Sizing the Welfare Compound for the Workforce

Twenty-eight on the Monday, and the cabins land on the Thursday

Welfare is the one part of a site that has to be finished before the work starts, and it is ordered by somebody reading a programme that has not happened yet. The enabling gang turns up on the Monday, twenty-eight strong, and the welfare schedule was written against a peak of a hundred and eighty in month fourteen. Order for the twenty-eight and you spend the next year adding units in ones. Order for the hundred and eighty and you pay hire on eleven empty cabins through a winter.

Neither is the answer, and the reason people get it wrong is that welfare hire is bought as a count of cabins when almost every constraint on it is a rate: basins per minute at half past twelve, seats per sitting, litres per head per day into a tank somebody empties, kilowatts of latent heat out of a room full of wet coats overnight. Get the rates first and the cabin count writes itself — including the month at which the second block has to already be standing, which is a date and belongs on the programme rather than in a hire manager's inbox.

The headcount you size against is not the one on the resource curve

The resource curve gives average operatives per month, and averages are exactly the wrong shape for welfare. What sizes a toilet block is the number of bodies inside the fence at one instant, which is higher than the curve for three reasons that all land at once.

The first is trade overlap. A curve that says a hundred and twenty in month nine says nothing about the fortnight when the first-fix electricians, the ductwork gang and the ceiling grid are all in the same building, and the subcontractor's own labour returns arrive after the fact. The second is shift changeover: if a night gang hands over to a day gang, for forty minutes there are two workforces washing and changing in one block, and that overlap is the design case for basins and lockers even though it never appears as a headcount anywhere. The third is everyone who is not on the labour return at all — delivery drivers, the testing engineer, the client's monthly walk-round.

So build the number rather than read it, and keep the working out. When the hire company asks why you want six sanitary units rather than four, the answer that gets the order approved is the arithmetic, not the assertion.

1. Take the peak month off the resource curve, not the average across the phase.
2. Add the subcontractors who report their own labour separately and are missing from your return.
3. Add the changeover overlap in full if two shifts are on site at the same moment, even briefly.
4. Add a counted visitor and delivery-driver allowance rather than a percentage.
5. Re-issue the number to the hire desk whenever the programme is re-baselined, not only when somebody complains.

The law hands you a duty and withholds the number

There is a persistent belief on British sites that a regulation somewhere states how many toilets a hundred people need. It does not. The Construction (Design and Management) Regulations 2015 put the obligation on the principal contractor and set out in Schedule 2 what has to exist — sanitary conveniences, washing facilities, drinking water, changing rooms and lockers, and rest facilities with the means to boil water and heat food — and describes all of it as suitable and sufficient. That is a duty, not a quantity, written that way because a tunnel headworks and a two-storey housing plot are not the same problem.

The table everybody actually uses comes from somewhere else: HSE's L24, the Approved Code of Practice and guidance to the Workplace (Health, Safety and Welfare) Regulations 1992, which scales water closets and washstations against the number of people at work. Construction sites sit outside most of those workplace regulations and are governed by CDM instead, so L24's table is a defensible benchmark rather than a rule you are being measured against. Using it is sensible; citing it as the legal minimum is wrong, and an inspector who hears that will ask what assessment you actually did.

The American position is the opposite way round: the number is written down and the washing duty is narrower than people assume. OSHA 29 CFR 1926.51 covers sanitation for construction, and its Table D-1 sets toilet provision directly against headcount — a single facility for twenty or fewer, then one toilet seat and one urinal per forty workers, and per fifty workers once the site passes two hundred. A footnote allows urinals to stand in for toilets where the facility will not be used by women, but it puts a floor under the toilet count rather than permitting a straight swap. The washing facilities paragraph in the same section, though, is tied to operations where contaminants may be harmful rather than expressed as a ratio per head.

For the appliance scales themselves, and for the floor a cubicle and its door swing genuinely need, BS 6465-1 and BS 6465-2 are the documents to have open when you are arguing with a cabin layout drawing. Whichever regime you are under, the count you can defend is the one where somebody wrote down the headcount, the ratio's source, and why a bigger or smaller number suited this site.

Basins run out before cubicles do, and they run out at two known minutes

Toilet use spreads itself through a day. Hand washing does not: it happens in the four minutes before the break and the six minutes before the gate, and it happens to everybody at once. That asymmetry is why a block sized correctly on cubicles still produces a queue, and why the queue produces the thing you were trying to avoid, which is people going to lunch without washing.

Treat a basin as a throughput. A proper wash is twenty to thirty seconds at the tap, and with approach, drying and turnaround a basin serves roughly one person a minute in practice. A hundred and twenty people converging on a five-minute window need something close to twenty-four basin-minutes per minute of window — which is to say the number of basins that clears that crowd is two dozen, not the four you get in one sanitary unit. You will not hire two dozen, and you do not have to: the fix is usually to break the peak rather than to buy throughput, by staggering break times between blocks or by putting a satellite wash point at the far work face so that half the crowd never walks to the main compound. What you cannot do is decide the queue is a discipline problem.

Ninety kilograms of water, and nowhere for it to go

A drying room is the most under-specified thing in a welfare compound, because it looks like an empty cabin with a heater in it and is in fact a piece of process plant. A soaked jacket, trousers and a pair of boots carry on the order of a kilogram and a half of water. Sixty people coming in off a wet November afternoon bring in something near ninety kilograms of it, and by seven the next morning all of it has to be somewhere other than in the clothing.

Evaporating water costs about 0.63 kWh per kilogram — the latent heat of vaporisation, in units a hire schedule can use. Ninety kilograms is around fifty-seven kilowatt-hours, and across a fourteen-hour night that is a continuous four kilowatts doing nothing but turning water into vapour. Then the vapour has to leave, because a hot sealed cabin reaches equilibrium at high humidity and the clothes stop drying while the walls start streaming. Carrying ninety kilograms away on ventilation air means shifting on the order of a thousand cubic metres an hour through a cabin of thirty — dozens of air changes an hour, not the two or three a trickle vent gives you — and heating that incoming air from a winter night to room temperature is another five kilowatts. The psychrometrics behind those figures are in the ASHRAE Handbook — Fundamentals. The point of doing them is that a drying room for sixty people is a ten-kilowatt appliance running all night, and it is routinely hired as a cabin with a three-kilowatt fan heater in the corner.

That number is also the argument for the other approach. A heat-and-vent room throws its latent heat out of the vent; a recirculating room with a refrigerant dehumidifier condenses the water instead and gives the heat back to the air, which is why the better hire units are built that way. It changes what you specify: ninety litres a day is around a hundred and ninety US pints a day of dehumidification, which is several domestic-sized machines or one commercial unit, and a machine rated at a warm-basement duty will not hold that rate in a cold cabin. Hanging space is its own constraint on top of that — clothing packed shoulder to shoulder on one rail dries at the rate of the outside of the bundle, not at the rate the arithmetic promised.

Schedule 2's changing-room paragraph carries the drying requirement where clothing becomes wet, and OSHA's change-room provisions are triggered by protective clothing worn against contaminants rather than by rain. Neither gives a kilowatt. The mass balance does, and it is the only thing that tells you in September whether the unit you booked can cope in January.

Seats are a sittings problem, and the microwave is the real queue

Canteen seating is the one figure people size by eye, and eye-sizing gives a room that is fine at half past eleven and unusable at half past twelve. The controlling arithmetic is sittings: seats times the number of times the room turns over inside the break is the number of people who get to sit down. A hundred and twenty operatives on a thirty-minute break with a room seating forty is three sittings, which needs a ninety-minute window. Without it, forty people eat and eighty stand in the drying room — which is where they will go, and which undoes the drying you paid ten kilowatts for.

The queue is rarely at the seats, though. It is at the microwave, and that one is arithmetic anybody can do on the back of a delivery ticket. Two and a half minutes of heating plus the shuffle in and out is about three minutes per person per machine, so one microwave serves ten people in a thirty-minute break. If seventy per cent of a hundred and twenty-strong workforce heats food, that is eighty-four heatings and something like eight microwaves. Nobody hires eight microwaves. They hire two, and the last man in the queue eats at ten past one, which is either lost production or a cold lunch, and both of them turn up later as a canteen that people have stopped using.

The same counting applies to the boiling water Schedule 2 asks for: a twenty-litre wall boiler holds perhaps sixty mugs and then has to reheat, so its recovery rate rather than its capacity decides whether the second half of the break gets tea. Every one of these appliances is also a nameplate on a receptacle circuit, and they all go on at the same moment, which is where the next section starts.

Week six, when the second gang lands and nothing has been re-counted

Welfare provision degrades without anybody deciding it should. The headcount steps up when a subcontractor mobilises early, a cubicle goes out of service and is not reported because there is another one, the drying room fills with stored gear until air cannot move through it, and the tanker interval that suited a hundred people is quietly serving a hundred and fifty. None of it shows up in a hire schedule, because the hire schedule has not changed — which is the point.

So book the re-count against the programme rather than against a complaint. Compare this week's labour with the number the block was hired against, and stand in the canteen at half past twelve rather than at four. Check what the tanker pumped last visit against what the tank holds, because a load that comes back consistently full means the interval has moved. Test the drying room by touching the clothes at seven in the morning, which is the only test of it that means anything. Where a number has moved, the fix is another unit and a date on the programme — considerably cheaper than the version where an inspector finds it first.