Does C = 0 have a expanded twist for problem vendors (suppliers)?

R

RI_Tammy

My company currently uses a C = 0 Sampling Plan. However because of the reduced sampling plan it isn't working for problem vendors. Is there some kind of expanded version of C = 0 that can be used? I used to use the standard ANSI sampling plan using reduced, normal and expanded charts.

Can anyone offer any guidance??

Thanks,
Tammy
 

mdurivage

Quite Involved in Discussions
Re: Does C = 0 have a expanded twist for problem vendors

We use a one colum shift to the left if we start experiencing supplier issues.
 

Statistical Steven

Statistician
Leader
Super Moderator
Re: Does C = 0 have a expanded twist for problem vendors

Increasing the sample size will reduce the AQL and LTPD levels. But that does not solve the problem. If a vendor consistently fails the current sampling plans, any increase in scrutiny will only increase the number of lots rejected. Need to improve the quality coming in from the vendor.
 

Bev D

Heretical Statistician
Leader
Super Moderator
The straight forward answer to your question is to simply go to a 'tightened' sample plan and the sample size that allows a c=0 accept number (If you are comfortabel with ANSI plans)

However, there are a lot of other approaches out there, so it might be helpful if you told us a bit more about your situation.

What type of 'problem' are you expereincing with your problem suppliers? high rejection rate at incoming? low rejection rate at incoming but high fallout in manufacturing?

Also remember that the AQL is the rate at which 95% of the lots will b accepted. The LTPD is the rate at which 90% of the lots will be rejected. Often people misinterpret the AQL...
 
C

CertifiedDataJunkie

It's obvious that the sampling plan alone cannot ensure that incoming lots are accepted at your facility (i.e., cannot inspect quality into a product). For example, these same lots are often sorted at the vendor site before shipment if found to violate the c=0 sampling plan to identify nonconforming items and ensure that the shipped lot is in compliance with customer requirements. Therefore, ensure that your vendor partner is measuring these critical attributes the same way they are measured at your site. Also, ensure that these specifications are well defined and understood at the vendor site. As Statistical Steven indicated, the failure of the sampling plan may be a symptom and the root cause for these failures needs to be identified. My experience is that measurement systems analysis (MSA) is one issue that often needs to be addressed so that lots released by the vendor are measured the same way as they are evaluated at the customer site.
 

Kales Veggie

People: The Vital Few
My company currently uses a C = 0 Sampling Plan. However because of the reduced sampling plan it isn't working for problem vendors. Is there some kind of expanded version of C = 0 that can be used? I used to use the standard ANSI sampling plan using reduced, normal and expanded charts.

Can anyone offer any guidance??

Thanks,
Tammy

Let me ask you this: Why would you keep accepting "NG" product? (You are expecting zero defects, right?)

This supplier needs supplier development support, containment actions in place, a visit from your Supplier Quality group or Purchasing, a corrective action request.

More inspection does not solve the problem.
 
R

RI_Tammy

We generally have no problems. There are some instances where a product has passed inspection at C = 0 sampling rates yet non conformances have been found down the line. Of course CAPAs were originated and implemented. But when this instance occurs once I would want the next incoming inspection to be more stringent than the standard sampling rate of C = 0 to look at I larger sampling rate. My question is how to write this in to the procedure. Maybe if a non conformance occurs a shift two columns to the left will be implemented for the next run or two. Also is the C = 0 sampling plan a good plan to use for validations and auditing? What level would you use for that - 1.0??
 
S

Sturmkind

Hi, RI Tammy!

I agree with Kales and related comments that direct effort at the supplier's facility may yield immediate results in the reduction of the probability of them shipping non-conforming product to you.

If the non-conformity causes disruption downstream, then consideration of sampling by variables could detect shifts within the supplier process or they could sample themselves and send the results with each shipment.

With either Attribute or Variable sampling plans the plan to draw as random a sample as possible is paramount though I have rarely seen that assumption carried out in actual practice.

The concept of switching 1 or 2 columns to the left on the C=0 plans makes perfect sense. Switching rules were deliberately omitted by Squeglia; from N. Squeglia, 3rd ed. C=0 page 16 "Considerable thought was given to 'tightened' and 'reduced' inspection...different operating characteristic curves are involved...Tightened inspection was viewed as a means to force corrective action by increasing the chance of lot rejection."

So, to modify your procedures for a leftward switch for 'x' number of lots should work and a switch of 1 column left for standard characteristics and 2 for significant characteristics may be practical.
 

bobdoering

Stop X-bar/R Madness!!
Trusted Information Resource
We generally have no problems. There are some instances where a product has passed inspection at C = 0 sampling rates yet non conformances have been found down the line. Of course CAPAs were originated and implemented. But when this instance occurs once I would want the next incoming inspection to be more stringent than the standard sampling rate of C = 0 to look at I larger sampling rate. My question is how to write this in to the procedure. Maybe if a non conformance occurs a shift two columns to the left will be implemented for the next run or two. Also is the C = 0 sampling plan a good plan to use for validations and auditing? What level would you use for that - 1.0??

Here's the thing: C=0 sampling plans are developed as a balance between processor and consumer risk, not a guarantee of detection. In fact, sampling plans become more lax when the capability becomes better BUT in actuality they need o be more rigorous to find fewer and fewer rejects. Think needle in a haystack. There is only one needle. But to detect it, would you check less? If you decide to check more, how much more? Don't expect sampling plan techniques to help...you will end up inspecting 100% to find that needle.
 

Statistical Steven

Statistician
Leader
Super Moderator
Bob...excellent point. I have given a talk called "The Odds are Against Auditing". Basically stating that the more compliant a system is the larger the sample size needs to be to find any issues.
 
Top Bottom