Direct Roving Tex: Choosing 2400, 4800 or 9600 for Pultrusion and Filament Winding

A pultrusion line that never quite runs clean usually has a tex problem, not a resin problem. The profile comes out with dry patches at the corners as soon as you push the speed up. Or the glass content sits a point below what the drawing asks for, and no amount of die temperature closes the gap. Or every position on the creel is occupied and the section is still short of fiber. None of these show up as a datasheet error, because nothing on the datasheet is wrong. They show up because tex was chosen from a catalog instead of from the line.

This guide works through the choice between 2400, 4800 and 9600 tex the way a process engineer actually meets it: what the number measures, why two rovings with the same tex are not interchangeable, how many ends a given profile needs, and what each step up costs you.

What tex actually measures (and the yield number behind it)

Tex is a linear density: grams per 1,000 metres. A 2400 tex roving weighs 2.4 g for every metre you pull; 4800 tex weighs 4.8 g; 9600 tex weighs 9.6 g. That is the whole definition, and it is worth stating plainly because tex gets used as shorthand for "thickness" in a way that hides two things it does not describe: the filament diameter inside the strand, and the strength of the laminate you will build from it.

If your drawings or your purchasing department work in US units, the same specification appears as yield in yards per pound, which expresses the same property from the other direction:

In one line: yd/lb = 496,055 ÷ tex.

tex	g/m	≈ yd/lb
1200	1.2	413
2400	2.4	207
4800	4.8	103
9600	9.6	52

The conversion is arithmetic, not an approximation of material behaviour: 1 lb = 453.592 g and 1 yd = 0.9144 m, and the constant falls out of those two facts. It matters because a US distributor quoting "206 yield" and a mill datasheet saying "2400 tex" are describing the same roving in two notations. Convert before you compare, and never put a tex figure and a yield figure in the same table without labelling which column is which.

Same tex, different strand

Two suppliers can both quote 4800 tex and still ship strands that behave differently, because tex is the product of two independent decisions: how many filaments the bushing carries, and how thick each filament is.

One E-glass filament at 17 µm carries roughly 0.58 tex; at 24 µm, about 1.15 tex, at an assumed glass density of 2.55 g/cm³. Run that arithmetic and a useful thing appears: a 4,000-hole bushing drawn at 17 µm lands near 2,315 tex, while a 2,000-hole bushing at 24 µm lands near 2,307 tex. Same tex, half the hole count, filaments about 40% larger in diameter.

The two strands are not equivalent downstream. A bundle of many fine filaments presents far more surface area to the resin and spreads into a flatter band, which helps wet-out and helps fill a wide, thin cavity. A bundle of fewer, coarser filaments drains and impregnates differently, resists opening at the die entrance, and translates its own strength into the laminate differently. Neither is better in the abstract; they are different tools.

The action this implies is small and cheap: ask for the filament diameter and the bushing hole count alongside the tex. Suppliers publish tex and diameter as separate columns for exactly this reason, and it is the fastest way to find out whether a cheaper 4800 is the same reinforcement or merely the same number.

The mass balance: from profile to ends

This is the calculation that turns a drawing into a creel layout. Three steps, and the assumptions matter as much as the arithmetic.

Step 1 — glass mass per metre. Take the cross-section area in cm², multiply by 100 to get the volume of one metre of profile, multiply by the target fiber volume fraction, then by glass density. For the density, 2.55 g/cm³ is the value used here.

Step 2 — ends. Divide the glass mass per metre by the mass one end delivers per metre, which is tex ÷ 1,000.

Step 3 — round and re-check. Ends are integers. Round up, then calculate the fiber content you are actually going to build, because that is the profile you will ship.

Assumptions, stated once so you can substitute your own: all reinforcement is roving (no mat, veil or filler in the section), the target Vf is a volume fraction rather than a weight fraction, and die clearance and resin bleed-back are ignored. Change any of those and the numbers move.

Worked example: a 100 mm × 6 mm flat profile at a target Vf of 0.60.

Area = 6.00 cm², so one metre of profile contains 600 cm³ of material

Glass volume = 600 × 0.60 = 360 cm³

Glass mass = 360 × 2.55 = 918 g/m

tex	g/m per end	ends required (exact)	ends to order	g/m delivered	Vf actually built
2400	2.4	382.5	383	919.2	0.601
4800	4.8	191.3	192	921.6	0.602
9600	9.6	95.6	96	921.6	0.602

All three tex options hit the target within rounding. Which is the point: at this stage of the calculation tex looks like a free choice. The differences that decide it are not in this table.

What you gain and what you pay when you move up a step

Three hundred and eighty-three ends become 192, then 96. That single ratio drives four separate accounts, and they do not all move in the same direction.

Creel positions and threading time. Creel cost is per position, and so is the labour of threading it. Halving the ends halves both. On a line where creel length is the binding constraint — which is most pultrusion lines — this is often the real reason to move up a step, and it has nothing to do with the material.

Package changes and splices. Half the ends means half the packages, so half the doffs and half the splices per shift. On a continuous run, fewer interruptions is worth more than a marginally better wet-out figure, because every splice is a place where something can go wrong.

Wet-out. This is where the catalog claim needs a qualifier. Higher tex does raise fiber throughput per end, and the product page says so directly: a higher tex number means a thicker strand, which can be used to increase fiber content and speed up production. That is true, and it is a statement about throughput per end. It is not a promise that your impregnation will keep up.

Residence time in the bath is bath length divided by line speed. A 0.6 m bath running at 0.3 m/min gives each strand two minutes in the resin; at 0.6 m/min it gives one minute; at 1.2 m/min, thirty seconds. A coarser bundle has more filament layers for the resin to penetrate, so the same residence time buys less penetration. Raise the tex and the line speed together — which is exactly what the throughput argument encourages — and the two effects compound against wet-out.

Fiber content resolution. One end of 2400 tex moves Vf by about 0.16 percentage points on this profile. One end of 4800 tex moves it 0.31; one end of 9600 tex, 0.63. If your specification is 60 ± 1%, all three are workable. If you are trimming inside a one-point window — 62% to 63%, or matching a customer's coupon — 2400 tex gives you roughly six steps across it, 9600 tex barely one.

Put together, the pattern is consistent: thin walls, small sections and tight fiber-content tolerances push you toward the finer tex, where the extra ends buy resolution and even spreading. Thick sections, long runs and creel-limited lines push you toward the coarser tex, where fewer ends buy simplicity. Very few lines are limited by all four at once, which is why the answer differs between two plants running the same profile.

Pultrusion and filament winding: same tex, two ceilings

The same tex behaves differently in the two processes because the limiting constraint is not the same.

In pultrusion, the ceiling is the die entrance. Every end has to be spread into a band that fills the cavity evenly, without gaps and without overlap. More, thinner ends spread more readily and fill a wide, shallow section better; fewer, thicker ends are stiffer, resist opening, and can leave the corners starved. So the practical tex is set by how many ends your creel holds and how well your pre-die spreading works — not by what the datasheet offers.

In filament winding, the ceiling is bandwidth and tension. At equal throughput, 48 ends of 2400 tex and 12 ends of 9600 tex deliver the same mass per hour. Twelve ends means a shorter tension path and fewer guides, which is easier to control and easier to rethread. But each of those twelve packages now carries four times the mass flow, so one snag, one bad package or one tension spike shows up as a band defect over a wider area. Winding also cares about band width per pass, which changes with tex: a coarser roving covers more width in one traverse, which changes how many passes close a layer.

A roving that is ideal at 383 ends through a pultrusion die can be the wrong choice at 12 ends on a mandrel. Same number, different ceiling.

The throughput number that sizes your creel

Before you commit to a tex, check what your creel has to deliver:

The arithmetic is one line: kg/h = ends × (tex ÷ 1000) × line speed (m/min) × 0.06.

At 25 m/min, delivering 172.8 kg/h takes 48 ends of 2400 tex, 24 ends of 4800 tex, or 12 ends of 9600 tex.

tex	ends for 172.8 kg/h at 25 m/min	ends for 345.6 kg/h at 25 m/min
2400	48	96
4800	24	48
9600	12	24

This is the number that sets creel size, package mass and doffing interval. It does not tell you whether the line will wet out — that is the residence time calculation above — and it does not tell you whether the profile will fill. Use it to size the creel, then use the other two to check that the creel you sized can actually run.

FAQ

Can I just buy the coarsest tex to cut down on ends?

You can, and on a creel-limited line it is often right. Check two things first: the residence time your bath gives at your target speed, and whether one end of that tex is too coarse a step for your fiber-content tolerance. If both pass, the coarser tex wins on simplicity.

Two suppliers quote the same tex — are they interchangeable?

Not necessarily. Ask for filament diameter, bushing hole count and the sizing system. Same tex with a different strand construction will wet out differently and will not translate strength into the laminate identically.

Is 4800 tex "in between" 2400 and 9600?

In mass, yes: exactly double and half. In process terms, no. It halves the ends without halving the number of adjustment steps, and it doubles the mass flow per package rather than quadrupling it. The trade-offs are not linear, which is why the middle option is often the one that fits a line with a fixed creel and a tight tolerance.

Choose the tex your line can hold

Tex is a line decision wearing a datasheet's clothes. Start from the section and the target fiber content, convert to grams per metre, divide by tex to get ends, then check the three things a catalog cannot tell you: whether the creel has room, whether the bath has time, and whether one end is a fine enough adjustment. Browse Ruifiber's fiberglass reinforcement products to see the reinforcement range alongside this roving line, and read the direct roving for pultrusion and filament winding page before you send an RFQ that names your process, section, target Vf and resin — the more of those you specify, the more of the answer comes back already matched to your line.