SolidWorks Mythbuster: Geometry Pattern

The Geometry Pattern option in feature patterns has been misunderstood from the very beginning. When I was a budding reseller application engineer, I remember SolidWorks tech support telling me that Geometry Pattern was intended for those people who made huge patterns of punched holes in sheet metal. The patterns were dead slow, and Geometry Pattern was supposed to help. Feature Statistics didn’t exist (for users) back then, and the patterns were so slow, with and without the option that no one thought to actually test the theory. It sounded plausible,and it was something everyone hoped was true, but it turned out to be incorrect.

There was another effect of the Geometry Pattern that some people noticed. When the feature you were patterning had an end condition that changed the feature, like the hole in the transparent part shown here has an Offset From Surface end condition, Geometry Pattern ignored the end condition, and just patterned the faces of the feature as they were. I guess we just thought this was a secondary function, but it turns out to be the primary function.

The truth about this feature is that the only time GP speeds up patterns is when there is a variable end condition involved, such as Offset From Surface. For regular Through All or Blind feature patterns, GP actually slows it down.

GP also has one unwanted side effect that the Help actually mentions. If the feature that you are patterning has a face that merges with a face that doesn’t belong to the feature, that may prevent the pattern from working.

For example, the one on the left will not work with GP, but the one on the right will. That is because of which face is merged. If the merged face can be merged in all pattern instances, then GP will work.

For details of testing done to determine speeds of pattern settings, refer to the SolidWorks Bible chapter on Patterning (Chapter 8), all versions have this information. For additional information, check out SoliDan’s blog. Commenter Charles Culp reminded me of this articlewritten by Dan Bertschi where Dan comes to the same results, but adds to it information about mirroring patterns that use GP, which in Dan’s test appear to be faster.

This chart also compares sketch patterns. But then anybody reading this blog knows better than to use a sketch pattern, right?

Verdict? The common belief has been that Geometry Pattern speeds up large patterns. The truth is that Geometry Pattern is slower for most pattern types, and its primary use is to pattern features with variable end conditions without using the end conditions.