Strong isa design decision.

Past the commodity plastics

The strength question arrives early in almost every quote we review, and it usually arrives attached to a material name: a commodity plastic for the first prototype, a high-temperature thermoplastic for anything hot, and a fiber-filled grade for the parts in between. However, strength in a composite part is not a property of the plastic you started with: it is a decision about reinforcement, and about where that reinforcement sits relative to the load your part actually sees. Get that decision right and a printed part stands in for machined aluminum. Get it wrong and it is a desk toy.

These two commodity materials built the desktop-printing world, and they earned it on cost and ease, because PLA and ABS are inexpensive, forgiving to run, and entirely adequate for a concept model or a fit check. Neither belongs on a production line. PLA creeps and softens near a warm engine bay; ABS warps and is mechanically unremarkable, and any environment with real heat, solvents, or repeated load rules both of them out. The interesting question starts one tier up.

The high-temperature trap

The pursuit of genuine thermal and chemical performance leads to PEEK and ULTEM (PEI), and both earn their reputation on the datasheet: PEEK carries a melting point around 343°C and, when fiber-reinforced, a heat-deflection temperature into the low 300s°C, self-extinguishing, chemically resistant, and widely used in aerospace.³ Every property in that list is real, and every one of them is conditional on printing the material correctly, which is where the trap actually sits: not in the polymer, but rather in what the polymer demands of the process.

In practice, printing PEEK well means a nozzle near 400°C, a heated chamber around 120 to 135°C so the polymer can crystallize in place, and a bed up near 230°C.³ Miss that thermal window and semi-crystalline shrinkage produces exactly the warping and delamination that plagued ABS for two decades, now on a spool that costs well over a hundred dollars a kilogram.⁴ The chamber is the piece most shops underestimate, because the polymer crystallizes as the part cools, and a chamber that drifts a few degrees may leave stress locked into the geometry, which is a process failure rather than a material one.

For a part that has to repeat, the barrier to entry is the real specification, because a material can be right on paper and still be unbuildable at your volume, on your equipment, within your tolerance. That is why so many “we tried PEEK” stories end with a warped part in a drawer. The polymer did not fail those parts: the process window did, and it closes faster than most build schedules allow.

PEEK is not hard because the polymer is exotic. It is hard because it will warp the instant your chamber drifts a few degrees.

The manufacturability tax on high-temp plastics

A better base: chopped-fiber nylon

A more pragmatic path splits the problem in two: choose a base that behaves on the machine, then add the strength separately. That base starts with Nylon 6, which brings an excellent surface finish, strong resistance to the oils and coolants of a real shop floor, and far friendlier dimensional behavior than PEEK. Nylon alone is flexible and soft, so the base is blended with chopped, short-strand carbon fiber. That composite, which Markforged calls Onyx, prints stable and stiff, resists the warping that limits high-temp plastics, and gives you a matrix genuinely worth reinforcing.⁵

This is the floor, not the ceiling. Chopped fiber is stiffening rather than structure, since the short strands raise stiffness and help the part hold its dimensions, though no single strand spans the distance a load has to travel. What the base buys you is a part that comes off the machine clean, flat, and stable enough to trust as a datum. What it does not buy you is aluminum-class strength, and that is the job of the next material in the stack.

Continuous fiber does the work

The step that lets a printed part rival metal is continuous fiber, laid into that matrix strand by strand along the load path. Defining the system first makes the rest of the decision straightforward: a thermoplastic matrix that holds geometry and surface finish, a continuous filament that carries tension along its own length, and a toolpath that decides which layers the filament occupies and which direction it runs in. The matrix is what you hold, and the fiber is what holds your part. Four fibers cover the work, and each one answers a different failure mode:

• Continuous carbon fiber is the strength-and-stiffness choice, and the one that stands in for machined aluminum. The strength-to-weight is superb, and the failure is brittle once the fiber is pushed past its limit, so the design carries margin.
• Fiberglass is stiffer than any base plastic, tolerant of intermittent loading, and inclined to deform before it fractures. It is the versatile default for the parts where carbon is more than the load case asks for.
• Kevlar (aramid) is ductile and impact-tolerant, in that it bends and comes back, which is what makes it right for robotic end effectors, soft jaws, and guards that take repeated abuse in daily handling.
• HSHT fiberglass is high-strength, high-temperature glass with a long elastic range, suited to heat and to clamping tools that must deform under pressure and snap back to shape.

As a selection rule, the load case decides the fiber and the geometry decides the routing. A part under a steady bending moment wants carbon along the tension face, while a part that gets dropped, clamped, or knocked into wants the ductility of aramid or glass, where energy goes into deformation instead of into a crack. Where a part sees both, the choice comes down to which failure would take your line down first, and that is a conversation worth having before the file is ever sliced.

Strength you place on purpose

The advantage over a traditional layup is selectivity. A hand layup reinforces everywhere, expensively, because the reinforcement arrives as a sheet and the sheet goes where the mold is. Composite printing routes fiber only where the bending moment actually lives: concentric rings around a bore, isotropic sandwich layers across a panel, and plain matrix everywhere the load case says nothing is happening. Strength becomes a dial you set layer by layer, rather than a bulk property you pay for across the whole part.

With the fiber routed, the remaining question is how load gets into the part at all. A composite carries load well along a fiber and comparatively poorly across a layer boundary, so the interfaces (bolted joints, threaded bosses, and keyed faces) decide how much of your layup ever gets used. Metal hardware is what bridges that gap: heat-set inserts, bolts through captive nuts, and machine keys move concentrated load off a small plastic feature and into the bulk of the part, and we work that hardware through in the brief on threads that hold. Routing and hardware are one design conversation, not two.

What it replaces, and for how much

In the right geometry, continuous carbon fiber laid into an Onyx matrix can perform in the same class as 6061-T6 aluminum. Published material data (Markforged’s composites datasheet, tested by ASTM methods) reports reinforced flexural strength around 540 MPa in the fiber direction against roughly 480 MPa for 6061-T6.¹² Those are coupon figures measured in the reinforced test direction, and they are not a forecast for your bracket. A real part’s performance depends on fiber routing, orientation, geometry, and the load case, which is why we ask for the load case along with the model.

For most of the parts we quote, the economics are the headline rather than the strength numbers. A complex bracket that would otherwise be machined from an aluminum billet, with the CAM programming, work-holding, and tool-clearance planning that implies, prints overnight with fiber placed exactly where the loads are, often at a fraction of the machined cost. You skip the CAM step and the tool-clearance geometry problems entirely, and both of those are engineering hours spent before a single chip is cut. And because it is fabricated domestically in composite, it avoids the freight, brokerage, minimum-order quantities, and applicable import duties that ride along with imported metal tooling; we work that cost math through in the cost-curve brief.

Overall, the four tiers answer four different questions: commodity plastics prove a shape, high-temperature thermoplastics survive an environment at a real manufacturability cost, a chopped-fiber base gives you a part that comes off the machine stable, and continuous fiber gives you strength along a path you selected. Only the last of those is a strength decision. The other three are decisions about process, environment, and finish, and each is worth making deliberately rather than by default.

Perhaps the more useful habit is to stop treating material selection as the strength decision at all: the base sets what the process can hold, the fiber sets what the part can carry, and the routing sets where that strength goes. Each of those is free to change in CAD and expensive to change on a running line. The parts that hold are the ones where that sequence happened before anyone opened the slicer.