The reemergence of the Economic Order Quantity (EOQ) policy as a good performer in a multistage system is explained in part by its excellent performance with respect to final product stock-outs. The EOQ policy places larger orders fewer times per year and is therefore exposed to the risk of stock-out less often. Tests by researchers also confirm that fixed EOQ policy does quite well when there is considerable uncertainly in the demand pattern. Based on the simulation experiments, managers may select policies based on their preference for a given weighting of the criteria.
These results must be regarded as preliminary. As with most early simulation experiments, actual shop conditions are only approximated. In this instance, the simulated shop was loaded well under capacity. No lot splitting was permitted, and when â€œdesiredâ€ production exceeded capacity, a priority index was used to determine which lots would not be processed. It is difficult to tell how the five lot size policies would have performed if these arbitrary rules had not been imposed and if the load had been varied over a range that included heavy loads. Nevertheless, testing alternative lot size policies within multistage environments is important and is a step toward resolving the issue of how different lot size policies actually perform.
We have already noted that dependent items are not subject to the kinds of random variations in demand that characterize primary product demand. The demand variability is largely planned and is lumpy in nature. There are sources of variation; however, for which buffer stocks is a logical counter measure. Buffer stocks in requirements systems are designed to absorb random variations in the supply schedule. The time required for processing orders through the system is variable because of such factors as delays, breakdowns, and plan changes. In addition, the actual quantity delivered from production is variable because of scrap. The result is that we need a cushion to absorb variations in supply time and in the quantity actually delivered.
If the lead time for manufacturing a component or a part is variable, then stock-outs could occur if a batch is not scheduled for production sufficiently in advance of actual requirements. In our example, suppose component B takes normally two weeks to manufacture. However, 10% of the time the shop is busy with other orders; therefore, the lead time is really three weeks. Rarely, say one percent of the time, production lead time could be as high as four weeks. Now in this situation, if management desires that stock-outs do not occur even 1% of the time, then a batch of component B must be put into production four weeks in advance of the actual net requirement. This would results in much higher inventory holding costs since most of the time production will take less than four weeks and excess inventory will have to be carried. The determination of lead time should therefore balance the cost of holding inventory and cost of stock-outs. Attempts should also be made to reduce the variability in lead time by proper scheduling and coordinated production control systems.
In some situations, per period requirements are variable because of sales forecast errors, customer order changes, or production variability in upstream departments. Alternatively, a larger quantity than required may have to be manufactured to adjust for rejects, scraps, and the like. A safety stock is needed to guard against such variability. Safety stock is determined by considering the cost of stock-outs and cost of holding excess inventories. Once a safety stock level is determined, the lot size policies discussed earlier are modified to start production when net requirements fall to the level of the safety stock.
Caution is warranted in the use of buffer stocks. If sales forecasts are inflated, production lead times are pessimistically estimated to be longer than normal, and orders are triggered when on-hand inventory is still sufficient, the compounding effects could produce very large raw material and work-in-process inventories. This defeats the primary purpose of material requirements planning systems. It is therefore recommended that sufficient efforts be expanded in identifying, isolating and correcting the causes of variation in-lead times or requirements so that safety lead times and safety stocks can be kept at a minimum.