JP, yes, you'd get the roundness, cylindricity, longitudinal straightness and position from the total runout. I suppose that in a sense you'd also get the line-segment longitudinal parallelism to the axis as well; interesting thought. As for how good the bottom of the groove needs to be, that's dependant on a number of factors including load magnitude & type, seal compound & deformability/durability. I had to work with some o-rings and custom seals that were made from some interesting polymers that were highly susceptible to abbrasion wear and seatng form; those required even better control of the surface. But to turn your question around, why go for a lesser, and more difficult to validate, control than one that gives you extras for free. If you buy a base-model car these days, it probably has a lot of "extras" like power locks & windows, CD player, etc ... would you tell them to strip the extras out because it doesn't add value to you personally? By more difficult, I mean that you have to verify each segment independently rather than as an entire surface; decision criteria must then be set for how many locations along the surface, etc. As Dave points out, it's a very short surface typically (1.5-2.5mm), so where is the economy and value of using the lesser control?
Personally ... wait for it, Dave ... I prefer a surface profile that goes from the tangency point on the major diameter to the other tangency point on the major diameter. It controls size, location, runout, form, orientation. I've gone thru the work of checking the difference vs traditional linear tolerances & runout ... less time to inspect the profile & well within the tolerances required. Considering that most such grooves are now done by form tools, it's even more likely that the same tolerance can be held on all aspects.
Jim Sykes, P.Eng, GDTP-S
Profile Services
TecEase, Inc.
