For round carbon fiber tubes, the process decision is driven by three numbers: dominant load direction, production length, and diameter range, with pultrusion covering 3–300 mm at ±0.05 mm tolerance and continuous runs up to 10 km [S3].
Filament winding is the right answer when the service load is hoop stress, hydrostatic pressure, or multi-axis torque, and the part is closed-form, typically circular, with diameters well beyond 300 mm [S1][S5]. The two processes overlap in the small-diameter range but diverge sharply in fiber angle control, throughput, and part length.
Pultrusion Process Window for Round Profiles
Pultrusion is a continuous, automated pull-through operation, with the fiber rovings impregnated and then dragged through a heated steel die that defines the cross-section and cures the resin in one pass; Epsilon Composite operates such lines on a 24/7 basis and treats it as the historical core process for carbon profiles [S3]. The output window for round sections is tightly bounded: outer diameter 3–300 mm, wall thickness 0.8–20 mm, geometric tolerance ±0.05 mm on diameter and ±0.01 mm on wall, with cut length or spooled length up to 10 km at ±0.2 mm cut tolerance [S3].
Fiber orientation in pultrusion is essentially uniaxial, with the rovings aligned along the pull axis, which is why axial tensile and compressive stiffness are maximised and why drive shafts, structural supports, and robotic-arm booms gravitate to this route [S3][S4]. Resin systems are typically epoxy for high-performance duty, with polyester and vinyl ester used for cost-driven industrial builds where chemical resistance matters more than peak mechanicals [S1].
Filament Winding Process Window for Round Profiles
Filament winding lays down continuous tows onto a rotating mandrel, with carriage translation controlled to set the wind angle, and the wound preform is then cured before mandrel extraction [S5]. The standard execution is wet winding through a resin bath, but prepreg tow is equally usable when tighter resin content and lower void content are required [S1][S4].
Production-scale winding machines handle large envelopes: 4-axis winders at 4 ft diameter × 6 ft length, 5-axis winders at 8 ft diameter × 15 ft length, and a 96-end pull-winder covers perimeters up to 16 inches for hybrid layups [S5]. The defining capability is angle control, because hoop (≈90°), helical (commonly ±45° to ±70°), and polar patterns can be programmed layer by layer, which is what gives wound tubes their balanced biaxial or multi-axial response [S1][S5].
Decision Matrix: Pultrusion vs Filament Winding on 4 Criteria
The two processes separate cleanly on four decision criteria, and the call should be made on the strongest criterion first: [S1]
1) Dominant load direction: pultrusion for axial tension/compression and bending stiffness along the length; filament winding for hoop pressure, torsional, and combined multi-axial loading on closed-form parts [S1][S2][S6].
2) Diameter and tolerance: pultrusion wins on small-to-medium rounds (3–300 mm, ±0.05 mm) and on length (continuous up to 10 km); filament winding is the only practical route for diameters above ~300 mm, including pressure vessels, large rollers, and poles [S3].
3) Production economics: pultrusion delivers near-100% yield on long runs because the die sets the geometry and the line runs continuously, with re-spooling allowing uninterrupted lengths beyond 5,000 ft in shippable form; filament winding has higher per-part labour and tooling cost but buys you fiber-angle tailoring that pultrusion cannot match without a hybrid step [S5].
4) Material consistency: pultrusion with resin injection holds very low void content, with epoxy and even high-viscosity elastomer resins processable; filament winding wet systems are sensitive to bath temperature and tow tension, so prepreg tow is often specified for aerospace-grade wound parts [S4][S5].
Hybrid and Pull-Winding Routes That Blur the Line
The sharpest 2024–2026 trend is hybrid pultrusion-plus-winding, sold by Epsilon as the K1 process for diameters above 300 mm: ultra-high-modulus pultruded profiles are embedded between wound layers, giving the part very high stiffness at optimised weight and extending pultrusion's reach into technical rollers and extra-large poles [S3]. Epsilon's stated upper diameter on standard pultrusion is 300 mm, so K1 exists precisely to push past that ceiling without abandoning the axial stiffness advantage of pultruded carbon.
Pull-winding is the second hybrid: standard fiber spools are laid down at programmed angles in-line with the pultrusion pull, replacing expensive woven fabrics for bias plies, with the winder wheels slaved to line speed so angle stays constant [S5]. This delivers torsionally balanced cross-sections at lower cost than fabric-based pultrusion and is the most cost-effective route when a part needs a few off-axis plies but not a fully wound preform [S5]. Related process research on coreless winding-pultrusion combinations is documented in composites literature, showing the same logic: use pultrusion for axial stiffness, winding for off-axis control, and a single continuous line for both [S7].
Use-Case Selection Guide
Specify pultruded carbon round tube when the application is a drive shaft, a robotic-arm link, a telescoping pole, a structural mast section, or any axial load path where the longest possible continuous length and tight OD tolerance matter; typical diameter window 10–200 mm in epoxy resin, with vinyl ester as the cost-down option for chemical-exposure service [S1][S3][S4]. For deeper context on how epoxy cure windows interact with pultrusion line speed, see the 2026 epoxy cure-time reference Epoxy Curing Temperature and Time Trade-off: 2026 Process Reference.
Specify filament-wound carbon round tube when the part is a pressure vessel, a chemical storage tank, a roller cover, a large-diameter pipe, or any rotating shaft carrying significant torque; expect to program ±45° to ±70° helical plies for shear and 90° hoops for pressure, with the option of wet winding for cost or prepreg tow for low-void aerospace grade [S1][S4][S5]. Skip pultrusion for any application where the principal stress is not axial; skip filament winding for any application where the run is long, the cross-section is small, and the tolerance is tight, because that is exactly where pultrusion's economics and dimensional control dominate [S2][S3][S6].
Track these two signals going forward: the 300 mm diameter ceiling on standard pultrusion, which currently forces every larger roller or pole into K1 hybrid or pure winding, and the continued migration of pull-winding into pultrusion lines as a lower-cost replacement for woven bias fabrics.