💻 computer science

Non-planar 3D Printing of Double Shells

This paper presents a multi-axis FDM robotic 3D printing methodology for fabricating double-shell structures by converting transversal strip networks into printable partitions that are individually printed and assembled, demonstrating the technique's versatility across various scales and geometric complexities.

Original authors: Ioanna Mitropoulou, Amir Vaxman, Olga Diamanti, Benjamin Dillenburger

Published 2026-06-23

📖 4 min read☕ Coffee break read

CC BY 4.0

Original authors: Ioanna Mitropoulou, Amir Vaxman, Olga Diamanti, Benjamin Dillenburger

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine you are trying to build a curved, hollow wall out of melted plastic, like a giant, intricate honeycomb. Usually, when robots 3D print these shapes, they build them layer by layer, like stacking flat pancakes. But if the wall is curved, this "pancake" method leaves ugly, stair-step edges on the side, and the wall is weak in one direction because the plastic strands only stick well along the length of the line, not across it.

This paper presents a clever new way to build these curved walls using a robot arm that can twist and turn freely. Instead of building a single, thin shell, they build a double shell—think of it like a sandwich with two slices of bread and a tiny bit of filling in the middle, but the "bread" is made of two separate layers printed in different directions.

Here is how they do it, broken down into simple steps:

1. The "Strip" Idea

Instead of thinking of the wall as a solid block, the researchers imagine it as a net made of two sets of strips crossing each other, like a woven basket or a plaid shirt. One set of strips runs one way (let's call them the "Blue Strips"), and the other set runs the opposite way (the "Green Strips").

2. The "Standing Shell" Trick

Normally, 3D printers need a flat table to build on. But this robot is smart. It prints the "Blue Strips" first, but it doesn't lay them flat. It prints them standing up, curving along the shape of the wall. It's like building a curved fence where each new piece of wood rests on the one before it, without needing a flat floor underneath. The robot arm twists and turns to keep the nozzle pointing exactly where the plastic needs to go.

3. Cutting the Puzzle

The problem is that the robot can't print the entire giant wall in one go; it's too big and the angles get too weird. So, the researchers use a digital "scissors" to cut the wall into smaller, manageable puzzle pieces.

The Topological Cut: They make sure the pieces are simple shapes that can be printed one after another without getting stuck.
The Geometric Cut: They make sure no single piece is too big for the robot's workspace and that the robot doesn't have to twist its arm into an impossible position to print it.

4. The "Double Shell" Sandwich

Here is the magic part:

They print all the "Blue Strip" pieces to make one half of the wall.
They print all the "Green Strip" pieces to make the other half.
Because the strips run in different directions, the two halves reinforce each other. If you push on the wall from one side, the Blue strips hold it; if you push from the other, the Green strips hold it. It's much stronger than a single layer.

5. Putting the Puzzle Together

Once the pieces are printed, they have to be assembled. Since 3D printing isn't perfect (the plastic shrinks a tiny bit as it cools), the pieces might not fit together perfectly.

The Gap Strategy: The researchers intentionally leave tiny gaps in the seams of the pieces. This acts like a "wiggle room" so that when the pieces are put together, they can slide slightly to fit perfectly without forcing anything.
The Ribs: Inside the hollow space between the two shells, there are small connecting bars (ribs) that act like the filling in the sandwich. These ribs are printed so they interlock, like puzzle teeth, holding the two halves together.
Screws: Finally, they use a few screws to lock the pieces in place, just to be safe.

Why is this cool?

Stronger: By printing in two directions and making it a double wall, the object is much tougher and less likely to break.
Smoother: Because the robot follows the curve of the wall, there are no ugly "stair-step" edges.
Versatile: They tested this on all sorts of weird, wavy shapes, from simple curves to complex, mathematically perfect surfaces (like the "Costa" and "Batwing" shapes mentioned in the paper).

In short, this paper shows how to use a flexible robot arm to print two interlocking, curved walls made of plastic strips, cut them into printable puzzle pieces, and screw them together to make a strong, smooth, and complex 3D object.

Technical Summary: Non-planar 3D Printing of Double Shells

Problem Statement
The paper addresses inherent limitations in single-layer Fused Deposition Modeling (FDM) of shell structures, particularly when scaled up using robotic arms. Traditional single-shell printing suffers from two primary issues:

Thickness Constraints: The shell thickness is limited by the nozzle diameter (typically 0.5–2.5 mm), leading to instability when scaling up.
Anisotropy and Surface Quality: Due to the one-dimensional nature of print paths, the surface approximation is smooth parallel to the paths but exhibits a "staircase" effect orthogonal to them. This results in anisotropic mechanical properties, with significantly higher strength along the print direction, and aesthetic degradation.

While non-planar slicing and multi-axis printing have been explored to reduce support structures and improve surface quality, the authors argue that printing a single shell in transversal directions still leaves these structural and aesthetic vulnerabilities.

Methodology
The proposed method fabricates double shell structures by printing two separate outer walls in transversal directions (aligned with two orthogonal strip networks, $U$ and $V$ ) and connecting them with sparse internal ribs. The workflow relies on a Strip-Decomposable Quad (SDQ) mesh as an intermediate representation. The process involves four main stages:

Topological Partitioning:
The continuous strip networks ( $U$ and $V$ ) are partitioned into simply-connected patches to ensure an unambiguous, non-branching print sequence (standing shell logic).
- Constraint: Cuts must not overlap between the $U$ and $V$ networks to allow for assembly.
- Singularity Handling: Around singularities (specifically D2 types), standard separatrix cuts can cause self-intersection or overlapping cuts on both sides. The authors introduce specific transversal cuts ( $C_U$ and $C_V$ ) at a distance $d_q$ from the singularity to interrupt overlapping cuts, ensuring the final assembled object remains connected.
Geometric Partitioning:
Patches are further subdivided to adhere to physical fabrication constraints:
- Size Bounds: Patches are cut if their principal dimensions exceed the robot's reachable bounding box (50 x 50 x 50 cm in the authors' setup).
- Angle Bounds: The variation in the print direction angle ( $\gamma$ ) during the buildup of a patch is monitored. If the cumulative angle variation exceeds a threshold (set to $\pi/2$ ), the strip is cut into smaller segments to maintain feasible print angles.
Digital-to-Physical Preparation:
- Offsetting: The original surface is offset by $t/2$ on both sides to create a double shell of total thickness $t$ (default $4n$ , where $n$ is nozzle diameter).
- Tolerance Management: To account for printing inaccuracies (shrinkage, over-extrusion), a single strip is removed from the boundary of patches on one side (the $V$ side) to create gaps. This prevents assembly failure due to cumulative errors.
- Orientation: The global print orientation for each piece is calculated to align the average print direction vector with the vertical $z$ -axis, ensuring the first path lies on a sacrificial base.
- Ribs and Interlocking: Rigidifying ribs are printed transversely. To interlock the two shells, ribs on one side are interrupted where they intersect ribs from the other side, creating a guide for assembly.
Fabrication and Assembly:
The pieces are printed using a 6-axis robotic arm (UR10) with a 2.5 mm nozzle. The extrusion rate is varied dynamically to match the changing layer height. The final structure is assembled by alternating pieces from the $U$ and $V$ sides, secured manually with screws at rib intersections.

Key Contributions

Double Shell Paradigm: A shift from single-shell to double-shell fabrication using transversal non-planar paths, which adds structural depth and distributes imperfections more evenly.
Partitioning Algorithm: A robust methodology for partitioning SDQ meshes into printable patches that simultaneously satisfy topological requirements (unambiguous print order), geometric constraints (size and angle limits), and assembly constraints (non-overlapping seams).
Tolerance-Aware Design: A specific strategy for managing fabrication inaccuracies by strategically introducing gaps in the seams, acknowledging that absolute precision is unattainable in large-scale additive manufacturing.

Results
The authors demonstrate the method's versatility and robustness through the fabrication of five prototypes of varying geometric complexity and scale:

Costa Minimal Surface: Partitioned into 7 ( $U$ ) and 11 ( $V$ ) pieces; total print time 51.5 hours.
Curved and Elongated Surfaces: Fabricated with 3–6 pieces per side.
Chen Gackstatter and Batwing Minimal Surfaces: Demonstrated handling of complex curvature and boundary constraints, with piece counts ranging from 5 to 16 per side.
The results confirm that the method can handle surfaces of different scales and geometric complexities, producing stable double-shell structures with reduced staircase artifacts compared to single-shell alternatives.

Significance and Claims
The paper positions its work within the theme of "Beyond Optimization" by prioritizing robustness and adaptability over hyper-precision.

Real-World Assembly: The authors claim that their method is significant because it proactively manages fabrication imperfections (via gap strategies) rather than attempting to eliminate them, making the system adaptable to real-world conditions where absolute precision is unfeasible.
Architectural Potential: The double-shell approach is presented as a solution to the low thickness and anisotropy of standard 3D-printed shells, offering potential for façade panels, molds, and pavilions. It also facilitates the embedding of functional equipment (tubes, wires) and reinforcement between the shells.
Aesthetic and Functional Control: The ability to control print paths in transversal directions allows for new design possibilities, including the manipulation of surface textures and the potential for multi-material or fiber-reinforced printing in future iterations.

The authors acknowledge limitations, noting that the double-shell approach doubles the material usage and fabrication time compared to single shells, and that the point-link connections (screws) are weaker than continuous inter-layer bonding found in other transversal printing methods. They suggest future work in optimizing support structures and developing smarter joint geometries (e.g., puzzle joints) to improve assembly quality.