Lower limit of discrimination with UT

Date 12-24-2013 04:11

1/64" is 0.015625.
Don't know what material is being examined, technique deployed, etc, but as a general rule, the smallest detectable flaw is 1/2 of the wavelength.
Wavelength = velocity/frequency.

Can UT reliably and repeatably find flaws that size? That answer is dependent upon many factors, but as a general rule, yes. Are the technicians always capable? no.

By jwright650 (*****)

Date 12-24-2013 12:27 Edited 12-24-2013 12:32

Wow...1/64" holy smokes that is a small defect.
Are they finding this with a straight beam transducer?

edit*

Gerald, How does one map out a defect that small to be able to say that is exceeds 1/64" or not?
Since there is no mention of how wide it can be, I guess you would have to assume 1/64" in any direction?

By welderbrent (*****)

Date 12-24-2013 14:30

John,

As I am just getting started with some study for UT this has me somewhat baffled. I can understand it being able to spot that size of an indication, but I thought that to most any code it had to be considerably larger than that to be rejectable and require repair.

Am I missing something here?

Have a Great Day, Brent

By jwright650 (*****)

Date 12-24-2013 16:09

Flaw sizing in D1.1 is way different than what has been proposed to us here Brent. Gerald has more UT experience in this than I do, so I have to draw off of his knowledge....IOW, I learn something new whenever he posts on this subject.

By CWI555

Date 12-24-2013 15:57

I didn't see anything in the post about materials, codes, techniques, industry etc. To quote the OP:

"Ihave a customer that just had to do a ton of rework because they were failing UT examinations. If a defect was found that was planar and exceeded 1/64" in dimension (if it's planar, must be in 2 dimensions, right?). Then it was a fail. Can UT Reliably and repeatably find defects of this size?"

My response;
"1/64" is 0.015625.
Don't know what material is being examined, technique deployed, etc, but as a general rule, the smallest detectable flaw is 1/2 of the wavelength.
Wavelength = velocity/frequency.

Can UT reliably and repeatably find flaws that size? That answer is dependent upon many factors, but as a general rule, yes. Are the technicians always capable? no. "
c
Given the lack of pertinent information, the answer stands as a theoretical yes. BTW, a 1/64" flaw in 'any' direction is rounded. :razz:

Calculate the wavelength for various materials, then consider the .015625" flaw size. Now lets match up a theoretical:

Lets pick titanium and list the wavelengths for longitudinal @ a velocity of 6100m/s:
1mhz = .24015748"
5mhz = .04803149"
7.5mhz = .032020997"
10mhz = .024015748"
15mhz = .016010498"

It becomes clear that it is definitely detectable. For aerospace applications in thin walled materials, it's not that unusual. Mapping btw isn't that hard or complicated for an automated system such as an automated immersion bed. It can be done manually as well. Remember, the typical 6dB drop method isn't 'the only' method. There are ID OD rolls, Time of flight crack tip diffraction, immersion delta technique, etc etc. Then there is also the assumption of wave form being made. Lamb waves, Raleigh waves, Flexular waves (Antisemmetrical lamb waves), etc etc.

Bottom line, we were not given enough information. When it comes to theory, yes in fact it can be done.

Now if we are talking a typical structural or pipe scans, I'd have to raise the flag.

By jwright650 (*****)

Date 12-24-2013 16:06

Thanks for the extra info Gerald :wink:

...I'm just not all that familiar with anything other than structural welding, so this had me dreaming back to my UT class days trying to figure this out. I took the formula and ran backwards through trying to find a material velocity that was on a chart of materials that I had...just baffled at looking for something that small, as I've never been challenged to do that before.

By fschweighardt (***)

Date 12-24-2013 17:24

I'm still trying to get my hands on the inspection procedure, and if possible some of the calculations as above that would validate the procedural assumptions.

Material is X70 bout 1.25 wall

By CWI555

Date 12-25-2013 04:33 Edited 12-25-2013 05:03

I am unaware of any piping code that calls out the necessary equipment spec that would find a 1/64" flaw. Strongly suggest digging deeply on that one and raising the flag.

Edit after some further research;
If this is an 1104 procedure, they are trying to use 1/64" for a reason. The last edition I have is 2005, but I can't imagine the criteria for cracks have changed. The only planar flaw that is an effective automatic defect is a crack (assuming no code year change). Given the acceptance criteria in 1104 for all other planar type indications, differentiating between a crack planar, and say a fusion planar becomes important.

The technician must see a crack tip diffraction, sin wave inversion, etc to proof the crack. In all, I suspect your getting hosed.

By fschweighardt (***)

Date 12-25-2013 06:11

I think it's an end user defined criteria, per contract, still want to get to the bottom of any proof that they can actually see the damn flaw. They have taken samples that supposedly contain the indication and ground .005 at a time and could never find it

By CWI555

Date 12-25-2013 14:17

High probability someone didn't realize what they were asking when that got put into a contract. A few things for you to look at. All based on 1/64" max dimension. Which btw, for it to be classified as a linear, the length has to be 3x it's width as a general rule of thumb. If 1/64 is the greatest dimension, then the smallest dimension should be <=0.005208".

Bandwidth on the transducers. A 5mhz ducer isn't actually 5. It can be narrow band, broad band, or some point between. The listed frequency is the center peak response of that bandwidth. Get the ducer certs and apply the frequency formula.

If it's phased array, what version, what was the head, what frequencies are present in the array, what kind of array was it, focal laws applied if any, or is it simply a phasor.

If they demonstrated this "1/64" flaw. Remember upper and lower bounds, width etc. Could be a good idea to call for them to demonstrate the procedure.

If it was automated, what system, cal standards, reference standards etc used. Certified? Demonstrated? etc. Encoding is a big one, particularly when it comes to time dilation. Was it actually automated or just a mouse encoder? I can stretch a 1/128" flaw found with 25mhz immersion and make it look like the grand canyon by dilating the time.

The list of variables is long one, but for piping, I don't recall a procedure, code, or specification I've ever seen that rejects 1/64". Even IGSCC EPRI nuclear UT isn't that stringent though it could find it. The ID or OD 'undercut', criteria for visual isn't that stringent.

The only two worlds I've ever seen something that stringent was aerospace and milspecs.

My opinion for what it's worth.

By CWI555

Date 12-25-2013 04:33

It's common in the aerospace world. Consider a craft skin that is only .032" thick out of the gate. What size flaw would fail something like that? From an aerospace standpoint, the size makes sense to me.

Regards,
Gerald