The supply chain approach to planning and procurement management

Hewlett-Packard Journal, Feb, 1997 by Gregory A. Kruger

Consider using part consumption within the factory versus build plan as the frame of reference. The function of statistical safety stocks here is to provide confidence that material is available to support the production plan. A factory with a steady-rate build plan would carry relatively little safety stock because there are only small fluctuations in actual part consumption. Of course, actual order fulfillment performance would depend upon finished goods inventory and the appropriateness of the plan. In this environment, the organization's SRT objective has no direct bearing on the safety stock calculations. The factors influencing the estimate of demand uncertainty and hence safety stock are fluctuations in actual builds from the planned build, part yield loss, and part use for reasons other than production.

If the point of reference calls for measuring demand uncertainty as the deviation between the forecast and real-time incoming customer orders, safety stock becomes a tool to provide sufficient material to meet customer demand. This factory is not running steady-state production but rather building what is required. Now the SRT objective should be included in the safety stock calculations since production does not have to build exactly to real-time demand if the SRT objective is not zero. From this perspective, statistical safety stocks, projected on-hand inventory, SRT, and service levels are all tied together, giving a picture of the investments necessary to handle marketplace uncertainty and still achieve order fulfillment goals.

In choosing between these two frames of reference for the definition of demand uncertainty it comes down to an analysis of factory complexity and timing. If factory cycle times are relatively short so that production is not far removed from customer orders, then demand uncertainty can be measured as real-time orders versus forecast. However, if factory cycle times are long so that production timing is well-removed from incoming orders, then demand uncertainty would best be measured as part consumption versus build plan.

SRT in Safety Stock Calculations

Appendix IV documents the mathematics for incorporating SRT objectives into the safety stock calculations. As has been discussed, using the SRT mathematics would be appropriate when measuring demand uncertainty as deviations of real-time customer orders from forecast. It is critical, however, that we understand how production cycle times affect the factory's actual SRT performance.

As stated in Appendix IV, if factory cycle time is considered to be zero, the SRT mathematics ensures that material sufficient to match customer orders will arrive no later than the desired number of weeks after the customer's order. Clearly, time must be allocated to allow the factory to build and test the completed product. In this paper, this production time is not the cycle time for building one unit but for building a week's worth of demand.

Care must be taken when using the SRT mathematics. Consider that the practice of booking customer orders inside the SRT window will place demands on material earlier than expected from the mathematical model given in Appendix IV. In practice, one should be conservative and use perhaps no more than half of the stated SRT as input to the safety stock model.