AboutJoe Martin Expertise I can answer questions about Quality Systems (AS9100, AS9103, Boeing D6-82479, etc.) Particular areas of expertise are Statistical Process Control (SPC), sampling inspection, design for Six Sigma, and Six Sigma tools in general.
Experience Quality engineer in aerospace for 20+ years. ASQ Certified Quality Engineer (#8920) since 1982.
Expert: Joe Martin Date: 9/24/2007 Subject: AS9100 8.2.4
Question While preparing for an AS9100 registration audit a question came up concerning paragraph 8.2.4. Would you interpret the requirement for zero defect sampling plans to be applicable to parts being received? The reason for the confusion is the "Note" in paragraph 1.1 stating that the term "product" applies only to the product required by the customer. There is also no such restriction under paragraph 7.4.3 "Verification of Purchased Product"
Answer Ah, nothing like being an AS9100 (or D6-82479, or whatever) lawyer!
As they say in the NFL, let’s break it down: “When the organization uses sampling inspection as a means of product acceptance, the plan shall be statistically valid and appropriate for use.”
What does “statistically valid” mean? Who decides when or if something is “appropriate for use”? The purpose of a specification is to provide clear, unambiguous requirements. That sentence does nothing of the sort—it’s just a bunch of extraneous “weasel words”. (Think maybe I’m not a big AS9100 fan?)
Now, on to your immediate question, which I presume refers to the next sentence: “The plan shall preclude the acceptance of lots whose samples have known nonconformities.” (The technical term is a c=0 sampling plan.)
This one’s clear enough: If you’re going to use sampling inspection, you have to use a c=0 plan. Period.
Although you don’t mention it, you apparently have some concern about using c=0 plans for incoming inspection. If you’re looking primarily to minimize the sample size (and thus the amount of inspection), why not just use one of Boeing’s c=0 plans? The sample sizes are quite reasonable; much more so than most of the ANSI Z1.4 (MIL-STD-105) plans. Or, you could pick and choose AQL and inspection levels (and thus sample sizes) with c=0 from the ANSI plans. Or you could develop some c=0 plans yourself, complete with associated risk factors, etc. (Not that hard, but another subject entirely.)
As a last resort, which I don’t recommend, you might argue the “term ‘product’ applies only to the product required by the customer” line. (Look at the note about “NOTE” on page 5.) Even if you convince your customers and/or the ISO registrar, you’re faced with maintaining some sort of dual sampling scheme based on whether or not the product will ultimately be shipped to a customer.
Bottom line advice: Avoid the potential for corrective action and/or real problems! Pick, or develop yourself, a set of c=0 plans that suits your needs and go with them throughout your operation.
