In Example 11.3, suppose you want to run five simulations, where the probability of passing inspection is varied from 0.6 to 1.0 in increments of 0.1. Use the RISKSIMTABLE function appropriately to do this. Comment on the effect of this parameter on the key outputs. In particular, does the probability of passing inspection have a large effect on when production should start? (Note: When this probability is low, it might be necessary to produce more than 25 batches, the maximum built into the model. Check whether this maximum should be increased.)
EXAMPLE 11.3 TRYING TO MEET AN ORDER DUE DATE AT WOZAC
The Wozac Company is a drug manufacturer. Wozac has recently accepted an order from its best customer for 8000 ounces of a new miracle drug, and Wozac wants to plan its production schedule to meet the customer’s promised delivery date of December 1, 2010. There are three sources of uncertainty that make planning difficult. First, the drug must be produced in batches, and there is uncertainty in the time required to produce a batch, which could be anywhere from 5 to 11 days. This uncertainty is described by the discrete distribution in Table 11.1. Second, the yield (usable quantity) from any batch is uncertain. Based on historical data, Wozac believes the yield can be modeled by a triangular distribution with minimum, most likely, and maximum values equal to 600, 1000, and 1100 ounces, respectively. Third, all batches must go through a rigorous inspection once they are completed. The probability that a typical batch passes inspection is only 0.8. With probability 0.2, the batch fails inspection, and none of it can be used to help fill the order. Wozac wants to use simulation to help decide how many days prior to the due date it should begin production.
Objective To use simulation to determine when Wozac should begin production for this order so that there is a high probability of completing it by the due date.
WHERE DO THE NUMBERS COME FROM? The important inputs here are the probability distributions of the time to produce a batch, the yield from a batch, and the inspection result. The probabilities we have assumed would undoubtedly be based on previous production data. For example, the company might have observed that about 80% of all batches in the past passed inspection. Of course, a discrete distribution is natural for the number of days to produce a batch, and a continuous distribution is appropriate for the yield from a batch.