Minor Nonconformity - Ruler Attribute Study

[anneliesehuss]

"We're going to hold onto this information because our control plans are going to be completely rewritten. We're moving from Excel to APIS, so we may or may not even include both of these in the new plans"

I would be cautious about excluding gauges that are required for the process from the control plan because that could also result in a finding. If the idea is to stop using the gauges altogether, then by all means, remove them.

 

[Bev D]

"A ruler, while it can provide continuous (variable) data, is not very precise...."
I'd argue that a ruler doesn't provide any data, it is an aid to the Mark 1 eyeball providing data. Unlike say a vernier where you can read off a measurement made by making specific contact with an item in 2 places, rulers are usually held against something and the user makes a balanced decision on how the features between which the measurement is being made align with the markings on the ruler - and those markings are fairly generous in size, so where are you "estimating" the measurement to be from - the middle of the marking or one side?

Semantics mostly. Anything that has a scale that the assessor must ‘read’ has the same problems. They all have angle of vision, positioning, size of pointer vs scale line, and interpolations etc. etc. the analog device doesn’t ‘send’ a data result like a digital device. But both types of devices can be repeatable and accurate enough or not depending on the actual characteristic being measured. Which is why we perform calibration and Guage R&R and other MSA studies.

And we don’t know if the ruler is a wooden one purchased at a big box home store or a calibrated fine increment metal ruler - like those on a ‘precise’ analog vernier…

I once solved a chronic problem with one of those metal pocket rulers you got at trade shows, and there was a dab of rubber cement on the end to avoid scratching the paint on the car. Another time my fingernail was a better measurement device than an expensive digital surface finish device. It all depends.

We would need a lot more information to determine if an attribute study or a continuous data study would be appropriate.

 

[Mordacious Enigma]

Oof, lots to unpack here, and I downloaded that MSA excel yesterday before making an account. Pretty comprehensive (if not a bit confusing on learning how to use). As previously stated, our control plan may or may not even have the rulers in the future. With that being said, our plan of action right now is to give some sort of data to the auditor. She wanted an attribute study but upon discussing with you all it seems more appropriate to do a variance study, then now I have to think about doing Kappa. With that being said, we're still "chewing" on what form this data will take in the end.

I can't speak for anyone else in my company but I do not do things to "check boxes", especially when audits are involved. I address them, learn from them, and think preventative. As far as the rulers themselves, they are six inch steel rulers from General brand. They are calibrated against a calibrated tape measure, which are calibrated by an externally calibrated master ruler. Bleh, that was not a fun sentence.

"We would need a lot more information to determine if an attribute study or a continuous data study would be appropriate."

I'm an open book. The material she was looking at has a margin of error of +/-2.0mm, and +/-1.0mm further down the line. There are a total of five dimensional values with their corresponding tolerances throughout the entire material process. The sewing where the rulers come into play later on are where the margins get extremely tight. Due to the material's shape, using something digital like a caliper is a bit of an issue. I know rulers of any kind are kind of an interesting subject in of itself but we're kind of stuck with what we got.

 

[Bev D]

Sorry about the spreadsheet - it’s not easy to make an EXCEL spreadsheet user friendly and obvious especially if you are new to the statistical technique. It was developed for an internal 4 week class before we had JMP for everyone. (Or cell phones or laptops…). It’s going on 20 years old now and frankly I’m surprised it still works given the changes EXCEL has made. But since stats software is going greedy to expensive subscription models maybe EXCEL will make a come back.

So you are dealing with fabric and sewing? You are correct that a caliper or hard fixturing is probably not appropriate. My first thought now is that a continuous data (variables) study is appropriate. But I would use 25-30 pieces, 2 operators and 2 measurements per operator/piece and not the ‘standard’ 10X3X3. You can always go back and augment readings and operators later once you get a passing result. But a 30X2X2 will provide a lot more insight for your first look. Also I would plot it on a Youden plot (it’s in the spreadsheet..) as this too will provide more insight than the statistical output stuff.,..good luck. You can always post your results here for us to help you with the interpretation…

 

[Mordacious Enigma]

Sorry about the spreadsheet - it’s not easy to make an EXCEL spreadsheet user friendly and obvious especially if you are new to the statistical technique. It was developed for an internal 4 week class before we had JMP for everyone. (Or cell phones or laptops…). It’s going on 20 years old now and frankly I’m surprised it still works given the changes EXCEL has made. But since stats software is going greedy to expensive subscription models maybe EXCEL will make a come back.

So you are dealing with fabric and sewing? You are correct that a caliper or hard fixturing is probably not appropriate. My first thought now is that a continuous data (variables) study is appropriate. But I would use 25-30 pieces, 2 operators and 2 measurements per operator/piece and not the ‘standard’ 10X3X3. You can always go back and augment readings and operators later once you get a passing result. But a 30X2X2 will provide a lot more insight for your first look. Also I would plot it on a Youden plot (it’s in the spreadsheet..) as this too will provide more insight than the statistical output stuff.,..good luck. You can always post your results here for us to help you with the interpretation…

Yeah, and it doesn't help that Excel was never my strong suit. I can whip up a CAPA in Trackwise and close it out in a flash, but give me cells to work with and I'm deer in the headlights. This is also my first experience in the automotive industry (I only started three months ago), so my past experiences in other industries only get me so far.

There's other materials such as metal that go into the whole thing, but it is largely sewing/fabric in nature. I agree with the amount of participants and sample amount. Unfortunately, due to my inexperience with excels and MSA studies (looking at this excel really puts things in perspective), I am kind of stuck and my boss is also stuck, lol. I guess that's why we got hit during the audit. The fact that we never did one when our records go back to 2015 was a miracle.

So, with that being said. If anyone wants to sit down and walk me through the excel, that would be greatly appreciated. Attribute study I got but this has a few more nuances to it. I can get the parts (possibly next week) and people to do the physical aspect but this is the last hurdle before I can begin closing out the audit. And not to assume anyone would be willing to help me but if someone were and wants a relative data to what I would be doing then it would be:

Target - 3mm
Tolerance Range +/-1.0mm

Unfortunately, the part that got us dinged is not in production at the moment, so I had to reference something else. It also needs a variables study at some point, but I need to coordinate with a few people that have other stuff going on at the moment. I'll still keep poking around in this excel to see if I can figure some stuff out.

 

[Golfman25]

Ah the old ruler. Has there ever been any "gage" that has caused more issues during audits?

So as I see it, the ruler with something like fabric is no way precise enough to have statistically "good" measurement. But is there anything better? Probably not, given the cost. I would do both a variable study and an attribute study. See which one works best and which one more reflects what your actually doing in production. My guess it the attribute study will give you "better" results. Let us know. Good luck.

 

[Tidge]

So as I see it, the ruler with something like fabric is no way precise enough to have statistically "good" measurement. But is there anything better? Probably not, given the cost. I would do both a variable study and an attribute study. See which one works best and which one more reflects what your actually doing in production. My guess it the attribute study will give you "better" results. Let us know. Good luck.

The Scope of ASTM F2203 (from an old copy I have lying around):

This test method covers the measurement of linear dimension of flexible packages and packaging materials. It is recommended for use with an allowable tolerance range of 3 mm (1⁄8 in.) or greater based on gage repeatability and reproducibility presented in the Precision and Bias section.

So perhaps not quite to the 1mm tolerance, but the method is there.

 

[Bev D]

An Attribute analysis can be done after taking the continuous data measurements IF there are at least 25 samples AND there are several out of tolerance parts (>5) These will probably be marginal. Teh assessors just need to record whether they would pass or fail the part along with the continuous data measurement.

 

[ChrisM]

OK, taking a step back and having a fresh look at this..... why use a ruler at all if it's fabric you are measuring? Why not have a gage made up - say a flat piece of petal with some pegs in it.... hold one end of the material against one peg, so long as the other end lies between 2 further pegs, it's of acceptable length (a sort of go/no-go gage) ? Get the gage calibrated (measured out accurately against dims and tolerances) and that should surely help, and probably be quicker too than individually "measuring" each piece

 

[Bev D]

They should perform the Guage R&R then decide based on the data (what a concept eh?) if they need a different system. +/- 1mm is a pretty tight tolerance for fabric…

As for the OP and their supervisor being ‘stuck’…this happens a lot when encountering something new that is not wel understood. The only real solution is to just dive in, ask questions, study and practice, practice practice.