Cable Structures

What is a cable structure?

A cable structure carries load in pure tension. A steel cable cannot resist bending or compression; it can only be pulled. That single limitation is the whole advantage, because tension stresses every fibre of the cross-section at once, where a beam in bending is fully stressed only at its extreme fibres and carries much of the rest of its section as dead weight.

• Beam in bending Only the top and bottom fibres reach full stress. The middle of the section is carried along as weight.

• Cable in tension Every wire carries the same stress, so none of the section is idle — and the wire itself is several times stronger than structural steel.

The consequence is a structure that can cross 15 – 120 m on very little material.

The terms get used interchangeably and shouldn't be:

• Cable structure — the general case. Any structure whose primary load path is cables in tension.
• Tensile structure — the same thing, more often used when the covering is a membrane and the form is sculptural.
• Membrane structure — the covering itself carries load in tension, with cables at the edges and ridges. The fabric is structural, not cladding.
• Cable net — cables tensioned in two opposing directions forming a stiff surface, carrying infill at the intersections. A distinct system, System 08.
• Cable-stayed — cables supporting a rigid element, like a mast supporting a beam. The rigid element still bends; the cables reduce how much.

In every tension-only system, stability comes from pre-stress, not from stiffness. A cable is only stable when it is already tight, which is why pre-tension is a design quantity rather than an installation setting.

Which cable system does your project need?

Four systems in this family, and they are not interchangeable. The question that separates them is what the cables are doing:

The practical distinction: on this page the cables carry the covering. On 06 and 07 they reinforce something else. On 08 the net is the surface.

• 05 Cable structures The cables are the structure, carrying a covering over open space.

• 06 Cable-supported curtain wall Vertical cables replace mullions; glass is point-fixed to them.

• 07 Cable-supported skylight Cables reinforce a steel lattice so its members stay slim.

• 08 Cable net structures Cables in two directions form the surface itself.

Form finding: why you cannot draw it first

Shape and force are solved together. A cable structure cannot be drawn and then analysed; the equilibrium shape is the design.

This is the single most important thing for an architect to understand about this system, because it inverts the normal workflow.

With a framed structure you draw a shape and an engineer verifies it. With a cable structure the shape is an output. Given the boundary conditions — where the anchor points are — and a chosen pre-tension, there is exactly one equilibrium shape the structure will take. Draw a different one and it will not stay there; it will move to the equilibrium shape the moment it is tensioned, or it will not be stable at all.

What this means in practice:

• You choose boundaries, not surfaces. The architect's control is over where the high points, low points and anchor positions sit. The surface between them follows.
• Anticlastic curvature is not a style. The saddle shape characteristic of tensile structures — curving up in one direction and down in the perpendicular one — is what makes them stable. Opposing curvature lets one set of cables hold the other set tight. A flat surface can only be stiffened by pulling it far harder, and one that curves the same way in both directions cannot be pre-stressed against itself at all without ballast or internal pressure.
• The model is the drawing. Form finding produces a geometry that has to be built to, and the setting-out comes from that model rather than from dimensions on a plan.
• Changing one anchor moves the whole surface. There is no local edit. Relocating a mast because of a services clash re-forms the entire roof.

• You set the boundary Mast heads and anchor points are the architect’s decision. The surface between them is what equilibrium makes of them.

• Move one point, the whole surface moves Raise one mast and every cable re-forms, not just the ones near it. There is no local edit.

Schematic. A real form-finding model also sets the sag of the edge cables and the pre-tension that holds the shape.

The design conversation is therefore about boundaries, sag, and how much curvature the architecture will accept — not about the surface itself. Bring that conversation forward and the geometry is yours; leave it late and it belongs to the equilibrium. The same form-found logic runs through free-form structures.

Pre-tension: a number with cost at both ends

Too little and the structure flutters; too much and the anchorage grows. The pre-tension range is a design decision with cost consequences at both ends.

That sentence contains the whole trade, and it is worth being explicit about both failure modes.

Too little pre-tension. Cables go slack under reversing wind. A slack cable is not a structure — it is a rope that goes taut again with a shock load. The result is movement, noise, snatch loading at the fittings, and fatigue at the connections. This is the failure mode that shows up as a complaint rather than a collapse.

Too much pre-tension. Every unit of pre-tension is a permanent force that must be resisted for the life of the building. It does not go away when the wind drops. Raise it and the cables get bigger, the fittings get bigger, and — most expensively — the anchorage and foundations get bigger. Pre-tension is the cheapest thing to increase on a drawing and the most expensive thing to resist in concrete.

The optimum keeps every cable in tension under the full envelope of load cases, including reversal, with as little margin as the analysis permits. Pre-tension is derived from form finding, not chosen — but the target range is a design decision made jointly with the structural engineer early, because the anchorage is sized from it.

1. 1 · Too little Cables go slack under reversing wind — movement, noise, snatch loads and fatigue at the fittings.
2. 2 · Too much A permanent force for the life of the building. Cables, fittings, anchorage and foundations all grow with it.
3. Target Every cable stays in tension under every load case, reversal included, with as little margin as the analysis allows.

Illustrative shapes, not to scale. The actual range comes from form finding on the project, agreed with the structural engineer.

Anchorage: where these projects are won or lost

Cable forces have to go somewhere. Foundations, ring beams or the primary frame must be designed for them from the outset.

This is the most common commercial failure on cable structures, and it is almost always a sequencing problem rather than an engineering one.

A cable roof is light. The reactions it produces are not. A tensile structure pulls its anchors inward and often upward, and those forces are permanent, whereas a conventional roof mostly pushes down and only occasionally lifts. Three consequences:

• The anchorage is frequently a larger structure than the cable roof it serves. Tension foundations resist uplift, which usually means mass, ground anchors or piles — not the pad footings a light roof suggests.
• It cannot be added later. If the primary frame has been designed and priced before the cable roof is engineered, the reactions arrive as a change to somebody else's completed design. That is the moment cable schemes get value-engineered out.
• Ring beams close the force loop. Where the cable structure can be tied to itself — a compression ring taking the pull of the cables — the horizontal forces never reach the foundations at all. Where geometry allows it, this is by far the cheapest anchorage strategy, and it is a geometric decision made at concept.

• Loop closed The cables pull against a compression ring. Their horizontal forces balance inside the roof, and the building below carries only its weight.

• Loop open Pre-tension goes to the ground. The anchors resist pull and uplift permanently, whether or not the wind is blowing — mass, ground anchors or piles.

Practical guidance: get indicative reactions from the façade or structural engineer before the primary structure is fixed. Even approximate figures at concept stage let the frame and foundations allow for them, and they cost little to produce.

Cable types: what to specify and why

Three constructions on this system, in ascending order of stiffness and cost.

Structural strand. Wire strands laid helically around a core. The most economical and the most flexible, with the lowest axial stiffness of the three, because the layered construction beds down and stretches under load before its stiffness settles.

Spiral strand. The general-purpose structural cable. Round wires in concentric helical layers, pre-stretched during manufacture to remove constructional stretch so the installed stiffness is predictable. The default for most façade and roof applications.

Full-locked coil. Z-shaped outer wires interlock to form a closed, smooth outer surface. Three advantages follow: a higher fill factor and therefore more steel in the same diameter; a sealed surface that resists water ingress into the core, which is where cables corrode from the inside; and higher axial stiffness. It is the choice for long spans, for high loads, and wherever the cable has to last with minimal maintenance.

• Structural strand Strands laid helically around a core. The most economical and the most flexible, with the lowest axial stiffness.

• Spiral strand Round wires in concentric layers, pre-stretched in manufacture so the installed stiffness is predictable. The general-purpose structural cable.

• Full-locked coil Z-shaped outer wires lock together: more steel in the same diameter, a sealed surface that keeps water out of the core, and the highest stiffness.

Cross-sections drawn to the same diameter. Diameter range 10 – 100 mm; locked coil to 180 mm.

Diameter range: 10 – 100 mm; locked coil to 180 mm. Wire tensile strength: 1,370 – 1,570 MPa to EN 1993-1-11. For comparison, structural sections are typically S235 or S355, with yield strengths of 235 and 355 MPa — the wire in a structural cable is several times stronger, which is the other half of why so little material is needed.

Design is to EN 1993-1-11 and EN 1993-1-1, with ropes to EN 12385-4.

End fittings, and how tolerance gets absorbed

Open and closed sockets, forks, and adjustable turnbuckles.

The fittings are not an afterthought — they are how a structure with no bending stiffness is made to fit a building that was built to construction tolerances.

• Sockets are the primary termination, the cable end resin- or zinc-cast into a conical socket that develops the cable's strength. An open socket is forked and takes a pin through a plate; a closed socket has a solid eye that sits between plates or inside a clevis.
• Forks connect to a gusset or bracket and allow rotation in one plane — important, because a cable must be free to align with its own force direction. Restrain that rotation and you introduce bending into a component that cannot resist it.
• Adjustable turnbuckles are where the tolerance lives. Cables are manufactured to length; buildings are not. Adjustment lets pre-tension be introduced, measured and corrected, and it is how the structure is brought onto its form-found geometry.

• Open socket The cable end is cast into a conical socket. Forked: the plate goes between the jaws and a pin passes through both.

• Closed socket A solid eye instead of jaws. It sits between plates, or inside a clevis, on a pin.

• Fork Pinned to a gusset or bracket, free to rotate in one plane so the cable can align with its own force. Restrain it and the cable is put into bending.

• Adjustable turnbuckle Opposite-handed threads draw the ends in or out. This is where the tolerance lives, and how pre-tension is introduced, measured and corrected.

Simplified to the working parts. Proportions vary with cable size and manufacturer.

Design the adjustment range deliberately. Too little and the structure cannot be tensioned to the design geometry on a building that is a few millimetres out. Too much and the fittings grow and become visible on a system whose entire appeal is that the structure nearly disappears.

Corrosion protection and service life

Galvanised Class A, Galfan, or stainless 1.4401 / 1.4462. Cable life is a coating and detailing question.

• Galvanised Class A — the heaviest zinc coating class for wire ropes. Economical, appropriate for inland and sheltered environments.
• Galfan — a zinc-aluminium alloy coating that outperforms plain zinc in the same thickness, particularly in marine and industrial atmospheres. Usually the best value in a moderately aggressive environment.
• Stainless 1.4401 — grade 316, the general architectural stainless. Chosen as much for appearance as for durability; it stays bright where a galvanised cable dulls.
• Stainless 1.4462 — duplex. Higher strength than 316 and substantially better resistance to chloride-induced attack. The choice for coastal and other chloride-rich environments, and where a smaller diameter is wanted at the same capacity.

Two details matter as much as the material:

Water must not sit at the fittings. Cables corrode at the termination far more often than along the length, because that is where water collects and where the coating is interrupted. Sockets are detailed to drain.

Inspection has to be possible. A cable that cannot be inspected is a cable whose condition is unknown. On a long-life structure, access to the terminations should be designed in, not improvised later.

Coverings

PTFE or PVC membrane, ETFE, glass, or metal panel. The covering changes the structure, because on a light roof the covering is a large share of what the cables are carrying.

PVC-coated polyester is the economical membrane. More flexible, easier to fabricate and repair, shorter service life, and it discolours over time. Right where budget governs or where the structure is not expected to be permanent.

PTFE-coated glass fibre costs more and lasts substantially longer. It performs far better in fire, is effectively self-cleaning in rain, and holds its appearance. Stiffer and less forgiving to fabricate, so the patterning has to be right.

ETFE is a foil rather than a woven fabric — the lightest option, the most transparent, and repairable in place. Rain noise is real and should be raised with the client before it is discovered.

Glass changes the problem entirely. It is rigid on a structure that moves, so every fixing must articulate, and the dead load transforms the cable sizing. Where the requirement is a glazed cable-supported surface, System 08 or System 06 is usually the right page.

Metal panel suits opaque roofs where the form is the point and daylight is not.

What drives the cost

Cable structures are priced differently from framed ones, and the intuition most people bring to them is wrong. The cables are rarely the expensive part. What costs money is everything that has to resist them.

In order of leverage:

1. Whether the force loop closes

This dwarfs everything else on the list.

If the geometry allows the cable tension to be balanced against a compression ring — the structure pulling against itself — the horizontal forces never reach the foundations at all. The roof becomes self-equilibrating, and the building below only has to carry its weight.

If it does not close, every unit of pre-tension has to be resisted by the primary frame or the ground. That means mass, ground anchors or piles, permanently, whether or not the wind is blowing.

The cheapest cable structure is the one whose pre-tension never reaches the ground. Whether that is possible is decided by the plan geometry at concept stage, by the architect, usually before anyone has asked a structural engineer. It is worth asking the question early, because the answer is often yes with a small adjustment and almost never yes after the plan is fixed.

2. Span, against the anchorage it needs

Anchorage does not shrink in proportion to the span. Even a short canopy needs anchor points, fittings and foundations sized for its pre-tension, and those costs have little area to be spread across.

Over a short span, a conventional frame will usually beat it. Across a long span the anchorage is spread thinly and the system pulls away from the alternatives. Below the bottom of this system's range, 15 m, it is usually the economics that stop working rather than the engineering.

3. The covering

On a light roof the covering is a large share of both the cost and the load, and the choice cascades. PVC-coated polyester is the economical membrane; PTFE-coated glass fibre costs more up front and usually less over a building's life. ETFE sits apart — the lightest and most transparent, and where it is used as inflated cushions it brings an air-supply system with it.

Glass is the step change. It is rigid on a structure that moves, so every fixing must articulate, and its dead load resizes the cables, the fittings and the anchorage beneath them. A glazed cable structure is not a membrane structure with different infill; it is a different project.

4. Cable construction and corrosion specification

Two ladders, both driven by the project rather than by preference. Structural strand, spiral strand, full-locked coil — ascending cost, ascending stiffness and durability. Galvanised Class A, Galfan, 316, duplex — ascending cost, ascending resistance.

Cable construction

1. Structural strand
2. Spiral strand
3. Full-locked coil

Cost, stiffness and durability rise →

Corrosion protection

1. Galvanised Class A
2. Galfan
3. Stainless 1.4401 (316)
4. Stainless 1.4462 (duplex)

Cost and resistance rise →

Neither ladder is where a scheme is won or lost, but specifying stainless where Galfan would serve, on a structure with hundreds of metres of cable, is an easy way to spend money on the wrong thing. Both are environment decisions, not aesthetic ones — except where the cable is seen at close range, in which case say so, because that is a legitimate reason to pay for 316.

5. How many fittings are unique

Same principle as panelisation on a free-form surface. A structure using one fitting type two hundred times costs a fraction of one using forty types five times each. Fitting families are worth rationalising for exactly the reason panel sets are.

6. Whether the primary structure was designed for it

Strictly this is not a cost of the cable structure at all — it is the cost of finding out late.

If the frame and foundations were designed and priced before the cable reactions existed, those reactions arrive as a change to somebody else's completed work, at the worst possible point in the programme. It is the single most common reason cable schemes are abandoned, and it is entirely avoidable: indicative reactions at concept stage cost little to produce and are enough for the frame to allow for them.

Deflection and movement

The specification gives the deflection limit as a serviceability limit agreed with the design team rather than a fixed ratio. That is deliberate rather than a gap.

A cable structure moves. That is not a defect — it is how it resists load, by changing shape until the geometry balances the force. How much movement is acceptable depends entirely on what the structure touches: a free-standing canopy can move a great deal, while the same structure abutting a glazed wall cannot move more than the joint between them will absorb.

So the limit is set by the interfaces, not by a code table. In practice that means:

• Agree the movement budget early, with everyone whose work meets the cable structure.
• Design the interfaces for the movement, not the movement for the interfaces.
• Dynamic behaviour is checked separately from static deflection. A light, flexible, pre-stressed surface can flutter or resonate under wind well below the static design load. Adequate pre-tension is the primary defence.

What to decide, and when

1. 01

   Concept

   What is being covered and how open it must be. Where anchor points can go, and whether the force loop can be closed with a ring beam rather than taken to ground. The covering, at least in category — membrane, ETFE or rigid. This is the stage where the architect has most control over the final shape.

2. 02

   Developed design

   Form finding, which produces the geometry and the pre-tension range. Indicative anchorage reactions issued to the structural engineer before the primary frame is fixed. Cable construction and corrosion specification against the environment. The movement budget agreed with the interfacing trades.

3. 03

   Technical design

   Cable schedule and fittings, adjustment range, the tensioning sequence, and the interface details that absorb the movement.

4. 04

   Before tender

   A form-found geometry and a defined anchorage strategy. A cable structure tendered on a drawn shape rather than a found one will be priced with a risk margin, because the tenderer knows the shape will change.