O-ring groove for 3D prints: depth and width from O-ring size

Cut the groove 70 to 80% as deep as the O-ring’s cross-section and about 1.3 to 1.5 times as wide. That squeezes the ring 20 to 30%, the static face-seal range in the Parker O-Ring Handbook (ORD 5700), and leaves it room to fill about 75% of the groove. On a printed part, the direction of the print lines under the ring and the density of the walls decide whether that groove holds.

How deep should an O-ring groove be in a 3D print?

An O-ring groove in a 3D print should be 70 to 80% of the O-ring’s cross-section deep, which squeezes the ring 20 to 30% in a face seal. For a 2 mm O-ring that is 1.4 to 1.6 mm. The Parker O-Ring Handbook lists 0.050 to 0.054 in (1.27 to 1.37 mm) for a 0.070 in ring.

Gland depth is the distance from the groove floor to the face the ring presses against. Width matters as much as depth. Parker calls for a gland fill (the share of the groove’s cross-section the ring takes up) of 60 to 85%, with 75% the optimum. At least 10% must stay empty for swell, heat and tolerance stack-up, so a narrow groove overfills even at the right depth.

Face seal: the ring is pressed between the groove floor and a flat lid

O-ring cross-section	Gland depth	Groove width	Squeeze
1.5 mm	1.0 mm	2.1 mm	33%
0.070 in (1.78 mm), AS568 -004 to -050	1.27 to 1.37 mm (0.050 to 0.054 in)	2.57 to 2.72 mm (0.101 to 0.107 in)	19 to 32%
2.0 mm	1.4 mm	2.5 mm	30%
2.5 mm	1.8 mm	3.4 mm	28%
0.103 in (2.62 mm), AS568 -102 to -178	1.88 to 2.03 mm (0.074 to 0.080 in)	3.45 to 3.61 mm (0.136 to 0.142 in)	20 to 30%
3.0 mm	2.2 mm	3.9 mm	27%
3.5 mm	2.7 mm	4.8 mm	23%
0.139 in (3.53 mm), AS568 -201 to -284	2.57 to 2.72 mm (0.101 to 0.107 in)	4.50 to 4.75 mm (0.177 to 0.187 in)	20 to 30%

The inch rows are AS568 sizes, the SAE standard for inch O-rings, from Parker’s Design Chart 4-3 with the groove width for liquids. Parker narrows the groove for gas or vacuum: 0.084 to 0.089 in (2.13 to 2.26 mm), 0.120 to 0.125 in (3.05 to 3.18 mm) and 0.158 to 0.164 in (4.01 to 4.17 mm). The metric rows come from Sealing Australia’s flange groove table, which is based on 70 Shore A rings; their squeeze is worked out from the nominal sizes.

Two metric rows sit past Parker’s limits. The 1.5 mm groove squeezes 33%, over Parker’s 30% maximum for static seals. The 2.0 mm row is shared with 1.9 mm rings, and a true 2.0 mm ring fills its 1.4 × 2.5 mm groove to 90%, over the 85% maximum. For a 2.0 mm ring, use the 1.5 × 2.8 mm groove from the worked example below.

Radial seal: the ring sits in a groove around a plug or inside a bore

O-ring cross-section	Gland depth	Groove width	Squeeze
1.5 mm	1.2 mm	2.1 mm	20%
0.070 in (1.78 mm), AS568 -0xx	1.27 to 1.32 mm (0.050 to 0.052 in)	2.36 to 2.49 mm (0.093 to 0.098 in)	22 to 32%
2.0 mm	1.65 mm	2.5 mm	18%
2.5 mm	2.1 mm	3.4 mm	16%
0.103 in (2.62 mm), AS568 -1xx	2.06 to 2.11 mm (0.081 to 0.083 in)	3.56 to 3.68 mm (0.140 to 0.145 in)	17 to 24%
3.0 mm	2.5 mm	3.9 mm	17%
3.5 mm	3.1 mm	4.8 mm	11%
0.139 in (3.53 mm), AS568 -2xx	2.82 to 2.87 mm (0.111 to 0.113 in)	4.75 to 4.88 mm (0.187 to 0.192 in)	16 to 23%

The inch rows are Parker’s Design Chart 4-2 for industrial static glands with no back-up ring; the metric rows are Sealing Australia’s cylinder groove table. Radial glands take less squeeze because the ring has to be pushed into the bore: Parker notes that face-seal squeeze in a radial gland “is quite difficult to assemble.” Here the gland depth runs from the groove floor to the opposite surface (the bore wall, for a groove on a plug), so it includes the gap between plug and bore.

How to design an O-ring groove

Pick a squeeze, get the depth from it, then size the width for the fill:

squeeze        = (cross-section - gland depth) ÷ cross-section
gland depth    = cross-section × (1 - squeeze)
groove width   = (π ÷ 4 × cross-section²) ÷ (fill × gland depth)

face seal:          groove depth = gland depth (lid closed flat on the face)
radial, on a plug:  groove diameter = bore diameter - 2 × gland depth
radial, in a bore:  groove diameter = rod diameter + 2 × gland depth

• Squeeze: 20 to 30% for a static face seal (Parker), 15 to 30% for static seals in general (Global O-Ring and Seal). Parker sets 30% as the static maximum.
• Minimum squeeze: at least 0.2 mm (0.007 in) on any ring. Below that, Parker says, “almost all elastomers quickly take 100% compression set.”
• Fill: 75%, inside Parker’s 60 to 85% range.
• Groove position on a face seal: Parker places the groove against the side the pressure pushes the ring toward. With pressure from inside, as in a pressurized container, make the groove’s outside diameter equal to the O-ring’s outside diameter. With pressure from outside, as in a box that keeps water out or holds a vacuum, make the groove’s inside diameter equal to the O-ring’s inside diameter.
• Rectangular face-seal grooves: round the inside corners to at least 3 times the cross-section, ideally 6 times (Parker). For a 2 mm ring, that is 6 to 12 mm.
• Radial stretch: choose a ring whose inside diameter is at most 5% smaller than the groove diameter. Parker does not recommend more than 5% stretch.

Worked example: a 2 mm O-ring in a printed box lid

The box keeps water out, so the pressure comes from outside and the groove’s inside diameter matches the ring. The ring is 50 mm inside diameter with a 2 mm cross-section.

1. Squeeze 25%, the middle of Parker’s face-seal range: depth = 2 × (1 - 0.25) = 1.5 mm.
2. Fill 75%: width = (π ÷ 4 × 2²) ÷ (0.75 × 1.5) = 3.14 ÷ 1.125 = 2.79 mm. Model 2.8 mm.
3. Groove inside diameter 50 mm, outside diameter 50 + 2 × 2.8 = 55.6 mm.
4. Squeeze is 0.5 mm, above Parker’s 0.2 mm minimum.
5. Match the layers. QIDI’s design guide says to make Z gaps exact multiples of the layer height. 1.5 mm is 10 layers at 0.15 mm. At 0.2 mm layers the slicer can only make 1.4 mm (30% squeeze, 80% fill) or 1.6 mm (20% squeeze, 70% fill); both stay inside Parker’s ranges.
6. Close the lid until the faces touch. Any gap between lid and box adds to the gland depth and takes squeeze away. In Prusa’s watertight tests, a clip-lid box leaked a few meters under water, partly because the lid did not press hard enough on the seal. A housing with a threaded lid held at 30 m.

Face seal or radial seal for a 3D printed part?

Use a face seal where the design allows. Its depth is set in Z, in whole layers below a flat top surface.

A radial seal’s squeeze depends on a printed hole, and holes are a printer’s weakest feature. QIDI’s fit guide cites a study of PETG (glycol-modified polyethylene terephthalate) parts in which 20 mm holes came out 1.4 to 2.59% off, about four times the error on outside dimensions. That is 0.14 to 0.26 mm per side, as large as the whole 0.35 mm squeeze on a 2 mm ring in a radial gland.

A face seal also takes more squeeze (20 to 30% against 16 to 24% for 0.103 and 0.139 in rings), so the same print error is a smaller fraction of the total. It keeps sealing even with the lid slightly off center. Parker’s radial glands allow 0.002 to 0.003 in (0.05 to 0.08 mm) of eccentricity and a total gap of 0.002 to 0.006 in (0.05 to 0.15 mm) between plug and bore.

When the part needs a radial seal, such as an end cap or a plug in a tube:

• Print the bore and the plug standing up, so each layer line circles the bore in the same direction as the ring.
• Size the groove from the printed bore. Print the bore, measure it, then model the groove at the measured bore minus twice the gland depth. Or correct every hole first with the hole compensation calculator.
• Keep the plug from wandering. Fit the plug to the bore next to the O-ring at 0.10 mm per side, the snug fit in the tolerance calculator. A loose plug can sit off center and unload one side of the ring.

Example: a 3 mm ring on a plug in a bore designed at 40 mm that printed at 39.7 mm. The groove diameter is 39.7 - 2 × 2.5 = 34.7 mm. The ring’s inside diameter should be 33.1 to 34.7 mm to stay under 5% stretch.

How much does print error change the squeeze?

Each 0.1 mm of depth error shifts the squeeze by 0.1 mm divided by the cross-section: 6.7 points on a 1.5 mm ring, 5 on a 2 mm ring and 3.3 on a 3 mm ring. Prusa states its printers are accurate to at least 0.2 mm. Here is what that does to a groove modeled for 25% squeeze:

Cross-section	Modeled depth	Squeeze at ±0.1 mm	Squeeze at ±0.2 mm
1.5 mm	1.13 mm	18 to 32%	12 to 38%
2.0 mm	1.50 mm	20 to 30%	15 to 35%
2.5 mm	1.88 mm	21 to 29%	17 to 33%
3.0 mm	2.25 mm	22 to 28%	18 to 32%
0.139 in (3.53 mm)	2.65 mm	22 to 28%	19 to 31%

Global O-Ring and Seal asks that every tolerance extreme stay between 5 and 30% squeeze. At ±0.2 mm, a 1.5 mm ring falls to 0.18 mm of squeeze, under Parker’s 0.2 mm minimum. At the other extreme it climbs to 38%. A 3 mm ring stays within 18 to 32%.

The ring adds its own tolerance on top: Parker lists ±0.003 in (±0.08 mm) for 0.070 and 0.103 in rings and ±0.004 in (±0.10 mm) for 0.139 in rings. So use the thickest ring the wall can take, 3 mm or 0.139 in where it fits. Measure the printed groove with the depth rod of your calipers before you order rings, and correct the model or the layer height if the depth is wrong. Radial glands start lighter, so they have less margin: the 3.5 mm metric row’s 0.4 mm squeeze drops to Parker’s 0.2 mm minimum at 0.2 mm of error.

How to print an O-ring groove that doesn’t leak

An O-ring seals along one line around the groove, and anything that crosses it is a leak path. Parker limits static sealing surfaces to 32 microinches rms (root mean square roughness, 0.8 µm), and 16 for face-type gas seals. It adds that direction counts: lathe marks that follow the groove seal even when rough, while marks that cut across the ring need the fine finish. On a print, that means:

• Print a face-seal groove facing up. Its floor is then a top surface and its side walls are perimeters that run around the ring. Never let the seal line cross layer steps, as it does in a groove cut into a vertical wall or a bore lying on its side.
• Set the top surface pattern to Concentric. In PrusaSlicer it is the Top fill pattern (Print Settings, Infill); in OrcaSlicer it is Surface pattern. Concentric copies the perimeter shapes, so the floor’s lines also run around the groove.
• Make the mating face smooth. Prusa advises printing surfaces that touch a rubber seal face down on a smooth sheet, or ironing them. In PrusaSlicer, set the ironing type to All top surfaces to reach a groove floor; Topmost surface only irons just the part’s top face.
• Make the walls dense. Prusa’s watertight guide asks for at least 3 to 4 perimeters (5 to 6 is rarely needed) and a nozzle 5 to 10 °C hotter than the preset. It also calls for 5 to 10% more extrusion and 0.15 to 0.2 mm layers. For parts under pressure, Prusa starts at 4 perimeters and 60% infill and calls a fully solid wall ideal. In its tests, PETG leaked “through the seams and contact points between perimeters and solid infill.”
• Grease the ring. Prusa applies silicone grease to the O-ring before every use.

Print direction alone can decide whether a groove leaks. In Processing Magazine, Kyle Timm of Valin Corporation describes a printed assembly that kept leaking past the O-ring in its circular groove. The print lines ran radially from the center outward and formed channels under the ring. Printing the groove in a circular path stopped the leak.

Sanding, epoxy and vapor smoothing

• Sanding: sand around the groove, never across it, to follow Parker’s rule on mark direction.
• Epoxy: Prusa made PETG watertight by coating it with epoxy resin. Coat only walls with 4 perimeters and 60% infill or more; at 1 to 2 perimeters, Prusa says, the coating “will most likely crack under higher pressure.” The coating adds thickness, so measure the groove afterwards.
• Acetone vapor smoothing: this works on ABS (acrylonitrile butadiene styrene) and ASA (acrylonitrile styrene acrylate). Prusa’s guide says smoothing may make the watertight settings unnecessary. In Prusa’s own test, smoothed ASA leaked slowly, probably around the seal or through bubbles. The ASA also “slightly expanded during the smoothing process,” so measure the groove after smoothing.

Can you 3D print an O-ring?

You can print an O-ring in flexible TPU (thermoplastic polyurethane), but a printed ring seldom seals. Parker recommends 70 to 80 Shore A for most O-rings, and common TPU filaments run 85A to 95A. Prusa printed gaskets in 92A and 98A filament and none held water, even screwed down hard. Buy a molded O-ring for any real seal.

Material	Shore hardness
Parker N0674-70 nitrile O-ring compound	70A (71 tested)
Recreus Filaflex 70A	70A
NinjaTek NinjaFlex	85A
Stratasys FDM TPU 92A	92A
Polymaker PolyFlex TPU95	95A

Hardness matters on printed parts because their surfaces are rough. Parker explains that softer rings flow more easily into fine grooves in the mating surfaces, which matters most in low-pressure seals that the fluid does not push into place. At very low pressures, Parker suggests 50 to 60 Shore A.

A printed ring also loses its grip faster. Compression set is the share of the squeeze a ring fails to recover; at 100%, Parker says, the ring “just contacts mating surfaces but no longer exerts a force.” Stratasys lists 21% for FDM TPU 92A after 22 hours at 23 °C and 44% after 22 hours at 70 °C (ASTM D395). Parker’s report on N0674-70 nitrile gives 12.6% after 22 hours at 100 °C (212 °F).

Prusa’s sealing tests found the same in practice. Its team printed gaskets in Flexfill 98A, Flexfill 92A and FilaFlex 40, with and without ironing. None held water between two printed parts, even when screwed down hard enough to damage the threads. The team found no material printable on a standard filament (FDM) printer “soft enough to work as a proper seal between two 3D-printed surfaces,” and called O-rings in correctly modeled printed parts “the best and most reliable method.”

A printed ring is still useful as a fit check while you size the groove, or as a placeholder until molded rings arrive. Recreus Filaflex 70A is the softest grade in the table above, but its maker does not recommend it for all-metal hotends or Bowden extruders. For any seal that must hold water or air, buy molded rings. Prusa points to diving shops, car shops and hobby markets.