# Ultrasound imaging of critical metal parts

#### Wired452

Joined Apr 6, 2021
10
Hi people.
I am trying to locate any information at all to use ultrasound to get 3D imaging of small mechanical devices (4 to 8 CM cube)
the purpose to detect flaws, metal fatigue, and functional operation in critical assemblies. Like a 3D snapshot resembling X-ray.
Historical applications were like detecting cracks, wear, and fatigue in large pins supporting Ferris wheels.
The devices are not very complicated, 16 moving internal parts. ( compare to a tiny Wankle rotary engine)
NO, the device cannot be dismantled for individual part analisis. A failure of the device would destroy a very expensive project.
TYIA

#### MisterBill2

Joined Jan 23, 2018
13,674
Years ago I deigned a machine for production checking of brake drums for cracks and flaws. That machine used a commercially offered ultrasonic inspection system that would probably do just what you seek. So the systems do exist, or at least did exist. I can offer one caution, which is that to calibrate the test analysis as to pass and fail you will need a known-good part to establish a good-part profile. With brake drums that was easy, all it took was some X-ray inspected parts.

#### Wired452

Joined Apr 6, 2021
10
These systems do exist, I have seen them used.
What of these mechanical devices to test are in working order or the failure would wipe out the entire sub assembly.
The idea is to spot potential failure before it happens. This is part of a 4 million dollar machine, an expensive loss.

#### MisterBill2

Joined Jan 23, 2018
13,674
OK, and it seems like you could use a portable model.. That project was done several jobs ago and we just built the machine to present the parts to the tester. Unfortunately I have no additional information about them

#### Wired452

Joined Apr 6, 2021
10
No reason this could not be portable and with today's electronics come in under a couple pounds.
A transducer could be square or round, 2 CM by 8 CM. a meter of cable into a front end/DSP hardly more than a smart phone.
The 3D image processing could be done on a powerful tablet or small laptop.
No law says this has to be ultrasound, I have been looking into reflex magnetometers.
Looking at ultrasound though, that seems the best way to go.
What I am lacking here is information for transducers and the needed signal processing hardware.
Thanks.

#### scorbin1

Joined Dec 24, 2019
103
We use a portable untrasound device to non-intrusively inspect spot welds in an automotive manufacturing environment. Its a somewhat bulky handheld instrument, about 2-3 times the size of a typical multi-meter. I don't have a lot of data on it, but I know that even though it's a portable device, it's strapped down to it's own dedicated $500 cart and one day ALL production was stopped to make room for a meeting to instruct ALL associates to make room and stay clear of it when it passes down the aisle, lol. I have a feeling that was taken a bit far, but that's just my perspective. 15 minutes of plant-wide downtime is extremely expensive so I am guessing that it is as well!! #### MisterBill2 Joined Jan 23, 2018 13,674 We use a portable untrasound device to non-intrusively inspect spot welds in an automotive manufacturing environment. Its a somewhat bulky handheld instrument, about 2-3 times the size of a typical multi-meter. I don't have a lot of data on it, but I know that even though it's a portable device, it's strapped down to it's own dedicated$500 cart and one day ALL production was stopped to make room for a meeting to instruct ALL associates to make room and stay clear of it when it passes down the aisle, lol. I have a feeling that was taken a bit far, but that's just my perspective. 15 minutes of plant-wide downtime is extremely expensive so I am guessing that it is as well!!
I can vouch for Scorbin, that at least when we built the machine those devices were expensive! There might possibly be a scheme to use one of the medical devices to search for cracks and flaws, but there is probably a big difference in the amount of power needed to send a signal through six inches of steel instead of six inches of human. But a very good starting point would be to contact a maker of the medical ultrasound systems and present the requirement to them. It may be that something is available now.
And still there is the issue of being able to understand whatever results come from an examination with ultrasound. It may be like looking at a boat's radar screen and take some insight to understand what is being seen. So even after finally obtaining a suitable ultrasound system there will still need to be a lot of learning done.

#### ericgibbs

Joined Jan 29, 2010
16,741

#### MisterBill2

Joined Jan 23, 2018
13,674
It seems that the challenge here is doing the examination without taking the thing apart. Any sort of X-ray equivalent demands access to send radiation through, and also, mostly, protecting everybody from that radiation.
Acoustic signature analysis might possibly be useful, listening for any change in the sound of the machine, but that analysis does not detect any changes until they happen. So there is a problem with that.

#### ericgibbs

Joined Jan 29, 2010
16,741
but that analysis does not detect any changes until they happen. So there is a problem with that.
Hi Bill
It detects changes before a critical failure of a mechanism

Surely that applies to all methods of detection of a potential failure, isn't that what the TS requires.?

So how does that constitute a problem.?

E

#### MisterBill2

Joined Jan 23, 2018
13,674
My admittedly limited experience with ultrasonic inspection was that while it was able to detect actual cracks it did not show the flexing strain effects prior to crack formation. (Also known as work-hardening)

#### scorbin1

Joined Dec 24, 2019
103
I think what MisterBill2 is trying to say is that it's not a simple measurement. You are not likely to be able to run a single test on a sinlge part and determine if there are any flaws unless you have a PhD in metallurgy and know exactly what all of the potential variables are. Even then that's a big ask. Our procedure for a given part is generally as follows:

1. Perform ultrasound measurements on a decent sized lot of samples(usually at least 20-50), some assumed good, some known bad.
2. immediately there after we do a destruct analysis on those same samples. Note this requires that you destroy the samples. Our destruct process generally looks like this:
1. cut out a cross section of the area under test
2. analyze visually with a microscope or other 'known good' test method to verify parts are indeed in spec
3. Once the destruct analysis is complete we compare the consistency of the results between the two tests
4. If we have consistent results, ie. good and bad parts give consistent results respectively, we take the pictures of the ultrasound curves for known good and bad parts and that curve gets approved to be used as a 'guage'
5. At that point, we can take measurements of further parts and simply compare the curves to the known good and bad curves.

Without this process, the data it will return will be arbitrary at best. For us it's really only feasible for mass production parts as the cost of setup is quite considerable in terms of sacrificial samples.

With this method, ultrasound can absolutely indicate even minute amounts of undesired stress in the material long before cracks form. However first you have to have known good, and known bad baseline samples to compare that to. You can watch for deviations from known good samples to get an idea that there MAY be a problem, but you have to have experienced and measured the defect repeatedly before you can confidently conclude that it is said defect.

Then efficiency comes into question and depends heavily on a few things. How controlled is your test environment? There are a lot of potential variables that can affect the readings, such as probe placement, probe orientation, the small inherent movement of even the most steady hand, quality of probe, quality of the meter, features of the meter, etc. Most commercial equipment employ algorithms that can clean up the effects of the aforementioned variables but that's not something you're likely to have in a home-brew setup. These variables can have even the best lab tech chasing their tail for hours or even days if not controlled.

It is absolutely possible, it's just a matter of cost vs benefit. How far are you willing to go to get these measurements, and how accurate do you expect/require these measurements to be.