I have a customer that requires all of our process to prove capability to 1.33Cpk or higher, but they want us to continue 100% inspection. I can not see the logic in requiring proof of capability if they are still going to require 100% inspection for the same characteristics. Anyone ever run into this problems before??

I have a customer that requires all of our process to prove capability to 1.33Cpk or higher, but they want us to continue 100% inspection. I can not see the logic in requiring proof of capability if they are still going to require 100% inspection for the same characteristics. Anyone ever run into this problems before??

Hi Lpitt56,

I think your confused with those two...cpk has nothing to do with 100% inspection. :nope:
Cpk pertains to the capability of your process/machine to produce products within specifications. high Cpk = more products within specs. if i remember it correctly Cpk = 1.33 is equivalent to 95% yield. 1.66 is around 99%. Cpk monitoring is a proactive approach. 100% inspection on the otherhand is a reactive approach...trying to inspect everything to ensure conformity to specs.

IMO, your customer is trying to tell you and help you build his confidence on the products you are delivering to him. In short he's saying:

Cpk = improve your process and while your doing that...
100% inspection = make sure you don't ship to me bad products. :whip:

later on, you may negotiate on relaxing 100% inspection to sampling after you demonstrate your process capability and conformity.

:topic: I don't believe there's such a thing as 100% inspection. except of course if it will be done by a machine or robot.:notme:

hope this helps.

:bigwave: EMTT

Thanks for the reply. I agree that 100% inspection does not add value or protection. The problem I have is that I am in the aerospace industry and sampling is almost never allowed. Every part we build must be measured and validated on a CMM, undergo ultrasonic inspection and many other steps. The industry will never allow sampling because of the high liability. For that reason, you see very little SPC in the aerospace industry. Now I have one customer that wants me to do a capability study (as you properly stated it’s a statistical prediction of your yield). But I'm thinking why do I need statistics when I have actual process date for every part produced? I already know exactly what my yield is so doing a study to determine a Cpk just does not seem to be worthwhile.

I'm I overlooking something?

Lpitt56

First off, welcome!

Cpk does tell you a little more than simply the yield - it tells how consistently you can produce the product. Just being in spec is not always good enough- often in it important to be close to target and/or to reduce variation. If you could have parts scattered from one spec limit to the other, you might have 100% yield, but that isn't as good as all the parts tightly clustered near the target. Cpk tells you something about how well you are clustered and how far the parts are from the specs.

Since you have all the measurements already, it presumably only takes 0.01 seconds of computer time to calculate the Cpk value, so it shouldn't be much of a hardship. ;)


Tim F

Thanks for the reply. I agree that 100% inspection does not add value or protection. The problem I have is that I am in the aerospace industry and sampling is almost never allowed. Every part we build must be measured and validated on a CMM, undergo ultrasonic inspection and many other steps. The industry will never allow sampling because of the high liability. For that reason, you see very little SPC in the aerospace industry. Now I have one customer that wants me to do a capability study (as you properly stated it’s a statistical prediction of your yield). But I'm thinking why do I need statistics when I have actual process date for every part produced? I already know exactly what my yield is so doing a study to determine a Cpk just does not seem to be worthwhile.

I'm I overlooking something?

Two small things.

- 100% inspection does not ensure that you wont reject the part that you inspected. SPC helps you to keep within limits. I dont think you would have any issues if your inspection efforts produce better results.

- SPC gives you advance warnings when you tend to start slipping, if used as a control tool. You can then do setting corrections and ensure that you stay within limits. (Not produce rejects)

Think of it, it may sound worthwhile.

Thanks for the reply. I agree that 100% inspection does not add value or protection. The problem I have is that I am in the aerospace industry and sampling is almost never allowed. Every part we build must be measured and validated on a CMM, undergo ultrasonic inspection and many other steps. The industry will never allow sampling because of the high liability. For that reason, you see very little SPC in the aerospace industry. Now I have one customer that wants me to do a capability study (as you properly stated it’s a statistical prediction of your yield). But I'm thinking why do I need statistics when I have actual process date for every part produced? I already know exactly what my yield is so doing a study to determine a Cpk just does not seem to be worthwhile.

I'm I overlooking something?

I think the customer is asking for some extra confort factor.

The use of SPC to control manufacturing, when done effectively, should help to reduce the occations where product is made close to limit.

When I was in aerospace, 10+ years ago, after work performed for a major end user, we found that 100% inspection (by people) was proven to be at a risk. From memory, it was concluded 100% inspection would only find 95% of defects and to find 99.95% you need three independant inspection sequences

So to actively reduce the ammount slipping through, you need to actively reduced the defect quantity produced.

All of the talk about the utility of SPC misses the point here, imo. If your customer tenders a contract that includes the requirement to not only measure each part, but to have it blessed by a priest in a Pittsburgh Steelers uniform, and your company agrees to it, then that's what you have to do. It's a matter of contract review. If the customer is making unreasonable demands, don't agree to them, and if the customer tries to invoke new requirements that weren't figured into the original quote, then your company has the right to reopen the negotiations.

All of the talk about the utility of SPC misses the point here, imo. If your customer tenders a contract that includes the requirement to not only measure each part, but to have it blessed by a priest in a Pittsburgh Steelers uniform, and your company agrees to it, then that's what you have to do. It's a matter of contract review. If the customer is making unreasonable demands, don't agree to them, and if the customer tries to invoke new requirements that weren't figured into the original quote, then your company has the right to reopen the negotiations.

:applause:

Jim is correct. While cpk is supposed to show process performance and, in theory, a soundly performing process shouldn't need 100% inspection the customer can still ask for it. If 100% inspection is in the contract or in documentation supplied as an explanation of performance expectations, you're stuck with it unless you can convince them it's not needed.

Thanks to everyone who has reply. Many useful opinions but I think my original delima has been overlooked. I am contractually bound to 100% inspection and testing for the part we produce. My question is do I have grounds to dispute the need for conducting capability studies. I have data already for 100% as mentioned in one reply but for years we measure six parts at a time on a CMM all from six different tools, so the data is not stratified. However, do I have a legitimate dispute to say we don't need the capability study if we are still going to be required to do 100% inspection. I think the customer's quality department just wants SPC for the sake of saying thier supplier use SPC. It just seem pointless to me to do a capability study to predict a conformace to specification if we are still going to be required to actual perform 100% inspection to specification. Do I have a dispute?

Let me ask you a slightly different question. You already havce have data for the parts - would you think it unusual for the customer to ask you to provide the mean and the standard deviation of those numbers instead of (or perhaps in addition to) each of the individual measurements?

In a sense, that is all that is being asked when they want Cpk.
Cpk starts with the mean and standard deviation and further summarizes that info into a single number (based also on the spec limits). Cp-k doesn't provide any information that isn's available from the individual data points, but it is much easier for "the manager" to understand. It is just one other way to summarize the data.

Raw data: most information, hard to interpret
Histogram: intuitive, but tough to analyze quantitatively
% conforming: critical in your application, but throws away info about distribution of the data
mean & st. dev: A common 2 number summary that provide info about the middle and the spread of the data, but it throws out info about specific data points. Also, it says nothing about specs therefore says nothing directly about % conforming.
Cpk: a one number summary that includes info about the spec limits, so it gives a good estimate of % conforming. It also gives an estimate of the amount of variation in the dataAnd there are many other summaries that could be calculated - Cp, Ppk, ... Note that from the original data, the customer could generate all the other summaries themselves. Each summary provides a slightly different piece of info.

So, I think it is reasonable for the customer to want the Cpk in addition to the % conforming. They might reasonably want the originnal data or any of the other summaries. Of course, since this is a bit mroe work, it might be reasonable to charge a little extra for this service. ;)


Tim F

Thanks to everyone who has reply. Many useful opinions but I think my original delima has been overlooked. I am contractually bound to 100% inspection and testing for the part we produce. My question is do I have grounds to dispute the need for conducting capability studies. I have data already for 100% as mentioned in one reply but for years we measure six parts at a time on a CMM all from six different tools, so the data is not stratified. However, do I have a legitimate dispute to say we don't need the capability study if we are still going to be required to do 100% inspection. I think the customer's quality department just wants SPC for the sake of saying thier supplier use SPC. It just seem pointless to me to do a capability study to predict a conformace to specification if we are still going to be required to actual perform 100% inspection to specification. Do I have a dispute?

Lpitt,

The customer's concerns today are quality, cost and delivery. If you are making 100 products, inspecting all, rejecting / reworking them and supplying 2 OK pieces at his receiving inspection stage, he is definitely bothered. Getting no rejections at receiving inspection is no longer sufficient for him. If your manufacturing produces rejects, the cost is high, and there may be supply delays. He cant afford these.

It is a routine in automotive industry that the customer asks for a Cpk study and asks for 100% inspection till a value of >1.67 is achieved. So, both SPC and 100% inspection continue.

And anyway, you have at your disposal the means to inspect. Are there any issues in conducting an SPC study for the customer?

Is your Cpk lower than what he wants? - I think you can only benefit by improving on it.

Does he want it to be done too often for your comfort? - highly unlikely. Here, you may have a dispute. But then, you can always negotiate on the frequency aspect.

Shortage of staff? - deploy someone temporarily. It is most likely a one time effort.

But one serious advice - Avoid basing your study on historical data. 99.999999% chance that it is useless for reaching any meaningful conclusion.

Let me ask you a slightly different question. You already havce have data for the parts - would you think it unusual for the customer to ask you to provide the mean and the standard deviation of those numbers instead of (or perhaps in addition to) each of the individual measurements?

In a sense, that is all that is being asked when they want Cpk.
Cpk starts with the mean and standard deviation and further summarizes that info into a single number (based also on the spec limits). Cp-k doesn't provide any information that isn's available from the individual data points, but it is much easier for "the manager" to understand. It is just one other way to summarize the data.

Raw data: most information, hard to interpret
Histogram: intuitive, but tough to analyze quantitatively
% conforming: critical in your application, but throws away info about distribution of the data
mean & st. dev: A common 2 number summary that provide info about the middle and the spread of the data, but it throws out info about specific data points. Also, it says nothing about specs therefore says nothing directly about % conforming.
Cpk: a one number summary that includes info about the spec limits, so it gives a good estimate of % conforming. It also gives an estimate of the amount of variation in the dataAnd there are many other summaries that could be calculated - Cp, Ppk, ... Note that from the original data, the customer could generate all the other summaries themselves. Each summary provides a slightly different piece of info.

So, I think it is reasonable for the customer to want the Cpk in addition to the % conforming. They might reasonably want the originnal data or any of the other summaries. Of course, since this is a bit mroe work, it might be reasonable to charge a little extra for this service. ;)


Tim F

Wow - that may the most concise and clear explanation of SPC in layman's terms that I have ever seen! :applause:

I love the responses but your missing my point. I am a big proponent of using SPC to reduce variation and refine processes and have used it in many applications. I also preach the vertues of using every statistical tool in the tool box.

However, not every process warrant a formal study. In my case we provide one part at low volume to a customer with an internal scrap rate of less than one 1oth of 1%. We average 100% performance to master schedule and have not had an escape in over 3 years.

When we ask the customer the purpose of the requirement to do a formal capability study they simply reply they were requiring all their supplier to have some form of SPC. The reason I keep mentioning the 100% inspection is to illustrate they are not requiring this as a process improvement project or to reduce the time and expense of testing and inspection for a refined process. Because of the good performance metric they have no plans to improve tooling, change the process, material or anything. My customer's quality department has fell into the trap of forgetting what his mission is about. It should be about producing high quality product and services and not about producing pretty charts with no value.

SPC is a fantastic tool but it should have a define purpose. We currently stage six different tools on a CMM and do not stratify to data so for us to do the study correctly we will have to build new staging fixture and rewrite our CMM program to seperate the parts by tool. In other words it will cost us money and time and for what? So, back to my original question. How do I approach the customer with my concerns without insulting him. After all, I love my customer, he sends me money!!!

why don't you just ask him what the purpose of the two requirements - in tandem - are?
you can position your inquiry as wanting to know why both are required in order to fully mean his needs. this way your coming from a point of clarification witht the intent to satisfy your customer, not to dispute teh validity of his requirement.
at the same time you can ask if calcualting a Cpk from the 100% data is acceptable? if he says no - you'll know his requirement is illogical and that further inquiry would be useless.

Hey, that pretty good. Thanks!

I love the responses but your missing my point. I am a big proponent of using SPC to reduce variation and refine processes and have used it in many applications. I also preach the vertues of using every statistical tool in the tool box.

However, not every process warrant a formal study. In my case we provide one part at low volume to a customer with an internal scrap rate of less than one 1oth of 1%. We average 100% performance to master schedule and have not had an escape in over 3 years.

When we ask the customer the purpose of the requirement to do a formal capability study they simply reply they were requiring all their supplier to have some form of SPC.

Lpitt,

Does yor customer want you to do Cpk or some form of SPC? A one time Cpk study done one a batch of 100 pieces only for final dimension after 6 stage processing in a CNC machining centre on the same setting would still count as SPC. You only check whether the dimension is normally distributed and the Cpk is within specified limits. No repeats of the study unless any change in process takes place.

For smaller batches, you cant really reach a representative value even with 100% inspection.

But you know so many tools from the SPC toolbox. Why, even a pie chart is SPC! Use something applicable to the situation and sell it to the customer.

Win - Win.

Hi,
I'm in the aerospace line myself and have heard about a standard for sampling plans - AS9108. Just wondering if anyone involved in this discussion has thought about using this or has any experience with it? I haven't been able to get my hands on a copy, so I'm not sure if it's published yet. :confused:

If this is a component, then maybe the interfacing parts at your customer will shown some light. Sometimes stacked tolerancing come into place. This may ask for tighter control.

100% within specification and yet different characteristics as attached.