Modern structural engineering is extremely good at using the least material necessary to carry a load. That is a genuine achievement. It is cheaper, it uses less timber, it spans further, and it lets you build a floor with no load-bearing walls underneath it. Under normal conditions these assemblies are perfectly adequate and often better than what they replaced. Under fire conditions they are unforgiving in a specific and predictable way, and the reason is that the efficiency was achieved by removing exactly the material that fire has to consume before the assembly fails. This lesson is about that trade. It applies to almost everything built in the last forty years, which means it applies to most new residential construction, most commercial fit-out, and a great deal of what is going up on your streets right now.

Three objectives. First, you should be able to describe how a lightweight truss and an engineered I-joist are actually made, because both failure modes follow directly from the construction. Second, you should be able to explain why they fail differently from traditional dimensional lumber, and in particular why the failure is sudden rather than progressive. Third, you should be able to look at a building and estimate the likelihood that it contains these assemblies, using era, span, and a handful of exterior clues, since you will almost never get to look at the floor system itself. The word to watch through this lesson is warning. The question is not only whether an assembly fails but whether it tells you first.

A lightweight wood truss is a set of thin members, often nothing more than two by four, arranged in triangles to carry a load across a long span. The members are held together at each joint by a stamped steel plate whose teeth press into the wood, typically penetrating something like three eighths of an inch. Read that number again, because everything follows from it. The strength of the truss is not in the timber. It is in those connections, and the connections are three eighths of an inch deep. Under fire, two things happen at once. The steel plate conducts heat straight into the small volume of wood holding its teeth, and it expands and distorts while the timber beneath it chars and shrinks away from it. The teeth lose their grip. When one joint lets go, the load redistributes to joints that were never designed to carry it, and the failure propagates across the truss and then to the trusses beside it. Testing has shown these assemblies failing in as little as five to ten minutes of fire exposure.

Engineered I-joists follow the same logic in a different shape. A thin vertical web of oriented strand board separates two small solid flanges at top and bottom. Structurally that is elegant, because in a beam under load the stress lives at the top and bottom and the middle does very little, so you can remove almost all of the middle. The problem is what is now in the middle: a web perhaps three eighths of an inch thick, made of wood chips and adhesive, positioned exactly where fire arrives from the floor below. It does not need to burn through much to lose the web entirely, and once the web is gone the flanges have nothing holding them apart and the joist has no depth left to work with. Worse, these webs are routinely drilled for ducts, wiring and plumbing, so the field-modified joist above a basement fire may already have had material removed at the point where it matters most.

So here is the rule that ties the lesson together, and it is worth memorising in exactly these words. Under fire, mass is time. A heavy timber section chars on the outside at a fairly predictable rate, something on the order of an inch and a half an hour, and the char layer itself insulates the sound wood underneath. So a large section under fire loses capacity gradually, sags, creaks, and gives you warning before it goes. A thin section has no reserve to char away. It reaches the temperature at which it fails and it fails, without the intermediate stage where somebody notices. That is the real difference, and it is not primarily about how long, it is about whether you get told. Old buildings with heavy members warn you and then fail. Modern lightweight assemblies fail. Any judgement about how long you can stay in a building has to account for which of those two you are standing on.

Four misreadings, and what is actually true. The first is that new means safer, which is true for ignition and for egress and false for structural endurance under fire. The second is that a finished ceiling tells you what is above it, when a layer of gypsum board conceals a truss floor and a solid-timber floor equally well, and its protection lasts only as long as the board stays attached. The third is that a floor that feels solid underfoot is structurally sound, when lightweight assemblies retain nearly all of their stiffness right up to the point of failure, so the floor feels normal until it is not there. And the fourth is that the roof and the floor are the same construction, when it is very common for a building to have a lightweight truss roof over conventional framing, or the reverse after a renovation. The through-line is that surfaces conceal assemblies, and the assembly is what fails.

Think about the buildings going up near you right now, and about the ones finished in the last twenty years. Which of them were framed with assemblies that nobody has laid eyes on since the drywall went up? For almost all of them the honest answer is all of them, and that is the point of the question. Then ask the useful follow-up: what would tell you? Long clear spans with no interior columns, a wide-open ground floor under living space above, garages with rooms over them, gable ends with a very shallow pitch, and the simple fact of the era, all point the same way. And ask who else knows. Pre-incident planning, building department records, and in some jurisdictions truss identification placards at the entrance exist precisely because this information cannot be recovered from looking at the finished building.

Three things to remember. Lightweight trusses fail at their connections rather than through their members, and those connections are steel teeth biting about three eighths of an inch into wood, which is why failure can arrive within five to ten minutes and why it propagates across the assembly instead of staying local. Engineered I-joists are thin exactly where fire reaches them, and the web is often already drilled through for services. And mass, under fire, is time and also warning: heavy sections sag and complain before they go, thin ones do not, so the absence of any signal from a modern floor is not evidence that it is sound.