Which 3D printing filament should you choose? A materials guide and selector

“Strong” can mean several different things

Stiffness describes resistance to deformation under load. Strength at failure concerns the stress sustained before failure under defined test conditions. Toughness relates to the energy absorbed before fracture, while impact testing studies rapid loading. A part can be stiff yet break with little warning deformation.

Ductility is the ability to undergo plastic deformation before fracture. That deformation need not disappear when the load is removed. Elasticity concerns recovery. A clip assembled once and a tongue that must return to its position thousands of times can therefore need different materials and designs.

Hardness measures local resistance to indentation using a defined method. A Shore number does not calculate how much a finished part bends. Thickness, shape, walls and infill all influence the result.

Another failure mode is creep, progressive deformation under sustained load. A bracket may pass a short test and move out of position after weeks. Heat can accelerate the process. Hanging a weight briefly is not sufficient validation for permanent service.

Replace “I need a very strong plastic” with a specific statement: “this bracket must keep the component aligned”, or “this latch must bend during assembly and recover”. That lets you compare relevant behaviour rather than search for an absolute winner.

PLA is a family, not a single recipe

Resin selection and formulation affect behaviour: modifiers, pigments, fillers, moisture and filament manufacturing all matter. The crop supplying the sugar does not, by itself, explain a spool’s toughness or quality. Renewable feedstock also does not mean a printed object will rapidly biodegrade in a garden or the sea.

Modified PLA formulations, including products based on Ingeo 3D850 or 3D870, can extend applications beyond particular conventional PLAs. Check which property improves, how it was tested and what treatment was applied. These grades are not equivalent to ABS for every application, and different commercial formulations are not necessarily identical.

Annealing can promote crystallisation in suitable PLAs and alter thermal behaviour. It can also cause shrinkage and distortion. Where a datasheet requires treatment to achieve its results, those results cannot be assigned to an untreated print. For mating parts, dimensional change may matter as much as a thermal improvement.

High-speed PLA follows another priority: enabling faster processing with appropriate conditions. It does not automatically establish better toughness, lower stringing or higher heat resistance. Material, hotend and volumetric demand still determine the usable process; a commercial name does not remove that relationship.

Compare the specific filament documentation and distinguish moulded specimens, printed specimens and treated results. A simple indoor model may not justify extra processing. For a functional part, a representative sample is more informative than the word “PRO”.

Nominal material properties and manufacturing results

Datasheet comparisons cannot capture every process defect. An ABS part with insufficient interlayer bonding may fail before a well-made PETG equivalent, even if ABS performs better in a different test. This is not universal superiority for PETG: the selected process must manufacture the actual geometry reliably.

PETG belongs to the copolyester family related to PET. Its formulation changes crystallisation tendency and processing behaviour. It should not simply be identified with a drinks bottle, nor should that analogy establish food-contact suitability for a print.

Natural filament may be transparent, but line interfaces, voids, geometry and pigments strongly influence a finished object. Transparent feedstock does not guarantee optical transparency in FDM.

Watch moisture condition, suitable printing temperature and sustainable flow. Changing retraction without checking moisture can make you chase a problem that is elsewhere. Drying is not a generic temperature setting either: follow the grade and spool requirements.

PETG can be useful for certain replacement parts on open printers where design and temperature allow it. Do not extend that idea to ducts near the hotend or components operating inside a heated chamber. Location and loading are part of the specification even when two replacement parts look similar.

Why deformation starts before printing finishes

Cooling material contracts. Uneven cooling, or restraint from a broad base, introduces stresses that may lift corners, bend walls or contribute to interlayer cracks. Adhesion helps hold the base but does not, on its own, correct the part’s thermal state.

Consider bed, adhesion, environment, orientation and geometry together. Long bases, abrupt thickness changes and sharp corners can make the job harder. A fillet or different orientation may help more than repeatedly adding adhesive without understanding the cause.

The part-cooling fan cools deposited material; room ventilation or extraction manages emissions. They are different functions. Too much local cooling can damage bonding or cause distortion, while particular bridges may require targeted adjustment. Avoid universal always-on or always-off rules.

ASA is not merely ABS with an outdoor label: the polymer formulation differs. Nor do all ABS grades respond identically to sunlight; stabilised formulations and protective coatings exist. For a straightforward choice, ASA deserves consideration where UV is a requirement, while standard ABS should not automatically be treated as UV resistant.

Custom duct adapters illustrate the freedom of FDM geometry, but their function may also require checking gases, chemicals and temperature. Making an adapter is not a service-life test for every installation. Similarly, assess an HPRO configuration by its actual equipment rather than transfer every option’s capability to an entire range.

Filament must transmit force before it becomes the part

The extruder pushes material towards the hotend. Soft filament may compress, buckle or escape through a gap before transmitting the required pressure. The harder it is to melt and force through the nozzle in those conditions, the more important confinement and a low-resistance path become.

A Bowden system adds length and friction. Firm flexibles such as 95A can work in suitable Bowden arrangements, but that does not support a recommendation for every machine or for 70A and 60A. Study extruder, guidance, hotend and profile carefully for these grades. A manufacturer’s recommendation for one configuration does not certify a different setup.

Shore A hardness measures indentation resistance using a defined method. Within a range it helps describe softness, but it is not a universal elasticity scale. Elongation at break does not mean all that strain is recoverable. Seals, wheels and bump stops may additionally depend on compression set, fatigue, abrasion or dynamic behaviour.

Geometry can dominate feel: thickness and walls can make a soft material into a firm component, while a lattice permits deformation through a different structure. Copying hardness and infill from another object does not reproduce its response in a new shape.

Sustainable flow depends on feeding, viscosity, nozzle and temperature. Increasing speed can raise pressure until the filament becomes unstable before the hotend exhausts its thermal capacity. Adjust for flow and quality rather than assign a universal speed to a Shore number. Moisture can also disrupt extrusion; storage and grade-specific drying belong to the process.

For FleX-920, separating hardness, toughness and rebound avoids misleading promises. Without a restitution test, a rebound coefficient or superiority over every TPU cannot be claimed. Compare equivalent geometry and a defined function, not just a piece of filament between your fingers.

Two engineering families, several things to verify

PA6, PA12 and copolyamides do not have identical moisture uptake, stiffness or thermal requirements. Water in filament during extrusion can cause bubbles and harm finish and properties. Moisture conditioning of an already printed part can alter stiffness and ductility. These are related but different situations.

Drying before printing and protecting the supply throughout a long job may be necessary. Follow the manufacturer’s time and temperature. A desiccant box preserves condition but is not always an effective drying process; check the spool’s limits too.

For gears, bushes and guides, wear resistance is not reduced to “choose nylon”. Mating surfaces, loading, speed, temperature, lubrication and finish matter. A fibre that improves stiffness does not necessarily improve every sliding contact.

PP is semicrystalline, and shrinkage and low surface energy make adhesion particularly important. Some formulations use dedicated surfaces or primers under conditions unlike another commercial PP. Shared initials do not justify copying a bed profile.

A living hinge shows the interest in repeated flexing. Yet selecting PP does not establish its service life: thickness, radii, orientation and continuous toolpaths influence strain concentration. A sample undergoing the real motion can reveal weaknesses before the whole component is made.

Chemical resistance needs a substance, concentration, temperature, contact duration and stress state. Neither PA nor PP should be recommended for an unidentified chemical by a general selector. Familiar use of PP in packaging does not validate food-contact or orthopaedic suitability of a print.

Fibre does not replace the base polymer

A composite combines a polymer matrix and a filler. Carbon-filled PETG still has a PETG matrix; carbon-filled PA retains requirements associated with its polyamide. Fibre alone cannot establish service temperature, a drying profile or universal chemical resistance.

Many FDM filaments contain short fibres. Their orientation during processing contributes to directional behaviour, while the print retains layer interfaces. A strong test result does not prove the same properties in every axis or around a screw near a thin wall.

Stiffness can increase while elongation at break decreases. That explains why a bracket may benefit while a flexible latch does not. Fibre percentage alone does not identify a superior formulation: matrix, fibres, distribution, bonding and processing all matter.

PETG-CF may offer an accessible engineering route where stiffness is useful without automatically taking on all PA-CF requirements. Some geometries print and finish conveniently. This practical observation does not make every PETG-CF identical or establish better bridges and bonding in every profile.

For a nozzle, check wear resistance and the hotend’s mechanical interface separately. A hardened nozzle of the wrong geometry remains incompatible. Minimum diameter depends on the product and particles, not a universal rule for every filled filament. Also inspect the drive system; an “abrasion-resistant nozzle” checkbox cannot describe the whole configuration.

Expanding by 50% is not saving 50% of the weight

Imagine the same mass occupying 1.5 times its original volume. Filling the original volume would ideally require mass divided by 1.5: roughly a one-third reduction, not half. This is a geometric relationship, not a guaranteed spool result. Walls, infill and required properties can change the actual saving.

A foaming filament can be relatively firm before the hotend and produce a softer part after expansion. This can help feeding, but does not establish a proportional gain in maximum flow or productivity. Melting and foaming need suitable conditions.

Oozing is a common challenge: material can keep leaving the nozzle after the extruder stops pushing. Temperature and expansion, toolpath continuity, retraction and travel must be adjusted together. Faster travel can reduce oozing time within machine capabilities; it does not correct an unstable profile on its own.

At HTA3D, our work on podiatry insole manufacturing has used 95A Foamy, rather than traditional Foamy. That distinction matters for profiles and properties. Describing fabrication and parameter development does not demonstrate clinical benefit or certify each finished component.

PLA-LW presents other compromises. Low weight may help model aircraft or display models, but walls, joints and loads need assessment. A lighter object with the same outer shape does not necessarily retain solid-material strength. Calibrate and check before extrapolating to more consequential parts.

Fillers can make passage through a small nozzle harder and alter finish or cohesion. Check the product’s minimum diameter and instructions. Decorative materials do not all use the same particle size or wear hardware like carbon- or glass-filled composites.

Thin walls and concentrated loads deserve attention. Geometry that works in unfilled PLA can be more fragile in another formulation. An attractive finish does not make a structural reinforcement. Print a sample with similar detail and wall thickness to check extrusion, appearance and handling.

An additional property does not certify a part

A flame-retarded material is not incombustible. A UL 94 result and an EN 13501-1 classification address different evaluations and are not interchangeable. Test, thickness, configuration and scope matter. Additives may also alter processing; do not assume the unmodified polymer’s profile applies.

Fire resistance, thermal behaviour under load and final-product suitability are different matters. A favourable filament classification does not establish certification of a device, enclosure or installation. For a regulated use, the applicable document and its scope belong in the project.

A conductive TPU can support investigation of resistive sensors or low-current signals. It does not replace metallic wiring: geometry, contacts and print orientation affect effective resistance. Measure the finished component and design around the result; do not extend the idea to power wiring or safety-critical electrical connections.

Measured antibacterial activity does not mean sterility, efficacy against every organism or universal protection in use. Nor does it remove cleaning, hygienic design or medical-application evaluation.

The selector therefore treats these as specialist enquiries. Price or ease cannot turn an undocumented function into an affirmative recommendation. If you need certification, identify which one and for which component before choosing a commercial label.

First ask whether the support is accessible. Mechanical access may make orientation or a breakaway interface sufficient. An enclosed cavity may justify soluble support and additional processing. Choose support alongside the main material.

Using a second material only at the interface can avoid making all support from a specialist filament. But the machine must perform the required changes and purge properly: contamination can weaken the part’s own layer bonding. The ability to pause a print does not solve every multi-material geometry.

Check both materials’ temperatures, adhesion throughout printing, moisture condition and removal procedure. Solvent-removable support also requires assessment of handling safety and used-bath disposal. Solubility is not an instruction to handle a chemical without its safety documentation.

Easy Off needs compatibility with the actual PLA; PVA needs particular moisture and storage attention. “Support” describes a manufacturing function, not a family with one common behaviour.

Recycled material may come from manufacturing recovery, unlike material reclaimed after consumer use. Identify origin and product control rather than assume every recycled grade exactly retains a virgin counterpart’s properties.

Colour transition happens while changing colour during production. Where colour is the only deviation and the spool meets its formulation’s specifications, the corresponding profile may be appropriate. This is different from mixing unrelated polymers.

Ask whether you need a specific, repeatable colour. It may not matter for a hidden part, but it may matter for a visible series or matching replacement. Savings and material recovery can be benefits without inventing a quantified environmental impact or universal mechanical equivalence.