Adaptive Sketch Lost After Parameter Change
You are refining a bracket sketch that started from a rectangular plate profile. The sketch has 14 dimensions and 22 geometric constraints. A drafter added a 3 mm fillet at each of the four corners, then changed the plate width parameter from 120 mm to 90 mm. Inventor now reports the sketch as fully constrained but the corner fillets have become tangent to each other and the sketch profile no longer closes. Undo is not available because the file was saved and reopened. The part must still be 90 mm wide, and the original 120 mm version must remain recoverable for a variant. You have about 20 minutes before the review meeting.
1. What is the most likely cause of the non-closing profile, and how do you diagnose it without deleting constraints?
Read the worked answer
The most likely cause is that the corner fillets were dimensioned or constrained relative to the overall width rather than to a local, stable reference. When width dropped from 120 mm to 90 mm, the two fillets on each short edge were driven toward each other until their tangent arcs overlapped or collapsed. Because tangency and equal-radius constraints still hold, the sketch can report fully constrained while the loop is self-intersecting; fully constrained does not mean geometrically valid.
Diagnosis order: first use Sketch Doctor or the profile error highlight to find which loop fails. Then drag-test each dimension by selecting it and watching which geometry moves; a dimension that moves several unrelated entities is a suspect. Turn on constraint visibility and inspect the corner region for stacked tangent and coincident constraints. Check the dimension display precision and look for driven (reference) dimensions that were meant to be driving. Do not delete constraints yet: suppress or temporarily convert suspect dimensions to reference to see whether the loop closes, then restore intent deliberately.
2. How do you preserve the 120 mm variant while delivering the 90 mm part, and what modeling change prevents recurrence?
Read the worked answer
Preserve the variant with a derived or iPart/iAssembly approach rather than by editing the same sketch back and forth. Concretely: save the current 120 mm file as the master, then create the 90 mm version as a derived part (Derive Component) or as an iPart member with the width as a table-driven parameter. The sketch should be driven by a named user parameter (for example PlateWidth) so the table controls it, not a hard-coded dimension.
To prevent recurrence, rebuild the corner treatment so it is not width-dependent: constrain fillet centers to the corner with a fixed offset from each edge, or move the fillets to a feature-level Fillet after extrusion, where a 3 mm radius on a 90 mm edge is geometrically safe. If the sketch must keep the fillets, add a dimension that ties radius to width with a rule or equation, and add a sanity check dimension for the flat length between fillets. Test both widths before saving.