API 653: Tank Shell - Ultrasonic Thickness Testing SOP
API 653: Tank Shell - Ultrasonic Thickness Testing SOP
(OP)
I see often in old reports (including mine) that have 4, 8, and even up to 16 vertical runs of UTT readings for the tank shell metal.
After going through heaps of API documents, i haven't figured out how we came to this standard of care. Am i missing something? maybe some of the gray-hairs remember how this all got started...
I can see the value to multiple lines when the tank is insulated, since Corrosion Under Insulation may be at play in different ways... but for an uninsulated tank, it seems like overkill once you get above 5 feet (or really above the 1 ft critical shell) and especially so if there is an internal inpsection happening too.
After going through heaps of API documents, i haven't figured out how we came to this standard of care. Am i missing something? maybe some of the gray-hairs remember how this all got started...
I can see the value to multiple lines when the tank is insulated, since Corrosion Under Insulation may be at play in different ways... but for an uninsulated tank, it seems like overkill once you get above 5 feet (or really above the 1 ft critical shell) and especially so if there is an internal inpsection happening too.





RE: API 653: Tank Shell - Ultrasonic Thickness Testing SOP
The API-653 Inspx checklist runs on for about 12 pages of line-tiems to evaluate. Demand that your report have that level of detail.+
On tanks up to about 40-ft, 4 lines of readings - traversed or 'drops' - are appropriate. On the floor and the lowest shell ring, the readings need to start about 1-ft apart, then increase in spacing as the readings are farther and farther from the 'chime' - the floor-to-shell corner joint.
If CUI is suspected, or found, have the insulation stripped until the extent can be visually established. Then establish the Tnom as per the API-653 averaging method.
If the floor has 'problems' found with the UT, then bring in a MFL scanner and do the whole floor. I would MUCH rather have a fully scanned 4-ft band of floor, at the shell, than 500 spot readings. I use a fairly large UT transducer, a 1/2-inch round. Think about how much floor gets missed with 100 or even 500 spots. This is the reason for a trained, experienced tank Inspector -- preferably an API-653.
All a bunch of thicknesses give is a small part of the 'picture'. A good Inspector will be able to interpret these results and decide if more testing, ore even cutting corrosion coupons from the floor are needed. Save your UT-thickness money for a real Inspector, and allow him to remove insulation for CUI evaluation. Also allow MFL scans or cutting a coupon out of your floor.
A whole bunch of spots in a grid is only information for refining the Visual, not really a proper Inspx by itself.
The QC/QA industry started using 'directed' sampling plans since WWII. Just sample your tank, until the sample thicknesses indicate a problem. And get someone experienced to do that Visual Inspx.