What if a bike were 3D-printed in one piece?

A one-piece printed motorcycle frame already exists in spirit: Airbus's APWorks printed a 6 kg frame in 2016. The whole bike in one print does not, and fatigue, not printing, is the reason.

Inside the chamber of a metal 3D printer while a laser melts powder layer by layer
Photo: René Volfík, CC BY-SA 4.0, via Wikimedia Commons

You could 3D-print a motorcycle's frame, swingarm and bodywork as one metal piece today, and Airbus's APWorks came close in 2016 with a 6 kg printed frame on a 35 kg electric bike. You could not print the whole working bike in one go: the tyres, magnets, battery cells and bearings need materials and processes no single printer combines. The hidden killer is fatigue, because the rough inside walls of a one-piece print can't be machined, and as-built roughness cuts fatigue strength by about 40%.

Key takeaways
  • 6 kg: the printed Scalmalloy frame of the APWorks Light Rider, a 35 kg bike.
  • ~40% lower fatigue strength for as-built printed aluminium surfaces, per a Ford study.
  • 5 × 10⁷ wheel turns in 1 lakh km, so a frame lives in the high-cycle fatigue regime.
  • 2–40 cm³/h is a single laser's build rate, so a frame takes days, not minutes.
  • 2.7× heavier tubes if you must design to as-built rather than machined fatigue strength.

3D printing has a reputation for making anything in one shot. Print a phone stand, print a drone frame, so why not print a motorcycle and ride it off the build plate? The idea is less silly than it sounds. Aerospace and racing teams already print load-bearing metal parts.

The useful question is what "one piece" would cost you. I split the bike into what can be printed together, what can't, and what would break first if you tried.

Has anyone already printed a motorcycle?

Yes, at least three times, and each one shows a different limit. None is truly one piece.

MachineYearWhat was printedMaterial and processMass
APWorks Light Rider (Airbus)2016Hollow frameScalmalloy aluminium, laser powder bed35 kg bike, 6 kg frame
BMW S 1000 RR prototype2018Frame and swingarmMetal, selective laser meltingNot released
BigRep NERA201815 parts: frame, fork, rims, airless tyres, seatPolymer filaments, large-format FDM60 kg

The Light Rider used a 6 kW motor and topped out at 80 km/h; APWorks offered 50 units. BMW showed its rough-surfaced S 1000 RR frame at its 2018 Digital Day without releasing any structural figures. The NERA printed everything except the electronics, but in 15 separate parts. Even J.Laverack's 2025 Speedform titanium bicycle is printed in three sections and welded.

Diagram of selective laser melting with a laser scanning a powder bed and a recoater spreading new layers
Selective laser melting: a laser fuses a thin layer of metal powder, the platform drops, a recoater spreads the next layer. Every part of a printed frame is built this way, tens of microns at a time.Photo: Chaolin Tan et al., CC BY 4.0, via Wikimedia Commons
The story

In July 2020, Porsche, Mahle and Trumpf announced something engine builders had said couldn't be done: aluminium pistons printed by laser metal fusion for the 700 PS engine of the 911 GT2 RS. The printed pistons were about 10% lighter than the forged ones in the production car, and Porsche said the engine could make up to 30 PS more. The trick was a closed cooling duct inside the piston crown, a shape you can't forge or cast. Zeiss measuring systems checked every part. Even so, the pistons were machined after printing, exactly like forged ones. Printing made the shape; machining made the surface.

VoxelMatters: Porsche 3D prints pistons for the 911 GT2 RS

Which parts of a bike can't be printed together?

Anything that needs a different material, a different process or a moving clearance. A laser powder-bed printer melts one metal alloy at a time. Polymer printers extrude plastic. No printer lays down steel, copper, rubber and lithium in the same build.

  • Tyres. Rubber is vulcanised, not melted. NERA's answer was airless printed polymer tyres, which work at walking speeds, not on a highway.
  • Motor and battery. Copper windings, rare-earth magnets and battery cells are made in their own factories.
  • Bearings. A printed bearing needs a gap to move, and printed surfaces are far too rough to run without grinding.
  • Brakes. Discs and pads are friction materials with specified compositions.
  • Fasteners and threads. Printed threads are possible, but a one-piece bike has none to service.

So a literal one-piece bike is out. The realistic version is a one-piece chassis: frame, swingarm, subframe, tank cover and footpeg mounts printed as one structure, with bought-in parts bolted on.

Printable as one piece vs bought inmotor, batterySolid orange: one printed metal chassis (frame, swingarm, subframe)Dashed: tyres, wheels, bearings, motor, cells, brakes, bought infork, bought in
A one-piece printed chassis is realistic; everything that moves, grips, stores energy or makes heat still comes from its own factory. Schematic, not to scale.

Why is fatigue the real problem?

Because a frame is loaded and unloaded tens of millions of times, and printed metal starts cracks more easily. Fatigue is failure at stresses well below a material's tensile strength, caused by repeated loading.

How many load cycles does a frame see?

A 17-inch motorcycle wheel with its tyre has a rolling circumference of about 1.95 m (my estimate). Over 1,00,000 km:

\[ N = \frac{1.0\times10^{8}\ \text{m}}{1.95\ \text{m}} \approx 5.1\times10^{7}\ \text{wheel revolutions} \]

Not every turn is a full stress cycle, but road bumps, engine pulses and braking add their own. The frame lives in the high-cycle fatigue regime, above 10⁷ cycles, where surface condition matters most.

Basquin's law, in plain words

For high-cycle fatigue, stress amplitude \( \sigma_a \) and cycles to failure \( N \) follow Basquin's law, a straight line on a log-log plot:

\[ \sigma_a = \sigma_f' \,(2N)^{\,b} \qquad\Rightarrow\qquad \frac{N_2}{N_1} = \left(\frac{\sigma_{a,2}}{\sigma_{a,1}}\right)^{1/b} \]

The exponent \( b \) is typically around −0.1 for metals (typical range −0.05 to −0.12, MetricGate). That makes life brutally sensitive to stress: 10% more stress at a crack-starting spot cuts fatigue life by roughly 60%, since \( 1.1^{10} \approx 2.6 \).

Calculated in this post from an exponent of −0.1, inside the typical range above: a stress ratio of 1.1 raised to the power 1/0.1 gives a factor of about 2.6.

Close-up of a metal fatigue fracture surface showing beach marks spreading from a crack origin
A fatigue fracture surface from NIST. The crack starts at a surface flaw and grows a little with each load cycle, leaving the curved "beach marks", until the remaining section snaps.Photo: National Institute of Standards and Technology, Public domain, via Wikimedia Commons

How much heavier would a one-piece printed frame be? The worked solution

Assumptions

  • Material: L-PBF AlSi10Mg aluminium, density about 2,700 kg/m³ (approximate; an EOS data sheet lists at least 2.67 g/cm³, EOS).
  • Fatigue strength: 154 MPa for machined and polished border-region samples and 96 MPa for polished-only, from Ford's study; as-built roughness cuts strength by about 40%.
  • A frame tube 40 mm in diameter carrying a cyclic bending moment of 150 N·m (my illustrative load).
  • Thin-walled tube, so the section modulus is \( Z \approx \pi r^2 t \).
  1. As-built fatigue strength. Apply Ford's 40% roughness penalty to the polished-only value:
    \[ \sigma_{as\text{-}built} \approx 0.6 \times 96 \approx 58\ \text{MPa} \]

    Calculated in this post from Ford's 96 MPa polished-only strength and its roughness penalty of about 40% (Lai et al., ASTM STP 1631).

  2. Wall thickness for a given stress. Bending stress \( \sigma = M/Z \), so, with these example inputs (a 40 mm tube and 150 N·m):
    \[ t = \frac{M}{\pi r^2 \sigma} = \frac{150}{\pi \times 0.02^2 \times 58\times10^{6}} = 2.06\ \text{mm} \]
    With a machined inside surface at 154 MPa, \( t = 0.78 \) mm.
  3. Mass per metre of tube. \( m' = 2\pi r t \rho \): with these example inputs, 0.70 kg/m as-built against 0.26 kg/m machined.
  4. The ratio. At constant diameter, mass scales with \( 1/\sigma \):
    \[ \frac{m_{as\text{-}built}}{m_{machined}} = \frac{154}{58} \approx 2.7 \]

That is the core of the problem. A one-piece print is valuable precisely because it has hollow, internal, organic shapes. But those inner walls can't be reached by a cutter, so they keep their as-built roughness, and the designer has to size every tube for the weaker surface. The Light Rider's frame is hollow, and hollow saves weight only as long as the inner walls are thick enough for their roughness.

Verdict: possible with changes

A one-piece printed metal chassis is possible today; APWorks and BMW have shown it. A one-piece whole motorcycle is impossible with current printers, because tyres, cells, magnets, bearings and brakes need different materials and processes. The deciding number is fatigue: unmachinable as-built inner surfaces cut fatigue strength by about 40%, which makes tubes up to about 2.7 times heavier than machined ones for the same life. The working design is a printed chassis plus bought-in running gear, with internal surfaces smoothed by chemical or abrasive-flow finishing.

How long would printing take?

Days for a frame, weeks for a whole metal chassis. A University of Sheffield paper puts a single laser's build rate at 2 to 40 cm³/h, depending on material, and cites 18 cm³/h for single-laser Ti-6Al-4V against 105 cm³/h for a four-laser machine.

The Light Rider's 6 kg aluminium frame is about 2,200 cm³ of metal:

Calculated in this post from the Light Rider's 6 kg frame (New Atlas) and aluminium at about 2.7 g/cm³ (EOS data sheet).

\[ t_{print} = \frac{2{,}200\ \text{cm}^3}{18\ \text{cm}^3/\text{h}} \approx 122\ \text{h} \qquad \text{or} \qquad \frac{2{,}200}{105} \approx 21\ \text{h} \]

That excludes powder handling, cooling, heat treatment and support removal. A bike also needs a build chamber longer than itself: the NS200 is 2,017 mm long, and the NERA measured 190 cm. I could not find a laser powder-bed machine with a two-metre chamber in the sources I opened, which is why large printed frames are made in sections.

Use this to check what you just read. Change the fatigue strength and watch the tube, and the print time, grow.

Printed frame tube: thickness, mass and print time

How sensitive is the answer to the assumptions?

  • Surface finishing. Chemical or abrasive-flow polishing can smooth internal channels. If it recovers even half the 40% penalty, the mass ratio falls from 2.7 to about 2.0.

    Calculated in this post from Ford's 40% penalty (Lai et al.): halving it gives 0.8 × 96 ≈ 77 MPa, and 154 ÷ 77 ≈ 2.0.

  • Porosity. Once the surface is smooth, Ford found subsurface pores take over, predicted by a modified Murakami model of defect size. Hot isostatic pressing closes many pores.
  • Fatigue ratio. Zhang and Xu's database puts AM alloys' fatigue-to-tensile ratio at 0.2–0.7, against 0.25–0.65 for conventional alloys. Good printed metal can match wrought; bad printed metal is far worse.
  • Stress concentration. Topology-optimised shapes avoid sharp corners. By Basquin's law, removing a 10% stress hot spot more than doubles life.

Where the model breaks: a real frame has combined bending, torsion and multiaxial stress, and welds or joints usually fail first. A one-piece print removes the joints, which is its real advantage, and I haven't credited that above.

Same 40 mm tube, same load, same lifeMachined inside154 MPa: wall 0.78 mm0.26 kg per metreAs-built inside~58 MPa: wall 2.06 mm0.70 kg per metre2.7× the metal
Rough inner walls that no cutter can reach force a thicker tube. Ring thickness drawn to scale relative to diameter, using my worked example above.
Several 3D-printed titanium horseshoes with a rough, grainy as-printed surface
3D-printed titanium horseshoes from CSIRO. The grainy surface is what an as-built part looks like up close, and it is where fatigue cracks like to start.Photo: , CSIRO, CC BY 3.0, via Wikimedia Commons
Printing makes the shape. Machining makes the surface. Fatigue only cares about the surface.The working above

What does the research say?

Fatigue database of additively manufactured alloysZian Zhang, Zhiping Xu · Scientific Data 10, 249 · 2023

The Tsinghua authors screened 3,415 records and built a database from 1,848 articles covering 116 AM alloys: 15,146 stress-life points in 1,610 datasets, plus strain-life and crack-growth data. They find the fatigue-to-tensile strength ratio of AM alloys spans 0.2–0.7, against 0.25–0.65 for conventional alloys, and that poor printed surfaces make S-N data scatter. Limit: it pools many machines and post-treatments, so a single value can't be read off for one part.

Read the paper (free)
Effects of Surface Roughness and Porosity on Fatigue Behavior of AlSi10Mg Produced by Laser Powder Bed Fusion ProcessWei-Jen Lai et al., Ford Motor Co. · ASTM STP 1631 · 2020

Ford engineers tested L-PBF AlSi10Mg in as-built, polished and machined-and-polished states. Roughness cut fatigue strength by about 40%, more than standard models for conventional metal predict; machined and polished samples reached 154 MPa against 96 MPa for polished-only. After homogenisation the difference between border and core regions disappeared, and a modified Murakami model predicted strength from pore size. Limit: specimens, not whole structures.

Read the abstract (free)
Diode area melting of Ti6Al4V using 808 nm laser sources and variable multi-beam profilesMohammed Alsaddah et al. · Materials and Design 215 · 2022

The Sheffield team tested a new multi-diode melting method aimed at LPBF's slow build rate. Their review of the field puts single-laser LPBF at 2–40 cm³/h depending on material, and cites a four-laser system at 105 cm³/h for Ti-6Al-4V against 18 cm³/h for one laser. Limit: their own diode method is at an early research stage.

Read the paper (free)

For how a conventional frame's geometry sets handling, read rake and trail explained; for what a frame has to survive on Indian roads, the used bike checklist shows where cracks hide. The Yantra page collects the rest of the engineering on this site.

What if next?

Questions people ask

Can you 3D print a motorcycle frame?

Yes. APWorks printed a hollow 6 kg aluminium frame for its 35 kg Light Rider in 2016, and BMW showed a metal-printed S 1000 RR frame and swingarm in 2018. Neither replaced conventional production frames.

Are 3D-printed metal parts as strong as forged ones?

In tensile strength they can be close. In fatigue they are often weaker unless machined and heat-treated, because surface roughness and pores start cracks. Ford measured about a 40% fatigue penalty for as-built AlSi10Mg.

Has anyone made a fully 3D-printed motorcycle?

BigRep's NERA (2018) printed everything except the electronics, including airless tyres, but in 15 separate parts using polymer filaments. It was a design study, not a road-legal bike.

Why don't manufacturers 3D print whole bikes?

Cost and speed. A single laser builds 2–40 cm³ an hour, so one frame takes days, while conventional frames are welded or cast in volume. Printing pays off for low-volume, complex parts like racing components.

Sources

  1. New Atlas: Airbus unveils 3D-printed Light Rider (2016) (source for 35 kg, 6 kg frame, Scalmalloy, 6 kW, 80 km/h, 50 units)
  2. Autocar India: BMW Motorrad reveals 3D-printed frame and swingarm (2018) (source for SLM process and no released properties)
  3. 3Dnatives: NERA, a fully 3D printed electric motorcycle (2019) (source for 15 parts, 60 kg, 190 cm)
  4. VoxelMatters: Porsche 3D prints pistons for the 911 GT2 RS (2020) (source for 10% lighter, up to 30 PS, cooling duct)
  5. Zhang and Xu: Fatigue database of additively manufactured alloys, Scientific Data (2023) (source for the 0.2–0.7 fatigue ratio)
  6. Lai et al.: Effects of surface roughness and porosity on fatigue of L-PBF AlSi10Mg, ASTM STP 1631 (2020) (source for the 40% penalty, 154 and 96 MPa)
  7. Alsaddah et al.: Diode area melting of Ti6Al4V, Materials and Design (2022) (source for 2–40, 18 and 105 cm³/h)
  8. ICF15: Fatigue life of L-PBF AlSi10Mg, effects of surface roughness and porosity (RMIT) (source for as-built roughness of 1.5–2 µm against 0.8–1.0 µm machined)
  9. Bajaj Auto: Pulsar NS200 owner's manual (source for the 2,017 mm length)
  10. Loughborough University: Speedform bike launch (2025)
  11. Visordown: New super alloy could make future motorcycles lighter (2026)
3D printingadditive manufacturinglaser powder bed fusionmetal fatigueBasquin's lawAPWorks Light RiderBMWwhat if
Sapta

Sapta (Saptarshi Sarkar) is a mechanical engineer turned marketer in Bengaluru and a solo touring rider. He writes the seven Saptaverse blogs. Work with him or read more of RoadHook Machines.

Keep reading in RoadHook Machines

From the other blogs