From: Stan Hilliard shilliard@samplingplans.com
Date: 12 Sep 2000
Time: 15:57:33
It sounds like you have an attribute plan. For each decision, your sampling plan is n=1, c=0. I have calculated the AQL and RQL for this plan using "TP105 -- Sampling Plans for Attributes". I used a producer's risk of 0.05 (5%) and a consumer's risk of 0.05 (5%).
For the sampling plan n=1, c=0: AQL= 5% defective and RQL= 95% defective. The consumer's point at (beta=5%, RQL=95% defective) does not inspire much confidence in the acceptance criteria.
The AOQL of the plan n=1, c=0 is 0.25 (25% defective. This is assuming that rejected lots would be 100% inspected and the defectives pcs fixed. Otherwise the concept of AOQL doesn't apply to the plan.)
A possible solution -- If you have measured data, this could be turned into a variables plan with known standard deviation? If so, and if the within-hour standard deviation is small enough compared to the specification range, you could do much better. Software to analyze the variables plan is "TP414 -- Sampling Plans for Variables".
www.samplingplans.com/programtp414.htm
Sincerely, Stan Hilliard