001011100110 10 10 11 001011100100 001011100100
8 001011100110 1 2 6 000010100110 000011100110
9 001010111101 3 1 3 001010111101 000010111101
10 001010111101 7 10 11 001010111111 101010111111
Table 1. Continued
Generation Minimum Cost Velocity Generation Minimum Cost Velocity
No. Rs. m/s No. Rs. m/s
1 2491.407 3.8099 11 112.256 1.8343
2 113.654 1.6694 12 112.116 1.7741
3 113.654 1.6694 13 112.116 1.7741
4 112.281 1.8369 14 112.116 1.7741
5 112.181 1.8369 15 112.116 1.7741
6 112.256 1.8343 16 112.116 1.7741
7 112.256 1.8343 17 112.116 1.7741
8 112.256 1.8343 18 112.116 1.7741
9 112.256 1.8343 19 112.116 1.7741
10 112.256 1.8343 20 112.078 1.8003
If we follow the routine method of optimization with Taylor’s equation, then first we need to determine
t c from equation (10). This is the optimum tool life for cost minimization. Now from equation (11)
calculate vc , which is the optimum cutting velocity for cost minimization. This value of cutting velocity
gives the minimum production cost when substituted in equation (8). The values calculated with the
help of Taylor’s equation are as follows, which can be compared with the result of Genetic Algorithms.
t c =12,600 sec
vc = 2.2936 m / s
C pr = 150.3676 Rs.
5. Conclusion
Difficulties arise when the optimization process involves many parameters that interact in highly non-
linear way. In this situation number of constraints over the problem may also be quite high. These kind
of problems can be efficiently solved by heuristic methods like Genetic Algorithms and Simulated
Annealing.
6. References
Amiolemhen, P. E. and Ibhadode, A. O. A. (2004). Application of Genetic Algorithms – determination
of the optimal machining parameters in the conversion of a cylindrical bar stock into a continuous
finished profile. International Journal of Machine Tools and Manufacture, 44, 1403 - 1412.
Boohroyd, G. (1985). Fundamentals of metal machining and machine tools. McGraw-Hill Book
Company Inc.
6
Onwubolu, G. C. and Kumalo, T. (2002). Multi pass turning optimization based on Genetic Algorithms.
International Journal of Production Research, 39(16), 3727-3745.
Rao, S. S. (1996). Engineering optimization : theory and practice. New Age International (P) Limited,
New delhi.
Saravanan, R. et al. (2003). Machining parameters optimization for turning cylindrical stock into a
continued fiished profile using Genetic Algorithms and Simulated Annealing. International Journal
of Advance Manufacturing Technology, 21, 1-9.
Shaw, M. C. (1999). Metal cutting principles. CBS Publishers and Distributors, New Delhi.
Sonmez, A. I. et al. (1998). Dynamic optimization of multi pass milling operation via geometric
programming. International Journal of Machine Tools and Manufacture, 39, 297-320.
Wang, J. et al. (2002). Optimization of cutting conditions for single pass turning operations using a
deterministic approach. International Journal of Machine Tools and Manufacture, 42, 1023-1033.