Y14.41 approach

I have experience with a MBD for design and manufacture of precision components and some tooling design, but do not strictly adhere to Y14.41. Some companies I work with use a strictly MBD approach but I do not know how strictly they follow Y14.41. It would mostly appear that they have a combination of company standards that at least MOSTLY adhere to it. I have not seem any claims that their design packages were to be interpreted IAW Y14.41, but have seen statements saying it should be interpreted IAW their own proprietary company standard, so I assume there are deviations.

I would very, very strongly resist the concept of "min, max, or mean" being the only options for modeling. I believe "NOMINAL" is the most appropriate choice. Nominal might be mean, max, or min. It's the most intuitive and easily understand for precision metals machining of which we mainly focus on. Hopefully when they write "for example minimum, maximum, or mean" they do not mean those are the only three options.

Yes, modeling at "nominal" results in some modeled conditions being outside the actual acceptable range (.5000" shaft size, when the acceptable range is actually .4994-.4998 for a random example) and I don't know anyone in the chain of manufacturing that would find that situation intuitive. It's a 0.5000 (g6) tolerance, by the way. For hardware components... one would buy a 1/2" dowel pin, not a .5002 dowel pin. I also find no benefit to having it to the max/mean/min for CNC programming as all CAM packages I've worked with make it -very- easy to create offsets or path compensation to allow for /intentional/ deviation from the as-modeled condition specifically for situations such as this. I understand it may create a CAD model that does not EXACTLY match acceptable physical condition, but the differences of ten-thousandths of an inch are what I speak of.

I've not run into models from anyone that modeled everything to the max/min/mean only.

Sorry to say, we've not "switched" so I can't comment directly on your final question.

I personally find great benefit in the model "reading" the same as it would if it were a drawing. It's meant to communicate to the reader/viewer/user and to deviate from traditional nomenclature just because it's now a CAD model seems to invite error.

The model condition is just so the fabricator/inspector has a clue as to what he is starting with. As such it doesn't matter which one when considered independently of conversations with those users. For cooperative fabricators they will do the work to adapt their CAM to the model to produce a usable part. For uncooperative ones, they will demand model changes to match the flaws in their processes.

In my experience, nothing works best. It is too dependent on downstream users, spine stiffness in management, and competency of all participants.

Other than that, MBD is based on some weak goals. For some parts the complex geometry, such as for compound curved surfaces, don't lend themselves to conventional 2D depictions and MBD is a way around that. However, I don't recall anything in 14.41 defining a particular round-tripable data format, such that an engineer using software X would produce the same results as another using software Y.

What I do recall from 14.41 is substituting all the work normally done on 2D drawings into whatever proprietary and non-interoperable 3D interfaces that various software vendors might create. So as long as one stays within a software vendor's pipeline it might work or if there are side-channels to make up for the gaps in the MBD.

At one time there was an effort to embed information into STEP models that would allow an automated factory to churn out parts on demand, including understanding the variable geometry allowed for by the embedded tolerance model, but I have never heard anything come of it. At least I haven't seen any CAD modelers use STEP as their native format.

Should I understand that we cannot ""legally" (per Y14.41) model a/some feature(s) at one condition and other features at different conditions on the same solidmodel?