RFET Signal 30 mm from a Support Plate
You are scanning the tubes of a carbon steel shell-and-tube heat exchanger with remote-field eddy current testing. Tubes are 19 mm OD, 2.1 mm wall; carbon steel support plates sit at 600 mm intervals and are the only tube supports. The written procedure calls for a 300 mm/s pull speed, an inlet through-wall EDM hole as the amplitude reference, and rejection of any signal whose phase falls in the wall-loss region of the phase-angle plot. On tube R-14 the probe shows a signal 1.35 m from the tubesheet, about 30 mm short of the second support plate. Amplitude is 0.9 V against a 1.2 V reject threshold. Phase reads 42 degrees; support-plate signals elsewhere in this tube and in adjacent tubes read 17 to 19 degrees. Pull speed was steady, the probe ran centred, and the strip chart shows only this one anomaly.
1. Do you accept the tube or report the indication? Justify the decision using the physics of the technique.
Read the worked answer
Report it; do not accept on the strength of amplitude alone.
Reasoning
- In RFET, signal amplitude is dominated by fill factor, probe centring, wall thickness and proximity to a support plate. It is a weak discriminator.
- Phase angle is the primary discriminator because it depends mainly on the through-wall position (ID versus OD) of the metal loss, and it is comparatively insensitive to lift-off and fill-factor changes.
- Here the phase of 42° sits far from the 17–19° support-plate signature reproduced in this tube and in neighbouring tubes, and it falls in the wall-loss region named by the procedure. That is a defect-like response, not a support-plate response.
- 0.9 V versus a 1.2 V threshold does not clear it either: the threshold was derived using an inlet through-wall hole under ideal conditions, and a support plate within 30 mm can depress the wall-loss amplitude.
Action: confirm by a slow, centred re-scan (about 150 mm/s) over the region before the report is issued, and treat the tube as suspect until the confirmation is complete.
2. Why is the inlet through-wall EDM hole an inadequate reference for this call, and what should the reference standard contain?
Read the worked answer
The inlet through-wall hole is a single-point amplitude reference taken in an ideal geometry (straight tube, no support plate nearby, easily centred probe). It tells you what a 100% through-wall opening looks like at one location; it says nothing about:
- what phase an OD versus ID loss produces in this material and at this frequency;
- what a support plate looks like, including the phase distortion at the edge of a plate, which is exactly where this signal sits;
- how amplitude degrades when a defect is close to a plate.
A suitable reference is a calibration tube of the same alloy, dimensions and heat treatment carrying:
- EDM notches at ID and OD surfaces at several depths (e.g. 20%, 40%, 60% through-wall) so phase can be tied to defect type and depth;
- simulated support plates at the same axial spacing as the exchanger, so the plate signature and its edge effects are characterised.
With that reference the 42° reading can be assigned a depth and a surface rather than merely flagged.
3. A colleague proposes raising the gain by 4 dB or dropping the test frequency to 'make the picture clearer'. Evaluate both suggestions.
Read the worked answer
Raising gain: this scales amplitude uniformly. It will push 0.9 V above the 1.2 V threshold, but it does not change the phase angle at all, and it amplifies support-plate, lift-off and noise signals identically. Gain is being used to manufacture a rejection rather than to resolve the call. It also risks saturating the channel over the plate entry and hiding genuine data. Reject the proposal as a means of disposition; a modest gain increase is legitimate only as part of a documented re-scan for characterisation.
Lowering frequency: in RFET the through-wall (indirect) coupling improves as frequency falls, which is why low frequency is used - but the same effect increases the support-plate signal and broadens it, making a plate 30 mm away more of a confounder, not less. Raising frequency instead sharpens resolution but shortens the indirect path length and weakens the through-wall signal RFET depends on.
The defensible options are: slow, well-centred re-scan at the procedure frequency; a two-frequency mix to subtract the plate signature; and comparison against the reference tube described above.