B16.5 is for Class designated flanges, not for PN10. PN designated is in the scope of the EN 1092-series.
Second, keep in mind B16.5 acceptance criteria are for
new flanges.
I suggest to get a copy of PVP2013-97813. The current acceptance criteria in B16.5 for flange imperfections are, at some parts, unrealistic.
The paper by W. Brown elaborates on how to overcome this problem. He also gives recommendations of what is considered a better acceptance criteria.
Here's what he has to say on this table;
The issue with this requirement (table 3) is that it allows a flaw across the full surface finish to half the serration depth. It also allows a flaw to full serration depth across anywhere from 0.12 to 0.5 inches in the radial direction, depending on flange size. It also allows a defect in the flange surface finish of between 0.06 and 0.25 inches deep, with the same radial width. The latter two limits correspond, respectively, to defects that are the serration depth and approximately one half of the gasket width for a spiral wound gasket and significantly deeper than the serrations and one quarter of the gasket width for a spiral wound.
That being said, applying PCC-1 2013 to find out how radial is to be interpreted, I think your 1st measurement is incorrect, the 2nd is correct (that one is actually the radial distance).
For radial imperfections, you have to measure along the path from center to the raised face end, radially.
That what u did in picture 2. In picture 1, you measured some sort of width in stead of depth.
Think of a leakage path, and the potential a flaw has for leakage to be initiated.
The deeper (hence less chance the gasket filler may fill the flaw) and longer (i.e. more radial) the path, the greater the potential for leakage.
