Let's say for a steel bar, you have the frequency information on each spatial location along the steel bar, could you make a decision that a defect/crack is exist on the specific location without having to inspect the steel bar first?

  • Are you refering to some kind of ultrasonic inspection technique? If so, can you add more detail on exact what kind of test you are referring to? – Daniel Kiracofe Oct 10 at 12:21
  • @DanielKiracofe Hey thanks for asking. Actually I'm referring to simple and dense accelerometers that were attached to the steel bar (It might sounds a little unpractical, but I'm trying to see how my experiment goes). From the information, I wonder could I determine the defect location from it's frequency spectrum. – Gregor Isack Oct 10 at 12:39
  • @DanielKiracofe I'm trying to comparing all the frequencies of each accelerometer at a specific time and wonder would that give me some useful information in detecting the defect. – Gregor Isack Oct 10 at 12:48

The answer is yes. Having information on natural frequencies and mode shapes, such as you could gather from a series of accelerometers, could allow you to detect defects. However the resolution might not be all that great. I.e. you may be able to detect a large defect but miss a small defect. Depends on what size defect is considered acceptable. If you need the best resolution, there are probably better methods such as ultrasonic. Also mounting a lot accelerometers will be time consuming. Ultrasonic would be much quicker.

  • Is there any identifiable frequency behavior that exhibit by the defect/crack that allows one to categorize it as a defect? All the frequencies data obtained are quite random to me and I'm not sure where to begin. If you don't mind me asking. Thanks. – Gregor Isack Oct 11 at 0:17
  • Probably not. In general what you would do is to test a good beam that is known to not have any cracks, and determine the "signature" of the part. Or better yet, test a large sample of many beams to determine an average signature. e.g. average nat frequency is 100 Hz and std dev is 1 hz. Then if you find a part with nat freq of 99 it's probably good and if nat freq is 80 it's probably bad. Without any known good part to compare to, it will be pretty hard to tell if there is a defect or not using this method. – Daniel Kiracofe Oct 11 at 0:58
  • this is why I said resolution with this method is not that great. with something like ultrasound, you'll be able to "see" a crack (more or less). With this method, you are much more limited. – Daniel Kiracofe Oct 11 at 1:09

Your Answer


By clicking "Post Your Answer", you acknowledge that you have read our updated terms of service, privacy policy and cookie policy, and that your continued use of the website is subject to these policies.

Not the answer you're looking for? Browse other questions tagged or ask your own question.