Page 181 - Orlicky's Material Requirements Planning

160 PART 2 Concepts

The variability of demand consists of nonuniformity (varying magnitude of period
demand) and discontinuity (gaps of no period demand). The length of the planning hori-
zon, that is, demand visibility, obviously affects the comparative performance of the var-
ious algorithms. Shorter planning periods (e.g., weeks instead of months) result in small-
er requirements per period, enabling the lot-sizing technique to get closer to the best bal-
ance between setup and carrying costs. The setup/unit-cost ratio directly affects the fre-
quency of ordering and thus the lot size.
There does not appear to be one “best” lot-sizing algorithm that could be selected
for a given manufacturing environment for a class of items and in most cases even for a
single specific item. For purposes of MRP, the lot-for-lot approach should be used wher-
ever feasible, and in cases of significant setup cost (typical in the fabrication of compo-
nent parts), LUC, LTC, PPB, or even POQ should provide satisfactory results. When it
comes to selecting a lot-sizing technique (or techniques) to be incorporated in an MRP
system, neither detailed studies nor exhaustive debates are warranted—in practice, one
discrete lot-sizing algorithm is about as good as another.
Apart from the inherent weaknesses and difficulty of meaningful comparison
between algorithms, the one fact of life that renders, and always will render, any lot-siz-
ing technique vulnerable is the possibility that future requirements will change. After the
planned order is released, the order quantity may prove to be wrong in light of a change
in the magnitude and/or timing of net requirements.
When this happens, it does not matter how elaborately and with what precision the
lot size had been computed. All the discrete lot-sizing algorithms are based on the implic-
it assumption of certainty of demand. This is the true Achilles’ heel of lot sizing because,
in most cases, the pattern of future demand is never certain. A more realistic assumption
would be that the requirements schedule against which the lot size is being computed
will change.
In comparing the relative effectiveness of one discrete lot-sizing algorithm with that
of another, it is possible to determine which of the two is better vis-à-vis a given sched-
ule of net requirements. When the period of time spanned by this schedule has passed
into history, however, it might develop that the algorithm originally judged less effective
would, in fact, have had the better performance in light of how the requirements actual-
ly turned out.
In the final analysis, it does not matter how elaborately and with what precision lot
sizes can be computed. All techniques are vulnerable to changing demand, and unfortu-
nately, this is found in most MRP environ ments. The best way to ensure maximum econ-
omy in ordering materials is to develop and maintain stable plans for procurement and
fabrication. The spurious precision of lot-sizing technique is invalidated by what actual-
ly happens as opposed to what had been planned to happen. The relative actual effec-
tiveness of a lot-sizing algorithm can be determined only in retrospect.