Shell Fails on Corners That Already Have a Fillet
A 3 mm wall ABS housing cover, 160 × 90 × 25 mm deep, is modeled as a solid block. The modeler fillets the four vertical outside corner edges at R2, then removes the open face with a Shell of 3 mm. The fillet builds. The shell fails with 'the shell operation could not be completed' and the preview highlights one corner. Reversing the order - shell first at 3 mm, then fillet the four outer vertical edges at R2 - builds successfully. A colleague proposes solving the original order by dropping the wall to 2.5 mm and keeping R2, arguing 'it's only 0.5 mm'. The part is injection molded and the cosmetic outer corner radius is fixed at R2 by the industrial designer.
1. Explain why the 3 mm shell fails on the R2 corners, with the arithmetic, and give the minimum outer radius for a 3 mm wall if the inner corner radius must be at least 1 mm.
Read the worked answer
A shell offsets every face inward by the wall thickness. On a convex corner fillet of radius R, the offset surface has radius R − t.
R = 2, t = 3→ inner radius =2 − 3 = −1 mm.- A negative radius is geometrically meaningless: the offset surface would have to pass through itself and invert. The kernel aborts. The failure is not a tolerance issue or a bad edge pick - the requested geometry does not exist.
Requirement: R > t. For an inner corner radius of at least 1 mm: R ≥ t + 1 = 4 mm. Many shops prefer an inner radius of about half the wall (1.5 mm here), which pushes the outer radius to 3 + 1.5 = 4.5 mm, and 5 mm is the practical round number.
Note that R = t gives an inner radius of 0 - a knife edge. That builds, but it is a zero-thickness sliver and a molding defect, so 'it builds' is not the same as 'it is correct'. Always check the value you are about to accept, not just whether the feature turns green.
2. The colleague proposes dropping the wall to 2.5 mm to keep R2. Does that fix it? What about 2.0 mm?
Read the worked answer
No, and no - but the two fail differently, which is the point.
t = 2.5, R = 2→ offset radius =2 − 2.5 = −0.5 mm. Still negative. The shell still fails. The 0.5 mm reduction bought nothing because the offset only becomes legal once t is strictly below R.t = 2.0, R = 2→ offset radius =0. The shell now builds, which looks like success, but it produces a zero-radius inner corner: a knife edge where two wall faces meet with no thickness. In a molded part that is a stress riser and a short-shot/sink-mark site; in CAD it is degenerate geometry that often exports badly to STEP and breaks downstream filleting and drafting edge selection.
So the answer is not a wall reduction. The design constraint that has to move is the outer radius (or the wall), and since the designer has fixed R2 cosmetically, the correct technical response is to push back with the numbers: 'R2 with a 3 mm wall cannot be shelled; either the outer radius goes to 5 mm, or the wall goes to 2 mm with an explicitly modeled inner radius.' Give the designer a buildable pair of numbers rather than a failed feature.
3. You reverse the order: shell the solid at 3 mm, then fillet the four outer vertical edges at R2. The assembly of features builds. Describe what happens to the wall at the corners and state which order you would ship, and why.
Read the worked answer
Shell-then-fillet does not reproduce a 3 mm wall at the corner. The shell creates a sharp inner corner; the L2 fillet is then cut into the outer edge, and at the 45° diagonal the fillet arc reaches past the inner corner. The result is a corner that is locally far thinner than 3 mm - with R2 on a 3 mm wall it is under a millimetre and may break through altogether. Do not take that on trust: cut a section view through the corner and measure it. The feature building is not evidence that the wall is correct.
Which order to ship: neither, as specified. For an injection-molded part the correct order is fillet-before-shell with R ≥ t, so the wall is genuinely uniform around the corner and the inner radius falls out as R − t. That is the order the molder's draft analysis and cooling assumptions depend on.
Use shell-then-fillet only when you want the outer edge rounded on a part where corner wall thickness is not critical and you have verified the remaining thickness with a section. The exam-level judgment is that feature order is a design decision, not a convenience: changing order changes the geometry, even when both orders build.