Checking Truss Shop Drawings

Two duration factors, printed a line apart

Wood gets stronger the faster you load it, and the National Design Specification handles that with the load duration factor. The values in NDS Table 2.3.2 are not a range to choose within: each belongs to a load whose cumulative duration the table names, and the governing check takes the factor for the shortest-duration load in the combination examined. Every allowable value on the sheet has been through that multiplication except two the NDS excludes from it: modulus of elasticity, and compression perpendicular to grain — which is the value the bearing check further down this page runs on.

The trap on a truss drawing is that there are two duration factors, printed on adjacent lines, and they are not always the same number. One adjusts the lumber design values. The other adjusts the connector plate values, and the increase permitted on plate grip for short-duration load is governed separately by ANSI/TPI 1 rather than copied across from the lumber column. On a wind case especially, one factor doing both jobs is worth a question before it is worth a seal.

The repetitive member factor is the other adjustment that travels where it does not belong. It exists because closely spaced bending members sharing load through a deck are collectively stiffer than any one alone, and it is conditional on that spacing and that sharing. A girder carrying the end of a hip system is a single heavily loaded member, not one of a repeating set, so a repetitive-member increase on a girder sheet is an error rather than a habit. Wet service and elevated temperature adjustments run the other way and belong on any truss for a pool hall, a wash bay, an unconditioned agricultural building, or a space that will run humid before the roof closes in.

The ceiling is the part the owner will actually look at

A truss can be entirely safe and still generate a phone call, and the bottom chord is where the two verdicts part company. The code's deflection table gives L/360 under live load where a member supports a plaster or stucco ceiling and L/240 where the ceiling is not plaster, so a sheet can satisfy the looser figure comfortably and still open a hairline the length of a corridor at a taped joint. Where the finish schedule shows a level-five flat ceiling, a skim coat or a shadow gap, test the sheet's stated criterion against that specification rather than against the code minimum. Plenty of truss designers already work tighter than the code floor; plenty do not, and only the ratio printed on the sheet tells you which.

There is a modelling trap here that catches engineers and not carpenters. The bottom chord of a plated truss is not a simply supported beam across the truss span. It is a continuous member restrained at every panel point where a web lands on it, carrying axial tension from the truss action along its length and bending only locally between those panel points. The simple-beam deflection expression describes that local bending, and the length it wants is the panel length, not the span.

Put a real case through it. A 38 by 140 mm bottom chord has a moment of inertia of about 8.69 million mm to the fourth. A gypsum ceiling with insulation over it and a limited-storage live allowance comes out near 0.75 kPa, which at 610 mm centres is roughly 460 N per metre of chord. Over a 2.4 m panel between webs that deflects a shade over 2 mm against a local limit of 10 mm and passes without argument. Give the same numbers an 8 m span and the expression returns about 257 mm against a 33 mm limit — a catastrophic-looking failure describing nothing, because no bottom chord spans 8 m in bending. That number belongs to a ceiling joist which does not exist.

The truss's own vertical deflection at midspan is a different quantity, produced by the axial deformation of the whole triangulated system, and the truss designer's software prints it as a millimetre value and a ratio. That is the figure to hold against the finish. Use the panel check for what it is good at: local sag between webs on a long open panel with a heavy ceiling on it.

Both figures then have to be aged. Wood creeps under sustained load, and the NDS handles it by multiplying the long-term component of deflection by a creep factor before adding the short-term component back — 1.5 for lumber installed seasoned, 2.0 for unseasoned. A truss acceptable on day one and unacceptable at 1.5 times its dead-load component looks fine at handover and gets photographed by the client in year three.

One more effect belongs to the bottom chord and appears on no sheet at all. In a cold climate the bottom chord sits buried in insulation while the top chords sit in the ventilated space above, and the moisture-content difference between them arches the truss upward through the winter and settles it back in spring. The ceiling lifts off the interior partitions, opens a line at the wall junction and closes it again by May. Nothing is broken, and the remedy is a detail rather than a calculation: hold the ceiling board on clips near the partitions, keep the perimeter fasteners back, let the trim float. An interior partition connection detail in the set is there for this.

Camber buys geometry, not stiffness

The camber figure on a sheet is a number the plant sets on the jig, and it is sized off dead load alone: the value quoted across the industry is around one and a half times the calculated dead-load deflection. That factor is not arbitrary — it is the seasoned-lumber creep multiplier from the NDS, appearing here under a different name. Live load has no business in it, since a truss cambered for snow would sit visibly bowed in July.

What camber does not do is buy compliance. Deflection limits are checked against calculated deflection, and a truss that fails a ratio does not pass because it was cambered; a marginal ratio beside a generous camber is two facts about the same truss, only one of which is a check. Where camber matters to your review is coordination — a cambered girder on face-mount hangers throws the carried trusses out of plane, a cambered bottom chord puts a flat ceiling grid out of level, and a long cambered truss over a glazed wall moves the head clearance the window schedule was drawn to.

Plates: two capacities, one callout, and the area that does not count

The plate callout is the densest thing on the sheet and the part most often skimmed. It names a manufacturer, a plate designation, a size, a gauge, an orientation and a dimensioned location relative to the joint, and all six are load-bearing information. A plate is not a generic part: its tooth pattern, tooth length, steel thickness and published values belong to one manufacturer's evaluation report, so a truss designed on one maker's plate values cannot be fabricated with another maker's plate of the same nominal size.

Two independent capacities have to be satisfied at every joint, and they fail differently. Tooth capacity is the grip into the wood; it scales with plate contact area on each member either side of the joint line, and when it governs the plate withdraws from timber that is otherwise intact. Steel capacity is the plate's net section in tension and its shear along the joint line; it scales with plate width crossing that line, and when it governs the plate tears. Adding area in a direction that adds no width across the line buys nothing, which is why sizing by area alone is a first pass rather than a design.

Grip values are not a single constant either. ANSI/TPI 1 tabulates them for combinations of plate orientation to load and load direction to grain, because a tooth resists differently along the grain than across it and a plate differently along its slots than across them. Values fall with wood density too, so a joint that works in southern pine may not work in the same geometry in spruce-pine-fir. Take the number for the orientation the sheet draws.

Then there is area that exists on the plate and does not count. Teeth within a short distance of the joint line or of a member edge cannot develop full value, and ANSI/TPI 1 requires a placement tolerance to be allowed so a plate landing slightly off position still has the area assumed. Gross area therefore always exceeds effective area: a plate that measures generous with a rule can be tight on paper, a plate drawn hard against a chord edge is a defect rather than a tight fit, and a plate overhanging the wood is not a plate at all across the overhang.

Both faces of every joint are plated, one each side, pressed together — which is why an area check produces a per-side figure rather than a total. It also explains the commonest plant defect, incomplete embedment on one face from a cold press, a worn roller or a crowned member. Fabrication under an approved quality assurance programme or third-party inspection agency, which the code requires, exists to catch that, and the agency's name belongs somewhere in the submittal.

If you are sanity-checking a heavily loaded joint, get the units right before the arithmetic. Grip values are published in pounds per square inch of plate contact area, usually in the low hundreds — but the figure moves with species, orientation and load-to-grain angle, so take it from your plate's own report rather than a remembered band. Converting once: 200 psi is 1.38 newtons per square millimetre, which is 0.138 kN per square centimetre. That is the shape of number the check below wants.

The heel takes the largest force in the least room

On a common truss the axial force in the top chord is largest at the heel, and the heel is where the least depth is available to plate it. That is the whole difficulty in one sentence. A conventional heel lands the top chord on the bottom chord at the wall line with a plate depth of a few inches, and the plate transferring the full chord force has to fit inside it. When the geometry runs out, the answers are a wedge block, a scab, a taller heel or a different truss — and whichever it is has to be drawn, because it is a component the plant must cut and the site must not remove.

A raised heel is the usual resolution, and it arrives for a different reason: insulation. The energy code allows a reduced ceiling insulation value to be deemed compliant where the full uncompressed depth carries over the wall top plate at the eave, which needs vertical room at the heel to hold it. The stack-up is that insulation depth plus the bottom chord's own depth — not the whole dimension, because the attic ventilation provisions require a clear airspace between insulation and sheathing at the vent, one inch in the residential code, and the baffle holding it open sits in the heel too.

Heel height then propagates through drawings that are not truss drawings: wall plate height for a given ridge, gable-end stud lengths, fascia depth, the point where cladding stops. A heel changed after the order to fix an insulation problem is not a note on the truss sheet; it is a revised sheet, a revised elevation and usually a revised window head.

On a scissor or vaulted profile the heel question changes shape rather than going away. The bottom chord slopes, the heel joint angle is no longer the familiar one, and horizontal thrust at the bearing becomes a real number the wall must be checked for. A scissor sheet showing no horizontal reaction has either resolved that thrust internally or forgotten it, and it is worth knowing which.