Cell formation considering sequence

of operations and workload

.
1 2 3 4 5 6 7 8
1 1 2 1 1
2 1 4 1 1
3 3 3 2
4 4 3 3
5 2 2 2 2 3
6 2 3 1 1 3 2
1, 2, 3 --- 1, 4, 6

4, 5, 6 --- 2, 3, 5, 7, 8

10 moves
Let C
ij
be the number of parts that visit machine j immediately after visiting machine i

or visiting machine i immediately visiting machine j. The C
ij
values can be computed

from a given machine component incidence matrix.

Let X
ij
= 1 if machines i and j belong to the same machine group.

Here X
ij
are defined for values i = 1,...,M-1 and j = i+1,..,M.
( )
1
1 1
1
M M
ij ij
i j i
C X

= = +

Minimize
1
1 1
1 1,..,
k M
ik kj
i j k
X X N k M

= = +
+ s =

1
1
1 1,.., 2, 1,.., 1, 1,..,
ij ik jk
ij ik jk
ij ik jk
X X X
X X X
X X X i M j i M k j M
+ s
+ s
+ + s = = + = +
0,1
ij
X =
The binary IP formulation has 15 variables and 63 constraints.

The optimal solution is given by X
13
= X
14
= X
34
= X
25
= X
26
= X
56
= 1

with objective function value = 11.
1 4 5 6 2 3 7 8
1 1 2 1 1
3 3 3 2
4 4 3 3
2 4 1 1 1
5 2 2 2 2 3
6 2 1 3 1 3 2
Nair and Narendran (1998) introduced the following similarity coefficient between

machines i and j:
ij
ij
ij
A
S
B
=
Here A
ij
represents the contribution of parts visiting both the machines i and j. Part k visiting
both i and j contributes 1 if it visits the machine as its first or last machine in the sequence and
contributes 2 if it is an intermediate machine. Similarly B
ij
represents the contribution of parts
visiting both either machine i or machine j or both. Part k contributes 1 if it visits the machine as
its first or last machine in the sequence and contributes 2 if it is an intermediate machine. The
machine groups are formed based on the S
ij
values.
New Similarity Coefficient
Machines
i-j
J
ij
Machines
i-j
J
ij
Machines
i-j
J
ij

1-2 2/9 2-3 3/9 3-5 0
1-3 9/10 2-4 3/6 3-6 5/10
1-4 7/9 2-5 8/12 4-5 3/11
1-5 3/11 2-6 7/12 4-6 3/12
1-6 6/13 3-4 7/8 5-6 11/16
Considering machines 1 and 2, we observe that part 6 visits both. A total of 7
parts visit either machine 1 or machine 2 or both. A
12
= 2 because part 6 visits
machine 1 as first operation and visits machine 2 as fourth (last) operation. Each
contributes 1 to the A
12
value. The contributions of parts 1, 2, 4, 5, 6, 7, 8 are 1,
1, 2, 1, 2, 1, 1 respectively. Therefore B
12
= 9 and S
12
= 2/9.
The two machine groups are [1 3 4] and [2 5 6] respectively.
There are six inter cell moves.
Cell load data
1 2 3 4 5 6 7 8
1 .3 .4 .2 .1
2 .3 .3 .2 .4
3 .5 .3 .3
4 .4 .1 .2
5 .3 .2 .2 .4 .3
6 .3 .2 .4 .1 .3 .5
1
1
M
ij ik
i
kj
M
ik
i
w X
L
X
=
=
=

Let us assume that there are K groups (k = 1,..K). Each machine should belong to one
machine group. Let X
ik
= 1 if machine i belongs to cell k.


The cell load in cell k from part j is given by
The total cell load variation is given by
( )
2
1 1 1
M K N
il ij jl
i l j
Z X w L
= = =
=

The mathematical programming formulation (Venugopal and Narendran, 1992)
is to Minimize Z subject to

1
1
K
ik
l
X
=
=

1
1
M
ik
i
X
=
>

X
ik
= 0,1.
[1 3 4] and [2 5 6]. Find the cell load variation
for these machine cells?
The average cell load in cell 1 through part 1 is 0.8/3. The sum of loads on part 1 in
machines 1, 2 and 3 is 0.8. The average load is .8/3 = -2666. The average cell load by
the parts on the two cells are given in Table
1 2 3 4 5 6 7 8
1 .266 .1 0 .233 .0666 .233 .0666 .133
2 .233 .166 .2 .033 .1 .0666 .233 .2666
The cell load variance is given by (.3-.266)
2
+(.5-.266)
2
+(.4-.233)
2
+(.3-.266)
2
+---- = 0.47222.

For the cells [1 3 4] and [2 5 6] the variance is 0.31444.
Similarity coefficient
w
jil
= 1 if w
ji
> 0 or w
jl
> 0.
1 2 3 4 5 6
1 -- .047 .42 .583 .125 .155
2 .047 -- .166 .166 .375 .355
3 .42 .166 -- .365 0 .2
4 .585 .166 .365 -- .071 .107
5 125 .375 0 .071 -- .313
6 .155 .355 .2 .107 .313 --
[1 3 4] and [2 5 6]
s
12
= .1/.3 7 = 0.00476.