Engineered timber as a structural material

Wood is a natural fibre composite: cellulose microfibrils wound through a matrix of lignin and hemicelluloses. That construction yields a material strong along the grain and weak across it, one that swells and shrinks with humidity and keeps deforming under a load that never changes. Engineered timber — glulam, cross-laminated timber, laminated veneer lumber — is not upgraded wood; it is wood with each of these properties measured and routed around. Understanding why it works, and where it stops, means taking the properties in turn.

What cross-lamination buys

Nearly all the axial stiffness of the cell wall sits in cellulose microfibrils aligned almost parallel to the grain. The consequence: the longitudinal stiffness of softwood is more than an order of magnitude higher than the transverse, and shrinkage from green to dry is a few tenths of a percent along the grain but several percent across it, with tangential roughly twice radial. A solid-sawn board therefore cups and splits as it dries. Cross-lamination interleaves layers at 90°, so every layer restrains the movement of its neighbours: panel movement in either direction approaches the small longitudinal value, and the differential shrinkage that would tear a solid panel apart is spread through the glue lines. Laminating does a second thing: defect averaging. Knots and grain deviation set the strength ceiling; slicing wood into lamellas or millimetre-thin veneers disperses them so no single defect governs, and graded layers put the stiffest material where bending stress peaks.

Moisture as the governing variable

The cell wall is hygroscopic: hydroxyl-rich hemicelluloses and amorphous cellulose bind water vapour until the wood equilibrates with the air. Up to roughly 30% moisture content — the fibre saturation point — water sits inside the wall itself, and every property shifts: stiffness falls, creep accelerates, dimensions move. Above it, water is free in the cell cavities and nothing structural changes. Hence lumber is kiln-dried to the moisture content of its future service environment, mass-timber plants check every board before pressing, and design codes specify service classes rather than one allowable stress. Most timber failures in service are moisture events: a detail that traps water creates a local gradient, the wet zone swells and shrinks against dry wood, and checks open that admit more water.

Fire as predictable charring

Wood does not melt and has no single failure temperature. In a fire the surface layer pyrolyses, ignites and burns out to charcoal, and the char — porous, low-density, a poor conductor — insulates the wood behind it while pyrolysis itself absorbs heat. The result is the strangest fact in structural fire engineering: loss of section proceeds at a nearly constant rate, about 0.65 mm per minute for softwood in the standard test, a figure written directly into Eurocode 5. The designer does not rate the material; the designer sacrifices a calculated outer layer and counts on the cool core retaining its room-temperature strength. Unprotected steel, by contrast, has lost roughly half its yield strength by 600 °C, well within a developed fire. The real weak points of a timber frame are its steel connections, which conduct heat past the char front and need separate protection.

Time under load

Wood creeps. Under a constant load, deformation continues for years as the matrix slowly yields between load-bearing microfibrils, and long-term deflection grows to match or exceed the instantaneous elastic value. Worse is mechano-sorptive creep: moisture cycling under load multiplies deformation several-fold, because each wetting softens the matrix and molecular slip accumulates. Codes answer with duration-of-load factors that cut allowable stress for permanent loads; the physics is not negotiable, only priced in.