Professional Documents
Culture Documents
problems. 6.4.3 Single-Pass Heuristic f OC n logn), making Although producing a good starting
solution, binary ordering ignores machine uti- lizations, group sizes, and exceptional elements, In
this section we propose a fairly simple, single-pass heuristic for determining groups. At each step
the next ordered part is assigned to a group along with the required machines. The process Will
con- strain groups to the desired number of machines. The procedure is relatively quick. What is
sacrificed is that the procedure may request extra machines. However, we can easily compute
Iower bounds on the number of machines of each type that are needed for any feasible solution
and compare these to those suggested by this heuris- tic. If the heuristic finds a solution that does
not require any additional machines, we have an optimal solution. In general, the heuristic forms a
starting point for sub- sequent mannar modification of the groups. The heuristic is designed for
the case where it is deSired to totally eliminate all intergroup moves, and an upper bound on the
size of any. group exists. (Clearly, we assume that no part uses more machines than the maximum
group size.)
184
PART 11 MATERIA. FLOW SYSIEMS
(Add part and machines to current group): Tf
—m + MG m„, then set
, mG U,nG + tdi,n, and ifUmG + un,n n,nG, n,nG n,nG + 1 and
go to 3.
(Start new group): G G + I. Ai G, nmG 1 for unm 0,UmG —
On completing this initial group assignment, we check for the total number of
machines of each type required, If we are ag the Iower bound, rejoice. For machine
types exceeding this bound consider combining groups. Only group combinations
where slack for this machine type exceeds 1.0 need be considered. Small groups may
also be combined to approach the mu limit.
EXAMPLE 6.3
Consider the problem in Yable 6.4, and assume mu 4. Parts and machines
are alreacly ordered and utilizations are noted. Use the single•pass heuristic to
form tentative groups.
Solution
Updating of results is shown in Table 6.5.
i 1. Partlgoes to group 1, taking along a machine A andB. This imposes
a utilization of 0.4 on both machines, leaving a slack of 0.6 as shown in Table
6.5.
i — 2. Part 2 fits into this group with no new machines needed and,
hence, is added. Slack time is reduced to 0 and 0.1 for machines A
and B.
i 3. Part 3 would require three new machines if it were added to group
l, since neither existing machine can handle the added load. Adding three
machines would exceed the allowable group size; henee, a new group is
started.
i 4. Part 4 Will not fit into either group without adding a machine D.
However, there is space in the current group (group 2) for a new machine.
We thus add 4 and D to the second group. Machines A and B are unaffected
by adding part 4. Available time on machine C is reduced from 0.5 to 0.1.
i 5. Machine C does not have capacity for part 5, and the group is full
(in terms of machines), and so we open group 3.
i 6. Part 6 Will fit only in group•3and is added accordingly
i 7. Part 7 can fit into group 3. by adding machines E and F. This is
préferable to opening a new group, since we can utilize the existing machine
D. Addingúe two machines Will not violate the constraint of four machines
per group.
i 8. Part 8 is then added to this group, since it requires machines E and
F and sufficient slack time exists.
Examining the solution we find that only machine type D exceeds its Iower
bound. Both groups 2 and 3 have a D allocated, but planned production re-
quires only one machine D for the shop. Our alternatives include combining
groups 2 and 3 into one seven-machine group; purchasing the extra machine D;