Reference

Span and cost charts, and how we got them

The calculators on this site publish a steel weight per square foot, which almost nobody in the industry does. That number is only worth something if you can see where it came from, so here is the whole derivation — including the parts the model cannot do.

The baseline table

Everything starts here: a clear span rigid frame at a 20 ft eave, 25 ft bays, 20 psf ground snow, 115 mph wind, and no crane. Steel is primary plus secondary framing.

Clear span rigid frame, baseline case
Span	Steel lb/sf	Package $/sf	Installed $/sf	What changes at this span
40 ft	2.20–3.20	$14–17	$22–32	Light frames. Span is rarely the cost driver at this width - eave height and snow load matter more.
50 ft	2.60–3.60	$14–18	$22–34	Still comfortably inside standard shipping and erection practice.
60 ft	3.00–4.20	$15–18	$23–34	The practical floor for commercial work. Below this width, small-building suppliers price better.
80 ft	3.80–5.20	$16–20	$25–38	The crossover band. Clear span and modular start to diverge in cost here - price both.
100 ft	4.60–6.40	$17–21	$26–40	Haunch depth begins to govern interior clearance. Check usable height, not just the eave.
120 ft	5.60–7.80	$18–23	$28–44	Deep haunches and heavier anchor-bolt patterns. Foundation cost stops being a rounding error.
150 ft	7.00–9.80	$20–26	$31–49	The practical ceiling for a standard single-span rigid frame. Past this you are buying custom engineering.
200 ft	9.50–13.50	$24–32	$37–61	Custom engineering, shipping splices, and heavier erection equipment. Modular usually wins on cost alone.
250 ft	12.50–17.50	$28–38	$43–72	Specialist territory. Justified by use, not economics - hangars, arenas, column-free process floors.
300 ft	16.00–23.00	$33–46	$51–87	The upper limit most manufacturers offer. Expect a bespoke frame design and a long lead time.

Values between the rows above are linearly interpolated. Outside 40–300 ft the calculators decline to answer rather than extrapolate.

What it is calibrated against

The table is anchored to figures that are published and checkable, not to proprietary quote data:

Building package cost of $14–22 per square foot, and $22–42 per square foot installed before interior buildout, as commonly quoted across the industry.
Standard single-span rigid frames to about 150 ft, with custom engineering extending to roughly 300 ft.
Clear span and modular cost diverging noticeably from about 80 ft of width, with the gap widening as width grows.
Eave heights from 10 ft on basic storage to about 30 ft on most commercial work, and higher on crane and hangar buildings.

Where our numbers exit the published band — and they do, at long spans and heavy loads — the calculators say so on the page rather than clamping the result to look tidy. A 250 ft clear span genuinely does not cost $20 per square foot, and pretending otherwise would make the tool useless exactly where it is most needed.

Sensitivity coefficients

Weight moves away from the baseline by these factors. Each is floored so that no combination of light loads produces a weightless frame, and they are summed rather than multiplied — multiplying five of them compounds to over 7× baseline at the extremes and produces weights no building has ever had.

Fractional change in framing weight per unit of input
Input	Per unit	Mechanism
Eave height	+1.8% / ft	Longer column, larger moment at the haunch, heavier frame throughout.
Bay spacing	+1.0% / ft	Fewer frames, but deeper purlins and girts carrying more roof each. Net weight rises.
Ground snow	+0.85% / psf	Gravity load direct into the rafters. Usually the governing case on a wide roof. Calibrated on the ASCE 7-16 basis — see which snow number the model wants.
Wind speed	+0.45% / mph	Uplift and lateral demand, felt in bracing and connections more than in the rafter.
Bridge crane	4.5% × √tons	Runway beams, brackets, and repeated lateral load over the life of the building.
Cost pass-through	55%	Share of a weight change that reaches package price. Panels, trim, fasteners, and freight do not move with frame weight.

Which snow number the model wants

Ground snow load is the one input on this site where you can hand the model a correct number and still get a wrong answer, because ASCE 7-16 and ASCE 7-22 report p_g on incompatible bases and the same city has very different values under each.

ASCE 7-16 maps a 50-year ground snow load, used with a snow factor of 1.6 in the strength combinations.
ASCE 7-22 maps reliability-targeted, strength-level values and drops that factor to 1.0. The mapped number is much larger; the factor applied to it is smaller.

Washington DC is 25 psf under 7-16 and 62 psf under 7-22. Feed that 62 into a model calibrated on the 7-16 basis and you overstate frame weight by about 16% with no change in real demand behind it — and by 19% for a metro like Anchorage. So the snow field on every calculator asks which edition your number came from, and the model converts to a single internal basis before using it.

The conversion is p_g ÷ 1.6, because equal factored demand means 1.6 × p_g,7-16 = 1.0 × p_g,7-22. The converted figure is displayed next to the field rather than applied silently.

This is a nominal basis conversion, not an exact equivalence. The two editions genuinely disagree about demand in places, and converting makes that visible rather than hiding it. On this basis 7-22 is materially more demanding in marginal-snow climates (Washington DC: 39 psf equivalent against a 7-16 value of 25) and less demanding where snow is heavy and dependable (Minneapolis: 36 equivalent against 50). That spread is the reliability targeting doing its job — variance is highest where snow is marginal.

Do not carry a 7-22 value into a 7-16 workflow that still applies a 1.6 factor. That double-counts and over-designs the roof by roughly 1.6×. Plenty of jurisdictions still enforce ASCE 7-16 through IBC 2018 or 2021, so check which edition governs at your site before you use any snow number — from us or anyone else. Where our load data comes from.

Why weight rises faster than span

Bending moment at midspan grows with the square of the span, so a longer frame needs a deeper section. The deeper section is itself heavier, which adds to the load it has to carry. The taller haunch needs more material at the knee. Each effect compounds the last, so doubling the span roughly triples the steel rather than doubling it.

That compounding is the reason the industry has frame types at all. A modular frame is not made of better steel; it simply refuses to let any single member span the full width, stepping back down the curve.

What this model cannot do

It is not engineering, and it does not become engineering by being specific. A number carried to two decimal places is still an estimate.

No member selection. The model has no knowledge of AISC section properties, plate sizes, or weld design. It estimates total weight; it cannot tell you what any member is.
No load combinations. Real design applies combinations from ASCE 7 with factors, and different combinations govern different members. This applies smooth sensitivity factors instead.
No deflection or serviceability check. On long spans and crane buildings, deflection limits frequently govern over strength, which can add steel this model does not see.
No drift or unbalanced snow. Ground snow goes in; roof snow, drift against walls and roof steps, and unbalanced cases are not modelled. On a multi-gable roof this is a significant omission.
No seismic detailing. Seismic Design Category changes connection and bracing design substantially and is not an input to the weight model at all.
No regional pricing or market conditions. Steel price, freight, local labour, and how busy the manufacturer is can move a real quote further than any input on these pages.
No foundation design. The installed multiplier assumes an ordinary spread-footing foundation on competent soil. Poor soils, piles, or deep frost can break it.

Use these tools to scope a building, to decide which choices are worth arguing about, and to spot a quote that is out of family. Then get a stamped design. Three dealers will price it for free.

Corrections

If you quote or build these buildings for a living and one of these figures is wrong, we would rather know. The model lives in two files — the table in includes/data.php and the coefficients in includes/calc.php — and both are single-source, so a correction propagates everywhere at once. Say which figure, at what span and load case, and what you see in the field.