Continous Flow (Chain) Sampling Plans


Fully vaccinated are you?
Chain (Continuous) Sampling

Subject: Re: Q: Chain sampling /Fackelmann/Isackson
Date: Mon, 26 Oct 1998 09:55:19 -0600
From: ISO Standards Discussion

Subject: Re: Q: Chain sampling /Fackelmann/Isackson

Peter Fackelman posted (in part):

>Could a kind soul give me a pointer to a (MIL?) standard about chain sampling?

I am unaware of a MIL-STD for Ch-Sp plans. There is (or was) a handbook H106, "Multilevel Continuous Sampling..." for special cases when the quality is known to be better than some AOQL. There are Bayesian procedures as well.

The original work on chain sampling was done by Horace Dodge who published several papers on it in the '50's. However, the ASQ published a quite respectable monograph on it by Kenneth Stevens, "How to Perform Skip-Lot and Chain Sampling." It is still available from that organization, 800 248 1946.

Kwena mzuri,

Frank Isackson

Don Winton

For anyone interested, additional info regarding chain sampling can be found at Juran's QC Handbook, 4th Edition at pages 25.37-25.39 or Grant and Leavenworth's Statistical Quality Control, 6th Edition at pages 509-512. I do not have any of Dodge's original papers right off hand, but if I find them, I will post availability.



Continous Flow Sampling Plans

I currently use a C=0 sampling plan with a default AQL of 4.0 for visual inspections on the production floor. Production is moving from batch processing to continous flow. Where can I find information for developing a sampling plan to use in a continous flow process as opposed to a batch process?

Johnny Jewel

There is an ASQ publication, “How to Perform Continuous Sampling (CSP)” by Kenneth Stephens. Some of the sampling schemes it considers are from the old MIL-STD-1235C, which, for some reason, was not carried over to be an ANSI/AQC specification as, for instance, MIL-STD-105E was. Juran’s 3rd Edition is good also; from the 4th on, too much condensation was done to be of much value.

My organization studied continuous sampling during the development of a new product line. Our historical assembly method was by 1 person start-to-finish. The new product was concurrently designed with a new assembly line production process, and testing was changed from batch method (105E) to a continuous sampling plan. CSP-V met our requirements and has been in use for about 4 years.

The scary part of CSP’s is the requirement of 100% inspection at the beginning and whenever a non-conformance is detected. This can make the concept a hard sell, and growing into the plan can be agonizing. Batching for this product/assembly line would have been completely artificial, considering the nature of part flow to the line and the way the product is adjusted during/after assembly. One feature that is designed into the plans is that there is no retrieval when a non-conformance is detected. Depending on the plan chosen, you either immediately re-sample for verification or go immediately to 100% inspection until some clearance number of continuous OK product is reached. Then you go back to sampling. There are trade-offs among the types of plans, the main one being that as the % sampled decreases, the clearance number gets larger and vice-versa.

The good news for us was that the product was robust for the characteristic tested. At least that is how some have viewed the experience. There also was a big incentive to develop the product – tune it, continuously improve it -- to keep it out of the dreaded 100% inspection. In other words, we may have more confidence in this product because of the sampling plan that was chosen.

Don Winton

This site <A HREF="" TARGET="_blank"></A> has some good stuff on sampling.

This site <A HREF="" TARGET="_blank"></A> has some standards for download that may be of use.



I was better but I got over it.
Top Bottom