Comments on paper on "Construction weld inspection procedures"
These criticisms of our paper Construction weld inspection procedures using ultrasonic phased arrays by Michael Moles and Jinchi Zhang, R/D Tech, were sent to ASNT. The criticisms and reply are below.
Bonjour Mr. XXXXXXX Mr. YYYYY ZZZZZZZZ told me to communicate with you on this matter.
I would like to submit a formal complaint about an article that has been published in different medias including ASNTs. The article was declined in different flavors, but always include over-simplistic and false information about weld inspection technique. In essence, the article says that phased array ultrasound beams HAVE to be perpendicular to the wall (bevel) if one wants to detect lack of side-wall fusion defects. It should have been PROPOSING that perpendicularity provides stronger specular echoes, but still tip diffraction echoes of lack of fusion defect provide sufficient signal-to-noise ratio for most weld inspections.
You will find the articleflavors at:
May I recommend you to talk with Mr. Greg Selby (EPRI, email@example.com) and Mr. Jonathan Buttram, Level III (InterWAV, firstname.lastname@example.org) about this article. They seem to have the same opinion I have. EPRI also qualified a procedure recently which proves to some extent that Mr.Moles is wrong. If you are unsuccessful, I can propose you other names.
I think the technical revision commitee should have filtered this article or should have at least required a modification in the wording. I think an Errata should be published in the next ASNT Journal. But I also understand sometimes things slip through unfortunately...
Reply by Michael Moles:
This article has received many comments and criticisms, invariably from people who didnt read it properly. This author obviously didnt read it correctly either, doesnt appreciate the differences between construction and in-service inspections, and has compounded it with a lack of understanding on tip diffraction. Defining the article as over simplistic and false is itself raising the tempo beyond the normal realms of scientific discussions.
The article itself is primarily based on modeling, which has known limitations. In this instance, modeling is likely to overstate the issue of mis-orientation, but this is acknowledged in the text.
False is certainly an unnecessary and incorrect statement; there is no doubt that the response from a defect depends strongly on its orientation relative to the incident ultrasonic beam. This dependence has been known for years and has been the basis of codes such as ASME1 (appropriate angles should be used). To quote API, A (second) limitation in the use of shear wave ultrasonic inspection is the failure to detect large two-dimensional (planar) discontinuities as a result of the inherent direction of the reflected beam.2 A simple understanding of weld geometry and the position of the array shows that a single S-scan cannot have well oriented beam angles on the weld bevel, root and cap, at all locations. The thicker the component, the worse the problem becomes.
The modeling shows that location and orientation of the defect also matters, which will be no surprise to any operator. In the modeling, midwall defects were particularly difficult to detect, which mirrors real life pulse-echo ultrasonics. Again, thicker components presented more problems than thin components.
The article does not say that phased array beams (or any ultrasound beams for that matter) have to be perpendicular to the weld bevel; it says that the closer the angle is to the weld bevel incidence normal, the better. It is clear that current codes (e.g. ASME) do not demand normal incidence. How can they? Typically a 45o and/or 60o transducer is used for conventional inspections. With a 30o bevel, incident angles will be normal on the bevel, but root angles will be off. However, if a 37.5o bevel is used, the angles will be off-normal by 7.5o on both bevel and root, yet this is permitted by code.
Relying on tip back-diffraction alone would be a disaster. The author seems to have little understanding of diffraction, something which Olympus NDT (formerly R/D Tech) has been using for ~15 years. The standard diffraction technique is TOFD (Time-Of-Flight Diffraction Technique), which runs at least 20-40 dB above typical pulse-echo scanning levels because tip diffraction signals are so weak. Even with TOFD, we sometimes have trouble detecting crack tips, especially with noisy material. Using standard pulse-echo codes and looking for tip reflections would result in almost no defect detection; of course, some manufacturers would like that!
To address the issue of backward tip diffraction mentioned here (i.e. pulse echo), like everybody else we see diffracted signals quite often. They are a very good method of sizing, especially with phased arrays where the full image is available and coupling is a non-issue. However, the tip signals are mostly low amplitude (usually below recording threshold), and are identifiable primarily because of their association with nearby crack signals. Identifying, calling and characterizing tip echoes under normal pulse echo scanning is going to miss a lot of defects.
The EPRI procedure referred to here is a manual S-scan code for reactor pressure vessels. (The procedure is EPRI-proprietary and not available.) However, EPRI states that the code is primarily for fatigue cracking, with the possibility of finding weld defects. (One seriously hopes that nuclear-quality reactor pressure vessels have very few unknown weld defects.) More to the point, fatigue cracks occur at known locations (the surfaces), and typically provide good corner reflectors. Weld defects occur at characteristic locations in the weld, and deterministic orientations, but often without corner reflectors; therefore correct incidence angles are important for detection. Different procedures are required for different types of defects. The work report in January 2005 was specifically for construction welds, as stated in the first word of the title not for in-service inspections.
Incidentally, another EPRI-approved procedure for weld inspections from Zetec uses S-scans and linear scanning, but uses three S-scans as we recommend. Though Zetec ultrasonics is an offshoot of R/D Tech, this work was done independently, and provides support for the multi-S-scan approach. (Being nuclear, however, this procedure also requires extensive lateral scanning for mis-oriented defects; this is not normally required in the workmanship-type codes.)
1. American Society of Mechanical Engineers, ASME Section V, Article 4, paragraph T-472.1.1, 2001.
2. American Petroleum Institute, Recommended Practice for Ultrasonic and Magnetic Examination of Offshore Structural Fabrication and Guidelinesfor Qualification of Technicians, RP 2X, Third Edition, Sept 1996, p. 7.
By Michael Moles
Olympus NDT (formerly R/D Tech)