AI-Augmented Concept Exploration Software for Mechanical and Thermal Parts

Cognitive Design offers three powerful generative design methods, each built for a different engineering intent:

• Topology Optimisation. The classic. Get the lightest, stiffest geometry the load path allows. Level-set engine with full control over volume fraction, refinement, boundary evolution, smoothing, and preservation zones.
• Topology Weaving. Material follows the stress. Concept-stage skeletons generated along load paths from any FRep field, at a fraction of the cost of full topology optimisation.
• Topology Enclosure. Conformal shells around any functional volume. Ideal for housings, covers, and protective structures.

In every case, the MDD module actively shapes the geometry during generation according to DfAM and DfM rules:

• Additive manufacturing: overhang correction, minimum wall thickness, self-supporting geometry, support minimisation.
• Die casting: draft angles, undercut detection, parting line placement, mold geometry generation.
• 5-axis machining: tool accessibility verified during generation.
• Injection molding: uniform wall thickness, draft compliance.

Cognitive Design is the only platform that can run the same part through multiple manufacturing processes simultaneously, comparing AM, die casting, and machining side by side on cost, mass, and manufacturability from the first generation run.

Lattices are where additive manufacturing reveals its full potential, and where Cognitive Design shines.

The platform generates TPMS structures (Gyroid, Schwartz P, Diamond) and lattice geometries natively, with variable density driven by simulation field outputs: denser where stress is high, lighter where it is low. Performance optimised distributions, not uniform fillings.

Because all lattice geometry is represented as implicit mathematical fields, not meshes:

• Smooth, continuous transitions between solid and lattice regions
• File sizes that stay manageable, regardless of complexity
• Density, thickness, and cell type are controlled in 3D space by stress, temperature, or any custom field
• TPMS structures with maximised surface area, ideal for heat exchangers, battery cooling plates, and lightweight aerospace structures

Structural simulation uses the Immersed Method of Moments, a meshless FEA approach that operates on implicit geometry without requiring a surface or volume mesh. This removes the meshing bottleneck entirely and allows simulation to run continuously during generation.

Mechanical, thermal and thermo-mechanical simulations are native in the same workflow: Von Mises stress, displacement, temperature fields, and heat flux. Simulation results can drive geometry adaptation in real time: lattice density, shell thickness, and ribbing distribution all respond to the computed field using the Simulation Driven Design method.

When the first generation pass is not enough, SDD refines geometry directly from simulation results. Run an analysis, extract a field (stress, displacement, temperature), feed it back into the design: lattice density, shell thickness, and ribbing adapt. The geometry literally responds to physics. Ideal for parts where every gram or every degree counts.

Geometry is defined as mathematical fields, not discrete meshes. Density, thickness, lattice type, and surface curvature can be driven continuously by any scalar or vector field: stress, temperature, displacement, or custom distributions. Geometry that responds to physics at every point. This is the foundation of TPMS, variable-density lattice, stress-guided Topology Weaving, and SDD.

Optimised geometries produced by topology optimisation or lattice design are inherently mesh-based outputs. Cognitive Design reconstructs these automatically into BRep geometry and exports clean STEP or IGES files, ready for detail design, tolerance specification, and manufacturing preparation in any CAD environment.

No manual surface reconstruction. No third-party conversion tool required. This closes the loop between concept exploration and the downstream CAD chain without any manual intervention.

A single dashboard consolidates mass, Von Mises stress, displacement, safety factor, cost per manufacturing process, CO2, and manufacturability score across all generated variants. Every design decision is timestamped with its generation parameters, solver inputs, and KPI outputs. Comparison is visual and tabular. The first export from the Design Explorer is already a complete, traceable design selection record.

Every successful exploration can be saved as a parametric template: rules, constraints, solver logic, and manufacturing settings encoded once, applied to subsequent parts by changing inputs only. The next bracket in a part family takes hours instead of weeks. Methodology grows with each program instead of resetting with each hire.