[ATM] Machine polishing simulator, looking for help

[wa4guu at verizon.net (Jerry)]

Not to be a stick in the mud, but this is a project that I believe has
little chance of success. An "opinion" of course. I'll cite a few reasons I
think it is beyond amateur implementation.

There are just too many variables and the best machine and even the best
human cannot control them all. And the ones that can be controlled can't be
controlled absolutely. There are a number of factors in the figuring process
that either can't be accurately quantified or do not remain at a single
quantity throughout the process.

The location of wear on the glass is not just a function of a stroke and the
location of the stroke. The present shape on the glass will have an effect
on the location of wear. Picture that low zone at 50% the lap passing over
will be supported by the highs on either side and wear will be little on the
low. The shape of the glass and the shape of the lap at every instant of the
process would have to be known as would the water and polishing agent
concentration at every point of contact between lap and glass.

If you use sensors to measure friction you would not know the component each
location on the lap is contributing to the total. Are you going to put
numerous sensors in your lap?

There are some results of figuring that will go so far and then stop going
that way even if the same stroke is continued. Just as when you are
polishing and using the same stroke the entire time. The mirror takes up a
shape, not awful far from a sphere (if your are doing it right). I guess for
most people it is a little oblate. You can continue that stroke for hours
and the shape changes little. If the mirror is not oblate and you do that
stroke you will see movement toward oblate. But then when it has gotten as
far oblate as it is going with that stroke it stops moving. What this will
mean in your wear modeling method is that under the same conditions you have
two different wear patterns. One that wore glass away in selected zones
(making the oblate) and then even wear in all zones (maintaining the same
oblate). I don't know how that can be handled by a direct "forces applied
equal this wear pattern" calculating program. And there are the fast mirrors
that the "normal" parabolizing stroke only gets part the way to the parabola
and then one must force it farther by accented pressure of sub diameter lap
or other forceful means.

I have more but that is enough for a start.

Now another opinion of mine on the subject is that an approach that might
work would be rather than the direct physics type calculation of the forces
and moments to predict outcome, a program that "learns" from experience
might be easier. But then isn't that what humans do?


Jerry