where expertise comes together - since 1996

Web's Largest Portal of Nondestructive Testing (NDT)
Open Access Database (Conference Proceedings, Articles, News), Exhibition, Forum, Network

All Forum Boards
Technical Discussions >
Resonance testing
Career Discussions
Job Offers
Job Seeks
Classified Ads
About NDT.net
Articles & News

1682 views
08:02 Dec-27-2003

S S Ananthan

Consultant,
RAA Tech Solutions,
India,
Joined Jun 2000
36
Resonance testing

HEllo,

I understand that there is a method called resonance testing which uses the natural frequency of a job to identfy whether it has any defect by comapring the natural frfequency of a defect free job.

Is it used in detecting defects in welding ? Any specific experience ? Can the method be used for detecting stiffness of structures ?

S S Ananthan


 
05:58 Dec-30-2003

Godfrey Hands

Engineering,
PRI Nadcap,
United Kingdom,
Joined Nov 1998
284
Re: Resonance testing ----------- Start Original Message -----------
: HEllo,
: I understand that there is a method called resonance testing which uses the natural frequency of a job to identfy whether it has any defect by comapring the natural frfequency of a defect free job.
: Is it used in detecting defects in welding ? Any specific experience ? Can the method be used for detecting stiffness of structures ?
: S S Ananthan
------------ End Original Message ------------

Dear S.S. Ananthan,
You are right. Resonance testing is widely used in inspecting components for defects, and the principal is to compare the resonances from good pieces to those that you are inspecting.
When you have a defective part, then some of the resonances will change.
However, many other factors change the resonances by an amount equal to the change from a small defect, such as temperature, dimensions, heat treatment etc. and this limits the sensitivity in many types of simple resonance testing.

Quasar International have developed a technique that compensates for these acceptable variations and increases the sensitivity of the test to a practical one again.

This can be economically applied in mass produced components to detect defects at production rates. However, component weights within the range 20 grams to 20 Kg are normally the limits that we have experiences with.
Within welded parts, we have quite some experience on electron beam welded joints between shafts and gears for the automotive industry.

Please contact me with more details of your application, and I will provide you with some information specific to your application.

You can also look on the Quasar International web site at www.Quasarintl.com

Regards,

Godfrey Hands



 
05:06 Jan-05-2004

Wolfgang Bisle

R & D, Project leader & Group Leader InService Inspect.
Airbus Deutschland GmbH,
Germany,
Joined Jul 2000
35
Re: Resonance testing Sometimes I feel challenged by such a question and such an answer...
Resonance Testing is a very wide field. Of coarse the way Quasar Int. is workins is one little facette in this big field.
Other are for instance equipments like Fokker Bond Tester (used for Bond Inspection of Aircraft) as well as the Stavely Bondmaster, the Acoustic Flaw Detector (AFD), the Zetek Sondicator, and other instruemnts like Bondascope, Laminar 2000, Bondchecker etc.

All these instruments work with resonant excitation of structures and measure the frequency spectrum or the acoustic impedance at a certain frequency, which is selected due to the acoustic/resonance properties of the part to be inspected. Typically these techniques are preferably used to inspect bonded/disbonded structures.

Today new ways of resonance testing are.... or let me say methods with vibration loading are evaluated. Nicely - another splendit feature - the pick up of the signal can be done optically so that you might have a full non contact technique.
Best would be if you could use an aircoupled excitaion of vibration, something like a loudspeaker, an air gun, a sparksource, or for instance you can use a CO2-Laser like in LaserUS, or if non contact is impossible a shaker, or a piezo, maybe with vacuum suction cups fixed, can be used.

It should be necessary, that you are able to tune the excitation in a certain frequency range, sometimes only some % of your mid frequency, in other cases you need some decades. Deopends on application.

With a vibrometer (a special interferometer) you can pickup the surface motions. But you can even use a shearography optical head with an adapted trigger scheme for the image processing to get complete vibration pattern of an area.

Finally for crack inspection you may also use a thermocamera to pick up the heat generated by friction in the tips of the cracks. (o.k. here resonance is not a must)

The soundgun / vibrometer approach for instance was realized by a group of researchers and companies, now more or less advetized by Honeywell under the acronym SAM (structural anomaly mapping), RAID (remote acoustic impact doppler). As this method needs similar spectroscopy tools like Quasars systems the circle is closed here.

But I know there are a lot of other resonance based methods available.

For instance all around things companies like Bruel & Kjaer are doing.

Or think about very simple approaches: The coin or tap test. Mechaninzed the best solution available still is the Mitsui Woodpecker.

Be sure I know a lot about it, but definitely it is only a small part of the hole field of resonance testing. Be shure there is a whole world of applications and instruments. So to get more precise information you should specify your field of interest more precisely. Otherwise you will get only small slices of knowledge presented as salesmen mostly don't know much more then just their instrument....and tell you thats good for everything.





 
08:40 Jan-05-2004
M. Kim Johnson
Re: Resonance testing ----------- Start Original Message -----------
: Sometimes I feel challenged by such a question and such an answer...
: Resonance Testing is a very wide field. Of coarse the way Quasar Int. is workins is one little facette in this big field.
: Other are for instance equipments like Fokker Bond Tester (used for Bond Inspection of Aircraft) as well as the Stavely Bondmaster, the Acoustic Flaw Detector (AFD), the Zetek Sondicator, and other instruemnts like Bondascope, Laminar 2000, Bondchecker etc.
: All these instruments work with resonant excitation of structures and measure the frequency spectrum or the acoustic impedance at a certain frequency, which is selected due to the acoustic/resonance properties of the part to be inspected. Typically these techniques are preferably used to inspect bonded/disbonded structures.
: Today new ways of resonance testing are.... or let me say methods with vibration loading are evaluated. Nicely - another splendit feature - the pick up of the signal can be done optically so that you might have a full non contact technique.
: Best would be if you could use an aircoupled excitaion of vibration, something like a loudspeaker, an air gun, a sparksource, or for instance you can use a CO2-Laser like in LaserUS, or if non contact is impossible a shaker, or a piezo, maybe with vacuum suction cups fixed, can be used.
: It should be necessary, that you are able to tune the excitation in a certain frequency range, sometimes only some % of your mid frequency, in other cases you need some decades. Deopends on application.
: With a vibrometer (a special interferometer) you can pickup the surface motions. But you can even use a shearography optical head with an adapted trigger scheme for the image processing to get complete vibration pattern of an area.
: Finally for crack inspection you may also use a thermocamera to pick up the heat generated by friction in the tips of the cracks. (o.k. here resonance is not a must)
: The soundgun / vibrometer approach for instance was realized by a group of researchers and companies, now more or less advetized by Honeywell under the acronym SAM (structural anomaly mapping), RAID (remote acoustic impact doppler). As this method needs similar spectroscopy tools like Quasars systems the circle is closed here.
: But I know there are a lot of other resonance based methods available.
: For instance all around things companies like Bruel & Kjaer are doing.
: Or think about very simple approaches: The coin or tap test. Mechaninzed the best solution available still is the Mitsui Woodpecker.
: Be sure I know a lot about it, but definitely it is only a small part of the hole field of resonance testing. Be shure there is a whole world of applications and instruments. So to get more precise information you should specify your field of interest more precisely. Otherwise you will get only small slices of knowledge presented as salesmen mostly don't know much more then just their instrument....and tell you thats good for everything.
------------ End Original Message ------------

Actually, the Quasar Resonant Inspection method goes a bit further than what you have described. It is best used for production line parts and has been very successful at welds. The testing is on the order of a few seconds per part. But, the extraordinary virtue of the Quasar approach is its ability to to pull a 0.1% flaw out of parts that are normally varying at 1 to 2 %, or even greater. This is done by looking at multiple resonances and their patterns, and requires some software training. The only information used is resonant frequencies, and since it is using frequency domain swept data, it can look at multiple higher frequencies providing significantly more resolution than other techniques which typically resolve only a few of the lower modes. Of course, it is not useful for most composits, since they do not resonate well. But, it works estremely well with most metals and ceramics. It is also not a one-shot measurement technique. It must betrained by looking at "good" versus "bad" examples.

Kim


 
09:17 Jan-06-2004

Wolfgang Bisle

R & D, Project leader & Group Leader InService Inspect.
Airbus Deutschland GmbH,
Germany,
Joined Jul 2000
35
Re: Resonance testing Kim,
yes, that's it. You must train it.
But this only functions if you know your parts best, and know very well what is "good".
But that is only true maybe for mass production.
For instance in the field of aeronautics these "learning" methods have mostly failed, because after a short time of application conditions occured, which had not been foreseen in the training. Or the bandwidth of "good" is not so unique, that the training algorithm functions well.

Within a research programm with other companies your System is just under evaluation for aircraft application, we will see what happens.

What you said about composites and their resonance ability is a little confusing if put into relation to reality of NDT.
Composites might not work well with your approach of Resonance testing, but the widespread use of other resonance testing methods in that field gives another view.
And the expression "composites" is not very specific.
I guess you talk about Fiber reinforced plastics. But for instance what is with fibre metall laminates, which can also be categorized as composites?

In general a system based on trasinig algorithms are always a risky approch in respect to inspection reliability. As you don't really controll the physics of the inspection process. It is only a block box, and according Murphy's law the chance is extremely high that you put in garbage and get out garbage, and you never know....
Wolfgang


 
00:49 Jan-07-2004
M. Kim Johnson
Re: Resonance testing ----------- Start Original Message -----------
Wolfgang,

Please let me reply between the lines, so to speak. I believe this is important for understanding.

Kim,
: yes, that's it. You must train it.
: But this only functions if you know your parts best, and know very well what is "good".

This is correct. In fact, we have found this to be one of the largest problems in all of manufacturing - the ability to separate structurally unacceptable from structurally acceptable parts using almost any method, especially when the flaws are small. This is probably one of the more difficult tasks that a manufacturer faces, regardless of the NDT technique used: what parts really are flawed and what parts really are good? Generally, there is some reasonable way to come to a first cut answer. However, on par, we have found that parts are very often misclassified using standard NDT techniques when one is concerned about performance flaws, and not just indicators. We had one case where 100% of all of the "bad" parts were not structurally flawed, and one case where 30% of the "good" parts were structurally flawed.

For essentially all parts we have tested, some such combination of misclassification has occurred. The way we find this out on the first training exercise is that the structurally misclassified parts stand out as outliers using our pattern recognition software.

: But that is only true maybe for mass production.

Partially true - this method is not really effective without a reasonable training base. But, when there are safety critical, or very expensive parts that are made in relatively small numbers, then our resonance inspection may be highly cost effective, depnding on amount of "good" variation allowed and the ability to find or replicate artificial flaws. This is highly situation dependent.

: For instance in the field of aeronautics these "learning" methods have mostly failed, because after a short time of application conditions occured, which had not been foreseen in the training. Or the bandwidth of "good" is not so unique, that the training algorithm functions well.

We have not really tackled aeronautical parts on any scale because we have concentrated our resources there is a higher probability of payout sooner. So, no one has tried our system and methods to be able to say that they won't work. Generalizing from other, totally different techniques is probably inappropriate at this juncture. However, we have done some work with aircraft. If you are talking of trying to train on a few parts, your statement, above is absolutely correct. We need a reasonable sample representing the normal, expected manufacturing variations that are allowable. If we have that, then our Resonant Inspection will work. What that magic number of parts is depends on the amount of manufacturing variation you see is. Sometimes changes over time are a bad thing. We will detect that. If it is not bad, then it is relatively simple to retrain, adding the changed parts into the data base and letting software take an hour, or so, to resolve the difference and generate a new recognition algorithm.

Also, we did a test with a helecopter consortium in the US in which we measured a helicopter spinle in situ regarding the test fixture. We were looking for change in a single, safety critical part. An edm notch was made in a high stress location, and the spindle was flexed in the test fixture. We looked at the pattern for change over time. As it turned out, we were able to predict a structural change occurring in the spindle well before either visual or small strain guage indications showed up. When the crack first started to propogate, we had already identified that a significant change was occurring. This is another way of using our Resonant Inspection methods on aircraft that is impossible without the resolution you get from continuous wave scanning to look at multiple frequencies.

: Within a research programm with other companies your System is just under evaluation for aircraft application, we will see what happens.

Is this the effort that Godfrey Hands is working, perhaps? If not, could you say what it is?

: What you said about composites and their resonance ability is a little confusing if put into relation to reality of NDT.
: Composites might not work well with your approach of Resonance testing, but the widespread use of other resonance testing methods in that field gives another view.
: And the expression "composites" is not very specific.
: I guess you talk about Fiber reinforced plastics. But for instance what is with fibre metall laminates, which can also be categorized as composites?

Actually, I should have been more specific. I was speaking of non-metalic composites such as carbon fiber/epoxy, Kevlar/epoxy, etc. Metal composites that resonate without critical dampening are candidates for our Resonant Inspection. We can only detect the grossest of defects in non-metalic composites since they will not support resonances at higher frequencies. But, all other resonant techniques will have the same problem on these materials because of the physics involved.

: In general a system based on trasinig algorithms are always a risky approch in respect to inspection reliability. As you don't really controll the physics of the inspection process. It is only a block box, and according Murphy's law the chance is extremely high that you put in garbage and get out garbage, and you never know....

I'm sorry. What you say about our Resonant Inspection simply is not so. The key here is understanding of the physics. I can say that we do have the fundamental understanding. In fact, our earliest pattern recognition software was based on that understanding, but used brute force calculational techniques to derive the patterns. This is not black box technology! We looked for, and found much more sophisticated pattern recognition techniques that work with great reliability and speed. But, we do know why and how it works. Of course, all of this is patented, but it is unlikely that the essence of the physics understanding coupled with the calculational tools and engineering required to put it all together can be easily replicated without significant resource committment. The real key is in the engineering of the apparatus - keeping the transducer contact points the same for each part so that boundary conditions are not effected, understanding and accomodating for temperature effects, making wide-band transducers that are robust and highly repeatable, etc.

I apologize to anyone not interested in this topic, but Wolfgang, please contact me privately if you wish to speak more. I have been working with the development of this refined version of Resonant Inspection for over 12 years, now, and like to share what I know.

Kim

: Wolfgang
------------ End Original Message ------------




 
06:40 Jan-07-2004
Wolfgang Bisle
Re: Resonance testing Let me just put in a short comment, until that I personally will end my contribution to this topic.

My concern about "training"-methods was a general feeling about systems which try to do the inspectors work, when they do "decisions" on what ever base. Until now we have non of them in our inspection manuals, as they have until yet not passed the necessary qualification processes.

I have not spoken especially about Quasars Approach. I only expressed a general concern about a way of automatic data evaluation where the inspector is sitting in front of a more or less black box.

The research project is not with Godfrey Hands - but I can't tell here details, maybe if you think a little bit more world wide you might get an idea.

But more on this might only be boring for other readers.



 
07:08 Jan-10-2004
Peter Mäckel
Re: Resonance testing

----------- Start Original Message -----------
: HEllo,
: I understand that there is a method called resonance testing which uses the natural frequency of a job to identfy whether it has any defect by comapring the natural frfequency of a defect free job.
: Is it used in detecting defects in welding ? Any specific experience ? Can the method be used for detecting stiffness of structures ?
: S S Ananthan
------------ End Original Message ------------

Dear S.S. Ananthan,
we can offer a method and equipment, which can either identify natural frequency of the structure/object itself but also natural frequencies of local defects itself due to the reduced stiffness of the defect areas (e. g. debonding or delaminations or cracks or impacts).

In combination with FEM it has been also used to identify the material parameters like stiffness (or damping) of defect CFK for examples after impacts. The method works full field and non contact. Some examples are shown on our internet pages:

http://www.shearwin.com/isi-sys-english/applications/ndt-dynamic/ndt-dynamic.htm

http://www.shearwin.com/isi-sys-english/applications/impact-carbon/carbon-plate.htm

http://www.shearwin.com/isi-sys-english/applications/glare/glare-plate.htm

The systems are applied for vibration measurements in industry typically in combination with laboratory shakers in harmonic excitation modes using the forced normal mode excitation method (e . g. for components like electronic boards, housings and others)

It is the first time that I discovered this forum, which we got via an Email from someone unknown. Thank you to the unknown person.

Peter Mäckel





 
06:01 Jan-13-2004
S S ANANTHAN
Re: Resonance testing Dear Friends,

Thanks for all the useful information on the resonance testing. Thank you Mr Kim Johnson and Wolfgang Bisle for the information. I will be in touch with both of you directly.

S S Ananthan
----------- Start Original Message -----------
: Let me just put in a short comment, until that I personally will end my contribution to this topic.
: My concern about "training"-methods was a general feeling about systems which try to do the inspectors work, when they do "decisions" on what ever base. Until now we have non of them in our inspection manuals, as they have until yet not passed the necessary qualification processes.
: I have not spoken especially about Quasars Approach. I only expressed a general concern about a way of automatic data evaluation where the inspector is sitting in front of a more or less black box.
: The research project is not with Godfrey Hands - but I can't tell here details, maybe if you think a little bit more world wide you might get an idea.
: But more on this might only be boring for other readers.
------------ End Original Message ------------




 


© NDT.net - The Web's Largest Portal of Nondestructive Testing (NDT) ISSN 1435-4934

Open Access Database, |Conference Proceedings| |Articles| |News| |Exhibition| |Forum| |Professional Network|