Most of the plastic in a typical desktop 3D print is doing very little work. A bracket that carries a five-kilogram load does not need to be solid anywhere near the middle of its span, and a shelf that must not sag does not care about the material buried in the middle of its cross-section. Yet that is exactly where the slicer puts it — a solid slab of infill running straight through the region of lowest stress. The result is a part that is heavier than it needs to be, slower to print, and in some cases weaker than a lighter version of the same thing.
This article is about the third layer of part design, after geometry and after plastic allocation. We covered how a single part's edges and corners concentrate stress in ribs, gussets, fillets and bosses, and how shell count and infill density divide the load in shell count, wall thickness and infill. The question left over is the one that decides whether a part feels engineered or improvised: where in the cross-section should the plastic actually go, and how much of it should leave the part entirely?
Bending Is Not a Material Problem, It Is a Distance Problem
The stiffness of a beam — how much it deflects under a given load — is governed by a quantity called the second moment of area. The name is off-putting, but the behaviour is simple: each bit of material contributes to stiffness in proportion to the square of its distance from the neutral axis. Material near the outside of a section is worth enormously more than material near the middle.
That single fact explains almost everything in this article. If a rectangular section is 20 mm tall and you move plastic from the middle to the top and bottom faces, the section's stiffness rises dramatically while its weight stays exactly the same. This is why an I-beam is shaped the way it is, why a cardboard box is strong along its folds, and why a printed bracket with a hollowed-out core can outperform a solid one.
| Section shape | Relative stiffness | Relative weight | Stiffness per unit weight |
|---|---|---|---|
| Solid rectangle 20 × 10 mm | 1.00 | 1.00 | 1.00 |
| Hollow box, 1.5 mm wall, same outer size | 0.62 | 0.42 | 1.48 |
| Two flanges joined by a thin web (I-section) | 0.71 | 0.40 | 1.78 |
| Flanges pushed apart by a taller section | 2.10 | 0.55 | 3.82 |
| Solid rectangle with infill only | 0.35 | 0.55 | 0.64 |
Read the second row carefully, because it is the whole argument in one line. Removing the inner core of a section cuts its weight by more than half while giving up only about a third of its stiffness. The fifth row is what happens when designers add "structure" without moving it to the outside: a partially filled core is less stiff and less efficient than simply thinning the section and putting full walls where they count.
Three Ways to Carry a Load, and When Each One Wins
There are only three practical strategies for making a printed part carry a bending load, and they are not interchangeable. Choosing the wrong one is the most common reason a redesigned part comes out worse than the original.
| Strategy | How it works | Best for | Fails when |
|---|---|---|---|
| Deepen the section | Increase overall height in the load direction | Anything you are free to reshape; gives the largest gain for the least plastic | Envelope is fixed by fit, or the part must stay flat |
| Hollow the core | Remove low-stress central material, keep full outer walls | Thick, chunky parts; boxes, blocks, chunky handles | Walls fall below three perimeters and buckle |
| Add ribs or a web | Connect two surfaces with thin walls in the load plane | Flat panels, lids, plates, anything that must stay thin | Ribs are added to a thick wall instead of replacing material |
The first row deserves emphasis because it is nearly free. A 30 mm tall bracket is more than twice as stiff as a 20 mm tall one at the same thickness, and it costs only 50 percent more material. Nothing about the material, the printer or the settings changed. The geometry did. Where designers get into trouble is fixing the envelope first — "it has to be 8 mm thick because it slips under the shelf" — and then trying to recover stiffness by other means. If the envelope is genuinely fixed, you are in row two or three and the budget is much tighter.
The Hollow Core: When Removing Material Helps and When It Backfires
Hollowing a chunky part is the most intuitive move, and it is also the one most often done badly. The rule that governs it is about wall thickness relative to the section, not about how big the part is.
| Wall thickness as fraction of outer size | What you get | Verdict |
|---|---|---|
| Below 1/20 | Walls too thin to print reliably; buckling dominates | Do not do this |
| 1/20 to 1/12 | Efficient but sensitive to any side load or impact | Only for pure bending, protected load paths |
| 1/10 to 1/6 | Near-optimal stiffness per gram; walls print as solid perimeters | The sweet spot for most printed parts |
| 1/5 to 1/3 | Still an improvement but wasting opportunity | Common default, can usually be thinned |
| Above 1/3 | Practically solid; the cavity buys nothing | Remove the cavity and thin the whole section |
For a typical 40 mm wide printed enclosure wall, that sweet spot lands at roughly 4 to 6 mm — which happens to be exactly where a well-configured shell count and infill combination lands anyway. This is why shell count and infill matter for stiffness: they are an approximate way of building a hollow section from the inside out, on a per-layer basis, without modelling the cavity explicitly.
The failure mode to watch for is wall buckling. A hollow box resists bending well but resists a sideways push or a drop poorly, because a thin unsupported wall has no lateral support. Two fixes work: keep walls at least three extrusion widths so they print as a continuous solid band, and add short internal cross-walls — one every 40 to 60 mm is usually enough — to tie opposing surfaces together and stop them from acting independently.
Ribs Are Not Added Material, They Are Relocated Material
The single most expensive mistake in ribbed designs is adding ribs to a wall that was already thick enough. Ribs are meant to let you thin a surface while restoring its stiffness, not to reinforce a surface that is already adequate. Done the second way, the part gets heavier, warps more, and picks up sink marks on the visible face.
The mechanics of a rib force the following constraints, all of which come from the same root cause — the rib is thinner than the wall it sits on, and that thickness mismatch causes differential cooling:
- Rib thickness: 50 to 65 percent of the adjoining wall. At 60 percent of a 2.0 mm wall, a rib is 1.2 mm — which is three 0.4 mm extrusion lines if you slice for it deliberately.
- Rib depth: up to three times the wall thickness before the gain flattens out. Beyond that, the rib's own bending becomes the limiting factor and the plastic is better spent elsewhere.
- Rib spacing: pitch no more than 4 to 6 times the wall thickness. Wider spacing means the unsupported panels between ribs sag between them, and the ribs stop cooperating.
- Root radius: always fillet where the rib meets the wall. A sharp internal corner is a stress concentrator, and it is also where the differential cooling crack starts. Add a radius at least equal to a quarter of the rib thickness.
- Draft: a small taper — 0.5 to 1 degree — on tall ribs prevents the print from becoming a warping lever as it cools.
Because rib geometry concentrates material into narrow bands, it also concentrates residual stress. Long ribs printed in ABS or ASA on a wide flat surface are a reliable recipe for lifting corners. The geometry fix is to break long ribs into shorter segments with small gaps rather than running one piece across the entire face. A second, complementary fix is to slow the cooling differential with an enclosed chamber, and the third is to accept a lower rib count and put the stiffness into section depth instead.
Rib or web geometry interacts directly with material choice, and the specific recommendations for where ribs actually help are worth reading alongside the ribs and bosses guide.
Slenderness: The Limit That No Amount of Plastic Fixes
Past a certain point, a part does not fail because the material is too weak. It fails because the shape becomes unstable. A long, thin column buckles sideways rather than crushing, and a narrow tall wall buckles out of plane rather than bending. Adding more plastic at the same slenderness ratio does almost nothing, because the failure is geometric.
| Slenderness (length ÷ thickness) | Expected behaviour under compression | What to do |
|---|---|---|
| Under 10 | Crushing governs; material strength is the limit | Add material, increase wall thickness |
| 10 to 25 | Mixed; both strength and stability matter | Split into two shorter members; add a mid-span brace |
| 25 to 50 | Buckling governs; extra thickness on the same shape helps little | Change the cross-section shape, not its thickness |
| Over 50 | Fails far below the material's rated strength | Redesign as a truss, a tube, or several short spans |
The practical lesson is counterintuitive: when a tall thin printed feature keeps failing, the instinct is to make it thicker and the correct move is usually to make it shorter — split it into two members with a cross-brace, or turn an open blade into a closed box or tube. A closed triangular or rectangular section resists buckling far better than an open one of the same weight, because the closed shape keeps the walls from moving independently.
Orientation Rewrites Every Rule Above
Nothing in this article means the same thing in every print orientation, because the inter-layer bond is the weakest link in any FDM part. The stiffness gain from moving plastic to the outside of a section only pays off if the load path stays inside layers rather than crossing between them.
| Load direction relative to layers | What carries the load | Effective stiffness | Design move |
|---|---|---|---|
| Bending along the layer plane (part lying flat) | Continuous extruded strands in the flanges | Full calculated value | Design and orient this way whenever possible |
| Bending across layers (part standing up) | Inter-layer adhesion in the flanges | Roughly 40 to 60 percent of the flat value | Reorient, or add a printed-through feature such as a dovetail |
| Tension along the layer plane | Strand tensile strength | Full material value | Safe; this is what printed parts do best |
| Tension across layers | Bond strength only | A fraction of the material value | Avoid entirely; no geometry fixes this cleanly |
This is the single most important interaction in the article. A well-shaped I-section printed in the wrong orientation can be weaker than a plain rectangular block printed in the right one. If you want the full picture of how orientation governs layer bonding, it is covered in the guide to inter-layer bonding and orientation.
Material Choice Shifts the Whole Budget
Stiffness-driven design assumes the material does not creep. That assumption holds for different materials for different lengths of time, and the distinction matters more than the headline stiffness numbers.
| Material | Stiffness character | Creep under sustained load | Best role in a stiff part |
|---|---|---|---|
| PLA | Highest stiffness of the common filaments | Poor over months, and softens near 55 °C | Short-term, cool, lightly loaded parts |
| PETG | Moderate, good elongation | Moderate | Balanced default for brackets and housings |
| ABS and ASA | Moderate; ASA for outdoor exposure | Better than PETG under sustained load | Warm environments and outdoor duty |
| PETG-CF | Markedly stiffer than plain PETG | Moderate | Stiffness-critical parts that stay indoors |
| ABS-CF | Stiff with improved thermal stability | Good | Stiff parts that see heat |
| PA6-CF20 | High stiffness and toughness together | Very good once dried | Serious structural parts and jigs |
| Lightweight PLA | Deliberately less dense, foaming on extrusion | Poor | Visual models where mass matters and loads do not |
| TPU-LW90A | Flexible, low density | Not applicable | Damping pads and lightweight flexible parts |
Two entries in that table look like contradictions and are not. Lightweight PLA and lightweight TPU achieve their lower density by foaming during extrusion, which reduces stiffness along with mass. They are the right answer when a part needs to be light and does not need to be stiff — a display model, a cosmetic shell, a damping layer — and the wrong answer the moment a structural load appears. Conversely, the filled engineering grades are the right answer when stiffness matters but toughness cannot be sacrificed, which is the usual situation for a real bracket.
Filled materials bring a workshop consequence that stiffness-focused designers forget: carbon-fibre and glass-fibre grades abrade the nozzle. Running these long-term through a brass nozzle erodes the orifice and slowly changes the extrusion width, which shows up as dimensional drift. A hardened nozzle such as the tungsten nozzle for Bambu Lab P1 removes that variable. Drying matters just as much, because moisture in a filled nylon turns into steam and produces the porous, low-density walls that undo exactly the stiffness you designed for — a filament dryer is part of the structural toolkit, not a separate hobby.
Where Added Plastic Stops Paying
Every strategy in this article has a point beyond which more plastic buys less stiffness than expected. Recognising that point is what separates a designed part from an over-built one.
- Section depth past roughly six times the width. A section that is very tall and narrow will buckle sideways before it bends, so the calculated stiffness is never realised.
- Ribs deeper than three wall thicknesses. The rib's own bending starts to dominate and the extra depth mostly adds weight.
- Walls past four perimeters in a hollow part. The section is approaching solid and the cavity has stopped earning its complexity.
- Infill above about 40 percent in a stiffness-critical part. The material is landing in the low-stress core, where its contribution is smallest.
- Any thickening of a slender open member. Buckling is geometric; change the shape, not the thickness.
- Adding ribs to a wall that already meets its stiffness target. The cost is weight, print time, warping and visible sink marks.
Assembly, Flatness and the Parts You Cannot Print in One Piece
Stiffness optimisation frequently pushes a design toward a longer or taller shape than the build volume allows. Splitting it is legitimate and often better than compromising the section, but the joint becomes the new weak point and must be designed as carefully as the section itself.
Three rules apply. First, place joints at points of low bending moment — near a support or a change in direction — not at mid-span where stress peaks. Second, make the joint geometry carry load in shear rather than peel, which means a stepped or dovetailed interface rather than a butt joint; the reasoning behind the shapes is set out in the guide to gluing printed parts. Third, allow for the flatness limits of printed surfaces, because a printed joint face that is 0.15 mm out of flat will not close consistently. Reliable fits come from measured clearance rather than from nominal dimensions, as covered in dimensional accuracy and tolerances.
Long stiff parts also accumulate warp, and warp directly attacks stiffness by changing the section's effective depth along its length. A 300 mm part that lifts 2 mm at one end is no longer the section you designed. The strategies that address this are in the guide to printing large parts without warping, and the general diagnosis is in the warping troubleshooting guide.
Mounting a Stiff Part Without Undoing It
A bracket's stiffness is worth nothing if the way it attaches to the world is soft. Three attachment rules follow from the same reasoning used throughout this article.
Keep the fixing points spread as widely as the load allows, because the fixing pattern sets the span the part actually sees; two bolts 20 mm apart behave like a 20 mm bracket no matter how large the plate is. Make the transition into each fixing point gradual, since a screw hole is a stress raiser and the local geometry around it decides whether the part cracks or the load reaches the section you designed — the details are in the guide to mounts that do not crack at the screw hole. And make sure the load path from the fixing point into the stiff section is continuous, with no thin flexible neck between the two, because the compliant link will absorb the deflection that the stiff section was built to prevent.
A Nine-Step Sequence for Stiff-for-Weight Parts
- Write down the load direction and where the part is supported. Everything else follows from this.
- Ask whether the envelope can grow in the load direction. If yes, deepen the section first — it is the cheapest stiffness available.
- If the envelope is fixed, decide between hollowing and ribbing based on whether the existing section is chunky or thin.
- Size walls at one-tenth to one-sixth of the outer dimension, and never below three extrusion widths.
- Size ribs at 50 to 65 percent of the adjoining wall, fillet the root, and break long ribs into segments.
- Check the slenderness ratio of every tall or narrow feature, and reshape rather than thicken anything above 25.
- Choose the print orientation so the load path runs along layers, and accept a taller or more supported setup to get it.
- Pick the material for creep resistance first and stiffness second; add a hardened nozzle and drying if the choice is a filled grade.
- Re-examine the fixings, the joints and the flatness, because those three interfaces decide whether the section you designed is the section that gets used.
None of this requires simulation software or a materials laboratory. It requires deciding where the plastic earns its place, and being willing to remove the plastic that does not. A part that is stiff because its shape puts material where the stress is will almost always be lighter, faster to print, and cheaper to make than one that is stiff because there is simply a great deal of it.