Reference image. The actual surface and colour depend on the resin and the finish.
Overview
Pultruded carbon fibre I beam, 30–200 mm deep, 20–120 mm flange width, wall thickness 2.5–8 mm, cut to length. Two flanges with a web between them put the material furthest from the neutral axis, at the same weight. Bending is one band above a solid rectangle or a square tube. Depth and flange tolerance ±0.2 mm (±0.3 mm above 100 depth), flange to web 90° ±0.5°. The table gives weight per metre and second moment of area: 100×50×4 is 1.19 kg/m with Ix 118.2 cm⁴. 200×120×8 is 5.26 kg/m with Ix 2185.8 cm⁴. The second moment of area is calculated, and can be recomputed from the section dimensions. Where the catalogue stops and the drawing starts, see custom composite parts.
Top structure carrying enclosure and housing loads
Stock sizes in detail
Read it as depth × flange × wall. Bending follows the depth and barely the flange; torsion is not on this table — that is square tube.
Code
Depth×flange
Wall
Length
Weight kg/m
Ix cm⁴
Tolerance
Note
IB-30-20-2.5
30×20
2.5
1000
0.25
2.2
±0.2
Small truss
IB-40-25-2.5
40×25
2.5
1000
0.33
5.3
±0.2
Equipment inner frame
IB-50-25-3
50×25
3
1000
0.44
10.4
±0.2
Light beam
IB-60-30-3
60×30
3
1000
0.53
18.6
±0.2
Robot arm beam
IB-80-40-3
80×40
3
1000
0.72
45.7
±0.2
Jig and fixture
IB-80-50-3
80×50
3
1000
0.81
54.6
±0.2
Measuring arm
IB-100-50-3
100×50
3
1000
0.90
91.4
±0.2
Frame beam
IB-100-50-4
100×50
4
1000
1.19
118.2
±0.2
Frame beam
IB-100-50-5
100×50
5
1000
1.47
143.3
±0.2
Load-bearing beam
IB-120-60-4
120×60
4
1000
1.44
208.4
±0.3
Main beam
IB-120-80-4
120×80
4
1000
1.69
262.2
±0.3
Gantry beam
IB-150-75-5
150×75
5
1000
2.25
508.7
±0.3
Long-span beam
IB-150-100-5
150×100
5
1000
2.63
640.2
±0.3
Long span, heavy load
IB-180-90-6
180×90
6
1000
3.24
1054.9
±0.3
Large-span gantry
IB-200-100-6
200×100
6
1000
3.61
1461.7
±0.3
Main load-bearing beam
IB-200-120-8
200×120
8
1000
5.26
2185.8
±0.3
To drawing, new die
The table lists stock and regular sizes. Sizes outside the table are produced to drawing; MOQ 200 pcs.
Three dimensions: depth, flange width, wall
The trade writes 100×50×4. 100 deep, 50 flange width, 4 wall. Depth decides bending. The point of an I section: material sits far from the neutral axis, and the second moment of area follows the square of the depth. Double the depth with flange and wall unchanged: Ix goes from 118.2 to 620.1, five times over. Weight goes from 1.19 to 1.81 kg/m, half as much again. Flange width decides two things: whether you can drill and mount on the flange, and whether the compression flange buckles first. Too narrow and a bolt head will not sit; too wide with a thin wall and the compression flange ripples. Wall governs web shear and flange buckling at once. A thin web shear-buckles first; a thin flange wrinkles first under compression. Wall is the constraint on both. The three dimensions are not mixed at random. Every combination in the table has been checked on depth-to-thickness and width-to-thickness ratios, both inside the usual bands. A direct selection does not land you in a hole.
Bending comes from depth, not from weight
This section is where an I section parts company with the others. The second moment of area in the table is worked out by breaking the section into rectangles of equal wall thickness, and you can recompute it:
Ix = 2 × [B·t³/12 + B·t·((H−t)/2)²] + t·(H−2t)³/12
Compare rows already in the table:
100×50×3 → 100×50×4: wall only, 0.90 to 1.19 kg/m (+32%), Ix 91.4 to 118.2 (+29%)
100×50×4 → 150×75×5: deeper and wider, 1.19 to 2.25 kg/m (+89%), Ix 118.2 to 508.7 (+330%)
150×75×5 → 200×120×8: 2.25 to 5.26 kg/m (+134%), Ix 508.7 to 2185.8 (+330% again)
The conclusion is blunt: for stiffness, add depth, not wall. Wall buys stiffness linearly; depth buys it as the square. For the same added weight, depth gives several times the bending capacity. The other way round: the space may fix the depth. Not enough structural height in the frame, no room for a deeper beam inside the housing. Then wall or a higher-modulus fibre is all that is left, and both cost far more weight. In that case tell us the height limit first and we work back to wall and flange width, which beats forcing a stock size. The same argument applies against square tube. A 50×50×5 square tube, 40 mm inside, has Ix 30.8 cm⁴. The I beam 100×50×4 has Ix 118.2 cm⁴ with 15% less section area, 768 against 900 mm². It is lighter and nearly four times stiffer in bending. The price is torsion. A square tube is closed on four sides, an I section is open. Torsional stiffness is an order of magnitude worse. Bending-led: I. Torsion or two-way stiffness: square tube.
Flange or web — which buckles first
An I beam rarely fails by pulling the material apart. It usually buckles locally first: the compression flange ripples, or the web bulges out in the shear zone. Compression flange is read on the outstand-to-thickness ratio, b/t. A wide, thin flange acts like a plate supported on one edge and buckles before you get near its strength. The outstand is the flange beyond the web rather than the full flange width. On the widest row, 200×120×8, that is (120 − 8)/2 = 56 mm, giving b/t = 7. Quoting the full width would give 15. That number flatters the section, and it is what you see on pages that have not thought about it. Both are inside normal limits, so nothing here is unsafe; the difference is whether the check is real. Web is read on the clear-depth-to-thickness ratio, h_w/t. A deep, thin web carries shear near the supports in bending and shear-buckles there. The deepest row, 200×120×8, has h_w = 200 − 2×8 = 184 mm and h_w/t = 23. Note the subscript: in the Ix formula above, H is the overall depth; here it is the web between the flanges. Keeping the two symbols apart matters, because 184 and 200 are 8% apart and the buckling limit is sensitive to it. The larger either ratio, the earlier buckling arrives. Carbon fibre has no ready table of limiting values the way steel does. The limit moves with fibre volume fraction, layup sequence and resin system. The same section in a different resin gives a different limit. So we do not publish a general limit. We publish the method. Send span, load, support conditions and allowed deflection. We work out the buckling margin for your section, thickening the web or adding vertical stiffeners where needed. Web penetrations need their own line. Cables, pipes and weight reduction all put holes in the web, and a hole edge is a stress concentration and the start of shear buckling. Keep the hole diameter within a third of the clear web depth. Leave at least twice the wall thickness from the hole edge to the flange. Load capacity is then generally unaffected. Beyond that, reinforce the edge. Mark the hole positions on the drawing and we drill to drawing at ±0.01 mm.
I and H: two names for one section
In English, I beam and h beam are used interchangeably in carbon fibre profiles, both meaning two flanges with a web between them. The difference is habit. Equal flanges, with web and flange the same thickness, are usually called I beam. Markedly wider flanges, or flanges thicker than the web, are usually called H beam, because the shape looks more like the letter. The only practical consequence for a buyer: say before the quote whether the wall is uniform over the section or the web is thin and the flange thick. 150×100×5 and 150×100×5×8 are different products. The second has an 8 mm flange on a 5 mm web. Weight and second moment of area both differ, by about a third if the two get mixed up. Every row in our table is uniform wall. A thickened flange is made to drawing on a new die; state the web and flange thickness separately and it is clear. Structural engineering also uses structural beam as a general term for a load-bearing beam, without fixing the section shape. Do not look for a designation under that word; select on depth and second moment of area.
Weight per metre, and how it compares with steel beams
Carbon fibre at 1.55 and steel at 7.85 gives about an 80% reduction at equal volume. Every band in the table carries kg/m. Against steel, quote a real section — these are the nearest Euronorm IPE sizes, which is what a buyer will look up:
100×50×4: carbon fibre 1.19 kg/m; IPE 100 is 8.1 kg/m
150×100×5: carbon fibre 2.63 kg/m; IPE 160 is 15.8 kg/m
200×120×8: carbon fibre 5.26 kg/m; IPE 200 is 22.4 kg/m
Those are weight at equal length, not at equal stiffness — the carbon section has not been resized yet. Read them as the ceiling rather than the answer. But the same size is not a direct replacement. Modulus depends on fibre grade and fibre volume fraction and changes with the material you select, so we do not publish one number. The equal-stiffness relation is I_carbon = I_steel × 210 / E_carbon. Steel is taken at 210 GPa, a published standard; E_carbon comes from your chosen layup. In practice the move is deeper, not thicker. The second moment of area follows the square of the depth. For a replacement, give us load, span and allowed deflection and we work back to depth; do not swap the steel designation straight across. Conversely, if the original design is governed by strength rather than by deflection, carbon fibre often drops a full size and still comes out lighter. Working out which of the two governs matters more than picking the designation.
Length, cutting and end facing
Standard lengths are 1000 mm and 2000 mm, with other lengths cut to order; longer needs a joint. Cutting: water-jet leaves a clean edge without chipping, the first choice for thin wall and visible parts. Thick wall and long runs take sawing, leaving 0.2 mm on the end to machine back. End squareness ±0.5°. An I beam end is two flanges plus a web, three faces. All three have to be square, or an end plate or a butt joint shows a gap. For a gantry beam, mark it especially: if the two end faces are not parallel, the two columns end up at different heights. Splicing: beyond 6000 mm the beam has to be joined. Place the joint where the bending moment is low, near a point of contraflexure or close to a support. An internal splice plate is bonded across it rather than a plain butt joint. Splice details are made to drawing.
Delivery states
Natural The pultruded surface as it comes: smooth, dark grey to black, unpatterned. The default for structural and internal parts.
Clear coat A coat over the natural surface, gloss or semi-matte. More resistant to soiling.
Matte The same coating system, flattened; less reflection, common on exposed frames and display work.
Painted Painted to a project colour reference; the colour and the gloss are yours, and we control flatness and adhesion.
Tolerance and inspection
Depth (≤100 mm)
±0.2 mm
Calliper + dimensional report
Depth (>100 mm)
±0.3 mm
Calliper
Flange width
±0.2 mm
Calliper
Wall thickness
±5% (thin wall counted by plies)
Calliper / ultrasonic
Flange to web squareness
90° ±0.5°
Engineer's square + feeler gauge
Flange symmetry
≤0.5 mm
Surface plate + height gauge
Inside radius
Determined by the wall thickness (of the same order)
Profile gauge / projector
Straightness
1 mm / m
Surface plate + feeler gauge
Twist
≤1° / m
Surface plate measurement
Length (cut to length)
±0.1 mm
Water-jet guided / CNC
End squareness
≤0.5°
Engineer's square
Parallelism of the two ends
≤0.5 mm
Surface plate + dial gauge
Hole position and slotting
±0.01 mm
CMM
Common questions
I beam or square tube — which should I choose?
Bending-led work takes the I; torsion or two-way stiffness takes the square tube. At equal weight an I beam bends far better — 100×50×4 and a 50×50×5 square tube have similar section area, with Ix 118.2 against 30.8 cm⁴ — while a square tube, closed on four sides, is an order of magnitude better in torsion. Gantry beams and long spans take I; columns and frames needing two-way stiffness take square tube. Mixing the two is normal: square columns, I beams across
Can it replace a steel I beam directly?
The dimensions match; the stiffness does not. Carbon fibre modulus depends on the layup, so at the same size it is always softer — by how much depends on your chosen material data. Give us load, span and allowed deflection and we work back to depth and wall; do not swap the steel designation straight across. Where strength governs rather than deflection, a size down is often enough
How is the Ix in the table calculated?
By breaking the section into rectangles of equal wall thickness, about the horizontal neutral axis: Ix = 2 × [B·t³/12 + B·t·((H−t)/2)²] + t·(H−2t)³/12 The two flanges are shifted by the parallel axis theorem and the web is taken at its clear depth. You can recompute from the table's dimensions; the difference comes from the inside radius, which is not counted. Unequal wall sections are calculated separately
Why is depth better value than wall?
Depth enters the second moment of area as a square, wall as a linear term. 150×75×5 has Ix 508.7 and 100×50×4 has 118.2, for 89% more weight and 330% more second moment of area. If the space allows, add depth first
Can you put holes in the web for cable?
Yes. Keep the hole diameter within a third of the clear web depth and at least twice the wall thickness from hole edge to flange, and load capacity is generally unaffected; beyond that, reinforce the edge. Holes are drilled to drawing at ±0.01 mm
Will it twist?
It will — the standard complaint about open sections — and we hold 1°/m. For long-span beams, put the twist requirement on the drawing and we check it over the working length. For zero twist, use square tube
Can the flange be wider or thicker?
Yes. A wider flange, a thicker flange (thin web, thick flange) and unequal flanges are all made to drawing on a new die, which takes 5–7 days. For a thickened flange, state the web and flange thickness separately; 150×100×5×8 is the shortest way to write it
Can you drill and tap it?
Drilling, notching and counterboring, all at ±0.01 mm. Carbon fibre is not threaded: for a bolted joint use through bolts with washers, or bond an insert in
Is there any carbon weave?
No. Pultrusion gives a unidirectional finish — a layer of resin, dark grey to black. Carbon weave means overwrapping 3K prepreg, or changing to a moulded part
What are the MOQ and lead time?
200 pcs. Normal batches ship in 15–20 days; a non-standard depth, flange or wall, or a new section, needs a die, which takes 5–7 days
Can you make glass fibre?
Yes, in the same construction — lower stiffness and lower cost, common for insulating parts and low-frequency structural parts. Quoted through the glass fibre product line
Send us the size
Section, length, quantity and service conditions — with these in hand the quote is accurate. Upload a drawing if you have one; without one, the same figures written out will do.
Company addressZhuyuan Tianhuan International, No. 76 Keji 2nd Road, High-tech Zone, Xi'an, Shaanxi, China
FactoryFengjing Industrial Park, Middle Section of Fengwu East Road, Huyi District, Xi'an, Shaanxi, China