Strength depends on how long the load lasts
Timber's published design values are quoted for a reference duration — conventionally ten years of accumulated load — and are then ADJUSTED for the actual duration of the load in question. The adjustment goes both ways.
A load lasting minutes, such as a wind gust, permits a value well above the reference: factors around one and a half to one point six are standard. A load lasting the life of the structure — dead load, a permanently stacked store — permits a value BELOW it, around nine tenths. So the same joist is genuinely stronger against a gale than against a bookcase.
The mechanism is creep rupture: wood under sustained stress accumulates damage and eventually fails at a load it would carry indefinitely if applied briefly. That is the same physics as static fatigue in glass, and it means duration is a material property rather than a loading convention.
The practical rule is that the governing case is whichever combination of load and its duration factor produces the worst demand-to-capacity ratio — not simply the largest load. A roof checked only for its snow case can be governed instead by its dead load at a lower factor, and a calculation that applies one duration factor to a combined load has answered the wrong question.
- C_D
- load duration — above 1.0 for short loads, below for permanent ones
- C_M
- wet service: a substantial reduction whenever moisture content stays high
- C_t
- temperature, for sustained elevated service
- C_F
- size factor — deeper members are proportionally weaker in bending
The factors multiply, and moisture is the one that bites
Because the adjustments are multiplicative, several individually modest reductions compound quickly. A member that is wet in service, warm, deep and permanently loaded can be operating at well under two thirds of the value printed in the table its grade came from.
Moisture is the largest of them in ordinary construction. Timber used where its moisture content stays high — external, in ground contact, in an unventilated void, or in a building that has not dried out — loses bending strength and stiffness by meaningful percentages, and the same condition is what permits decay to start at all.
Size is the factor that surprises people, because it points the wrong way. A deeper member is proportionally WEAKER in bending than a shallow one of the same grade, because a larger volume is more likely to contain the defect that governs. It still carries more, because depth is squared in the section modulus, but not by quite as much as the geometry alone suggests.
And grading is the foundation under all of it. A design value belongs to a species, a grade and a grading method, and timber that has lost its grade stamp has lost its design value — which is why reclaimed or unmarked timber cannot simply be assigned a value from a table.
Notching and boring: geometry, because the failure is a crack
The rules for cutting into a joist, rafter or stud are expressed as fractions of the member's depth and as zones along its length, and they are not derived from a stress calculation. They cannot be, because the failure they prevent is not a stress failure.
A notch creates a sharp re-entrant corner, and at that corner tension develops ACROSS the grain — the direction in which wood is weakest by an order of magnitude. A crack starts there and then runs along the grain, splitting the member far beyond the notch. The loss of capacity is therefore much greater than the proportion of material removed, and it does not scale smoothly with notch depth.
That is why the rules distinguish so sharply between a notch and a HOLE. A round hole has no sharp corner, so it removes material without creating the crack initiator, and a hole in the middle third of the depth away from the supports is comparatively benign. A notch of the same depth at a bearing is not.
Position along the span matters for the same reason as in the section-capacity paper: shear is highest at the supports and bending highest at mid-span, so a notch near a bearing attacks the member where shear governs. The permitted zones in the codes encode exactly that, which is why a cut that is legal in one part of a joist is prohibited a metre away.
Fire: timber loses section, steel loses strength
Wood burns at a rate that is remarkably predictable. A charring layer forms on the exposed face and insulates what is behind it, so the fire front advances into a solid member at roughly a fixed number of millimetres per minute — commonly quoted around two thirds of a millimetre for softwoods.
The residual section is what carries the load: the original dimensions less the char, less a thin zone immediately behind it whose strength has been reduced by heat. Everything inside that remains at ambient temperature and at full strength, which is the whole reason heavy timber performs as it does.
The comparison with unprotected steel is the point. Steel does not lose section in a fire; it loses STRENGTH throughout, and by around five to six hundred degrees it retains roughly half of it. A large timber member meanwhile still has most of its section at full strength. That is why exposed heavy timber and mass timber can achieve fire resistance periods by calculation while unprotected steel of equivalent capacity cannot.
The method has clear limits. It applies to members large enough that a residual section is left, which is why it is a mass-timber and heavy-section technique rather than a light-framing one; light frames are protected by their linings instead. It also assumes the member chars on the faces exposed and that connections — which are usually steel, inside the timber — are detailed to be protected, because a fully adequate residual section joined by an exposed plate fails at the plate.
Deck spans are the easy part; the ledger is the dangerous one
Joist span tables answer a bending-and-deflection question with a lookup, and for a deck they are rarely what governs safety. Almost every catastrophic deck failure is a COLLAPSE AWAY FROM THE BUILDING, and it is the ledger connection that lets go.
The reason it is dangerous is that the connection is hidden, it is loaded in a direction people do not visualise, and the commonest method of making it — nails, or worse, a nailed ledger relying on the cladding behind it — has a fraction of the capacity of a properly bolted or screwed connection into the rim board.
Water makes it worse over time. A ledger without flashing traps water against the building, the rim board behind it decays, and the fasteners end up anchored into rotten timber while looking exactly as they did on the day of installation. The failure arrives years later, under a crowd load, which is the one time the deck is fully loaded.
So the honest reading of a deck joist calculation is that it sizes a member and says nothing about whether the deck will stay attached. The pages here return the span; the connection is a detail from a prescriptive table or a designed one, and it is the part worth the attention.
Cross-laminated panels behave differently from solid timber
A cross-laminated panel is built from layers at right angles to one another, so only the layers running WITH the span contribute their full bending stiffness. The effective section is therefore smaller than the panel's thickness suggests, and it depends on the lay-up rather than on the depth alone.
The crosswise layers introduce a failure mode that does not exist in solid timber. Shear across those layers acts perpendicular to their grain and tries to roll the fibres over one another — ROLLING SHEAR — and its strength is a small fraction of ordinary shear strength. On short, heavily loaded spans it frequently governs, which is a check a solid-timber intuition does not prompt.
At the other end, long-span floor panels are usually governed by neither strength nor deflection but by VIBRATION. Mass timber floors are light and stiff, which puts their fundamental frequency and their response to footfall squarely into the range the deflection-and-vibration paper describes, and it is the commonest reason a panel is thicker than its load requires.
Panels are also a dimensional and logistical product before they are a structural one. Sizes are governed by press dimensions and by what can be transported and craned, so a layout is a packing exercise against available panel sizes — and the calculators here return a panel count against those, not a structural design.
Timber in contact with the ground
Decay needs moisture, oxygen, a temperature range and food. Timber supplies the last; the detailing decides the first two, and that is where durability is actually won or lost.
Preservative treatment is graded by use class, and a treatment adequate for a member kept dry and ventilated is not adequate for one in ground contact. Cutting a treated member exposes untreated timber at the cut face, which is why end cuts are re-treated and why a skid or a sole plate cut to length on site is at its most vulnerable exactly where it meets the ground.
The detail beats the chemistry. Separating timber from the ground with a bearer, a damp-proof membrane or a metal foot, keeping it ventilated on all faces, and shedding water away from horizontal surfaces each do more than an upgrade in treatment class. A skid sitting directly on soil will fail whatever it has been treated with; the same skid on a gravel bed with air beneath it will not.
The calculators here size and count. Durability is a specification and a detailing decision, and the pages say so rather than implying that a member sized correctly is a member that will last.
Calculators that use this method
Basis
- National Design Specification for Wood Construction (NDS) — the adjustment factor chain, load duration factors, wet service and size factors, and the notching and boring provisions.
- Eurocode 5 (EN 1995-1-1) for the equivalent treatment through modification factors and service classes, and EN 1995-1-2 for the charring method.
- NDS Chapter 16 and the effective char depth method, including the zero-strength layer behind the char front.
- International Residential Code R502.8 and R602.6 for permitted notches and bored holes in joists, rafters and studs, and R507 for deck ledger attachment.
- AWC Prescriptive Residential Wood Deck Construction Guide (DCA 6), for ledger fastening schedules and flashing.
- ANSI/APA PRG 320 for cross-laminated timber, including lay-up, effective stiffness and rolling shear.
- AWC/CLT Handbook and SCI guidance on mass timber floor vibration, the usual governing criterion on long panel spans.
- AWPA use category system and EN 335 service classes for preservative treatment, including the requirement to re-treat field cuts.