 |
 |
 |
|||
 |
|
||||
 |
 |
 |
|
||
 |
|
||||
 |
|
||||
 |
|
||||
|
Vol 27 - No 09 - November 2001 Letter to the editor I enjoyed Jeremy Mant's article about safety stocks in the October issue of Control, but suggest readers should be cautious of two matters. Firstly, the use of the distribution of forecast errors alone, without taking any account of the frequency of stock replenishments, is likely to give erratic results. For example, by the definition of service given, and based on a 90% level, a product replenished once per month will stock-out once every ten months – poor service. The same product, with the same distribution of errors, but replenished once every year (with a much larger amount of stock, of course) will stock-out once every ten years – pretty good service! The safety stock calculation originally introduced in IBM's IMPACT system in the late 1950s and found as standard in all software, incorporates the term A/N, where A is the product's annual demand and N is the number of replenishments per year. The workings out involved are reproduced in all text books dealing with this subject. The second matter relates to the distribution of forecast errors for ‘fast moving consumer goods'. While the distribution of forecast errors for general industrial goods are almost always normal, the distribution for consumer goods is very commonly found to be gamma or exponential. This is hardly a pedantic point –safety stock calculations involve the tail end of distributions where the difference between them are very considerable. Yours sincerely David Crabtree, FIOM Related Topics:
|
|
|
|
|
||
|
|
|