wrote in message
Can anyone provide any information on the accuracy with which Pro-E or other CAD programs calculate
the volume and/or surface area of complex solid objects?
Also, is it possible to increase the number of displayed digits of these calculations?
For displayed dimensions, there's a lot of rounding/truncating going on. This is controlled, as a
practical consideration, by configuration options, generally for display or manufacturing purposes.
In the background, most CAD programs calculate to 10 decimal precision. If you've ever used a CAM
package and looked at a tool path, the unfiltered numbers for x,y,z coordinates are to that level of
precision. Obviously, the machine controller has to do its own rounding/truncating as machine
precision is only to 4-5 decimal places. But it shows that the program is calculating to a far
higher level of precision, precisely to avoid accuracy issues (errors in position, size and shape of
tiny features on huge bodies).
On the other hand, to save memory and computing resources, Pro/e allows artificially limiting
precision with an accuracy value called "part accuracy" that can generate an error if you exceed its
ability to accurately calculate a small feature's size and/or position. In that case, it needs to be
made finer to give the program greater precision in doing its calculations. BTW, Pro/e is always
looking at feature size/position in relation to overall part size which is the longest diagonal that
can be drawn through a box completely surrounding the part. And then it's just a ratio of the
feature dimensions divided by that diagonal. When that turns out to be smaller than the set or
default part accuracy, an error is generated and the feature fails. In fact, I've had features fail
long after they were created because the part grew significantly in one or more directions.
In any case, there are lots of examples of the troublesomeness Pro/e's dependency on accuracy,
particularly part accuracy. One remedy is the use of absolute accuracy which skips the bounding box
of the part and uses the bounding box of the default datums, Pro/e "world size" of over 600x600x600
units and its bounding box which is a fixed, "absolute" value. Then the question of accuracy boils
down to what is the smallest feature you might or typically do create and setting absolute accuracy
as the ratio of that smallest size divided by the world size diagonal, say 1000 units. Then you set
the default absolute accuracy to that value and accuracy lower boundary to say another decimal of
precision, giving it some computational overhead. Once done, all parts are equal and all features
live and die by the test of world size. This will go a long way to eliminating feature failures,
merge/cutout failures and problems with import failures.