A design methodology where algorithms, driven by engineering constraints, loads and manufacturing rules, automatically explore and generate multiple geometry candidates, with the engineer selecting among results.
Generative design (GD) extends topology optimization by running many parallel explorations with different objective trade-offs (minimize mass, maximize stiffness, minimize cost) and manufacturing constraints (overhang limits, minimum feature size, symmetry). Platforms such as nTopology use field-driven implicit modeling: a scalar field (e.g., von Mises stress from FEA) drives strut diameter, wall thickness or lattice density continuously across the part.
GD is most effective for highly loaded, geometrically unconstrained parts: aircraft seat brackets (Airbus A320 test showed 45% mass reduction), prosthetic socket custom inserts, robot end-effectors and heat-sink geometries. The workflow is: set loads + keep-out zones + AM constraints → algorithm generates N solutions → engineer picks one → validate + print.
GD outputs assume ideal AM conditions: always validate with actual process rules (wall min, overhang angle) before accepting a candidate. Multiple equally valid solutions can confuse teams without clear selection criteria. Computational time scales with problem complexity; large design spaces may need hours on a workstation.
Related terms: Topology Optimization, Lattice & Gyroid Structures, Lightweighting / Mass Reduction
| Theme | optimization |
|---|---|
| Also called | AI-driven design, algorithmic design, field-driven design |
| Source | Wiki/concepts/course-ntopology.md, Wiki/concepts/dfam.md |
Engineer, author of The Big Book of 3D Printing and additive manufacturing expert