Water Supply Pipe Sizing: WSFU & Hunter's Curve Explained

Here is the problem every water supply design has to solve: a building may contain hundreds of taps, showers and flush valves, but they will never all run at once. Size the pipes for the impossible everything-open case and you waste money and create new problems; size them for too little and the top floor loses pressure every morning.

The method that solves this — fixture units and Hunter's curve — is nearly a century old and still underpins the sizing tables in every major plumbing code. This guide explains how it works, how fixture units become a design flow, and how the pressure budget turns that flow into actual pipe sizes.

Why You Can't Just Add Up the Taps

Each fixture has a known flow rate, but fixtures are used intermittently and briefly. The chance of every fixture drawing water simultaneously is effectively zero, and the more fixtures a pipe serves, the smaller the simultaneous fraction becomes. Diversity, in other words, grows with scale — a riser serving 200 apartments does not need 200 times the flow of one apartment.

In the 1930s and 40s, Roy Hunter at the US National Bureau of Standards turned this into a usable method: treat fixture use as a probability problem, assign each fixture a weight, and compute the demand that will only be exceeded rarely. That weight is the water supply fixture unit (WSFU), and the demand curve his mathematics produced is Hunter's curve.

Water Supply Fixture Units (WSFU)

A WSFU is not a flow rate — it is a load weight combining how much water a fixture draws, for how long, and how often. Typical values (codes differ slightly; always use your governing code's table):

• Lavatory (private): about 0.7–1 WSFU
• Shower or bathtub (private): about 1.4–2 WSFU
• Kitchen sink (private): about 1.4–1.5 WSFU
• Water closet, gravity tank (private): about 2.2–2.5 WSFU
• Water closet, flushometer (public): 5–10 WSFU — flush valves dominate any system they are in

Fixtures also carry separate cold, hot and total ratings, because the cold main, hot main and the pipe feeding the water heater each see different loads.

Hunter's Curve: From WSFU to GPM

Total up the WSFUs served by a pipe, then convert to a probable peak demand using Hunter's curve — in practice, the demand tables printed in the code appendices. The crucial property of the curve is that it is non-linear: demand grows much more slowly than fixture count.

Approximate anchor points from code charts, for tank-type systems: 20 WSFU corresponds to roughly 14 gpm (about 0.9 L/s); 100 WSFU to roughly 44 gpm (2.8 L/s); 500 WSFU to roughly 100+ gpm (6.3+ L/s). Note that 25 times the fixture units produced only about 7 times the flow — that is diversity at work. Systems dominated by flushometer valves follow a separate, higher curve at low fixture counts.

"The probable demand, not the possible demand, is the rational basis for pipe sizing." — after Roy B. Hunter, NBS Report BMS 65

Velocity Limits

Flow alone doesn't fix the size — velocity caps it. Design practice keeps domestic water around 5–8 ft/s (1.5–2.4 m/s) maximum, and many engineers hold hot water to about 5 ft/s (1.5 m/s) or less. Too fast brings noise, water hammer and erosion–corrosion, which attacks copper hot-water systems in particular; the material's limits matter as much as the code's.

The Pressure Budget

The second constraint is pressure. Sizing is ultimately an accounting exercise: start with what you have, subtract what you must keep, and spend the remainder on friction.

1. Start: available pressure from the mains, booster pumps or roof tank
2. Subtract static lift: 0.433 psi per ft of height (9.8 kPa per metre) — in a tall building this is the dominant loss
3. Subtract equipment losses: meter, backflow preventer, filters, water heater
4. Subtract the residual the worst-placed fixture needs to work: commonly around 8–20 psi (55–140 kPa) for ordinary fixtures, more for flushometers
5. What remains is the friction allowance. Divide it by the developed length of the critical run (including fitting equivalent lengths) to get an allowable pressure drop per 30 m / 100 ft — then pick, from the friction charts, the smallest pipe sizes that stay inside it and under the velocity cap

A Worked Sketch: Small Three-Storey Office

Take a modest office building: three floors, each with a toilet block of four lavatories, four tank-type water closets and a cleaner's sink, plus a small pantry per floor. Counting from a typical code table, each floor lands at roughly 20 WSFU, so the cold-water riser serves about 60 WSFU — call it, from the demand tables, in the region of 30–35 gpm (around 2 L/s).

Now the budget. Mains pressure at the connection: say 60 psi (about 4.1 bar). Static lift to the highest fixture, 10 m: 14 psi gone. Meter and backflow preventer: another 8 psi. Residual at the worst fixture: keep 15 psi. That leaves roughly 23 psi to spend on friction along the critical run. If the developed length — pipe plus fitting allowances — is 60 m (about 200 ft), the allowable gradient is around 11 psi per 100 ft, and the friction chart hands you the sizes: a riser in the 1½–2 in (40–50 mm) range, floor branches stepping down as the WSFU count falls, nothing below the code minimum at the fixture. Fifteen minutes of arithmetic, and every number in it defensible.

Material matters more than it looks

Friction charts are material-specific — PPR, copper, galvanised steel and PEX each have their own roughness, and the same nominal diameter carries different real bores. Hot water adds the recirculation loop: a separately sized circuit whose only job is to keep hot water moving so the far taps never run cold, and which is routinely forgotten until the complaints arrive.

The Cost of Getting It Wrong

• Oversized pipework wastes capital twice — larger pipe and larger insulation — and creates a slow-moving system where water ages in the pipe, which is exactly the stagnation condition Legionella guidance warns against
• Undersized pipework announces itself daily: pressure drops when two fixtures run together, showers that scald when a toilet flushes, whistling valves and water hammer
• Unbalanced hot water — correct sizes but no thought given to the recirculation loop — delivers cold "hot" water to the far fixtures every morning

Conclusion

Modern codes are slowly evolving past Hunter — fixtures use far less water than they did in 1940, and newer estimation methods (such as the IAPMO Water Demand Calculator for residential work) produce leaner flows. But the structure of the method is unchanged, and the engineer's judgement sits in the same places: honest fixture counts, the right curve for the fixture mix, and a pressure budget built from the building as it will actually be operated.

Hunter's insight still holds: design for the demand that will actually occur, not the one that theoretically could. Count the fixture units, read the probable flow, respect velocity and the pressure budget, and the system delivers quiet, steady water at every floor for decades. Supply sizing pairs naturally with drainage design — and if you want both reviewed or designed from scratch, that is precisely the work we do remotely, as part of full MEP engineering scope or as a standalone plumbing design service.