Why Triangles? Stress, Topology & 3D-Printed Cast Geometry
TL;DR: For 150 years, orthopedic immobilization has relied on heavy, unventilated plaster casts that cause skin maceration and stiffness. In 2014, our CUHK team built an automated 3D-printed cast pipeline using Voronoi meshing, sparking a fundamental question: would a triangular truss lattice have provided superior mechanical rigidity and anatomical fit? In this deep dive, we explore Maxwell's rigidity criterion, surface Gaussian curvature, and a decade of empirical FEA literature—showing that while triangles guarantee gapless conformal tiling and isotropic shear stability, hexagonal honeycombs excel under compression, and topology optimization beats all fixed geometries. Interactive 2D lattice FEA simulators and tiling sandboxes are provided below.
"A triangle is the only geometric polygon that cannot be deformed without altering the length of its edges. That single invariant anchors bridges, geodesic domes, and modern biomechanical lattices."
1. The Orthopedic Bottleneck & The 2014 Grant Pipeline
Traditional orthopedic plaster casts (invented by Antonius Mathijsen in 1851[^1]) have remained virtually unchanged for over a century and a half. While effective at gross bone immobilization, continuous circumferential plaster suffers from critical clinical drawbacks: - Cutaneous Complications: Trapped sweat and humidity lead to skin maceration, severe pruritus (itching), bacterial and fungal colonization, and contact dermatitis[^2]. - Weight and Stiffness: Bulky plaster and fiberglass restrict natural muscle twitching, accelerating disuse muscle atrophy. - X-Ray Attenuation: Serial radiographs require higher radiation doses or cast removal to monitor callus formation.
Between 2013 and 2015 at the Chinese University of Hong Kong (CUHK), our team—Defeng Wang, Lin Shi, and myself—secured an Innovation and Technology Fund (ITF) grant to replace plaster with an automated scan-to-print personalized orthopedic cast pipeline[^3].
flowchart LR
Scan["1. 3D Scan (CT / MRI / Optical)"] --> Dilate["2. Anatomical Dilation (+5% Clearance)"]
Dilate --> Decimate["3. Surface Mesh Optimization"]
Decimate --> Inset["4. Face Insetting & Strut Generation"]
Inset --> Shell["5. Shell Offset & Edge Filleting"]
Shell --> Print["6. Additive Manufacturing (SLS / FDM)"]
style Scan fill:#1e40af,stroke:#3b82f6,stroke-width:2px,color:#fff
style Dilate fill:#0284c7,stroke:#0369a1,stroke-width:2px,color:#fff
style Decimate fill:#7c3aed,stroke:#8b5cf6,stroke-width:2px,color:#fff
style Inset fill:#d97706,stroke:#f59e0b,stroke-width:2px,color:#fff
style Shell fill:#059669,stroke:#10b981,stroke-width:2px,color:#fff
style Print fill:#dc2626,stroke:#ef4444,stroke-width:2px,color:#fffIn our original pipeline, the mesh decimation tools (such as 3ds Max's ProOptimizer) generated an irregular Voronoi-style polygon mesh by default. This raised an intriguing fundamental question: Was this organic Voronoi mesh mechanically optimal, or would a purely triangulated lattice have delivered superior performance?
2. First Principles: Maxwell's Rigidity & Conformal Geometry
Why do civil engineers build bridges out of triangles rather than squares or pentagons? The mathematical foundation lies in structural mechanics and differential geometry.
1. Maxwell's Rigidity Criterion & Laman's Theorem
A pin-jointed structural network consisting of \(j\) joints (nodes) and \(b\) bars (edges) in two dimensions is isostatic (statically determinate and kinematically rigid without internal bending moments) if and only if it satisfies Maxwell's Rigidity Criterion[^4]:
In a triangular lattice, any applied external load is resolved into pure axial tension or compression along the struts (\(N_i = \sigma_i A\)). Because struts do not undergo bending moments, the structure is stretching-dominated, scaling its effective elastic modulus linearly with relative density:
In contrast, quadrilateral, hexagonal, and irregular Voronoi lattices are bending-dominated (\(b < 2j - 3\)). When loaded in shear, the struts flex and rotate at the nodes, scaling quadratically with relative density:
At high porosity (\(\rho^*/\rho_s \approx 0.15–0.25\), typical for ventilated casts), stretching-dominated triangular lattices are theoretically 4 to 6 times stiffer in shear per unit mass than bending-dominated honeycombs[^5].
3. Interactive 2D Lattice FEA & Stress Simulator
Test this fundamental mechanics principle in real time. Select different lattice architectures below, adjust the load mode (pure shear, axial compression, or bending moment), and inspect the real-time Von Mises stress distribution and elastic modulus response.
🔬 Finite Element Lattice Deformation & Stress Engine
K · u = F (Direct Stiffness Method)4. Differential Geometry: Conformal Tiling of Anatomical Surfaces
Beyond structural load bearing, a cast must conform precisely to complex human anatomy—the styloid process of the ulna, the thenar eminence of the palm, and the malleoli of the ankle.
According to Gauss's Theorema Egregium, the Gaussian Curvature \(K\) of a two-dimensional Riemannian surface with principal curvatures \(\kappa_1\) and \(\kappa_2\) is an intrinsic invariant:
Gaussian Curvature Classifications on Human Limbs:
1. Cylinder (Forearm Shaft) 2. Sphere (Joint / Heel) 3. Hyperbolic Saddle (Wrist Web)
K = 0 (Developable) K > 0 (Synclastic) K < 0 (Anticlastic)
╭──────╮ ╭────────╮ ╲ ╱
│ │ ╱ ╲ ╲ /\ ╱
│ │ │ ● │ ╳ ╳
│ │ ╲ ╱ ╱ \/ ╲
╰──────╯ ╰────────╯ ╱ ╲
- Planar & Developable Surfaces (\(K = 0\)): Forearm shafts can be unrolled into a flat sheet without stretching or tearing. Here, regular squares or hexagons tile easily.
- Double Curvature (\(K \neq 0\)): Anatomic joints (wrists, elbows, ankles) have non-zero Gaussian curvature.
- If you attempt to map a rigid square or hexagonal grid onto a sphere (\(K > 0\)) or a saddle (\(K < 0\)), the polygons must warp out of plane, creating non-planar faceting, pinching, or structural kinks.
- Triangles (\(3\) vertices), by definition, always define a unique flat 2D plane in 3D Euclidean space. A triangular mesh can conform to any arbitrary doubly-curved anatomical surface without geometric distortion or edge buckling[^6].
This is why every modern 3D scan and standard STL file is natively tessellated as a triangular polygon mesh!
5. What the Literature Actually Discovered (2014–2025)
When we launched our 2014 CUHK project, the intuitive working hypothesis was: "Triangular lattices must be strictly superior in every clinical dimension."
A decade of empirical biomedical and additive manufacturing literature has revealed a far more nuanced and fascinating reality:
graph TD
Hypothesis["Initial 2014 Hypothesis:<br/>Triangles are superior across all cast criteria"]
Hypothesis --> Fact1["1. Surface Tiling & Conformance<br/>✓ Confirmed: Triangles tile K ≠ 0 surfaces without warping"]
Hypothesis --> Fact2["2. Mechanical Shear & Isotropy<br/>✓ Confirmed: Triangular trusses yield isotropic shear stiffness"]
Hypothesis --> Fact3["3. Compressive Peak Resistance<br/>✗ Hexagonal honeycombs out-perform in compressive plateau"]
Hypothesis --> Fact4["4. Global Geometric Optimization<br/>☆ Topology Optimization (SIMP) beats all fixed patterns"]
style Hypothesis fill:#1e293b,stroke:#475569,stroke-width:2px,color:#fff
style Fact1 fill:#0284c7,stroke:#0369a1,stroke-width:2px,color:#fff
style Fact2 fill:#10b981,stroke:#059669,stroke-width:2px,color:#fff
style Fact3 fill:#d97706,stroke:#f59e0b,stroke-width:2px,color:#fff
style Fact4 fill:#7c3aed,stroke:#8b5cf6,stroke-width:2px,color:#fff1. The Stiffness vs. Isotropy Trade-Off
A 2023 study by Badini et al. evaluated 3D-printed wrist-hand orthoses across candidate lattice unit cells[^7]. They found that while a \(45^\circ\) triangular cell exhibited perfect elastic isotropy (uniform mechanical response regardless of wrist rotation angle), a non-triangular reticular cell achieved a significantly higher tensile modulus (\(12.56\text{ MPa}\) vs. \(7.50\text{ MPa}\)) and compressive modulus (\(7.48\text{ MPa}\) vs. \(5.52\text{ MPa}\)). For clinical orthoses where directional reinforcement along the volar forearm is needed, anisotropic reticular cells were selected over pure triangles.
2. Hexagonal Honeycombs Excel Under Compression
Suksawang et al. (2025) compared 3D-printed triangular, hexagonal, and rectangular lattice architectures for orthopedic fixation[^8]. Under axial compression, hexagonal unit cells sustained the highest peak load and longest plateau stress, out-performing triangular cells at identical relative densities (\(\rho^* = 0.30\)). The progressive bending collapse of hexagonal cells absorbs impact energy without catastrophic brittle strut snapping[^9].
3. Topology Optimization Outperforms All Fixed Patterns
Mian et al. (2023) tested circular, square, triangular, and hexagonal perforation cutouts on upper-limb splints[^10]. They proved that differences between regular cutout shapes were second-order: topology optimization (SIMP) reduced splint weight by 26.4% while preserving stiffness, compared to only 12.1% for the best uniform pattern.
4. Collaborative Full-Circle: The 2020 Clinical Study
In 2020, my original CUHK collaborators—Defeng Wang and Lin Shi—co-authored a landmark study in BioMed Research International[^11], performing full integrated finite element analysis on 3D-printed forearm casts for distal radius (Colles') fractures. The study validated that patient-specific printed casts maintained stable anatomical fracture reduction with zero skin complications across 20 clinical patients.
6. Biomechanical & Multi-Lattice Benchmark Matrix
| Lattice Architecture | Mechanical Dominance | Elastic Isotropy (\(E_x / E_y\)) | Shear Modulus (\(G_{xy}\)) | Compressive Energy Absorption | Conformal Tiling (\(K \neq 0\)) | Clinical Recommended Use Case |
|---|---|---|---|---|---|---|
| ▲ Triangular Truss (Kagome) | Stretching (\(b = 2j - 3\)) | 0.98 – 1.00 (Full) | High (\(12–16\text{ MPa}\)) | Moderate (Brittle buckling) | Flawless (3-point planar) | Torsional & multi-axial joint immobilization (Ankle / Wrist) |
| ⬡ Hexagonal Honeycomb | Bending (\(b < 2j - 3\)) | 0.65 – 0.75 (Anisotropic) | Low (\(4–7\text{ MPa}\)) | Superior (Long stable plateau) | Moderate (Facet saddle distortion) | Impact-resistant protective shells & pads |
| 🌀 Voronoi Reticular (2014) | Stochastic Bending | 0.85 – 0.90 (Quasi-isotropic) | Moderate (\(8–11\text{ MPa}\)) | High (Distributed stress) | Good (Relaxed dual Delaunay) | Organic ventilated forearm & humerus casts |
| ⊞ Orthogonal Square | Bending (Shear) | 0.40 – 0.50 (Highly Anisotropic) | Very Low (\(2–4\text{ MPa}\)) | Moderate | Poor (Out-of-plane buckling) | Unidirectional spinal or leg bracing |
| 🧬 Topology Optimized (SIMP) | Heterogeneous Stress Paths | Variable (Functionally Graded) | Optimal (Load-specific) | Optimal | Excellent (Isosurface extraction) | High-performance elite sports & fracture reduction |
7. Direct Stiffness FEA Formulation
To model any lattice unit cell computationally, we solve the classical finite element global equilibrium equation:
where \(\mathbf{K} = \sum_e \mathbf{T}_e^T \mathbf{k}_e \mathbf{T}_e\) is the assembled global stiffness matrix, \(\mathbf{u}\) is the nodal displacement vector, and \(\mathbf{f}\) is the external load vector.
View 2D Direct Stiffness FEA Solver (Python)
import numpy as np
import scipy.sparse as sp
from scipy.sparse.linalg import spsolve
def solve_2d_truss(nodes: np.ndarray, elements: np.ndarray, E: float = 2000.0, A: float = 4.0,
fixed_nodes: list = None, force_node: int = 0, force_vec: tuple = (0.0, -100.0)):
"""
Direct stiffness method for 2D pin-jointed truss lattices.
- nodes: (N, 2) array of nodal coordinates (x, y) in mm
- elements: (M, 2) array of node connectivity indices
- E: Young's modulus (MPa)
- A: Strut cross-sectional area (mm^2)
"""
num_nodes = len(nodes)
dofs = 2 * num_nodes
K = np.zeros((dofs, dofs))
for (n1, n2) in elements:
x1, y1 = nodes[n1]
x2, y2 = nodes[n2]
L = np.hypot(x2 - x1, y2 - y1)
c = (x2 - x1) / L
s = (y2 - y1) / L
# Element stiffness in local/global coordinates
k_local = (E * A / L) * np.array([
[ c*c, c*s, -c*c, -c*s],
[ c*s, s*s, -c*s, -s*s],
[-c*c, -c*s, c*c, c*s],
[-c*s, -s*s, c*s, s*s]
])
elem_dofs = [2*n1, 2*n1+1, 2*n2, 2*n2+1]
for i in range(4):
for j in range(4):
K[elem_dofs[i], elem_dofs[j]] += k_local[i, j]
# External forces
F = np.zeros(dofs)
F[2*force_node] = force_vec[0]
F[2*force_node + 1] = force_vec[1]
# Apply Dirichlet boundary conditions (fixed ground nodes)
free_dofs = np.ones(dofs, dtype=bool)
for fn in (fixed_nodes or []):
free_dofs[2*fn] = False
free_dofs[2*fn + 1] = False
u = np.zeros(dofs)
u[free_dofs] = np.linalg.solve(K[np.ix_(free_dofs, free_dofs)], F[free_dofs])
# Compute element axial stress
stresses = []
for (n1, n2) in elements:
x1, y1 = nodes[n1]
x2, y2 = nodes[n2]
L = np.hypot(x2 - x1, y2 - y1)
c = (x2 - x1) / L
s = (y2 - y1) / L
u_elem = u[[2*n1, 2*n1+1, 2*n2, 2*n2+1]]
delta = np.dot([-c, -s, c, s], u_elem)
stresses.append(E * delta / L)
return u.reshape(-1, 2), np.array(stresses)
8. Summary & Key Engineering Takeaways
- Triangles Guarantee Conformal Geometry: Because 3 points define a plane, triangular meshes naturally tile doubly curved anatomical surfaces (\(K \neq 0\)) without out-of-plane saddle distortion.
- Stretching vs. Bending Mechanics: Triangulated trusses satisfy Maxwell's criterion (\(b = 2j - 3\)), eliminating internal joint bending and maximizing shear stiffness per unit mass.
- Isotropy Over Peak Strength: Triangles provide uniform, direction-independent elastic response, whereas hexagonal honeycombs maximize compressive plateau energy absorption.
- Topology Optimization Is King: Fixed tessellations are now giving way to functionally graded, load-specific topology optimization (SIMP), minimizing weight while reinforcing clinical fracture vectors.
Data and Code Availability
- Original Project Pipeline: 2014 ITF grant archives (CUHK Department of Imaging and Interventional Radiology / Department of Electronic Engineering).
- FEA & Visualizers: Interactive simulation routines embedded directly in this page; Python direct stiffness solver provided in Section 7.
- Literature Repository: All cited clinical and biomechanical papers are indexed with DOI links below.
References
[^1]: Mathijsen, A. Nieuwe wijze van aanwending van het gips-verband bij beenbreuken (New method of applying plaster of Paris bandages to bone fractures). Haarlem (1852). [^2]: Large, P. A 'new focus' in casting — an introduction to the concepts of focus rigidity casting. J. Orthop. Nurs. 5(4), 176–179 (2001). https://doi.org/10.1054/joon.2001.0181 [^3]: Wang, D., Fung, K. M., Shi, L. An Intelligent Pipeline for Designing Personalized Orthopedic Casts. International Biomedical Devices Workshop (IBDW), CUHK (2014). [^4]: Maxwell, J. C. On the calculation of the equilibrium and stiffness of frames. Philos. Mag. 27(182), 294–299 (1864). https://doi.org/10.1080/14786446408643668 [^5]: Deshpande, V. S., Ashby, M. F., Fleck, N. A. Foam topology: bending versus stretching dominated architectures. Acta Mater. 49(6), 1035–1040 (2001). https://doi.org/10.1016/S1359-6454(00)00379-7 [^6]: Botsch, M., Kobbelt, L., Pauly, M., Alliez, P., Lévy, B. Polygon Mesh Processing. CRC Press (2010). https://doi.org/10.1201/b10688 [^7]: Badini, S., Regondi, S., Lammi, C., Bollati, C., Donvito, G., Pugliese, R. Computational Mechanics of Form-Fitting 3D-Printed Lattice-Based Wrist-Hand Orthosis for Motor Neuron Disease. Biomedicines 11(7), 1787 (2023). https://doi.org/10.3390/biomedicines11071787 [^8]: Suksawang, B., Chaijareenont, P., Silthampitag, P. Effect of Unit Cell Design and Volume Fraction of 3D-Printed Lattice Structures on Compressive Response and Orthopedics Screw Pullout Strength. Materials 18(6), 1349 (2025). https://doi.org/10.3390/ma18061349 [^9]: Gibson, L. J., Ashby, M. F. Cellular Solids: Structure and Properties. Cambridge University Press (1997). https://doi.org/10.1017/CBO9781139878326 [^10]: Mian, S. H., Umer, U., Moiduddin, K., Alkhalefah, H. Finite Element Analysis of Upper Limb Splint Designs and Materials for 3D Printing. Polymers 15(14), 2993 (2023). https://doi.org/10.3390/polym15142993 [^11]: Chen, Y., Lin, H., Yu, Q., Zhang, X., Wang, D., Shi, L., Huang, W., Zhong, S. Application of 3D-Printed Orthopedic Cast for the Treatment of Forearm Fractures: Finite Element Analysis and Comparative Clinical Assessment. BioMed Res. Int. 2020, 9569530 (2020). https://doi.org/10.1155/2020/9569530