Machining UHMW: Why Tolerances Slip, and How to Design Around It

UHMW is the material that outlasts everything on a conveyor and then fails inspection for being 0.010" out — measured at 4 p.m., in a part that was on size at 8 a.m. Nothing went wrong on the machine. The material did exactly what its datasheet says it does, and the drawing asked for something the material was never going to give.

Here is what UHMW is, what it does after the cutter leaves, and how we hold the tolerances that can be held.

What is UHMW?

UHMW is ultra-high-molecular-weight polyethylene (UHMW-PE, PE 1000 in the European naming). Chemically it is the same polymer as the HDPE in a milk jug; the difference is chain length. Where HDPE's molecular weight runs in the hundreds of thousands, UHMW's runs in the millions, and those very long, tangled chains are the whole story:

• Its abrasion resistance is the benchmark. UHMW is the reference material in the ISO 15527 sand-slurry test — everything else is scored against it. Long chains resist being torn out of the surface, so it survives sliding abrasion and repeated impact where harder-looking plastics wear through.
• It absorbs impact instead of fracturing. Unnotched impact specimens do not break. Notched results are finite and depend on the method — MCAM reports notched Charpy at 115 kJ/m² with a partial break (ISO 179-1/1eA) — so "unbreakable" is the wrong word for a notched or sharp-cornered part. It yields and stretches rather than cracking: elongation at break is 300% by ASTM D638, 50% by ISO 527 on the same sheet.
• It is soft, waxy and dimensionally restless. The same chain structure that makes it tough makes it low-modulus, easy to deflect under a clamp, and among the most thermally expansive of the engineering plastics.

Virgin natural is the grade most machined parts come from, and everything below describes it. Filled and reprocessed grades (glass, oil-filled, anti-static, UV-stabilized) trade some toughness for a specific property, and they do not all machine the same way — MCAM's own guidance for glass- and carbon-filled grades reverses the unfilled rule, calling for reduced speed and increased feed on carbide, and its flame-retardant BurnGuard grade must be run dry because the FR package reacts with water- and oil-based coolants. Tell us the grade, not just "UHMW".

UHMW properties at a glance

Stock-shape datasheets, every row marked with its method. The two tensile figures disagree by a factor of two and both are right — one is yield, the other is ultimate at 300% elongation. A comparison that quotes 5,800 psi for UHMW next to a yield figure for another material is comparing different events.

The same warning applies to rows that look like independent confirmation. ISO and ASTM results on the same material routinely differ — tensile modulus below is 750 MPa by ISO 527 and 80 ksi (550 MPa) by ASTM D638, a 36% spread, because the test speeds differ. Two numbers agreeing across methods is a coincidence to check, not a validation.

Treat these as typical published values, not a specification; grade, color and stock size all move them. Work from the datasheet for the exact stock on a critical part.

Two rows explain nearly every UHMW tolerance complaint.

The number that matters: 110 µin/in·°F

Steel expands about 6.5 µin/in·°F. Aluminum about 13. UHMW: about 110 — roughly seventeen times steel.

Work it through on a real part. A 12" wear strip machined at 68 °F and measured at 78 °F has grown:

12 in × 110 µin/in·°F × 10 °F = 0.0132"

Ten degrees of shop drift — a door open, a machine warming up, sun on the inspection bench — is worth thirteen thousandths on a foot of material. A ±0.005" length tolerance on that strip is not a machining problem; it is a thermometer problem. The part will pass or fail depending on when and where it is measured — which is why a tolerance this tight on UHMW is meaningless unless both parties agree on the inspection temperature.

The same arithmetic applies in service. A 10-ft UHMW liner bolted down at 60 °F in a building that reaches 100 °F wants to grow more than half an inch (120 in × 110 µin/in·°F × 40 °F = 0.53"). Slotted holes, not round ones, or it buckles.

What moves the number — the levers, not a description of every job:

• Machine it, then leave it. Letting roughed stock come back to shop temperature before the final cuts, and again before inspection, is the single biggest lever on a tight dimension. Cutting puts heat into the part and UHMW sheds it slowly. It also costs time on the floor, which is why it is a thing to ask for rather than a thing that happens by default.
• Agree the inspection temperature before the part is quoted. Drawing tolerances are defined at 68 °F (20 °C) under ASME Y14.5 and ISO 1 — while the datasheets above report at 23 °C (73 °F), which on a foot of UHMW is another 0.007" of difference all by itself — so on a tight UHMW dimension, where and when it is measured is part of the specification. Tell us a dimension is critical and we will let the part come to a stated temperature, inspect it there, and record that temperature on the report. If your incoming inspection then happens in an un-conditioned warehouse in August, the disagreement is predictable and it is not a machining fault.
• Design the fit around the expansion, not the tolerance. Clearance where it slides, slots where it is bolted, and a tolerance band that reflects the service temperature range. We would rather have that conversation before the drawing is released than after the parts are rejected.

The second problem: it moves after the cut

UHMW stock — sheet especially — carries internal stress from extrusion or compression molding. Machining removes material asymmetrically, the stress rebalances, and the part relaxes into a new shape over hours or days. A flat plate goes to a slight bow; a slot closes up a few thousandths.

The mechanism is stiffness, not heat. UHMW's tensile modulus is about 80 ksi (ASTM D638) against roughly 10,000 ksi for 6061 aluminum — call it one part in 125. It deflects under a fixture clamp, gets machined flat while deflected, and springs back when released. (Its heat-deflection temperature of 116 °F at 264 psi is often quoted here; that is a 0.25 mm deflection under load, not a softening point, and it is not the mechanism behind extrusion-stress relaxation. UHMW does creep under sustained load at modest temperature — but that is a service-life caution, not the reason a plate bows on the bench.)

The levers, again — some are free, some are not:

• Rough, rest, finish. Take the bulk of the material off, let the part relax, then bring it to size. On flat parts with a lot of material coming off one face, machining both faces balances the stress. Both add operations, and both are worth asking for when flatness is the thing that matters.
• Fixture on the part, not on the vise. Vacuum tables, low-pressure toe clamps, double-sided tape on a sacrificial plate. Anything that squeezes UHMW distorts it, and vise jaws leave a part that is parallel only while it is in the vise. This one changes the setup, so tell us before we quote it, not after.
• Sharp, positive, and polished. UHMW does not chip — it cuts like wax and it smears. Positive rake, a polished flute so the chip clears instead of welding, and a real chip load. MCAM's own figures for UHMW-PE are 0–10° rake turning, 0–15° milling, 3–5° drilling, in HSS or carbide — sharp HSS twist drills work well — and its troubleshooting table names too much positive rake as a defect cause, so "positive" is the rule and "as much as possible" is not. Rubbing (light finishing passes at low feed) generates heat, and heat is the enemy. Sharp tooling matters more than feed rate.
• Air blast is the sensible default; coolant where the source calls for it. UHMW absorbs almost no water (0.1%), so coolant is not the stress-cracking risk it is with amorphous plastics like Ultem, and pressurized air is genuinely effective — it also keeps the stringy chips from wrapping the tool and re-cutting the surface. But "never use coolant on UHMW" overstates it: MCAM recommends a non-aromatic water-soluble coolant for drilling — especially deeper than 2× diameter — for parting off, and wherever surface finish or close tolerance matters. Heat is the enemy, and on a deep hole air alone does not clear it. (The exception runs the other way for flame-retardant BurnGuard grades, which must be run dry.)
• Deburr with the drawing in hand. UHMW throws long, tough burrs that do not snap off; they have to be trimmed. Edge condition should be called out, because "break all edges" on a 300%-elongation material means something different than it does on steel.

What tolerances are realistic

One table, and one honest caveat under it. These are routine numbers — ordinary fixturing, no rest cycle, cut and shipped, one setup, nothing special bought. It is the case worth publishing because it is the one that surprises people, and because it is the baseline every other number moves away from.

This is not what the material is capable of. It is what you get when nobody spends anything on it, and it is where rejections come from.

Tighter is available, and the levers described above are what buy it — a rest cycle, both faces machined, vacuum or low-pressure fixturing, inspection at an agreed temperature. We have deliberately not published a second set of numbers for that case. There is no credible published tolerance table for machined UHMW to cite, and a plausible figure invented to fill the gap is exactly what this post exists to warn against. Send the print and we will tell you what we can hold on your part — its size, its shape, and how much material comes off one side all change the answer.

Why the numbers are where they are

Three things eat the budget, and only one of them is fixable by a thermometer.

1. The part is not at room temperature when you cut it. UHMW's thermal conductivity is 0.4 W/(K·m) — it holds the heat the cutter puts into it, and sheds it slowly. At 110 µin/in·°F, a part that is 10 °F warm from machining is 0.013" long per foot while you measure it. This is a real budget item and a temperature-controlled room does not fix it; only letting the part rest does.

2. It moves after the cut. Extrusion and molding stress rebalances when material comes off asymmetrically, and the part relaxes over hours or days — permanently. It does not come back when the part returns to 20 °C. Bow on a plate scales with span and thinness, not with the dimension being toleranced, which is why a thin 24" plate is harder than a chunky one of the same length.

3. It deflects while you hold it. At ~80 ksi, UHMW machines flat while clamped and springs back when released. Vise jaws produce a part that is parallel only in the vise.

What thermal expansion does not justify

It does not justify a looser number on the drawing. A drawing tolerance is defined at 20 °C (68 °F) under ASME Y14.5 and ISO 1 — so a part in spec at 20 °C is in spec, and machining it 0.015" undersize to survive a warm incoming inspection does not make it pass. It makes it a worse part that still fails.

What the expansion arithmetic justifies is a conversation and a drawing note, not a wider band:

• Put the reference temperature on the print. "Dimensions apply at 20 °C per ASME Y14.5" costs nothing and settles the argument before it happens.
• Tolerance the service range, not just the bench. A 10-ft liner that grows half an inch across a 40 °F building swing needs slotted holes and clearance, not a tighter ±.
• Say which dimensions are critical. We will let those come to a stated temperature, measure them there, and record it on the report.

If a print carries ±0.001" on a UHMW dimension longer than an inch or two, the fastest thing we can do is call. Sometimes it is a copy-paste from a steel drawing. Sometimes the fit really is that critical — and then the answer is usually a different material for that feature (acetal holds a fit; UHMW takes the wear) or a design change that lets UHMW float.

When UHMW is the right call

• Sliding wear and impact: chute liners, wear strips, guide rails, star wheels, timing screws, dock bumpers. It is the default choice for these for good reason.
• Food contact: virgin natural is FDA compliant and takes washdown; low friction means product does not stick.
• Cold: usable to −200 °F, the bottom of the range on our UHMW material page, and it stays tough where many plastics go brittle. Note the direction of travel, though: impact strength falls as temperature falls, and the datasheet floor is set by how much impact the part actually sees. Tell us the service temperature and what hits it.
• Chemical exposure: broadly resistant, which is one of the reasons it earns its place in washdown and process environments. It is a polyolefin, so the caution is swelling in hydrocarbon solvents. Environmental stress cracking is very much a polyethylene failure mode — ASTM D1693 is literally titled Environmental Stress-Cracking of Ethylene Plastics — but UHMW's enormous chain length gives it outstanding resistance to it, which is a large part of why it is the grade that survives in service. Send us the fluid, the concentration and the temperature and we will check it against the manufacturer's resistance data for the specific grade — a general "resistant" is not something to design a seal or a liner around. For sustained outdoor exposure, specify a UV-stabilized grade.

When it is not

• Above ~180 °F continuous. It softens long before it melts (275 °F); load-bearing parts creep. Nylon for warm — it runs to 210 °F where acetal stops at the same 180 °F UHMW does, so acetal buys no headroom here — and PEEK for hot.
• Tight fits and precision bores. See everything above.
• Stiffness. ~80 ksi modulus — about 1/125 of aluminum; it flexes. If the part has to span an unsupported gap and stay straight, it wants a rib, a backing plate, or a different material.
• Bonding and painting. Polyethylene is nearly impossible to glue; design for mechanical fastening.

How we quote UHMW parts

Send the drawing and tell us the service temperature range and what the part does — wear, slide, or take impact. We carry UHMW in sheet and rod, cut to size or machined to print in Linden, NJ; ask us for the sizes and colors currently on the shelf. And if a tolerance on the print is going to fight the material, we would rather say so on the quote than after your inspection.

Request a quote · UHMW material page