What is the best method for NDT of cast iron? Application is paper machine dryer cylinders. We have historically just used standard pulse echo probes, then went to an automated time of flight (TOF, also known as hartford wedge I believe?) method due to inaccuracies found with . We have found the TOF to be very difficult to schedule as it scans the length of the roll by running on a rail with position feedback etc. Processing is done through special software by contractors.
We would like to go back to a manual measurement but want the accuracy of the TOF.
Are there any standard instruments out there that will work with a TOF probe.
Is there a smart way a standard instrument can be used, i.e. just collect both signals ( surface and thickness) and use the surface actual/measured factor to adjust the thickness measurement?
Are there other NDT tecnologies out there for measureing materials such as cast iron.
Its been a while since I have worked with this gear so go easy on me jargon wise. I have just been asked to reseach possible alternatives.
Any help much appreciated.
Norske Skog Tasman,
21:31 Jul-22-2009 Roger Duwe NDT Inspector, API-510, 570, 653 MISTRAS, USA, Joined Jan 2009 148
Re: best method for ndt cast iron thickness?In Reply to Mike Plant at 02:35 Jul-22-2009 (Opening).
We've had good-to-excellent results with conventional UT-straightbeam. Our setup consists of a 1.0 MHz 1/2" to 1" dia. 'ducer on a top-quality flawmeter. Panametrics Epoch 3 or 4, Krautkramer 50 or 60 series. Digital thickness-only meters don't work, dual transducers don't work, medium frequency 'ducers work poorly.
Set the meter to measuring peak-to-peak, NOT 'main bang' to peak. Calibrate on a sample of the material, single point cal. works well [assuming the cal block and the dryer are at the same temperature]. This is because you are measuring peak-to-peak on a linearly accurate meter, so the only calibration item is the actual velocity of your material.
What you are measuring on Peak-to-Peak is the time of flight of the UT pulse between the ID & OD of the drum. The 'quality' of the signal is easily verified, as a good signal has 4 to 7 reflection peaks on the meter screen, in a steadily decreasing series. If you cannot see 3 peaks, the data is probably not valid. Might be a 'tracking' error of the crawler, or not enough couplant.
One B-scan of one cylinder should only take about 1-1/2 to 2 minutes, just long enough for the linear encoder on the crawler 'buggy' to drive from one end to the other. The advantage of the B-scan over several spot measurements is that you now have a profile drawing of the interior of your drum, and can exaggerate the height measurements to easily show banding or dishing of the drum ID.