Transfer correction on a thick ferritic nozzle weld
A 60 mm thick ferritic steel nozzle-to-shell weld is to be examined from the shell surface with 2 MHz, 45° shear-wave probes. The written procedure requires a DAC from 2 mm side-drilled holes in a 25 mm thick reference block. The production surface is blast-cleaned and has a thin primer; the reference block is ground. At the first sensitivity check, the 2 mm SDH echo at 40 mm sound path is 5 dB lower on the component than on the block; 1 m away it is 2 dB lower. Couplant is glycerine and probe pressure is controlled. Three planar indications, 12-18 mm long, are found at 30-45 mm depth close to the fusion face. The client wants to know whether the weld can be accepted without further work. You must decide how to handle transfer correction and whether the indications are acceptable.
1. What transfer correction should be applied, and how does the 2 dB versus 5 dB variation affect the scan?
Read the worked answer
Transfer correction is the gain difference needed to make the same reflector produce the same screen height on the component as on the reference block. It must be measured at several representative points on the actual scan surface, at the sound paths used, with the same probe and couplant. Here the measured loss is 2 dB at one metre and 5 dB at the first point. A single correction based on the lower loss would be unconservative: the 5 dB area would be scanned 3 dB too insensitive. Either apply 5 dB over the whole area and confirm that grass does not become unacceptable, or divide the weld into zones and apply the measured correction in each. Because the variation is 3 dB, I would zone the correction and re-check at each zone boundary. The correction must be added to the DAC reference gain, and maximum echo amplitudes already recorded must be corrected before evaluation. If no correction was applied during the scan, the data are invalid for acceptance.
2. The three indications are planar, 12-18 mm long, at 30-45 mm depth. What is your acceptance decision and what sizing and characterisation would you perform?
Read the worked answer
The indications cannot be accepted on the present data. First correct amplitudes for transfer correction, then compare each maximum echo with the evaluation level for the specified quality level. All three are planar and close to the fusion face, the classic location for lack of fusion; that geometry is less tolerable than a rounded inclusion. Measure length by the 6 dB drop method with the probe moved along the weld, and check through-wall position and tilt with a second angle, for example 60°. A planar indication whose length exceeds the acceptance limit, or which gives a crack-like response, is rejectable regardless of a modest amplitude. If the echoes are below the recording level after correction they need not be sized, but the backwall and root should still be checked. My decision would be 'not acceptable pending confirmation': reject if planar length is above the limit, repair, and re-examine.