Cable-Net and Tensile Membrane Roof Structures
Most roof structures resist load through a mix of bending and compression, a truss chord in compression here, a beam bending there. A cable-net or fabric tensile roof refuses that entirely: every structural element in the system, the cables and the membrane skin alike, is meant to carry load purely in tension. A cable can be astonishingly efficient in tension, using a fraction of the material a bending member would need for the same span, but a pure tension structure only stays a structure if every element stays in tension under every load case it will ever see, which is a much harder design condition than it sounds.
Why Shape Isn't a Free Choice
A cable under a given load settles into exactly one equilibrium shape, the funicular shape for that specific load, and it can't resist any load pattern that doesn't match the funicular shape without changing its own shape to match. A single cable spanning under its own weight sags into a catenary; loaded at discrete points it breaks into straight segments between load points. Because roofs have to survive many different load cases, dead load, unbalanced snow drifting to one side, wind suction lifting part of the roof, a single-layer cable can't be shaped to be funicular for all of them simultaneously, so the actual engineering problem is prestressing the net so it stays taut and stable across every load case rather than snapping to a different shape and going slack.
This is why cable-net roofs are built with two families of cables curving in opposite directions, one set sagging like a suspension cable, the other arching the opposite way in plan and pulling against the first set. The doubly curved, saddle-shaped surface that results, an anticlastic surface where the two principal curvatures point opposite ways, is what gives the net enough prestress and geometric stiffness to resist both uplift and gravity load cases without any individual cable ever losing tension, similar in concept to how prestressed concrete uses locked-in internal force to keep a nominally weak material working the way the design intends across its full service load range.
Form-finding, the process of calculating the specific doubly curved geometry that puts a cable net or membrane into equilibrium under its target prestress, isn't a shape an architect sketches and an engineer verifies afterward; it's usually solved iteratively as part of the design itself, because an arbitrary sculpted shape generally has no valid tension-only equilibrium at all.
Membrane Skins Carry Load the Same Way, in Two Directions
A tensile fabric membrane, PTFE-coated fiberglass or PVC-coated polyester being the common choices, behaves like a continuous two-directional cable net rather than a plate; it has essentially no bending stiffness and resists load through in-plane tension distributed across its doubly curved surface. Fabric roofs are prestressed during installation specifically to guarantee that no region of the membrane ever goes slack and flutters under wind, since a fluttering membrane fatigues rapidly at its seams and clamped edges, the connection details that in practice govern the life of the roof far more than the fabric's bulk tensile strength does.
Edge Members Absorb Enormous Concentrated Force
Every cable in the net and every strip of membrane eventually has to anchor into something rigid, a compression ring, an arch, or a perimeter mast, and those edge members carry the accumulated tension from the entire tensioned surface as concentrated force. A circular tension-roof stadium, for instance, often uses a compression ring around the perimeter specifically to convert the inward-pulling radial tension from the cable net into ring compression, closing the load path without needing massive individual anchor foundations at every cable termination, a strategy conceptually related to how a dome's thin shell dome resolves meridional thrust into a ring beam at its base.
Design guidance for tensile and membrane structures is consolidated in guidelines published by the American Society of Civil Engineers, alongside material-specific design data from fabric manufacturers, since much of the governing behavior, fabric creep, seam strength, and long-term prestress loss, comes from material test data rather than from a single unified structural code the way steel or concrete design does.