How the two differ in kind
Cut a rolled section along a zigzag or a series of circles, slide the two halves apart and weld them back together, and the result is a beam substantially deeper than the one you started with, made of exactly the same weight of steel. Depth enters bending capacity as a SQUARE and stiffness as a CUBE, so that is a large gain for no extra material — and it is the entire argument for the product.
The second argument is usually worth more than the first. The openings left by the cut are a route for services, so ducts and pipes pass THROUGH the structural zone rather than beneath it. On a multi-storey building that can remove a few hundred millimetres from every floor-to-floor dimension, which compounds into a shorter building, less cladding and less of everything vertical. That saving frequently exceeds the steel saving by a wide margin.
What is given up is the web. A plain beam resists shear with a continuous plate; a castellated one has a series of posts between holes. So the shear check is against the NET section at an opening, the tees above and below each opening carry local bending as the shear crosses the hole — Vierendeel action, which a plain-section calculation never sees — and the web posts between openings can buckle, a failure mode that does not exist in a solid web.
The factors that actually differ
Show
| Castellated or cellular beam | Plain rolled girder | |
|---|---|---|
| Depth for the same steel | Substantially deeper — commonly around half again — for the same weight. Bending and stiffness both rise sharply with it. | Depth costs weight. The next section up is heavier everywhere, including in what the columns and foundations carry. |
| Services through the depth | The reason most of them are specified. Ducts pass through the openings, so the structural and services zones overlap instead of stacking. | Services pass under, which adds their depth to the beam's for every floor in the building. |
| Shear | The governing check, at the net section through an opening. Short heavily loaded spans are exactly where this bites. | Rarely the problem on a normal span — a solid web has a great deal of shear capacity. |
| Failure modes it invents | Vierendeel bending in the tees at each opening, and web-post buckling between openings. Both need checking; neither is intuitive. | Lateral-torsional buckling and the ordinary section checks. Well understood and tabulated. |
| Where the openings can be | Fixed by the fabrication geometry, though cellular beams allow some variation and infilled cells can be specified at high-shear locations. | Anywhere, if designed — but every hole is a designed and usually reinforced detail rather than a standard product feature. |
| Fire protection | More surface area per unit of steel and an awkward geometry, so intumescent coating costs more and board encasement is harder. | Simple geometry, standard protection, well-documented thicknesses. |
| Lead time and supply | A fabricated product with a fabrication programme. It is made for the job, so a late change is a real change. | A stock section. Available, substitutable, and forgiving of a design that moves. |
| Shape of the cost | More fabrication per tonne, less tonnage, and a saving elsewhere in the building that may dwarf both. It is a whole-building calculation rather than a steel one. | Lower unit cost, more tonnage, and the knock-on of extra depth carried through the whole building height. |
Which one, and when
Choose castellated or cellular beam when…
- The span is long and DEFLECTION governs, which is where depth is worth the most and shear demand is proportionally low.
- Services have to cross the beam and floor-to-floor height is constrained or expensive.
- The floor plate is repetitive, so the fabrication setup is spread across many identical members.
- Weight matters downstream — a transfer structure, a long span over an existing building, a crane limit.
Choose plain rolled girder when…
- The span is short and heavily loaded, so shear governs and a perforated web is the wrong section.
- There are large point loads, particularly near supports, where a web post would be carrying them across a hole.
- The programme is tight or the design is still moving, and a stock section that can be swapped is worth more than an optimised one that cannot.
- Fire protection cost is significant, and the simpler geometry is cheaper to protect.
Now run your own numbers
This page holds no prices on purpose — a national average is wrong for almost every real project. Quantify both options with your dimensions and your local quotes.
Frequently asked questions
- What is Vierendeel action and why does it matter here?
- In a solid beam, shear crosses the section through the web. At an opening there is no web, so the shear has to travel around the hole through the tee sections above and below it — and in doing so it bends them locally, as the members of a Vierendeel frame bend. That local bending is additional to the global bending in the beam, and it is largest where the shear is largest, which is near the supports. It is the reason a castellated beam's design is done opening by opening rather than at the section of maximum moment, and the reason openings are infilled or omitted near supports on heavily loaded beams. A plain-section calculation applied to a perforated beam misses it completely and returns an answer that looks fine.
- Can I just cut a hole in a plain beam instead?
- Yes, and it is routine — but it is a designed detail rather than a site decision, and the position matters more than the size. A hole in the middle third of the depth near mid-span, where bending is high and shear is low, is comparatively benign. The same hole near a support, where shear governs, is not. Large openings are usually reinforced with plates or rings around them, and the design does exactly the Vierendeel check described above. What must never happen is a hole cut on site to get a duct through, discovered by the structural engineer afterwards — that is a common and entirely avoidable cause of a beam that was adequate becoming inadequate without anyone touching the loads.
- Is a cellular beam the same thing as a castellated beam?
- Same idea, different cut. A castellated beam is cut on a zigzag line, producing hexagonal openings; a cellular beam is cut on a pair of offset semicircular lines, producing round ones. Round openings suit circular ductwork better and have a smoother stress distribution around them; hexagonal ones are the older, cheaper cut. Cellular fabrication also allows the opening diameter and spacing to be chosen more freely, including leaving solid web where a high-shear zone needs it. Structurally both are checked the same way, and the choice between them is usually driven by what the services need to pass through and by what the fabricator does well.
- Does it really save money?
- Sometimes on steel, often on the building, and the two calculations are different. On steel alone it can go either way: less tonnage, more fabrication per tonne, and a longer lead time. The saving that usually decides it is elsewhere — if the openings let services share the structural zone, every floor-to-floor dimension shrinks, and on a multi-storey building that means less cladding, shorter risers, shorter columns, less stair and lift travel and sometimes a storey gained under a height limit. That is a whole-building comparison and it is the one worth doing. A single-storey beam with nothing passing through it rarely justifies the product on steel savings alone.
- Why is fire protection more expensive?
- Because protection is applied to surface area and paid for by the square metre, while the structural benefit is measured in section properties. Cutting and offsetting a beam increases its surface area substantially without increasing its weight, so the ratio of protected surface to steel mass — the section factor that determines how quickly it heats — moves in the wrong direction. Intumescent coatings are therefore thicker and dearer, and board encasement has to negotiate the openings rather than boxing a simple prism. It is not a reason to reject the product, but it belongs in the comparison, and it is frequently left out of the steel-only sums that make the decision look obvious.
- How do I know whether shear or deflection governs my beam?
- Compare the span to the depth. Deflection grows with the FOURTH power of span while bending moment grows with the square, so long members are stiffness-governed and short ones are strength-governed — and shear capacity grows only with the web area, barely moving with depth at all. So as a beam gets shorter relative to its depth, shear demand rises against a capacity that has not changed, and at some span-to-depth ratio shear overtakes bending as the governing action. Long and deflection-governed is where a castellated beam earns its keep; short and shear-governed is where it is the wrong member, because the openings are precisely where the shear has to cross.