Hexagonal modular storage

You want to fill a drawer with cells for screws, resistors or drill bits, and the hexagon is tempting: it tiles without gaps and spreads space evenly — the honeycomb look. You model a grid of hexagonal prisms, print it as drawn, and run into two surprises. The cells you meant to slot into one another won't go in, or go in only by force and split the wall. And the walls that two cells share come out with a soft seam running down the middle, as if the slicer — the software that carves the model into layers — had not known what to do with it. Neither is bad luck: both follow directly from how an FDM printer lays down filament, and both can be designed out with numbers.

Why the hexagon fills the drawer

The regular hexagon tiles the plane without leaving a single gap. It is one of only three regular shapes that manage this — triangle, square and hexagon — and of the three it is the closest to a circle. That closeness to a circle is what you're after, because a circle spreads load equally in every direction, but packed together, circles waste about 9% of the area in gaps. The hexagon gives you almost that same isotropic spread and fills the drawer to 100%.

There is a second advantage, and this one is measured in filament. For the same cell area, the hexagon needs nearly 7% less perimeter than a square (the perimeter/√area ratio is 3.72 versus 4.00). Less perimeter per cell means less wall to extrude, less time and less material for the same usable pocket. It's the same reason bees build in hexagons rather than squares: wax is expensive.

That leaves load. When three walls meet at a honeycomb vertex, they meet at 120°, not the 90° of a square grid. That obtuse angle is a balanced knot: the three walls pull symmetrically, there is no square corner at which stress concentrates, and no edge where an impact can start peeling the layers apart. A hexagonal grid stands up to being crushed a good deal better than a square grid of the same wall thickness.

The shared wall is a whole number of perimeters

Here is the first printing failure. Between two adjacent cells there is a single wall, and the slicer traces that wall as a strip of fixed thickness, filled with beads. The bead width isn't set by the nozzle — the slicer fixes it: with a 0.4 mm nozzle it sits around 0.42–0.45 mm by default, though you can pin it yourself to a round value. What matters is not the exact figure but that the wall thickness be a whole multiple of that width. If it is, the slicer fills it with solid perimeters that weld to one another and out comes a solid wall. If it isn't, a narrow strip is left down the centre, too thin to take another full perimeter.

What the slicer does with that strip depends on its version. Variable-width perimeter generators — Arachne, on by default in PrusaSlicer and Cura for several versions now — adjust the width of each bead and fill the wall with no gap left. Fixed-width ones, by contrast, plug the strip with a thread of gap fill that barely bonds: that is the soft seam you sometimes see splitting the wall down the middle, a line that cracks at the first squeeze. Sizing the wall in whole beads spares you the problem in either mode.

It's worth being honest about what "the standard" means here. Unlike Gridfinity, the family of hexagonal drawer organisers has no single normative dimension sheet: they are community systems, often parametric, and each author picks their own across-flats and height. What is fixed is the geometry above — the relations between side, across-corners, area and pitch are those of any regular hexagon — and the wall thickness, which your nozzle imposes, not the designer. Those are the figures you can reason from without inventing anything.

Whole block or loose cells

You have two ways to build the grid, and each one hands you a different problem.

If you print the whole block in one piece, the cells share a wall for real: a single 0.8 mm wall between each pair of neighbours. You save the material and the time of the second wall, and the part comes out as rigid as a continuous honeycomb. It's what you want whenever the grid fits on the bed and you don't need to reconfigure it.

If you print loose cells to slot into one another — because you want to change the drawer's layout, or because each cell won't fit any other way — every cell carries its own wall and you join them with a male-female fit. And that fit is where FDM bites you. As explained in Real printed clearances, in FDM holes print undersize and pegs print oversize: every surface creeps inward. So a fit drawn to nominal size — zero clearance — comes out as interference: it won't go in, or it goes in splitting the wall between layers.

That's why the male-female fit is designed with clearance per side, not to nominal size. For a fit you assemble and disassemble by hand, the reference in PLA is 0.10–0.15 mm per side, so the total gap across the joint is double that. It is the location fit from that same article: it holds, stays centred and pulls apart with your fingers. But be clear about what that geometry buys you. A positive, uniform clearance on every face gives a friction location fit, not retention: it doesn't click or latch; it stays put because it rubs. If you want it to truly retain, you need a localised interference — a barb or lip with its own negative dimension — that snaps and grips, while the rest of the cell keeps its per-side clearance.

If you come up short, the wall splits along a layer line when you force it; if you overdo it, the cells rattle and you won't know why the block lost rigidity. Err on the loose side: almost every FDM bias tightens, so the safe error is the loose one. And this number is for PLA: in PETG, which oozes more and holds dimensions on holes worse, or in ABS/ASA, which shrink appreciably as they cool, the fit ends up tight, so recalibrate it upward before reusing the same value.

Print the prism upright

Orientation is not a finishing detail; it decides the geometry of the fit. Print the prism with its axis vertical, cell mouths facing up and the bottom resting on the bed.

That way the hexagon walls print vertical, traced as clean perimeters layer by layer — the orientation in which FDM is most dimensionally accurate. The cell bottom is the first layer, with no overhangs. No support is needed anywhere.

Lay it on its side and the opposite happens: the bottom becomes an overhang that has to be propped, and the top of the cavity, unsupported inside, sags and warps the mouth. A fit calibrated upright stops working the moment you turn the part.

Honeycomb works on the wall, too

The hexagonal cell doesn't live only in the drawer. The same tile, stood on edge and pinned to a vertical panel, is the basis of honeycomb-type wall systems, where each tile hooks to its neighbours and to the backing plate along the same 120° edge. The plane changes — from the drawer's horizontal to the wall's vertical — but the logic is identical: hexagons that tile without gaps, shared walls sized in whole beads, and fits you have to open per side to offset the FDM bias.

If you're going to organise vertically, the fit is no longer just cell against cell but tile against wall, with its own clearance and its own print orientation. That's developed in Honeycomb Storage Wall: a continuous grid of hexagonal tiles.

Start by deciding how you build the grid — monolithic block with a 0.8 mm shared wall, or loose cells with a 0.10–0.15 mm per-side fit — and then print a single test cell before you print the rest. A fit-tolerance tower, like the one in Real printed clearances, gives you the exact number in an afternoon and saves you a drawer full of cells that don't fit.