- Joined
- Jan 4, 2021
- Messages
- 2,014
That was my point: unless the material itself or the original packaging is marked, you can't tell what it is by appearance.Yes, very true, but many other tool steels are also wrapped in paper, it proves nothing.
That was my point: unless the material itself or the original packaging is marked, you can't tell what it is by appearance.Yes, very true, but many other tool steels are also wrapped in paper, it proves nothing.
I agree, not that my agreement means much, but in our industry all materials for critical uses require certs which list what it is, where it came from the batch number, etc... and even then we send in coupons to test to ensure it will meet requirements for strength and cycle test. This would not be required if a) there was an easy way to determine this information and B) such methodology were not cost prohibitive.Fifty three years ago when I took freshman chemistry, the way to identify materials was done through qualitative chemical analysis. This meant dissolving the sample in an appropriate solvent, usually something like aqua regia, and adding various chemical solutions to precipitate out product or create a distinctive color change. Typical metals, besides iron, alloyed in steels are manganese, chromium, nickel, cobalt, molybdenum, and a smattering of others. The definitive work on the subject was "Introduction to Semimicro Qualitative Analysis" by C. H. Sorum.
Beyond that, another analytical method was emission spectroscopy which would yield semi-quantitave results. It required a very expensive and huge spectrograph which created a plasma from the sample with an electric arc and analyzed the emitted spectrum via a photographic plate. Atomic absorption spectroscopy followed and was able to provide accurate compositions of metals. And the x-ray fluorescence spectroscopy mentioned by @graham-xrf is the latest in the arsenal.
The bottom line is that it is difficult, if not impossible, for a lay person to accurately identify a steel alloy. I realize this first hand as I have several tons of unknown steel alloys.
I have found that spark properties vary with the grit of the grinding wheel and the amount of pressure exerted so I'm not sure that AI would be able to help much. The best bet would be to build a library of known alloys and compare the unknown.Keeping in mind the practical realities of any test involving getting stuff taken apart enough (i.e. chemistry) or hot enough force it's atoms to expose its contents (plasma), as explained by @RJSakowski, I am not sure there is that much mileage in the colours in the spark shower. Sparks are, after all, little bits of the alloy made temporarily kinda hot. It's the length of the sparks, from the sizes of the chunks that fly off, giving them relative longevity.
That said, a kooky notion might be to get some images, or video, and submit to one of those AI engines that has trawled billions of images, including from anyone who has ever done tool grinding on YT. This would be the longest of long shots, and I'm almost ashamed to suggest it, but it may do better than me invoking radioactive attempts to get a rise out of a chunk of unknownium.
What I want is some trick app where I can just point the "smartphone" camera at the view of the sparks for a couple of seconds, and have it list what steel it thinks they are best looking like.I read an interesting paper that studied the difference between 4140 and 1045 sparks. It quoted people saying that an idiot could figure out the difference in 5 minutes. So the researchers grabbed a few people off the street and gave them 5 minutes of training. It turned out that they did pretty well, although not perfectly.