AS9100 Sampling Plan Requirements - 8.2.4

[stogsdill1]

Hello All!

Once again, thanks for being here. After a successful ISO/TS16949 certification, we are on to AS9100.

We are a medium sized contract manufacturer with a wide range of customers. We currently use an AQL (.010 - 10.0) C=0 sample plan for our ISO and TS receiving and in process inspections. I was told that this sample plan will not be acceptable for AS9100 and that ANSI Z1.4 was required.

The Specification reads:

"When the organization uses sampling inspection as a means of product acceptance, the plan shall be statistically valid justified on the basis of recognized statistical principles and appropriate for use (i.e.,
matching the sampling plan to the criticality of the product and to the process capability)"

I have always assumed the AQL was based on "recognized statistical principles". Can someone tell me definitively whether an AQL C=0 sample plan is acceptable for AS9100?

Thanks in advance!

 

[Sidney Vianna]

Re: AS9100 Sampling Plan Requirements - 8.2.4 - Sampling Plans

In principle, according to AS9100 Rev. C, you could even have an acceptance criteria other than C=0, if the product is noncritical.

Looking at the table available here, I think that, in most cases, you could justify an AQL of 0.01%, but not an AQL of 10%.

Also keep in mind that some aerospace customers have restrictions on sampling. For example, Boeing Commercial Airplanes document X31764 - BCA Quality Purchasing Data Requirements stipulates:

Seller's Inspection Options: Seller shall perform 100% inspection for in-process and final inspection or Seller shall conform to requirements of document D1-8007 "Requirements for Supplier Statistical Plans" as may be amended from time to time. With the exception noted herein Seller statistical sampling procedure/plan conformance to D1-8007 will constitute Boeing Quality approval.
Note: Any characteristics identified in the design documentation as "Safety" or "Critical" (or "Safety Critical," et al.) characteristics shall not be accepted using statistical product acceptance methods unless prior written authorization is granted by the specific Boeing design authority, or the method for acceptance is specifically defined in the design documentation. A "Safety" or "Critical" (or "Safety Critical," et al.) characteristic is defined as a characteristic designated by the design authority, where the responsibility for its definition is outside the scope of recommended practice ARP9013.
Buyer reserves the right to disallow a supplier's statistical methods for product acceptance for specific sites/programs, parts or characteristics, and to conduct surveillance at Seller's facility to assess conformance to the requirements of document D1-8007, available at https://suppliers.boeing.com within the "Supplier Quality" webpage.

 

[Big Jim]

Hello All!

Once again, thanks for being here. After a successful ISO/TS16949 certification, we are on to AS9100.

We are a medium sized contract manufacturer with a wide range of customers. We currently use an AQL (.010 - 10.0) C=0 sample plan for our ISO and TS receiving and in process inspections. I was told that this sample plan will not be acceptable for AS9100 and that ANSI Z1.4 was required.

The Specification reads:

"When the organization uses sampling inspection as a means of product acceptance, the plan shall be statistically valid justified on the basis of recognized statistical principles and appropriate for use (i.e.,
matching the sampling plan to the criticality of the product and to the process capability)"

I have always assumed the AQL was based on "recognized statistical principles". Can someone tell me definitively whether an AQL C=0 sample plan is acceptable for AS9100?

Thanks in advance!

I believe that you understand the requirement and what you have in mind can be appropriate.

Don't forget, however, that a customer requirement to go deeper can trump that. I believe that is what Sydney demonstrated.

In the absence of a customer requirement to go deeper, your auditor is over-reaching. Ask him to explain his position. He can't.

 

[Mike S.]

stogsdil1, you are okay. Ask whomever told you that "ANSI Z1.4 was required" to show you WHERE IS THE SHALL in AS9100. It is not there. They are full of "it".

The standard's text as you quoted it makes sense. No need to complicate matters by adding "requirements" that are not there!

BTW, SAE ARP 9013 has a definition for "statistically valid". It is pretty loose. This standard also allows C=0 plans.

We are AS9100C registered, use a C=0 plan in the ballpark you are talking about, and pass audits from the FAA, massive aerospace customers, and registrars with no problems.

 

[Project Man]

I had an auditor try to explain to me that ANSI Z1.4 has mathematically been proven so it is the only one that can be used. The sampling plans are slightly different at certain levels. BTW - If you are using AQL 1.0 c=0 it is identical to ANSI Z1.4

 

[Bev D]

I had an auditor try to explain to me that ANSI Z1.4 has mathematically been proven so it is the only one that can be used. The sampling plans are slightly different at certain levels. BTW - If you are using AQL 1.0 c=0 it is identical to ANSI Z1.4

ANSI Z1.4 is a version of the old Mil-Std.105 the MilStd has a basis in statistically 'correct' formulas, but there was negotiation involved.

Your auditor is making stuff up tho. there is nothing statistically invalid with a plan that has an acceptance number of zero that is based only an AQL.
it may not be the most effective or rational choice and it certainly doesn't make for thorough justification for an inspection plan. In order to be thorough you should consider effect of the defect, inherent true capability (not Cpk) and what defect rate is NOT acceptable to ship (RQL or LTPD) along with the rate that is acceptable to ship (AQL).

I find many suppliers that select a plan that comes from a statistically valid, 'proven' table from a 'proven' standard, but the selection is too often arbitrary - this is what gets you into trouble with inneffective sample plans.