Determine the distribution of a porous material

[ChocolateTerrain]

SO, basically you are ‘playing around’?
I’m sorry but you must understand that these types of analyses are complicated. Quality engineering itself is worhty of a 4 year degree and probably and advanced degree. It is not a look it up on the interwebs (or even CHATGPT) cut and paste, push buttons in software package. That is of course why this forum exists but it is free and staffed by very skilled experienced professionals. If you have questions we can help cut thru the BS and give really helpful answers.

That said: yes you need to use the individual values and not the averages. In most cases the average just artificially attenuates the real variation that Customers care about. BUT you really need to look at the variation within a pice compared to the variaiton of the seprate pieces. Use histograms or a ‘variability type chart. Dont’ use boxplots..

You could calculate the overall SD from the individual values you already have and then use the standard sample size formula. Look it up on the inter webs. Its simple. Now you may have a complex situation. For example if all of your samples are from a single ‘explosion’ event you don’t have truly representative samples. Also any non-homogeneity would require a very specific sample plan and since the non-homogeneity can take several forms it is too complicated to describe here. But the individual piece histograms of the individual readings and overall SD calculation should be enough for your purposes.

Well, I'd not really say that I'm "playing around" this is a continuon of my bachelor thesis I wrote for that company but then it was more about establishing methods just to measure.

Unfortunately the company don't have any expertise on what I'm trying to do there, they know that it works but as I've come to realise they don't always know why. This is why I'm trying to do the more "advanced stuff" since they have no guidance or opinions on what to do per say. Just that they want to know what I've described and I'm trying diffirnet things in order to fulfill that goal.

I did not mean that I take this lightly, that was not my intention, I guess I just try to write in a "light hearted way". I've tryed to stay off AI with the analysis part since I want to know what I'm doing. And hence I've turned to this forum. What "BS" are you meaning I should avoid?

Thank you for the input, I'll think about it.

 

[Bev D]

Well that makes much more sense and helps us help you.

It is always better to be open, detailed and specific, usually hyper focused questions on something like a specific stats test indicate the poster has a much bigger question behind the posted one - even if they don’t know it. (The old “you don’t know what you don’t know” problem). No offense

Have you performed an MSA?

Come back with your histograms as described above with the data you have and I think we can really help you.

 

[ChocolateTerrain]

Hi!
No, I've not preformed a MSA (Measurement System Analysis ?), this was roughly what we stated to do in the bachelor thesis, trying to focus on the process (the test method) of the measurements. Although to be honest, we more tried things rather than evaluated them and only one method where used (we were also trying measuring air speed at different heights and widths from the air inlet). And now I'm also trying to derive a Control Plan if i understand it correctly.

I've made a variability type chart (2 variants, don't know if the "scatter one" is better, just looks messy in my opinion):

https://imgur.com/a/1qCisTp

(I don't know if you want for both the material sizes, but let me know if you do!) Since I think I posted the histogram before, or did you mean for each material measured? (I name them 1B.12, 2B.12 etc. 1B.14, 2B.14 etc.)

 

[Bev D]

First charts aren’t supposed to be pretty or neat. The scatter of individual points is what you want this far more informative. Remember analysis about insight not prettiness. And the average and standard deviation plots are all but useless. In order to understand your mind has to draw the ‘messy’ chart in your head which it cannot do.

Yes you need to perform an MSA first.

 

[Miner]

@ChocolateTerrain Now that you have more than 5 posts in, please attach your files. Many companies block file share sites, as does mine, so I cannot see the files nor assist.

 

[ChocolateTerrain]

Hi again!
Alright, I went and checked MSA and if I understand it correctly a "Crossed Gage R&R" would be the best fitting for our purposes.
I'll aslo repost the histrogram and the variability type charts above:

I also would like to read up a bit on quality management, I found this book from a Course syllabus for a Quality Management: "Fundamentals of Quality Control and Improvement. John Wiley & Sons Inc." Would that be any good? Or any other recommendations?

 

[Tidge]

It looks like the composition of the B.12 and C.12 materials is such that the 'explosion' (to make the porous material) is resulting in a wildly uneven distribution. I wouldn't be surprised if there was just as much variation in the weights of the samples tested.

 

[Bev D]

Hi again!
Alright, I went and checked MSA and if I understand it correctly a "Crossed Gage R&R" would be the best fitting for our purposes.
I'll aslo repost the histrogram and the variability type charts above:

View attachment 31316
View attachment 31317


I also would like to read up a bit on quality management, I found this book from a Course syllabus for a Quality Management: "Fundamentals of Quality Control and Improvement. John Wiley & Sons Inc." Would that be any good? Or any other recommendations?

Better to look at the references in my resource “Essential References for Quality Engineering”. That book is basically pablum

Also reposting charts we’ve already seen and basically dismissed as not sufficient isn’t all that helpful for you…

 

[Miner]

Also reposting charts we’ve already seen and basically dismissed as not sufficient isn’t all that helpful for you…

That was at my request, since file-sharing sites are blocked at my company.

 

[ChocolateTerrain]

It looks like the composition of the B.12 and C.12 materials is such that the 'explosion' (to make the porous material) is resulting in a wildly uneven distribution. I wouldn't be surprised if there was just as much variation in the weights of the samples tested.

Yeah, I would assume the same. Between batches there is a big variance, within batches though it is quite small. It is when we compare on a batches to batches we get theiese big variances.