I would go with Total-Runout, both because of functionality and because of the ease of inspection.

Suppose you keep the tolerance value of .001".
With position, it will be written in the feature control frame as ⌀.001, and will create a tolerance zone of that diameter in the center of your workpiece, around datum axis A. In that tolerance zone the derived axis of the groove diameter will be evaluated. Therefore, only the location and orientation (coaxiality) of the groove diameter will be controlled, but not it's form. The measurement of position will not indicate how round or cylindrical the groove is, but you do care about it for the sealing quality.

The Total-Runout tolerance will be specified without the '⌀' symbol, and it will create a boundary between two concentric perfect theoretical cylinders perfectly centered to datum A with a radial gap of .001 between them (the diameters of those tolerance boundaries could be anything). In that zone (the gap), the actual groove surface must fit. The measurement can be by a reading of the dial indicator swept along the groove while rotating the shaft as it is chucked or supported in a V block by datum feature A.
This will control the coaxiality as well as the form (roundness and cylindricity) of the groove diameter within the .001 tolerance, which is probably tighter than the diameter tolerance on the ⌀.950 that is considered for rule #1, and I think is preferable for the O-Ring application.

Suresh KS,

I would first recommend asking yourself the following question: Was the original application of concentricity tolerance really needed from the functional standpoint? If you are able to convince yourself that controlling location of "mid-points" established from pairs of diametrically opposed points was the true design requirement, then, once a switch to Y14.5-2018 has been made, there is technically no symbolic way to replace this exact requirement.

If it is otherwise, then, as Burunduk stated, other geometric characteristics can be used. In that case you should ask yourself if it is OK to control form (circularity and/or cylindricity) and location of the groove bottom within a single tolerance of .001, and if yes this would justify usage of circular or total runout. But maybe it is fine to let the groove bottom have larger form error than it is location error? Or maybe it should be vice versa - larger location control and tighter form control?

Also, some industry standards, such as AS568 for o-ring groove design, show a direct linear dimension between the OD and the groove bottom. I am not saying this is something you should do in your case (especially that the dimensioning and tolerancing standard does not provide an interpretation for this type of dimensions). Just saying it is an option.

Thanks for your valuable suggestions Burunduk and pmarc,
Burunduk, I have another doubt from your answer, Why do you want to go with Total Runout? Since the groove has small surface can't we go with Runout which will be sufficient enough to measure the feature considering its surface, Yes if it has lengthy one then I would have gone with Total Run!!

Correct me if I'm wrong!!

Thanks for your valuable suggestions Burunduk and pmarc,
Burunduk, I have another doubt from your answer, Why do you want to go with Total Runout? Since the groove has small surface can't we go with Runout which will be sufficient enough to measure the feature considering its surface, Yes if it has lengthy one then I would have gone with Total Run!!

Correct me if I'm wrong!!

Suresh KS,
Yes, you could use Runout, but it will control the form of the surface differently. It will only control that each cross-section is circular enough, but not cylindricity. For example, your groove could be conical within the size tolerance*, and then the O-Ring will be pushed deeper on one side of the groove than on the other.

*BTW, what is the size tolerance on the groove diameter? My whole suggestion is based in the assumption that it's looser than .001".

Neither one is what one cares about, which is the amount of compression at every point along the perimeter.

Concentricity and runout and total runout do this indirectly and try to control either the variation of variation (concentricity) or variation with respect to a non-functional datum (runout and total runout).

The direct geometric tolerance control for this is profile of line applied at each location around the perimeter with a tangent plane at each location around the perimeter and an alignment to the axis. This control and control of the bore and piston diameters controls the compression of the o-ring directly to ensure it is stretched the correct amount and does not have too much exposure allowing the o-ring to squeeze out.

Gland depth is the control specified by o-ring makers.
https://www.globaloring.com/o-ring-groove-design/
https://www.mnrubber.com/tools-resources/design-gu...
https://o-ring.info/en/technical%20manual/ERIKS%20...
https://www.sealanddesign.com/technical/o-ring-gro...
https://www.applerubber.com/src/pdf/section4-seal-...
https://www.parker.com/content/dam/Parker-com/Lite...
https://dichtomatik.com.mx/catalogos/diseno-de-alo...


No surprise, concentricity, runout, and total runout are not mentioned by the companies that supply o-rings as a geometric tolerance control, though it is mentioned as a conventional English term.

There are groove depth measurement tools. They sit on the tangent plane for external features.

If one looks only at runout or total runout, it is possible to have an out-of-round condition in the cylinder out of phase with an out-of-round condition or being off-center of the o-ring groove, leading to both excessive depth and insufficient depth that would be ambiguous, even if the range of diameters for both features was reported. It would be necessary to measure the groove/gland depth to see if the part was usable.

Concentricity is a worse control for this case than runout and total runout as it does not detect all out-of-round cases.

I don't imagine anyone using total runout on a face seal thought they have the same compression requirements.

This was the previous discussion: https://www.eng-tips.com/viewthread.cfm?qid=467663

The Parker book is touched on here: https://www.eng-tips.com/viewthread.cfm?qid=263466

The main concern in the Parker book is as follows:

Runout (Shaft): Same as gyration; when expressed in inches
alone or accompanied by abbreviation “T.I.R.” (total
indicator reading), it refers to twice the radial distance
between shaft axis and axis of rotation

rather than the groove runout to the shaft. This is to control variation in compression as the shaft rotates. It is also mentioned in regards to a special sort of seal between a fitting fillet and a chamfer in a tube fitting boss seal. This makes some sense as there is no "depth" that can be measured on a chamfer. Since there is no further explanation and there is another precise diameter for the o-ring to engage, it might be to simply control the forcing cone action rather than the squeeze.

There is also mention of "*Total indicator reading between groove and adjacent bearing surface." as "Max. Eccentricity," not between the groove diameter and the axis of the entire bearing diameter. They want the compression to be uniform. A uniform compression will avoid the o-ring from trying to creep along the groove away from the high compression zones towards the low compression zones and changing the local o-ring section area in unwanted ways.

It is certainly worthwhile to consider cylindricity on the non-gland/groove surfaces involved and runout if there are guide bearings to locate one part relative to the other, but for the o-ring gland/groove the makers of o-rings appear unanimous in not using geometric tolerance symbols or referring to standards for them. But this isn't the forum for their nonsense.