While I don't necessarily disagree with metengr's advice, I shudder at the prospect of what might come from it. You are plowing new ground here, and they don't know either, so I fear any criteria that they might impose on you. It could be impossible to meet, if not even ridiculous.
When Hx tubing is eddy current tested, the eddy current probe if first passed down a check piece of tubing with known and well defined defects in it so as to get a look at how these known indications will show up on the screen. It establishes a known baseline. That is what you lack here.
With that in mind I suggest the following.
Test your welded tubesheet with other well known and accepted testing methods. At a minimum, dye penetrant and air or even better, hydrogen leakage methods. Others on the forum might want to suggest additional methods.
Surely if the statistical bell curve is right, there will be some small number of detectable defects that can be determined by those methods. If any are found, then shoot those, and see what the film looks like. Fix those defects test again, and if the defect is gone, then shoot it (them) again, and compare the film to the previous result.
Short of that, the only other thing you can do is prepare a mock-up tubesheet with some known defects and shoot those to establish your baseline. I don't know how to accomplish this, as, while it is hard to prevent a small percentage of defects, I don't know how you make a defined defect this way. It is different from drilling a hole or simulating wall loss in an eddy current check piece.
Maybe these thoughts will give you a place to start with your unusual requirement.
rmw
