TPMS & Advanced Cellular Architectures
Triply periodic minimal surfaces (TPMS) — gyroid, diamond, primitive and others — offer continuous-surface cellular architectures with potential advantages in load transfer, surface area and energy absorption. This article explains TPMS conceptually, contrasts sheet-based and skeletal architectures, and warns against choosing a cell for its visual appeal rather than its structural, manufacturing and inspection behaviour.
Triply Periodic Minimal Surfaces
Triply periodic minimal surfaces (TPMS) are smooth, continuous surfaces that repeat in three independent directions and have zero mean curvature at every point. In structural engineering, they are used as the basis for cellular architectures that fill a volume with a connected network of surfaces rather than discrete struts. The most commonly referenced TPMS architectures in additive manufacturing are the gyroid, the diamond and the primitive surface, though others exist. Each has a characteristic geometry: the gyroid is a chiral surface with two interpenetrating labyrinths; the diamond is a network of tetrahedrally connected nodes; the primitive is a network of cubically connected nodes. These descriptions are conceptual; the precise geometry is defined mathematically by implicit equations. The important point for the structural engineer is that TPMS architectures offer continuous surfaces, distributed load transfer and high surface area — properties that can be advantageous for stiffness tailoring, heat exchange and energy absorption, but that must be understood structurally, not chosen visually.
COMPLEX CELL GEOMETRY IS ONLY USEFUL IF ITS STRUCTURAL, MANUFACTURING AND INSPECTION BEHAVIOUR CAN BE UNDERSTOOD.
Sheet-Based vs Skeletal Architectures
A TPMS architecture can be realised in two ways. A sheet-based (or solid-network) architecture thickens the minimal surface into a thin sheet, producing a cellular structure of connected walls separating two void networks. A skeletal (or strut-based) architecture traces the network lines of the surface into discrete struts, producing a structure that is closer to a conventional strut lattice but with the connectivity of the TPMS. The two have different structural behaviour: the sheet-based architecture distributes load through continuous surfaces and avoids the sharp nodes of a strut lattice, while the skeletal architecture concentrates load in discrete members and may be more amenable to conventional lattice analysis. The choice between them is a structural and manufacturing decision, not a matter of preference. Sheet-based architectures can be more forgiving for fatigue because they avoid node stress concentrations, but they can be harder to inspect and their wall thickness is subject to the same manufacturing limits as any thin feature.
TPMS Architectures in a Common Design Volume
The diagram below shows clean cutaway examples of three TPMS architectures — gyroid, diamond and primitive — filling an identical outer design volume. Each is shown at the same cell scale, with wall thickness and relative density annotated, and with the manufacturing build orientation indicated. The purpose is to make clear that the choice of cell architecture changes the internal geometry significantly, and that the choice must be driven by the structural, thermal and manufacturing requirements — not by the appearance of the surface.
[DIAGRAM: Three TPMS architectures shown as clean cutaway renderings within an identical rectangular design volume, side by side. Left: gyroid — a smooth, chiral, continuous surface separating two interpenetrating void networks. Centre: diamond — a network of tetrahedrally connected smooth surfaces. Right: primitive — a network of cubically connected smooth surfaces. Each is annotated with cell scale (unit cell size), wall thickness, relative density (ρ*/ρ_s) and the build orientation arrow. A callout notes that the outer design volume and the relative density are identical across the three; only the cell architecture differs.]
TPMS vs Strut Lattice
The table below contrasts TPMS and strut lattice architectures across the factors that matter to a structural engineer. Neither is universally superior. A gyroid is not universally better than a strut lattice; a strut lattice is not universally better than a solid. The right choice depends on the structural requirement, the manufacturing process, the inspection method and the qualification basis. The table is a comparison, not a ranking.
| Aspect | TPMS (sheet-based) | Strut lattice |
|---|---|---|
| Architecture | Continuous smooth surfaces; no sharp nodes; two interpenetrating void networks | Discrete struts meeting at nodes; open cellular network |
| Surface continuity | Continuous — no re-entrant corners or node junctions in the ideal geometry | Discontinuous at nodes; strut junctions are stress concentration sites |
| Load transfer | Distributed through surfaces; membrane-dominated in the sheet | Concentrated in struts; axial or bending-dominated depending on cell type |
| Manufacturing | Smooth surfaces can be self-supporting over a range of orientations; wall thickness subject to process minimum | Strut diameter subject to process minimum; nodes are prone to porosity and build defects |
| Inspection | Continuous surfaces can be difficult to inspect by conventional radiography due to overlapping walls; CT often required | Struts and nodes are more discrete; inspection is challenging but geometry is more localised |
| Fatigue behaviour | Potential advantage from the absence of sharp nodes; surface roughness and wall thinning still dominate | Node stress concentrations are primary fatigue initiation sites; node integrity is critical |
| Homogenisation approach | Representative volume element of the unit cell; effective properties derived from the sheet geometry | Representative volume element of the unit cell; effective properties derived from strut geometry and connectivity |
| Typical applications | Heat exchangers, energy absorbers, stiffness-tailored cores, multifunctional structures | Lightweight cores, structural reinforcement, energy absorption, local load distribution |
Relative Density, Printability, Fatigue and Inspection
As with any cellular architecture, the behaviour of a TPMS structure is governed by its relative density, its wall thickness and its cell size. At low relative density, the walls are thin and the structure is dominated by wall buckling and bending; at higher relative density, the structure approaches the behaviour of a solid with distributed porosity. Printability is governed by the minimum manufacturable wall thickness, which is set by the process, the material and the build orientation — the same constraint that applies to any thin feature in AM. Fatigue behaviour is governed by the surface condition of the walls (as-built roughness is a crack-initiation site), by wall thinning at high-curvature regions, and by any porosity or lack-of-fusion in the wall. Inspection is a particular challenge: the continuous, overlapping surfaces of a TPMS can defeat conventional radiography and ultrasonic inspection, and computed tomography is often the only practical method for internal inspection. The inspection method must be chosen and its capability demonstrated before the architecture is committed to a fatigue-critical application.
Choose the Cell for the Requirement, Not the Appearance
TPMS architectures are visually striking, and it is tempting to choose a gyroid or a diamond because it looks sophisticated. This is not engineering. The choice of cell architecture should follow the structural, thermal and manufacturing requirements: whether the load is distributed or concentrated, whether the application needs surface area (for heat transfer) or minimal surface (for weight), whether the inspection method can cope with continuous overlapping surfaces, and whether the manufacturing process can produce the wall thickness at the required quality. A gyroid is not universally superior to a strut lattice any more than a strut lattice is universally superior to a solid. The right answer is the one that meets the structural, manufacturing and inspection requirements at the lowest mass and risk — and that can be qualified.
THE CHOICE OF CELL ARCHITECTURE SHOULD FOLLOW THE STRUCTURAL, THERMAL AND MANUFACTURING REQUIREMENTS — NOT THE VISUAL APPEAL OF THE GEOMETRY. A gyroid is not universally superior to a strut lattice any more than a strut lattice is universally superior to a solid.