SPC for Eddy Current Testing

#11
a. temel: so for clarity, you are using eddy current testing after you form the metal? if the eddy current tester indicates a crack the part is scrapped? and the reason a control chart (in your opinion) is not useful is that the material is from a constrained set of suppliers and (in your opinion) you have no leverage to improve their material quality? Also you are accepting that the eddy current testing is catching most - if not all - of the cracked material as your Customers have not complained about any cracked material escaping your facility to their facility or their customers - correct? (just as an aside, there is a way to perform an MSA on the eddy current testing, you are probably simply not aware of it). And can you confirm that your auditor asked specifically for a control chart or were they asking for other forms of statistical analysis (MSA, Capability, Yield trending, etc.) sometimes people conflate control charts and other forms of statistical analysis into one big lump of SPC....

Mike S: % of cracked material is attribute data and not variable data.
1) Eddy current machine is at the end of the line so yes.

2) If bars are separated,they can be reworked or sold as is to another customer which doesn't require crack controlled bar.So we have no scrap due to separated material.

3) Our suppliers are hot rolled steel manufacturers and there are not many of them.(And they are not happy when we found non conformances during supplier audits. They basicly say take it or leave ). We also cannot buy raw material abroad due to local money-dollar/euro parity.

4) We never had any complaints for controlled bars.Only some of the customers order non controlled bars and then complain about surface cracks (I guess this is a sign that machine catches non conformant ones)

5) Auditor said we should be using data gathered from crack control operation to improve the capability of the process (Prior to that we only used diameter of the bars for determining process capability)
 

Ninja

Looking for Reality
Trusted
#12
From what you've described above, and this:
5)Auditor said we should be using data gathered from crack control operation to improve the capability of the process
It sure does seem like something did not get communicated well. Whatever it is you choose to do, I would take care during the next interaction with the auditor to fix that communication. My two cents...
 

Mike S.

An Early 'Cover'
#13
My point is you can take attribute data (i.e. cracked or not cracked) and control chart the percentages, right? Let's say you have data for have 100 lots -- you have 100 numbers ranging from (using the OP's numbers) 5% to 50% cracked in time-ordered format. Why can you not control chart that, see if the variation is common cause or special cause, and go from there?

As for not being able to achieve an improvement..... Maybe you could find one or two or even three supplier(s) significantly better than the rest and buy more from the better supplier, or investigate what makes one better than another.
 
#14
It sounds like he's 100% inspecting the product with an end of line tester for cracks. So ... why would SPC needed? You don't need statistical control when you are doing 100% ACTUAL control.

Having said that, statistical data would be valuable in other ways:
1) You could use it to go after the suppliers (Why can't you give me less than 10% cracks when this guy over here does it all day long?)
2) You could use it to plan your capacity - knowing that a particular supplier has an average scrap rate, you can get better time estimates as to how long to run the job.

IATF DOES have sections stating you must be controlling your suppliers. Sounds like OP is caught between two large companies. Not fun.
 

Mike S.

An Early 'Cover'
#15
It sounds like he's 100% inspecting the product with an end of line tester for cracks. So ... why would SPC needed? You don't need statistical control when you are doing 100% ACTUAL control.
.
I disagree. I think SPC can still teach you about the nature of your process and maybe help you eventually eliminate the 100% inspection.
 
#16
I disagree. I think SPC can still teach you about the nature of your process and maybe help you eventually eliminate the 100% inspection.
Actually in this case,customer is paying us to do %100 control.Cause this bars are used in suspension and other important parts.
 
#17
I disagree. I think SPC can still teach you about the nature of your process and maybe help you eventually eliminate the 100% inspection.
If we assume he is RECORDING the 100% inspection results, how would this be the case? If I have full recorded data that I can analyze, how would taking pieces of the population be better?

Now, if he is NOT recording the data on some server and 100% inspection is just giving him a pass/fail light, I agree, because he wouldn't have trend data to analyze. Don't forget SPC is a statistical guess based on limited samples. That will never be as valuable as a full data set.

Arguably, you could say the goal being removing the 100% inspection. But I'm going to guess from the description he has a machine that has been bought and paid for. He can't return it at this point for money. Now, if it's a bottleneck and he can't make production, you could use the SPC to justify a skip check approach.
 

Mike S.

An Early 'Cover'
#18
If we assume he is RECORDING the 100% inspection results, how would this be the case? If I have full recorded data that I can analyze, how would taking pieces of the population be better?

Now, if he is NOT recording the data on some server and 100% inspection is just giving him a pass/fail light, I agree, because he wouldn't have trend data to analyze. Don't forget SPC is a statistical guess based on limited samples. That will never be as valuable as a full data set.

Arguably, you could say the goal being removing the 100% inspection. But I'm going to guess from the description he has a machine that has been bought and paid for. He can't return it at this point for money. Now, if it's a bottleneck and he can't make production, you could use the SPC to justify a skip check approach.
Perhaps I was being unclear.

Let's say I had data on 100% of the parts in 100% of the lots. 50 parts per lot, 30 lots. Lot number 1 had 5% cracks, lot #2 had 9% cracks.....lot #30 had 22% cracks. If I plotted this data on a I-MR chart could I not possibly learn something about my process? Is the process in control/predictable? Does it show common cause or special cause variation? Is there something time related (i.e. in the summer months lots are worse)? Any identifiable trends? Etc.
 

Bev D

Heretical Statistician
Staff member
Super Moderator
#19
these are excellent discussion points and some of the misunderstandings and myths of SPC are coming through.

first this type of 100% test is absolutely necessary for automotive use as described by the OP. a crack in these parts can get people killed. Ideally, yes, I would want the material supplier to guarantee a 100% crack free material. given some of the very real limitations of this in the industry we are far from the ideal state.

One can of course apply a control chart to the run data of the 100% inspection (SPC doesn't have to be on samples on only) and gain great information about the patterns in the material quality as well as an understanding of the performance of the various suppliers. the 100% inspection is an enumerative screen and the application of 'control chart' analysis is a perfect application for this analytic type study. This information could be invaluable to procurement and to supplier quality improvements

The next critical question is if there is leverage to actually use that information for those purposes. If the situation is as the OP has stated there appears to be very little leverage to improve the supplier quality. It sounds like the OP may be a small company in between a couple of very large suppliers and a large Customer. his ability to drive improvements at his suppliers may be limited.
 

Ninja

Looking for Reality
Trusted
#20
People who know a lot more than me posting above...so I have only this logic:

Problem is, ratio of material seperated during crack control operation is highly dependant on the raw material (%5 to %50 percent seperated). Since we buy our steel from companies that we cannot change due to many reasons, data gathered from the process cannot be used to evaluate suppliers
You'e analyzed enough to know that rawmat and rawmat suppliers have a direct affect on output. Have you shown that this is the ONLY cause?

We never had any complaints for controlled bars.
So it's a pure cost/benefit...as posted above, it seems like the auditor did not understand this.

customer is paying us to do %100 control
So 100% control is a given, and the idea of sampling is off the table.

If you have two suppliers that vary from 5% (assumed from vendor 1) and 50% (assumed from vendor 2), I suppose you mainly buy from Vendor 1.
If this assumption is correct, you can plot all of Vendor 1 material separate from Vendor 2 material and see if there is something in your process that can take 50% down to 40%, or 5% down to 3%.
If you already have the data...its just a matter of analyzing it...a pretty cheap enterprise. Typically its the data gathering that costs the most.

5% to 3% might not be worth pursuing, but I'd guess 50% to 40% likely is. But again, at that point it's all cost benefit.
The suggestions above on SPC or at least data mining can help you understand if there is a potential benefit...then you'll have to run it against the cost to see if it's worth doing.

HTH
 

Top Bottom