Fixture units are a probability model wearing a unit's clothes
A fixture unit is not a flow rate, although it is used as though it were. It is an index that encodes three things about a fixture at once: how much water it uses when it runs, how long it runs for, and how often it is used. Two fittings with the same flow rate get different fixture units if one is used briefly and rarely and the other continuously.
The point of the index is that fixture units can be ADDED and the total converted to a peak demand through a curve — and that curve is emphatically not a straight line. It rises steeply at first, where a handful of fixtures really might all run together, and then flattens, because the probability of everything in a large building running at the same instant becomes vanishingly small.
So a hundred fixtures need nowhere near a hundred times the flow of one. The demand per fixture falls continuously as the building gets larger, which is why a tower's riser is not proportional to its apartment count and why a small building's service is proportionally the most generous one in any estate.
This is a statistical statement, and it has a confidence attached that the single number conceals. The curve was constructed to give a flow that will be exceeded only rarely — not never — so a peak demand figure is a design value with an accepted probability of being exceeded, rather than a ceiling.
- n_i, u_i
- count and fixture-unit value of each fixture type
- f
- the published demand curve — probabilistic, and fitted to observed use
- k
- a calibration constant for the fixture mix; flush valves and tank types differ
Low-flow fixtures broke the curve, and the consequence is not just cost
The demand curves in daily use were fitted decades ago, to buildings whose toilets used three or four times the water that current ones do and whose taps and showers flowed far harder. The probability model still holds; the FLOW attached to each fixture unit does not.
Applying an unadjusted curve to a building of modern fittings therefore oversizes the supply, sometimes substantially. The obvious cost is capital — larger meters, larger mains, larger risers — and it is the smaller half of the problem.
The larger half is water quality. An oversized main holds more water and moves it more slowly, so the RESIDENCE TIME between the treatment works and the tap rises. Disinfectant residual decays over that time, cold water warms towards room temperature on its way through a heated building, and both changes favour bacterial growth — including Legionella, whose risk is governed by temperature and stagnation rather than by anything a sizing calculation reports.
So an oversized water service is not a conservative error. It is a different error with a different failure mode, and it is why modern guidance pairs demand estimation with turnover and temperature criteria, and why the calculators here say which vintage of assumption their curve carries.
A trap is a water seal, and it has four ways to fail
The barrier between a building and its drain is a small quantity of water sitting in a bend. It costs nothing, needs no power and works perfectly — until it is not there.
It leaves four ways. EVAPORATION, in a drain that is rarely used, which is a matter of weeks in a warm dry building. SIPHONAGE, where flow past or through the trap pulls the seal out — self-siphonage from the fixture's own discharge, or induced siphonage from another fixture on a shared, under-vented branch. BACK-PRESSURE, where a surge downstream pushes the seal up and out into the room. And CAPILLARY ACTION along a thread of lint or hair bridging the seal.
Three of those four are design problems answered by venting and by trap geometry. The first is not — an unused floor drain will dry out however well the system is designed, and that is the entire reason trap primers exist.
A primer's duty is therefore set by the evaporation rate, not by the drain. It must deliver enough water often enough to replace what leaves, across every trap it serves, and the failure mode when it is undersized is silent: nothing drips, nothing backs up, and sewer gas enters the building through a drain that looks exactly as it always did.
A grease interceptor is sized on time, not on flow
Grease separates from water because it is less dense and floats. That is a slow process, and the design quantity is the RETENTION TIME the interceptor gives it — the volume divided by the flow through it.
So a device sized on peak flow alone can be entirely wrong. Push the design flow through a vessel too small and the water leaves before the grease has risen, and the interceptor passes it downstream while appearing to work: it is full, it smells, and it is doing nothing.
Temperature is the variable that makes it worse in practice. Hot water from a dishwasher keeps grease emulsified and liquid, so it travels through the interceptor without separating and then congeals in the cooler pipe beyond it — which is why a flow of hot water is harder on a grease system than the same flow cold, and why dishwasher discharge is treated separately in several codes.
Maintenance is part of the sizing rather than an operational afterthought. Captured grease occupies the vessel, so retention time falls as it fills, and an interceptor is sized so that it still works at the maximum accumulation its service interval allows. A correctly-sized unit on the wrong service schedule is an undersized unit for most of its cycle.
On-site treatment: the tank settles, the ground treats
A septic tank is not a treatment plant. Its job is to give solids time to settle and grease time to rise, holding both while the liquid in the middle passes on, and to digest what settles slowly enough that the tank needs emptying in years rather than months. The actual treatment happens afterwards, in the soil.
So the tank is sized on daily flow times a retention period, and the drainage field is sized on something else entirely: the soil's ability to accept water, measured by a PERCOLATION test. Those two numbers come from different places and neither substitutes for the other — a generous tank on unsuitable ground is a system that will surface.
The daily flow figure carries the widest scatter of any input on this page. Published per-person allowances differ between jurisdictions and between decades, and real household consumption varies several-fold around any of them with occupancy, age and habit. A system sized at an average occupancy fails at the upper end of the distribution, which is why codes size by bedroom count rather than by declared occupants.
Consumption figures deserve the same caution. A dual-flush toilet's rated volumes describe two buttons; the average volume per use depends on which one people press, and observed averages sit well above the low figure. A usage calculation is a comparison tool — this fitting against that one, this habit against that one — and not a prediction of a bill.
Calculators that use this method
Basis
- Hunter, R.B. (1940), Methods of Estimating Loads in Plumbing Systems, NBS Report BMS65 — the probability basis of fixture units and the demand curve.
- International Plumbing Code and Uniform Plumbing Code fixture unit tables and demand curves, for both supply and drainage, and the venting requirements that protect trap seals.
- ASPE Plumbing Engineering Design Handbook, including the modern critique of Hunter's curve applied to low-flow fixtures and the oversizing it produces.
- ASHRAE Standard 188 and CDC/WHO guidance on Legionella control in building water systems — residence time, temperature and stagnation as the governing risks.
- ASME A112.18.1 / CSA B125.1 for fixture flow rates, and EPA WaterSense specifications for current low-flow values.
- PDI G101 and IAPMO/ANSI Z1001 for grease interceptor sizing by flow and retention time, and the effect of discharge temperature on separation.
- EN 12566 and US EPA Onsite Wastewater Treatment Systems Manual for septic tank retention and drainage field sizing from percolation rate.