What Friction Loss Does Over a Long Run

Why this matters

A long run behaves perfectly until it does not, and the transition is abrupt. Friction loss is a rate per unit of length multiplied by the length, and the rate climbs steeply with flow, so a branch that gives up a few psi serving one outlet gives up most of the supply pressure serving two. Nothing broke in between. Techs replace fixtures, regulators and valves chasing that transition, because everything they test in isolation passes, and it passes because the loss only exists while the flow does.

Before anyone opens the system

Get pressure readings from existing openings rather than making new ones. A hose bibb, a boiler drain, an existing gauge port: each gives you a tap without breaking into a pressurized line. Where a fitting genuinely has to come off, isolate, open a low outlet to relieve, and confirm zero on an installed gauge before a wrench moves. A closed system that has been heated holds pressure that shutting the supply valve does not remove.

Hot fluid scalds. Let the section cool before opening it, and crack fittings with the opening pointed away from you and from anyone else in the room.

If the fix means joining plastic pipe, solvent cement vapor is an inhalation hazard and the control is ventilation and the respiratory protection called for on the product's safety data sheet, not gloves. In a crawlspace, a closet or any poorly ventilated space, ventilate mechanically before you start rather than working faster to get out. Never apply a torch or heat gun to PVC or CPVC to soften or bend it: heated PVC releases hydrogen chloride and other irritants, and the joining method for that material is solvent cement or a listed mechanical fitting.

If the fix means soldering copper, flux and alloy fume are also an inhalation route, so ventilate and keep your head out of the plume, and hot work near framing needs the combustibles cleared or shielded plus a fire watch that stays after you stop.

Old pipe insulation on the existing run is presumed asbestos-containing until sampled where the building predates the 1980s. Do not cut, tear or abrade it to reach the pipe; disturbing it is regulated work under 29 CFR 1910.1001 in general industry or 29 CFR 1926.1101 in construction.

Opening a wall to reroute means unknown wiring behind it. De-energize the circuits serving that area and verify dead before cutting; that is energized electrical work territory under 29 CFR 1910.333(b)(2), and the proving sequence is live-dead-live per NFPA 70E-2021, 120.5.

The signal

A two-year-old addition off the back of a house. It holds a full bathroom and the laundry. The complaint is that the far shower is fine alone and unusable the moment the laundry fills.

History on the account: a shower valve cartridge replaced, then the whole valve, then the pressure regulator at the service. Three parts, three visits, complaint unchanged.

Candidates eliminated on evidence

The fixture. Already replaced twice, and taken out of the argument entirely by testing at a hose bibb on the same branch instead. The bibb showed the same collapse. A fixture that is not in the flow path cannot cause it.

The source or the regulator. Static at the front of the house, everything closed, settled at 62 psi and recovered to 62 psi within a few seconds of every flow test. A source that returns fully and quickly to the same static has head available; the losses are downstream. The method for reading static, residual and recovery is worked in the sibling article on pressure and flow, and it is what cleared the regulator that had already been replaced for nothing.

A partial obstruction in the branch. This was the plausible one. An obstruction is concentrated: it produces most of the loss across a short segment, so the loss per foot along the run is wildly uneven. Gauges at the branch start and the branch end, with an intermediate reading at the midpoint of the run, showed the drop splitting close to evenly with distance. Loss distributed in proportion to length is what pipe does. A blockage does not do that.

What the tables said before anyone measured

The branch is a measured 140 ft of pipe. Fittings do not add length, they add resistance, and the trade convention is to convert each one into the length of straight pipe that would produce the same loss, its equivalent length, read from the table for that size and material. The count on this branch, six elbows, two tees taken through the branch, and a valve, came to 45 ft of equivalent length. Effective length is therefore 140 + 45 = 185 ft, and the fittings alone are 45 / 185 = 24 percent of it.

The friction table for this pipe size and material gives a loss rate of 6 psi per 100 ft at 4 gpm. Read that rate from the table for your own size and material rather than carrying this one; it is the input, not the finding.

At 4 gpm, the shower alone: 1.85 hundreds of feet x 6 psi = 11.1 psi.

At 8 gpm, shower plus laundry on the same branch: for water, friction loss rises with flow raised to about the 1.85 power in the Hazen-Williams relationship the common tables are built on, and with about the square of velocity in the Darcy-Weisbach form for fully turbulent flow. Doubling flow therefore multiplies the rate by 2 to the 1.85 power, which is 3.6: the rate goes from 6 to 21.6 psi per 100 ft, and the branch loses 1.85 x 21.6 = 40.0 psi.

Same pipe, same fittings, same everything. Four times the loss for twice the flow.

What the gauges read

Three gauges, read at the same minute, with the laundry filling and the shower open.

Trunk loss from the service to the branch start: 62 - 56 = 6 psi. Branch loss: 56 - 15 = 41 psi. Total loss to the far fixture: 47 psi, leaving 15 psi where the shower is.

The 41 psi measured against the 40.0 psi predicted is about 3 percent apart. At the low-flow test, 12 psi measured against 11.1 predicted was about 8 percent apart. Both within the range a friction table promises.

That agreement is the finding, and it is worth saying plainly to whoever pays the bill: nothing is wrong. The branch is doing exactly what a run of that length, that diameter and that fitting count does at that flow. Three parts were replaced looking for a defect that never existed.

Where the loss actually sits

Of the 47 psi the far fixture gave up, 6 psi happened in the trunk that everything in the house shares, and 41 psi happened in the branch that only the addition uses. That is 13 percent shared and 87 percent local.

A fixture on the original house piping, about 25 ft of effective length off that same trunk, read about 55 psi during the same run: it paid the shared 6 psi and almost nothing else, because its own length is a seventh of the addition's.

This is where the phrase about one branch stealing from another needs correcting. The laundry is not taking pressure from the shower. Both are on a 185 ft branch, both raise its flow, and the branch converts flow into loss at a steep rate. The genuinely shared component, the trunk, accounted for about an eighth of the problem. The customer had been blaming the appliance for seven eighths of a length problem.

Sizing the fix before selling it

Four options, each run against the same 41 psi measured branch loss at 8 gpm so the comparisons are like for like.

Upsize the branch. At fixed flow, friction loss scales roughly with the inverse fifth power of inner diameter, because a larger bore both slows the fluid and shortens the length-to-diameter ratio at once. Going up one nominal size on this material gives about 25 percent more inner diameter, and 1.25 to the fifth power is about 3.05, so the loss falls to about a third: 41 psi becomes roughly 13 psi, and the far fixture would see about 62 - 6 - 13 = 43 psi with the laundry running. Two cautions in the same breath: nominal size is not inner diameter, and inner diameter differs by material and wall thickness, so pull the actual bores from the table for what you are installing rather than assuming a size step delivers 25 percent.

Reduce the fitting count. The 45 ft of equivalent length is 24 percent of the effective length. Removing half of it takes effective length from 185 to 162.5 ft, which is 0.878 of the original, so the loss falls from 41 psi to about 36 psi. That is a real improvement and it is nowhere near enough, and it is worth computing precisely so nobody spends a day straightening a route expecting it to solve the complaint.

Sequence the demand. Not running the laundry during a shower puts the branch back at 4 gpm and about 12 psi of loss, which is why the system has worked for two years except in one combination. This costs nothing and it genuinely works. Offer it as a real option rather than a brush-off, especially where the run is buried in finished space.

Add a booster on the branch. This raises the source curve instead of lowering the resistance curve. Correct when the run cannot be replaced, and the expensive answer when it can, because it spends energy every day to push through a loss that upsizing would have removed once.

Confirming it after the change

Re-run the same test that produced the finding: same two gauge points, same 8 gpm total, same measurement of static and recovery before and after, on a day nothing else in the building is drawing.

The prediction to check is branch loss near 13 psi. Landing there confirms the diameter relationship applied cleanly to this material and this bore. Landing well above it means either the actual inner diameter step was smaller than 25 percent, which the table will tell you in a minute, or the new run picked up fittings the old one did not have.

Take one intermediate reading at the midpoint again. Loss splitting evenly along the length is the signature of a run that is behaving as pipe; loss concentrated in one segment after a rebuild usually means a fitting installed with a restriction in it, or debris moved during the work and settled somewhere convenient.

The structure of this generalizes wherever a fluid moves through a path: an effective length that includes fittings, a rate per unit length pulled from the right table, and an exponent between about 1.8 and 2 tying that rate to flow. Duct runs, hydronic loops, refrigerant suction lines and vacuum hoses all use the same three inputs with different tables and different units, and in every one of them the far end pays for the whole length while the near end pays only for what it shares.

References

• Hazen-Williams and Darcy-Weisbach friction loss relationships as presented in plumbing and hydronic design references, and the manufacturer or trade friction tables for the specific pipe material and size
• Equivalent length tables for fittings and valves, published by pipe and fitting manufacturers and in model plumbing code appendices
• 29 CFR 1910.1001 and 29 CFR 1926.1101 for asbestos-containing insulation; 29 CFR 1910.333(b)(2) for energized electrical work; NFPA 70E-2021, 120.5 for live-dead-live proving
• See related: How Pressure and Flow Relate in a Real System; Water Pressure + Flow Rate Reference