I reply here to a very recurrent question about usability of models and the alternative to translating features.

A customer asks :

I am worried about the construction and use of the models.

In CATIA we have a hierarchy of elements, the parents of which are normally solids or volumes, these elements can be fully manipulated if fully translated into Solid works, in this example we have only surfaces, faces, and basic elements therefore cannot add/subtract solids

First of all SolidWorks have now several options to manipulate surfaces directly. However they may not be first class citizen in the menus. This could be the subject of a training and could be followed by formalizing best practices to deal with this to be feed into your internal HowTo’s database.

Second, all the mentioned solids, and wires can be translated into SolidWorks as well. This is controlled by the following options:

• GetMaskedEntities (default is false)  ( “Import Hidden Entities” in dialog box )
• ReadIsolatedCurves (default is false) (not available in dialog box)
• ReadWireframeIFF  (default is false) (not available in dialog box) force to read wires if no solid is found.
• ImportIsolatedSurfaces (default is true) (“Import Isolated Surfaces” in dialog box)

These options can be defined as a policy in an xml setting file near the files to be translated. Like this it’s possible to define options based on what is in the file. Some parts can be translated as small as possible (typically light group models or fasteners) while others like surface design can be translated with all their intermediate constructions and tools.

Now in the part considered most solids are built as sweep of curves sets. As intuitive as it looks the sweep surface construction is far from obvious. In the early days (and CATIA V4 was there) sweep were typically implemented as a product of poles and weights. Very easy to implement, looks good enough, but not good for anything because shapes have a tendency to be flattened in the centre. Modern implementations are done with multiple cross section generators followed with an interpolation. How the sections are generated and how the interpolation works is really a feature of the software. As opposed to Nurbs, cylinders, cones or torus there is no common ground for this computation. So if you rebuild a sweep in SolidWorks with the same base curves you may not lead to the same exact surface. Usually implementators are kind enough to enforce end conditions. So if you start with a section and end with another, the sections are part of the resulting surface. But this is not true for sections anywhere in between. So if you used booleans, trim or fillets to connect the surfaces and redo the same work based on newly generated shapes you may end up with a slightly different result. You may accept this deviation or not, depending on where on the design chain you sit. If you are a subcontracting or manufacturer you can’t, if you are the original designer you can, if you are designing a replacement of the part you can’t.

This is among the reasons that makes feature based translation inapplicable to many situations and is in favour of translating models as dumb solids, provided the translator includes a post translation deviation check as RadialSoft translators offer.

Establishing corresponding good practices on SolidWorks leading to surfaces as similar as possible could be the subject of a training with lessons learned extracted at the end.

Correspondingly it is likely that the same design intend would be done totally differently using SolidWorks. This could be explored as well in a dedicated training.