Page 176 - Orlicky's Material Requirements Planning

CHAPTER 8 Lot Sizing 155

ods owing to a lack of horizon. It eventually will have to be recomputed and increased so
as to pick up an additional 60 part-periods. This will more than offset the net saving of
30 part-periods in periods 1 through 9. But the first three lots then will have covered a
larger time span than had they not been adjusted. At this point, the basis has been lost for
making a valid comparison of the alternative strategies.
The look-back proposition appears, if anything, even more dubious than look-ahead
proposition. It suffers from the same shortcoming as look-ahead in that it fails to exam-
ine the consequences of the adjustment throughout the planning horizon. In our exam-
ple, the look-back produces a net saving of 75 part-periods but adds a fourth setup that
is worth 100 part-periods (Figure 8-12). If setup cost was larger than unit cost, the EPP
would be higher and the lots spaced further apart. The last period demand covered by
any prospective lot then almost always would entail more part-periods than the first peri-
od of the subsequent lot. Look-back then would be more consistently operative, resulting
in more and smaller orders, which would subvert the logic of the LTC technique on
which the look-back is grafted.

Wagner-Whitin Algorithm
This technique embodies an optimizing procedure based on a dynamic programming
4
model. The procedure is too mathematically involved to be suitable for a detailed
description here. Basically, it evaluates all possible ways of ordering to cover net require-
ments in each period of the planning horizon. Its objective is to arrive at the optimal
ordering strategy for the entire net requirements schedule. The Wagner-Whitin algorithm
is elegant in that it reaches this objective without actually having to consider, specifical-
ly, each of the strategies that are possible. The Wagner-Whitin solution to the net require-
ments schedule used in all preceding examples, except for PPB, is shown in Figure 8-13.
The Wagner-Whitin algorithm does minimize the combined (total) cost of setup and
of carrying inventory, and it is used as a standard for measuring the relative effectiveness
of the other discrete lot-sizing techniques. Its disadvantages, usually mentioned in the lit-
erature, are a high computational burden and the near impossibility of explaining it to
the average MRP system user.
The first of these two arguments is somewhat exaggerated. While it is true that there
are typically tens of thousands of inventory items in an MRP system for which planned

FIGURE 8-13
Period 1 2 3 4 5 6 7 8 9 Total
Wagner-Whitin
algorithm. New Requirements 35 10 40 20 5 10 30 150
Planned-Order Coverage 40 65 40 150

4 H. M. Wagner and T. M. Whitin, “Dynamic Version of the Economic Lot Size Model.” Management Science. 5(1),
1958.