Three things take your pressure, and only one of them is the pipe
Work out where a run's pressure goes and it separates cleanly into three parts. Friction against the pipe wall, which depends on how fast the water is moving, how rough the bore is and how far it has to go. Elevation, which costs 0.433 psi for every foot the water has to climb and gives it back if the run falls. And devices — filters, softeners, backflow preventers, meters — which have a published loss at a given flow and are frequently the largest single item in the list.
The reason people misjudge runs is that they think about the first one and forget the other two. A second-floor bathroom is roughly 4.3 psi down before a drop of water has moved, and a whole-house filter with a loaded cartridge can take more than the entire pipe run does.
The Hazen-Williams equation and where it applies
Head loss (ft per 100 ft) = 0.2083 × (100 ÷ C)1.852 × GPM1.852 ÷ ID4.8655
Multiply the result by 0.4335 to get psi. The exponents look arbitrary because they are: Hazen-Williams is an empirical fit rather than a derivation from first principles, developed for water at ordinary temperatures in ordinary pipe. It is the standard method for water distribution and it is what most plumbing tables are built on. It is not appropriate for other fluids, for very hot water, or for flow so slow it is no longer turbulent, and Darcy-Weisbach is the more general method when you leave those conditions.
The term that dominates is the diameter, raised to nearly the fifth power. Going up one nominal size roughly halves the velocity and cuts the friction loss to somewhere around a sixth of what it was — going from 1/2 inch type L copper to 3/4 inch at the same flow drops the loss to about 17 percent of what it was. That is why almost every real pressure problem is solved by a larger pipe rather than by anything cleverer.
C is the roughness coefficient — higher is smoother. The defaults here are 150 for PVC, CPVC and PEX, 140 for copper, and 100 for steel. Treat those as the middle of a range rather than as constants. Copper is quoted anywhere from 130 to 150 depending on age and water chemistry; steel is quoted from 100 down to 60 or worse once it has scaled. The override field is there because the honest answer for an existing system is often lower than any published new-pipe figure.
Fittings, expressed as pipe
Every fitting is a small obstacle, and the tidy way to account for one is to ask how many feet of straight pipe would cost the same pressure. That is its equivalent length, and it scales with the bore — a 90-degree elbow is conventionally about 30 diameters, so it is roughly 1.4 ft of equivalent pipe in half-inch and 5 ft in 2-inch.
| Fitting | Equivalent length, in bore diameters | What that is in 3/4" copper |
|---|---|---|
| 90° elbow, standard | 30 | about 2.0 ft |
| 45° elbow | 16 | about 1.0 ft |
| Tee, flow straight through the run | 20 | about 1.3 ft |
| Tee, flow turning into the branch | 60 | about 3.9 ft |
| Ball valve, full port, fully open | 3 | about 0.2 ft |
| Gate valve, fully open | 8 | about 0.5 ft |
| Swing check valve | 100 | about 6.5 ft |
| Globe valve | 340 | about 22 ft |
These are the conventional handbook ratios and they are approximations. Real fittings vary with geometry — a long-sweep elbow is considerably better than a hard 90, a reduced-port ball valve is nothing like a full-port one, and a partly closed valve of any kind is off the table entirely. Take the last row seriously though. A globe valve is worth twenty feet of pipe in a size where the whole run might be forty, which is why they are found on hose bibs and not buried in supply mains.
Velocity, and the two different reasons to care
Velocity in feet per second is 0.4085 times the flow in GPM divided by the square of the bore in inches. It matters for two separate reasons that get conflated.
The first is noise. Fast water in a pipe is audible, and fast water that gets stopped abruptly by a quick-closing valve produces water hammer, which is a pressure spike rather than a sound problem and can damage things. The second is erosion-corrosion, which is specific to copper: above a certain velocity the moving water strips the protective oxide film off the inside of the tube, particularly at the outside of elbows and just downstream of them, and the wall thins until it perforates. It is slow, it is invisible from outside, and hot water accelerates it.
The design figures generally quoted sit around 5 to 8 ft/s as a ceiling for cold water in copper and something lower — commonly in the 2 to 5 ft/s region — for hot, with recirculating loops treated more conservatively still because they run continuously rather than in bursts. These are industry design guidance with genuine variation between sources and between water chemistries, not a single threshold. Plastics do not erode the way copper does and are usually allowed higher, but the noise argument still applies. Below about 2 ft/s there is no pressure penalty worth mentioning, but the line is slow to flush and slow to bring hot water to a distant fixture.
Reading the answer honestly
What this page gives you is the pressure cost of a route at one flow rate. Real systems do not run at one flow rate — the loss climbs steeply the moment a second fixture opens, because the exponent on flow is nearly two. If you are checking whether a run will be adequate, run it twice: once at the flow of the fixture you care about, and once at the flow you would see with the demand you actually expect at the same time. The second number is the one that decides whether the shower goes cold when someone starts the dishwasher.
Sizing a supply system properly means fixture unit loading, simultaneous demand, the pressure available at the meter and the minimum required at the highest fixture, and it is governed by the code your jurisdiction has adopted. Plumbing is code-regulated and the requirements are not the same everywhere. Drain slope, venting, fixture unit loading, minimum pipe sizes and which materials are approved for which use are set by the plumbing code your local authority has adopted, along with whatever local amendments came with it. Most permanent work needs a permit and a licensed plumber. What is described here is common practice, not a rule you can hold up to an inspector.
Questions people ask
How much pressure does a run of pipe actually lose?
Enough to matter only when the pipe is undersized for the flow, and then a great deal. At 5 GPM through 3/4 inch type L copper the friction loss is about 3.1 psi per 100 feet, which is barely noticeable over the length of a house. Push the same 5 GPM through 1/2 inch and it is about 18 psi per 100 feet, nearly six times as much for one step down in size, because the diameter term in the friction equation is raised to nearly the fifth power. That single fact explains most of what people observe about their plumbing: undersized pipe is fine at low flow and falls apart the moment two things run at once. Elevation is more predictable — 0.433 psi per foot of rise, always, regardless of pipe size or flow.
What is a good water velocity in a pipe?
For supply piping, most design guidance lands somewhere in the 2 to 5 feet per second range as the comfortable zone, with ceilings commonly quoted around 5 to 8 ft/s for cold copper and lower for hot. There is real variation between sources, and the reasons differ: noise and water hammer apply to any material, while erosion-corrosion of the tube wall is a copper problem that gets worse with temperature and with continuous flow. A brief peak on a cold branch is a different risk from a recirculating hot loop running all day at the same speed. Treat these as design guidance ranges. If you are near the top of them and the line is hot copper, size up.
Do fittings really matter or is that just table padding?
On a short run with many fittings they can be most of the loss. A 20 foot run of half-inch type L copper with eight elbows and two tees carries about 13 feet of equivalent length in fittings against 20 feet of pipe, so the fittings are close to 40 percent of the total. On a 200 foot run with six elbows they are noise. The one to watch is the globe valve at 340 diameters, which is worth about 15 feet of half-inch pipe or 22 feet of 3/4 inch on its own — add a stop valve to that 20 foot run and it more than doubles again — and any valve that is partly closed, which is worth an amount nobody can calculate. Long-sweep elbows instead of hard 90s are the cheap fix on a run where the fitting count is high.
Why does my PEX system flow worse than the calculator says?
Almost certainly the fittings. PEX joined with crimp or clamp rings uses an insert fitting that sits inside the tube, and the opening through that insert is noticeably smaller than the tube bore — the restriction is at every joint, not just at the ends. The equivalent-length figures used here are for fittings that do not narrow the passage, so a heavily fitted crimped system loses more than this predicts. Expansion-type fittings, where the tube is stretched over the fitting and shrinks back onto it, keep a much larger opening. If you are working out a PEX run in advance, count the fittings and treat the answer as optimistic, or use the manufacturer flow data for the specific fitting system.
Can I just use the pressure at my hose bib as the starting pressure?
Only if you measure it with the water running, and even then it is the pressure at that point rather than at the meter. A gauge on a closed bib reads static pressure, which is the number utility pressure sets and which tells you nothing about what happens under flow. The useful measurement is a gauge on the bib with a known flow leaving somewhere else in the house, because the difference between the two readings is the loss in everything upstream of the gauge. If static reads 65 and it falls to 40 the moment a shower opens, the problem is already between the main and that gauge and no amount of new pipe downstream will fix it.