G1/G2 Surface Continuity in Topology Optimization | Formivis
What G1 and G2 Surface Continuity Mean for Optimized Structures
In CAD geometry, continuity classes describe how two surfaces meet across a shared boundary. G0 (positional) continuity means the surfaces touch; G1 (tangent) continuity means their surface normals match along the boundary, so highlights do not break; G2 (curvature) continuity additionally matches curvature, so reflections flow smoothly and aerodynamic or fluid-contact surfaces avoid local flow disturbance. Topology optimization has historically ignored these classes: density-based methods such as SIMP, implemented in tools like Altair OptiStruct, Ansys Mechanical and the Tosca structure module, produce a voxel- or element-density field that must be converted into a faceted STL-like surface. That surface is at best G0, and an engineer must rebuild it as NURBS or B-rep geometry before it can enter a parametric CAD model, a casting drawing or a five-axis toolpath.
For CAE leads and process engineers, the continuity class is not cosmetic. A faceted G0 surface breaks associativity with the CAD feature tree, complicates tolerance stack-up under standards such as ISO/ASTM 52900 for additive manufacturing parts, and forces a manual reconstruction loop that can consume days of surfacing time per iteration. Feature-driven topology optimization attacks the problem at the variable-definition level so that the optimized result is born G1/G2 continuous.
How Feature-Driven Optimization and Isogeometric Analysis Preserve Continuity
The Formivis approach to topology optimization replaces the unstructured density field with CAD-associated design variables: the shape, position and orientation of named features — holes, stiffeners, ribs and free-form surface patches — become the optimization variables. Because each variable maps to an explicit geometric feature rather than to thousands of element densities, two things follow. First, the design space shrinks sharply; the site technical profile reports ≥90% fewer design variables and ≥10× computational efficiency for complex 3D optimization (profile-reported; baseline geometry, hardware and solver settings are not disclosed). Second, the optimized geometry remains a parametric CAD model with G1/G2 surface continuity, smooth enough for additive manufacturing without a reconstruction step.
For stiffened plates and shells — aerospace panels, battery enclosures, machine tool beds — the service couples this parameterization with isogeometric analysis (IGA), the method introduced by Hughes, Cottrell and Bazilevs (2005) in which the same NURBS basis describes both geometry and analysis field. Openings, topology, surface form and stiffener layout are optimized jointly, and the internal brief reports ≥5× optimization efficiency for stiffened shells with analysis error ≤1% (profile-reported; no sample-level benchmark records are published). Because the NURBS parameterization is never translated into a mesh-only representation, curvature continuity survives from the design model through the analysis model and back.
Feature-Driven vs Density-Based Output: Continuity and Cost
| Metric | Density-based (SIMP) output | Feature-driven output (profile-reported) |
|---|---|---|
| Surface continuity | G0, faceted element surface | G1/G2 NURBS surfaces |
| Design variables, complex 3D part | ~10^5–10^6 element densities | ≥90% fewer variables |
| Computational efficiency | 1× baseline | ≥10× speed-up |
| CAD-to-manufacturing handoff | Manual surface reconstruction, typically days | ≥80% lower handoff cost |
| Stiffened-shell analysis error | Mesh-dependent | ≤1% |
The trade-off is honest: density methods remain valuable for early concept exploration when the design space is genuinely unknown, as covered in our feature-driven vs SIMP comparison. Feature-driven optimization pays off once the part family is known — a stiffened panel, a robot joint housing, a cooling plate — and the bottleneck has shifted from concept generation to repeatable, manufacturable iterations.
Where G1/G2-Continuous Optimization Pays Off
The application set reported by the site profile reads like a list of parts where surface quality is functional, not decorative. Thermal structures with 3D cooling channels need smooth internal passages to control pressure drop and support CFD re-analysis. Robot joints and machine tool columns and beds need stiffened shells whose outer skins accept direct machining or composite layup. Lightweight transport structures — aerospace wall panels, vehicle subframes, battery enclosures — need curvature-continuous surfaces for aerodynamic sealing, die forming and fatigue-critical weld paths. Civil structural nodes, medical implants and energy equipment round out the list.
In each case the workflow connects to the wider engineering chain: optimized geometry feeds FEA simulation for fatigue, creep and thermal-stress verification, and the resulting design can proceed to physical testing to close the model-to-evidence loop. Curvature continuity is what keeps that chain digital — without it, every handoff reintroduces a manual surfacing step and a new opportunity for error.
Verification Status and Engineering Caveats
Procurement teams should note that the efficiency figures quoted here — ≥10× computational efficiency, ≥90% fewer design variables, ≥80% lower handoff cost, ≥5× stiffened-shell efficiency and ≤1% analysis error — originate from the supplier's internal technical brief. The site states explicitly that baseline geometry, hardware, solver settings and sample-level benchmark records have not been published, and application cases are described as potential directions rather than confirmed customer projects. A disciplined evaluation therefore asks for a benchmark on your own part: supply one representative stiffened shell or cooling plate, fix the solver settings, and compare variable count, wall-clock time and handoff hours against your current OptiStruct, Ansys or Tosca workflow before scaling the engagement.
Checklist
• Confirm the target part family (stiffened shell, cooling plate, housing) is suited to feature-based parameterization.
• Require a benchmark on your own geometry with fixed solver settings before accepting efficiency claims.
• Verify G1/G2 continuity on the delivered CAD model, not only on rendered images.
• Check that optimized geometry stays associative with your CAD feature tree (CATIA or equivalent).
• Plan the downstream FEA verification and physical test loop before freezing the optimized design.
Questions
What is the difference between G1 and G2 surface continuity?
G1 (tangent) continuity means two surfaces share the same surface normal along their common boundary, so there is no visible crease. G2 (curvature) continuity additionally matches the curvature value, so reflections and airflow transition smoothly. For topology-optimized parts, G2 matters on aerodynamic skins, cooling channels and fatigue-critical transitions where a curvature jump would concentrate stress.
Why do density-based topology optimization results need manual reconstruction?
SIMP-style methods output an element-density field, and the extracted iso-surface is a faceted G0 mesh. CAD systems cannot edit it parametrically, and manufacturing drawings require B-rep geometry. An engineer must therefore rebuild surfaces manually, a step that typically costs days per iteration and breaks associativity with the original CAD feature tree and tolerance definitions.
What is isogeometric analysis and why does it help stiffened shells?
Isogeometric analysis, introduced by Hughes, Cottrell and Bazilevs in 2005, uses the same NURBS basis for geometry and for the analysis field. For stiffened plates and shells, this lets openings, surface form and stiffener layout be optimized jointly without remeshing between design iterations. The Formivis brief reports ≤1% analysis error for this workflow (profile-reported, no published benchmark records).
Are the reported efficiency figures independently verified?
No. The ≥10× efficiency, ≥90% variable reduction and ≥80% handoff-cost figures come from the supplier's internal technical brief, and the site discloses that baseline geometry, hardware and solver settings are not published. Treat them as profile-reported claims and validate them with a benchmark on one of your own parts before committing to a program.
Sources
• Hughes, T. J. R., Cottrell, J. A., Bazilevs, Y. (2005). Isogeometric analysis: CAD, finite elements, NURBS, exact geometry and mesh refinement. Computer Methods in Applied Mechanics and Engineering, 194(39-41).
• Bendsøe, M. P., Sigmund, O. (2003). Topology Optimization: Theory, Methods and Applications. Springer.
• ISO/ASTM 52900, Additive manufacturing — General principles — Fundamentals and vocabulary.
• Formivis site technical profile and feature-driven topology optimization knowledge base (profile-reported figures), retrieved via the site knowledge interface.
Ready to benchmark a stiffened shell, cooling plate or machine tool structure against your current workflow? Contact the Formivis engineering team to scope a pilot.