Printing Custom TPU Gaskets and O-Rings: Durometer Selection, Compression Set, and Design Tips

Needing a gasket or O-ring for a one-off enclosure, a replacement seal for a discontinued appliance, or a custom sealing profile that doesn't exist in a standard hardware store size is a common enough problem that it's worth knowing when a 3D printer can actually solve it. TPU (thermoplastic polyurethane) is flexible enough to compress and seal, printable on the same machines already covered for general flexible filament work on this site, and cheap enough to iterate on quickly. It is not, however, a drop-in replacement for real rubber or silicone in every application — this guide covers what actually determines whether a printed TPU gasket will hold a seal, and where it will let you down.

Durometer: The Number That Actually Matters

Shore hardness (durometer, typically given on the Shore A scale for flexible materials) measures how resistant a material is to indentation — lower numbers are softer and more compliant, higher numbers are stiffer. Standard TPU filaments span a useful range for gasket work:

Shore HardnessFeelGasket Use ~85A-90A (common "TPU 95A" filaments)Firm, rubber-band-likeLight sealing where the mating surfaces are flat and well-machined; better dimensional stability for thin sections, but needs more compression force to seal well against an imperfect surface ~60A-70A (softer TPU blends)Noticeably softer, closer to a pencil eraserConforms better to slightly uneven or rough mating surfaces at lower clamping force; the better default choice for most enclosure gasket work Real silicone/rubber O-rings for reference~40A-70A typical, task-dependentStill generally outperforms printed TPU for long-term sealing under sustained compression — see the compression set discussion below

Softer filament is harder to print reliably — it flexes in the extruder path (especially on Bowden setups) and is more prone to stringing and poor bridging — so there's a real tradeoff between printability and sealing performance. A direct-drive extruder, as covered in this site's direct drive vs Bowden guide, makes softer TPU meaningfully more manageable and is close to a requirement for reliably printing anything below about 85A.

Compression Set: The Reason This Isn't a Universal Solution

The property that makes or breaks a gasket's long-term performance is compression set — how much a material fails to spring back to its original thickness after being compressed for an extended period. A material with low compression set keeps sealing reliably for years under sustained clamping force; a material with high compression set gradually takes a permanent "squished" shape and stops applying sealing pressure, even though it looks fine sitting unloaded on a bench.

Printed TPU generally has noticeably worse compression set than cast silicone or EPDM rubber, especially over weeks to months under sustained load, and especially at elevated temperature. This is the single biggest reason a printed TPU gasket is a reasonable choice for some applications and a poor one for others:

Good fit: occasional-use enclosures (a project box opened periodically, not sealed for years), temporary or prototype seals, low-pressure dust/splash protection, gaskets that get replaced as routine maintenance anyway.
Poor fit: permanently sealed enclosures expected to hold a seal for years without inspection, anything under sustained high compression (a tightly torqued bolted flange), high-temperature environments (near a heat source, in a hot attic or vehicle engine bay) where compression set accelerates dramatically, or any application where a seal failure has real consequences (pressure vessels, anything holding back liquid under head pressure over the long term).

For a permanent, high-reliability seal, a printed TPU part used as a mold or master to cast a real silicone gasket (covered in this site's silicone mold making guide) often gets better long-term results than printing the final gasket directly in TPU.

Design Considerations for Printed Gaskets

Design for a defined compression ratio, not a loose fit. A gasket that's too thin for the groove or gap it seals won't generate enough sealing force; a common starting point is designing the uncompressed gasket cross-section 15-25% larger than the gap it needs to fill, then confirming actual compression in a test fit rather than guessing.
Round profiles print and perform better than sharp-cornered ones. A circular or D-shaped cross-section (closer to a true O-ring profile) distributes compression more evenly than a printed square or rectangular gasket cross-section, which tends to concentrate stress at corners.
Print orientation affects sealing surface quality. Printing a flat gasket lying on the bed gives the smoothest sealing face on the bottom layer but a rougher, layer-lined top surface; for a gasket sealing on both faces, consider whether a slight post-print surface treatment (light sanding, or a brief heat-gun pass to slightly reflow the surface) is worth the extra step for a critical seal.
Wall thickness and print settings matter more than for rigid parts. Under-extrusion or gaps between perimeters that would be a minor cosmetic issue on a rigid bracket become a leak path on a gasket — a slightly higher extrusion multiplier and 100% or near-100% infill (small gasket cross-sections print fast enough that infill percentage barely affects print time anyway) reduce the risk of a porous, leaky part.
Retention features beat friction fit for anything handled repeatedly. A groove, lip, or light adhesive bead holding the gasket in place prevents it from shifting or falling out during assembly and disassembly — designing the gasket to be captured rather than merely sandwiched saves real assembly frustration.

When to Reach for a Real O-Ring Instead

Standard O-ring sizes (AS568 in the US, metric equivalents elsewhere) are inexpensive, available in real sealing-grade rubber compounds (Buna-N, EPDM, silicone, Viton depending on chemical and temperature needs), and dramatically outperform printed TPU on compression set and long-term reliability. Before designing a fully custom printed gasket, it's worth checking whether the sealing gap can instead be designed around a standard O-ring groove — the printed part becomes just the housing with a correctly-dimensioned O-ring groove, and the actual sealing element is a five-cent, purpose-built rubber part that will outlast the printed housing around it. This hybrid approach — print the housing, buy the seal — is often the better engineering choice over printing the seal itself, and is what most commercial enclosures actually do even when everything else about the design is 3D printed or otherwise custom-fabricated.

A Practical Workflow

Determine whether this is genuinely a case where a standard O-ring and groove design will work — if so, that's usually the better answer.
If a fully custom profile or size is unavoidable, choose TPU hardness based on how compliant the mating surfaces are (softer for rough or slightly warped surfaces) and the direct-drive vs Bowden capability of the printer available.
Design in a defined compression ratio and a retention feature rather than a loose friction-fit gasket.
Print a test piece, check actual compression and seal quality (a simple air or water test, depending on the application, beats visual inspection alone), and iterate on cross-section dimensions before committing to a final design.
For anything needing long-term reliability under sustained load, either plan for periodic gasket replacement as routine maintenance or use the printed part as a mold master for a cast silicone version instead of the final sealing element.

Printed TPU gaskets are a genuinely useful tool for the right job — quick prototypes, low-stakes seals, and situations where no off-the-shelf part exists and waiting for a custom rubber part isn't an option. Knowing where compression set will quietly undermine a "permanent" seal, and when a standard O-ring groove is simply the better design, keeps this technique in the category of a smart shortcut rather than a seal that looks fine on day one and fails six months later.

Related Guides

Loading_