Page 229 - Orlicky's Material Requirements Planning

208 PART 3 Managing with the MRP System

The preceding two objectives and the techniques used to achieve them will be reviewed
separately in the discussion that follows.

Disentangling Option Combinations

Under the MRP approach, product variations or optional features must be forecast at the MPS
level; that is, it must be possible to forecast end items rather than their individual component
items. When a product has many optional features, their combinations can be astronomical,
and forecasting these combinations becomes impractical. A valid MPS could not even be
established and stated in terms of such BOMs. This problem is solved by means of a modu-
lar BOM. Instead of maintaining BOMs for individual end products, under this approach, the
BOM is restated in terms of the building blocks, or modules, from which the final product is
put together. The problem and its solution can best be demonstrated by an example. The farm
tractor discussed in the preceding section had 11 optional features and a total of 25 individ-
ual choices, making it possible to build 6,912 unique product configurations.
There is no special difficulty in writing a BOM for any one of these configurations,
but it is not practical to store and maintain thousands of BOMs for a single product fam-
ily. Many of the 6,912 possible configurations may never be sold during the life of the
product, and thus their BOMs would never be used. Furthermore, design improvements
and engineering changes could add additional BOMs to the file. Consider this: The trac-
tor, as described, has only one type of fender, but if the engineers create an option of spe-
cial fenders with mudguards, the number of possible option combinations will double
from 6,912 to 13,824. This means that another 6,812 BOMs would have to be added to the
file. This is one reason why BOMs for end products should not be maintained in this case.
But the other reason mentioned earlier is equally important; that is, with these thousands
of BOMs, it would not be possible to state a valid MPS in terms of end products.
If the tractor manufacturer produces 300 of this type of tractor per month, which 300
of the 6,912 possible configurations should he or she select as a forecast for a particular
month? This is simply not a practical proposition. Note that volume is part of the prob-
lem. A product family with 100 possible option combinations constitutes a problem if vol-
ume is 20 per month. If volume were 10,000 per month, the forecasting problem would
not be nearly as serious.
The solution of this problem lies in forecasting each of the higher-level components
(i.e., major assembly units such as engines and transmissions) separately and not
attempting to forecast end products at all. This amounts to forecasting the various choic-
es within the optional product features and translating such forecasts into the MPS.
Specifically, if 300 tractors of the type in question are to be produced in a given
month, 300 so-called basic tractors (including fenders, hoods, rear wheels, etc.) would be
scheduled. A BOM for this module would be required to match the schedule. There are
two choices of transmission, however, and let us assume that past demand has averaged,
say, 75 percent stick shift and 25 percent automatic. Applying these percentages to the
transmission option, 225 and 75 units, respectively, could be scheduled. But the batch of