3D printing infill: what is the strongest infill pattern?

A slicer splits an FDM part into three regions: solid shells, solid top and bottom layers, and infill. The shells carry bending and tensile load because they sit at the outer surface where stress is highest. Infill ties those shells together and resists local buckling of the walls. When a part fails in bending, the shell usually cracks first.

Think of the cross-section like an I-beam. The flanges do the work; the web keeps them apart. Increase wall count from 2 to 4 and you add material exactly where bending stress peaks. Increase infill from 20% to 40% and you mostly add material in the neutral zone, which does little for stiffness.

This is why the answer to the strongest infill question has a condition attached. Pattern choice matters, but only after wall count, layer orientation, and print temperature are already correct. On short, stubby parts in pure compression, infill shares more of the load and the pattern becomes more visible in test results.

For most functional brackets and housings, walls do the heavy lifting. Infill stops the walls from collapsing inward and supports top surfaces. That is a real job, just not the same job as carrying the main bending load.

• 1
  Walls firstBending stress is highest at the surface, so shell thickness drives stiffness more than infill density.
• 2
  Infill secondIt ties shells together and prevents wall buckling under compression.
• 3
  Anisotropy alwaysLayer lines are weak planes. Orientation beats any infill setting.

Grid and lines are the fastest to print. They cross at right angles and create a stiff in-plane lattice with strong directional bias. In compression along the Z axis, grid at 40% density performs well because the strands stack and share load. In shear, the same pattern is weaker.

Gyroid and tri-hexagon distribute load in three dimensions. Gyroid has no straight continuous paths, so stress cannot run along one line of strands. That makes it more uniform in torsion and multi-axis loading. It also prints slower and uses more material per unit of stiffness.

Honeycomb gives high in-plane stiffness at low density. Load applied to the flat face is carried well. Load applied to the point of the hexagons can crush the cells early, so orientation matters. Cubic and octet infills add vertical strands between layers. This helps Z-direction compression but adds print time.

Triangles and star patterns are stiff in one plane and compliant in others. They suit parts with a known load direction, such as a handle or a lever. If the load direction is unknown, a symmetric pattern like gyroid is the safer pick.

• 1
  Compression along ZCubic or octet adds vertical strands; grid also works at higher density.
• 2
  Torsion and multi-axisGyroid spreads stress with no single failure path.
• 3
  In-plane bendingHoneycomb and triangles give high stiffness for their mass.
• 4
  ImpactGyroid and honeycomb absorb energy more gradually than grid.

Strength does not scale linearly with density. Going from 20% to 40% adds noticeable stiffness in compression. Going from 60% to 80% adds much less, because the shells already carry most of the bending load and the extra material sits close to the neutral axis.

If a part is failing, raising infill is often the slowest fix. Check wall count, print orientation, and layer adhesion first. A part printed flat with 3 walls at 25% infill will usually beat the same part printed upright with 4 walls at 60% infill under bending.

There is a practical ceiling. Above roughly 50% density, many patterns stop improving stiffness per gram and start adding mass, print time, and warp risk. For a load-bearing bracket, 40–50% with 3–4 walls is a common working range.

Very low densities below 15% are for visual models and light covers. They save material but the walls can buckle inward under compression, and top surfaces may bridge poorly. If the part needs to hold a thread or a bearing, keep the local region solid.

• 1
  0–15%Visual models, light covers, no structural role.
• 2
  15–35%General housings and brackets with 2–3 walls.
• 3
  35–60%Load-bearing parts with 3–4 walls; diminishing returns above 50%.
• 4
  Local solidBosses, threads, and bearing seats should be solid regardless of global infill.

FDM parts are anisotropic. The bond between layers is weaker than the strand itself. A part loaded across layer lines will fail at the interface before infill density matters. Rotate the part on the build plate so the main tensile load runs along the strands, not across them.

If the required stiffness exceeds what a printed polymer can deliver, no infill pattern will close the gap. That is when the part should move to a machined or molded process. We machine 6061-T6, 7075, 17-4PH, and PEEK when printed parts hit their ceiling.

Short-fiber filaments improve stiffness in the strand direction but still bond weakly across layers. They help in-plane bending and compression. They do not fix Z-direction weakness, and they wear nozzles faster.

For prototypes that need to be isotropic, vacuum casting in polyurethane or CNC machining from plastics like ABS, PC, or POM gives consistent properties in all directions. Printed infill is a tool for the right job, not a universal substitute.

• 1
  Across layersRotate the part so tensile load follows the strand direction.
• 2
  High stiffnessMove to CNC machining in aluminum, stainless, or PEEK.
• 3
  Isotropic prototypesVacuum casting or machining removes layer directionality.