|
Home
Overview
Projects
Publications
Presentations
People
Research
Opportunities
Courses
Sponsors
Contact
Us
|
 |

PROJECT |
|
|
Parametric Solid Modeling
|
|
Limitations in Current CAD Systems
|
| |
|
Overview |
Current CAD systems for part modeling
exhibit severe limitations as far as parametric modeling
is concerned. Not only are the parametric
definitions inconsistent in different modeling
systems, but also the algorithms for constructing and
maintaining a part family are usually ill-defined and
ambiguous within any single system. Users have
little (if any) control over the behaviour of a part
family, once a nominal or generic part from that family
has been constructed and parametrized. The following
examples should help to illustrate these inconsistencies
and ambiguities. These examples were constructed in
the several recent versions and types of the more popular
commercial CAD systems.
|
|
| |
|
Inconsistencies in parametric updates |
| As a first example consider the behaviour
of a simple part constructed in one system as shown
in figure 1(a) below. It has two parameters d and t
(other parameters are not shown here for the sake of
clarity). Let us see how this part shape behaves for
changes in the value of parameter t. In particular
consider a simple parametric change which when executed
should result in the solid shown in figure 1(b). In
this system we first created the solid in figure 1(a)
as a Base (rectangular block) in which a Hole (big cylinder)
is made and then a Cut (small cylinder) is made. The
position (controlled by the parameter t) of the hole
dependent on the base and the position of the cut (controlled
by the parameter d) is in turn dependent on the edge
e1 which is an intersection edge of the base with the
hole. When the parameter value of t is edited so that
it now corresponds to that of the solid shown in figure
1(b) (essentially collapsing the face f1 and merging
the edges e1 and e3), what results in this system is
the solid shown in figure 1(c), with the cut jumping
off the the solid. When the solid in figure 1(a)
was constructed as a Base-Cut-Hole combination and the
same parametric edit executed, the resulting solid is
shown in figure 1(d), with the hole jumping on the same
face f2' but refering to the wrong edge. Finally when
the original solid was created using a Base-Cut-Cut
combination and the same parametric edit executed it
resulted in an error message from the system indicating
that the parametric edit could not be executed. Thus
for the same solid and the same parametric edit, even
though a correct update clearly exists (as shown in
figure 1(b)), this one system produces three different
results depending on the combination of the type of
features used to construct the solid.
Figure 1 |
|
| |
|
Ambiguity in updates |
| As another example consider the following
part which is constructed in another system as an union
of a block and a cylinder, with the parameter t controlling
the position of the cylinder with respect to the block.
When the parameter value of t is changed to 0, thus
translating the cylinder to the near end, the blend
jumps from one side of the face f1 to the other. Should
this parametric update be considered valid? If one considers
this parametric edit as a continuous change then the
face f3 continuously deforms taking along with it the
adjacent edges and vertices. Note however there are
many continuous transforms (such as a rotation about
x-axis by 180 degs or a rotation about the z-axis by
90 degs) between the two solids due to the symmetry,
but these transforms do not correspond to the semantics
of the parametric change (which will require that the
face f1 map to f1' for instance).
Figure 2
Finally consider the following part shown
in figure 3(a) below. It was constructed as a union
of a small block over a large block. Subsequently a
hole is made on the small block. This parametric solid
was edited (which corresponded to changing the values
of the parameters p1 and p2 to 0) in two different CAD
systems. The resulting solid in the first system is
shown in figure 3(b) (which is again a disconnected
solid) and the result solid in the second system is
shown in figure 3(c). Neither of the resulting solids corresponds to the intuitive result that one would expect.
Figure 3
|
|
| |
|
Conclusion |
| These examples clearly demonstrate
that the semantics of a parametric edit is ambiguous
and inconsistent both within and also between any two
systems. The inability to handle correct updates even
when one exists makes this situation impalatable.
|
|
|
|