PYTHON · NUMPY

Designing for the process (DFM)

Modify the part's geometry to remove ≥2 flagged DFM violations while preserving its functional dimensions, and verify the inspector now passes.

01Challenge

You found these problems in 1.1 and chose molding in 1.2. Now fix at least 2 of the 3 violations by editing the part's shape parameters, without touching the locked functional dimensions (BORE_D 12.00±0.05, HOLE_SPACING 30.00, PLATE_THK ≥ 6.00). Change a number, re-run the inspector, watch violations clear. You haven't been told the target values, so adjust and converge.

02Model

A manufacturable part isn't the designer winning or the process winning, it's a negotiation. The part wants its function; the process has hard limits. Good DFM finds the small changes that give the process what it needs while the part keeps what it needs.

Draft: tilt the walls one degree, almost invisible, but now the mold lets go, and the bore never moved. Wall: thicken it to the minimum so the cavity fills, growing it outward so the Ø12 bore stays exactly 12 mm. Fillet: round the sharp inside corners so plastic flows and stress doesn't pile up. Three tiny changes, and an impossible part becomes a cheap one, with function locked.

Reshape features so the process can make them.

03Guided practice

Pass A (worked): set DRAFT_DEG = 1.5 (was 0.0). The bore is generated separately and stays cylindrical, so 'no draft' clears with function intact.
Pass B (hint): molding wants wall ≥ 1.50 mm; you have 0.80. Raise WALL but grow it OUTWARD (boss radius = BORE_D/2 + WALL) so BORE_D stays 12.00.
Pass C (independent): add FILLET_R ≥ 0.50 but smaller than the wall; too large eats the bore and throws a functional violation.

04Feedback

PASS WHEN ≥2 DFM violations cleared for the chosen process AND every functional dimension preserved (BORE_D 12.00±0.05, HOLE_SPACING 30.00, PLATE_THK ≥6.00) AND the solid regenerates, on seed 1103.

If your run fails, check:

Both DFM violations cleared, but BORE_D regenerated at 11.4 mm (target 12.00±0.05): you thickened the wall inward and ate the bore. Grow it outward (boss radius = BORE_D/2 + WALL).
Draft is now 0.50°, better, but molding needs ≥1.00° to release cleanly. Raise DRAFT_DEG and re-run.
1 of 3 cleared (draft). Need ≥2. The thin wall is still flagged at 0.80 mm: raise WALL to ≥1.50 mm (outward).
The edited script didn't build: FILLET_R = 7.0 is larger than the wall it's filleting, so the solid self-intersects. Choose a fillet ≥0.50 but smaller than the wall.

05Retrieve & space

From 1.1: which original defect did fixing not require here, and which 1.2 process would make it matter? → the unreachable rib matters for CNC reach, not molding.
From 1.2: if you'd chosen CNC instead of molding, would draft still be required? → no; draft is a molding/casting release rule.
From Design course: make a one-line edit (WALL = 2.0) and predict before re-running whether the bore stays in tolerance → yes (outward growth).

06Mastery & project

In sim, the edited part clears ≥2 DFM violations for the chosen process AND preserves every functional dimension AND regenerates as a valid solid, on seed 1103: the L0→L1 module gate.

Outputs the actual buildable part bracket_v3 (Ø12.00 bore intact) that the entire rest of the course manufactures: M2 toolpaths it, M3 tolerances the same bore, and M5's yield depends on the draft/walls/fillets negotiated here.

GO FURTHER · THE LIVE PLAYGROUND

See DfAM in a real printed body ↗ — a fully-3D-printed humanoid, manufacturing Physical AI for real

← Choosing a material and a process From geometry to toolpath →

PYTHON · NUMPY

Designing for the process (DFM)

In the full Bench this regenerates a build123d solid; here a numpy model checks DFM thresholds AND functional-dimension preservation. Edit the shape params and run.

You read

the arrays and values already in scope

You change

the code you write in each cell

Fixed

the dataset and the checks that grade you