How Neuman Aluminium automates grain size analysis to ASTM E112
The challenges at Neuman Aluminium Strangpresswerk GmbH
In the materials laboratory of Neuman Aluminium Strangpresswerk GmbH in Lilienfeld/Marktl, it is determined how fine the grain is in cast aluminium bars and the parts made from them. The evaluation is to ASTM E112.
Until 2025 this was manual work. In the intercept method a test grid is laid over the micrograph, and every place where a grid line crosses a grain boundary has to be clicked individually. Johanna Höller of Research & Development describes it like this:
“With a very fine microstructure it takes correspondingly longer. It is also possible to miss one.”
Two things make this manual work laborious:
- The finer the grain, the more intersections there are to count. How much work a sample makes is decided by the microstructure.
- One analysis after another. Every intersection was counted individually by hand.
Why aluminium makes image evaluation particularly demanding
Because the grain in aluminium is not visible by itself. The sections are etched electrolytically to Barker. The anodic film that grows in the process is built up differently depending on grain orientation, and only under crossed polarisers do the interference colours arise from it by which grains can be told apart at all.
For an automated evaluation this is an unusual starting point: what becomes visible is not the grain boundary but a difference in orientation. And the same material looks completely different depending on whether it is in the as-cast or the heat-treated condition, and on whether a high-alloy or a low-alloy variant is being examined.
The solution
What interested the laboratory was one specific point: whether the laborious manual evaluation could be avoided.
Instead of one model that handles all these conditions half well, several were trained separately, each on the microstructures it actually gets to see. Three of them cover almost all of the conditions that come up in this analysis today.
The sequence itself has stayed the same: the sample is prepared and etched, photographed under the microscope, and the images go into MiViA together. The software outputs the result as a PDF, which is used for the documentation.
The difference lies in the word “together”. Several images run in one pass, and while they do, something else is being worked on in the laboratory.
What is left are the special cases. They are still evaluated by hand.
The results
The evaluation is still carried out to ASTM E112, and the results are reviewed by a specialist. What is gained is time, and precisely at the point where it was previously tied up: the counting.
Where that time goes is decided by the day: sample preparation, tensile tests, a report that needs writing. It does not flow back into one particular step, but into whatever is due.
For laboratories with a comparable task
Where the customer specifies the test method, the room for manoeuvre lies not in the result but in the way there. An evaluation that keeps to the standard and takes over the counting does not change what is measured in the end. It changes how much time metallography ties up in the laboratory.
Johanna Höller would recommend MiViA to a laboratory with a comparable task. In her view, the more analyses come in at once, the more efficient the evaluation with MiViA becomes.