41.2 THE DETERMINISTIC CASE: SELECTION OF REPLENISHMENT QUANTITIES IN A
FAMILY OF ITEMS.
In this section we consider the case where there is a family of coordinated items
defined in the following manner. There is a major setup cost (A) associated with
a replenishment of the family. In the procurement context this is the fixed (or
header) cost of placing an order, independent of the number of distinct items
involved in the order. In the production environment this is the changeover cost
associated with converting the facility from the production of some other family
to production within the family of interest. Then there is a minor setup cost (a)
associated with including item din a replenishment of the family. In the procurement
context a; is often called the line cost, the cost of adding onc more item or line to
the requisition. From a production standpoint a; represents the relatively minor
cost of switching to production of item ifrom production of some other item within
the same family. In general, we are concerned with items sharing capacity, such as
a container on a ship or a piece of production equipment, or with items sharing
costs, such asa major fixed ordering cost. For a review of the literature on determinis-
tic and probabilistic models see Goyal and Satir (1989), van Eijs (1993b) and
Simpson and Erengue (1995).
11.2.1 Assumptions
‘All of the assumptions behind the derivation of the economic order quantity (Sec-
tion 5.1 of Chapter 5) are retained, except that now coordination of items is allowed
in an effort to reduce setup costs. We recapitulate the assumptions:
1. The demand rate of each item is constant and deterministic (this deterministic
assumption will be relaxed in Sections 11.4 to 11.6)?
2, ‘The replenishment quantity of an item need not be an integral number of
units (the extension to integral units is as discussed for a single item in
Chapter 5).
3. The unit variable cost of any of the items does not depend on the quantity;
in particular, there are no discounts in either the unit purchase cost or the
unit transportation cost (we will relax this assumption in Section 11.3)
‘Fora related problem of finding groups of raw materials that should be ordered together, see Rosenblatt
and Finger (1988); Chakravarty (1984); Rosenblatt and Kaspi (1985); Chakravarty and Martin (1988)
van Eijs, Heuts and Kleijnen (1992); and Hong and Hayya (1992).
"The sitvation of mevorying but deterministic demand patterns with no capacity constraint (as was the
case in Chapter 6) has been analyzed by Kao (1979), and Lambrecht, Vanderveken, and Vander Eecken
(1979). In particular, the later presenta heuristic procedure that efficiently determines the timing and
sizes of replenishments of the various items involved. See Section 11.62 for the case with a capacity con-
scraint.426
COORDINATED REPLENISHMENTS AT A SINGLE STOCKING POINT
4. The replenishment lead time is of zero duration; but the extension to a fixed,
known, nonzero duration, independent of the magnitude of the replen-
ishment, is straightforward.
No shortages are allowed. (Sections 11.4 to 11.6 will deal with the situation
where shortages can occur.)
6. The entire order quantity is delivered at the same time.
11.2.2 The Decision Rule
In the independent EOQ analysis of Chapter 5 we showed that the EOQ expressed
as a time supply was given by
2A
Dor
Note that Tzog increases as the ratio A/Duincreases. Within our context this indicates
that an item i with a high setup cost a, and a low dollar usage rate D,x; should
probably be replenished less frequently (higher time supply) than an item j having
a low a, and a high Dy,. Because of the assumptions of deterministic demand, no
shortages permitted, and instantaneous delivery, it makes sense to include an item
in a replenishment only when its inventory drops to the zero level. Therefore, a
reasonable type of policy to consider is the use of a time interval (7) between
replenishments of the family and a set of m’s where m;, an integer, is the number
of T intervals that the replenishment quantity of item é will last. That is, item é will
be included in every mth replenishment of the family.’ For example, if m, = 3,
item 17 will only be included in every third replenishment of the family, with a
replenishment quantity sufficient to last a time interval of duration 87. Each time
it will be replenished just as its stock hits the zero level. We wish to select the values
of Tand the m’s to keep the total relevant costs as low as possible.
Figure 11.1 contains a more detailed example. Items 1 and 3 are ordered every
time an order is placed (that is, m, and m, = 1). Item 2 is ordered every other
time, so that m = 2
‘As shown in the appendix of this chapter, the integer ms must be sclected
to minimize
Teog
(4432) mos wl
Jackson, Maxwell and Muckstadt (1985) have developed an efficient procedure for the more restrictive
Situation where m = 1, 2, 4, 8, etc. (chat is, 2 multiple of two) and T itself is also such a multiple of
‘some basic time period (such asa day or a week). This type of restriction is commonly found in practice.
Moreover, Jackson et al. have shown that their solution is, at most, 6 percent more costly than the
“optimal” solution to the problem. We do not present the details here since their solution is more
difficult conceptually than is ours, See also Roundy (1985); Roundy (1986); Atkins and lyogun (1987);
Jackson, Maxwell, and Muckstadt (1988); Zoller (1990); Atkins (1991); Adkins and Sun (1995); and
‘Section 11.4, where we return to the “powersoftwo" solution. Klein and Ventura (1995) present @
simple procedure for finding the optimal cycle times if Tis restricted to take on one of a set of possible
discrete values.11.2 THE DETERMINISTIC CASE: SELECTION OF REPLENISHMENT QUANTITIES 427
tte 1 mem tiem 2 ems tem 3
Inventory level
Time
Figure 11.1. Behavior of Inventory over Time
where
A= major setup cost for the family, in dollars
minor setup cost for item i, in dollars
demand rate of item 4, in units/unit time
v= unit variable cost of item i, in §/unit
n= number ofitems in the family (the items are numbered 1, 2, 3,
n-1,)
m,= the integer number of Tintervals that the replenishment quantity of item i
will last
Also, once the best m/s are known, the corresponding value of T'is given by
T*(mis) 112
Itis worth noting that r does not appear in Eq. 11.1; that is, the best values of
the m’s do not depend on the carrying charge.
Goyal (1974) has proposed a search procedure for finding the best set of m/s.
See related research by Naddor (1975); Chakravarty (1985b); Bastian (1986); Goyal
(1988); Kaspi and Rosenblatt (1991); Goyal and Deshmukh (1993); van Eijs (19982);
Hariga (1994); and Viswanathan (1996). We suggest, instead, the use of the following
much simpler (noniterative) procedure which is derived in the appendix of this
chapter:
Srep 1 Number the items such that
Dy,
is smallest for item 1. Set m = 1
Srep 2 Evaluate
mi
aie 13
Dy, At a “
rounded to the nearest integer greater than zero.