Thats probably why the codes don't refine the requirement more. A CVN PQR is allowed to be applicable to many thicknesses in which there will be great diversity in grain refining patterns. Bevel angles can change these patterns. Welder technique can change these patterns. Etc., etc.
There are a variety if ways to handle these things if there is a specific concern. One good way is to make sure that you have notches to represent the full thickness with allowance for unrefined root ID's and unrefined cap beads, as Greg spoke of. Though the cap can be covered by more careful temper beading type techniques (not so easy to really control in practice) or PWHT.
The general idea is to verify that you have an overall microsturcture capable of arresting fractures at a certain level. It isn't an either or situation. There are 'ALWAYS" trouble zones that can initiate cracks. Carbides, nitrides, grain triple points, microstructural discontinuities, intermetallics, etc., depending upon the alloy involved.
But if the general microstrucutre is 'tough' the weld will be viable.
The other thing is, as always, 'sound engineering practice'. If you have to 'refine' your evaluation too much, perhaps you should go with a more forgiving alloy. For example Ni steels if your pushing the envelope with C steel, or even austentics if the ferritics are risky.
The problem here is everybody is trying to save money, so sometimes they try to put the burden on the backs of a CVN PQR. This is not without risks because there are just too many variables inthe real world.
That's basically my point. If the parameters are outside the allowable ranges qualified by a PQR, then, in my opinion, the welding is not qualified and tests should be run in the ranges needed. Or the welding parameters should be adjusted so that they are within the ranges.
One common problem I have run into is when a single pass fillet weld can't be run within the groove PQR test ranges. The fillet needs enough heat input to penetrate to the root particularly 5/16" 2F. Yet a groove weld can tolerate a wider range and still get adequate fusion.
Unfortunately, I believe many 5/16" 2F FCAW fillets do not fuse to the root, but we cannot see that in production. I have given many fillet tests and found the average welder (better than 90%) can't do that size on the first attempt, but almost all believe they can.
Some say that doesn't prove anything because production welds that size are in multiple passes- but then the heat input is out of range and we're back to unqualified welding.
Again, I am speaking from my experience with D1.5 and no other codes. And I know that test conditions often do not duplicate actual production situations, but they should be nearly the same because that is the point of the testing.
As far as not counting root and cap passes, I have not seen anything saying it is OK to do that. Will that matter? I can't say but it seems to me that a weld in service usually starts cracking from the outside and not the insde.
One thing that would be nice is if AWS could address PQR testing in a more definitive manner. There is a huge amount of information about the requrements of testing, but very little about how to actually conduct the testing.