Checking Whether a Floor Will Take a Heavy Load

Weigh It as It Will Stand, Not as It Arrives

Delivery weight is the wrong number and it is the number everybody quotes, because it is the one printed on the crate. What the structure below actually sees is the assembled, filled, commissioned weight, and for anything holding water the difference is an order of magnitude. Fresh water at ordinary room temperature runs close enough to 1.0 kg per litre for a load check; sea water at reef salinity is about 2.5 per cent denser, which on a large marine system is another 30 or 40 kg that nobody added because the tank was quoted in litres and the litres were assumed to be fresh.

Build the figure line by line rather than in one lump, because the lines that get forgotten are always the same ones. The sump under the cabinet is full at its working level, not its rated level, and on a power cut it is full at whatever drains back into it — that is the case the floor sees, so use it. Substrate is bought dry by the bag and lives wet. Dry rock gains weight for a fortnight after it goes in. The cabinet on a large display is a braced structure in its own right and weighs what a chest of drawers weighs before anything is put in it.

The same discipline applies to anything else that gets described by its empty weight: a gun safe with the contents in, a plant room buffer vessel at operating level, a file compactor with shelving loaded, a hot tub with four people in it. The rule is that you weigh the thing in the state it will spend its life, and you add the maintenance state on top if that is heavier — a tank being refilled after a rock rebuild, a vessel with the isolating valves shut and nowhere for the water to go.

A Weight Is Not a Loading

Having a weight in kilograms does not yet let you compare anything, because floors are not specified in kilograms. They are specified as a pressure — a uniformly distributed load over the whole area, in pounds per square foot or kilonewtons per square metre — and the comparison only means something once both sides are expressed the same way. The IRC's minimum uniformly distributed live loads in Table R301.5 put habitable rooms other than sleeping rooms at 40 psf and sleeping rooms at 30 psf. The equivalent metric route runs through BS EN 1991-1-1 (Eurocode 1, Part 1-1), whose Category A covers domestic and residential floors with the National Annex fixing the value the design actually used, and AS/NZS 1170.1 does the same job in Australia and New Zealand. Commercial work goes through IBC Table 1607.1 or the minimum live load table in ASCE/SEI 7 — Table 4.3-1 in the recent editions, Table 4-1 in older ones.

Then comes the trap. A 40 psf design floor is not a floor onto which you may place 40 psf anywhere you like. That figure is a design abstraction applied over the whole panel to size the members, and a concentrated load parked over the middle third of one span is a completely different structural case with a completely different answer. This is why ASCE/SEI 7 carries a separate concentrated live load requirement alongside the distributed one, applied over a small square — a 2.5 ft (760 mm) square — and requires the floor to be checked for whichever governs. AS/NZS 1170.1 likewise pairs a distributed value with a point value for the same occupancy. If the adopted table for your occupancy pairs the two, both apply.

Work out the contact footprint honestly while you are at it. A tank cabinet does not bear over its whole plan area — many bear on a perimeter rail or on four feet, and a cabinet with a 1,800 by 600 mm plan that touches the floor along two 1,800 mm rails 40 mm wide is delivering its whole weight through 0.144 square metres. That is not a floor loading question any more, it is a bearing question about floorboards, and it is why a heavy safe on adjustable feet dents a floor that carries a piano with no complaint.

One more subtraction before you decide there is margin. Part of a commercial floor's live load allowance is already spoken for: ASCE/SEI 7 requires a partition allowance — commonly 15 psf — wherever partitions can be relocated, and that allowance is not spare capacity waiting for your equipment. Nor is the difference between the design live load and whatever happens to be in the room today. Empty floor area is not credit.

Permanent Load Behaves Differently From Live Load

An aquarium is not a live load. Nothing about it is temporary, transient or occasional — it goes on the floor once and stays there, and in structural terms it is a dead load that arrived after the design was finished. That distinction is not pedantry, because timber's allowable stress depends on how long the load stays. The ANSI/AWC NDS assigns a load duration factor by cumulative duration in its Table 2.3.2, and permanent load attracts the lowest one: below the factor a normal ten-year floor live load earns, and far below what a wind or seismic case is allowed. The floor is being asked for its capacity in the least generous column of the table.

Deflection goes the same way, only worse. Timber under sustained load creeps, and the NDS handles this with a long-term loading provision that multiplies the deflection from the permanent part of the load before adding the short-term part — the multiplier being larger for unseasoned or wet-service timber than for dry seasoned material. BS EN 1995-1-1 (Eurocode 5) reaches the same place by a different route, with kmod adjusting strength for load duration and service class and kdef doing the creep. The practical consequence is a floor that passes a first-day check and then, over three or four years, develops a dish under the tank that keeps going and never comes back.

That creep matters for reasons beyond appearance when the thing on the floor is a glass box. A rigid tank on a floor that has sagged 8 mm across its footprint is no longer evenly supported, and glass tanks fail at the seams when they are supported unevenly. So the deflection limit that applies here is not the one that keeps a ceiling from cracking; it is whatever the tank's own maker states about the flatness of what it stands on, and that is usually the stricter of the two.

One Joist Takes More Than Its Share

Nothing about a heavy object divides itself neatly among the joists beneath it. Orientation decides most of it. A long tank set across the joists spreads its weight over five or six members and each takes a slice; the same tank turned through 90 degrees and set along a single joist line loads one member and its two neighbours, with the boards doing whatever sharing gets done. Turning the furniture is by a wide margin the cheapest structural intervention available on a domestic floor, and it costs nothing at the planning stage and everything afterwards.

Position along the span matters as much as orientation. A load landing near a support goes almost straight into the wall below and produces little bending; the same load at mid-span produces the maximum moment the member will ever see. So a tank against the party wall and a tank in the middle of the room are two different problems even though the weight on the delivery note is identical. Work out the moment on the single worst joist before doing anything else — the shared portion as a distributed load, plus the point contribution from any cabinet foot that lands within the span rather than over a support.

Check the ends as well as the middle. Bending gets all the attention, but a joist end bearing 40 mm onto a crumbling brick corbel or into a hanger that was never fully nailed will fail at the bearing long before the section runs out of bending capacity. Compression perpendicular to grain governs at the bearing, and the area available there is small, fixed, and frequently already damaged by an earlier trade.

A Floor Above the Floor

A raised access floor changes the question completely, because the thing under the wheel is not structure. It is a proprietary panel on adjustable pedestals over the real slab, and it has its own published capacity that has nothing to do with what the building frame can take. Both have to work. A slab in a converted 1970s office may be the weaker of the two once a row of battery cabinets goes on it, and the panel says nothing about that.

Panels are rated by test, and the rating that governs anything on wheels is the rolling load, not the concentrated load. The CISCA Recommended Test Procedures for Access Floors define the rolling test by passing a defined wheel over a panel a set number of times and publish two results — a low pass count and a high one — because the failure mode under repeated traffic is accumulated damage rather than a single overload. The high-pass figure is markedly lower and it is the one that applies to any route a trolley uses regularly. BS EN 12825 classifies panels for the European market and keeps the rolling case separate from the concentrated and distributed ones for the same reason.

On an existing floor with no paperwork, the rating is the whole problem. Lift a field panel and photograph the underside: most systems carry a maker's name and a type stamp there, and that stamp is generally the only route back to a published test certificate for a floor installed before anyone currently in the building started. Failing that, you have an unknown, and an unknown gets treated as one — a temporary deck, or a survey by the system's supplier, not a guess based on how solid it feels underfoot.

Two derating conditions catch people out even when the certificate is in hand. Cut panels at the perimeter and around risers are not full panels and carry no published rating at all, which is exactly where the pallet truck dropped through in the story at the top of this page. And a panel with a cutout in it has lost material from the part that has to carry the wheel. Add to both the fact that the certificate describes a new panel on an undamaged understructure, and a floor that has been lifted and relaid through three fit-outs is not that floor.