EN

Pipe Head Loss Calculator (Darcy-Weisbach)

A fluid flowing through a pipe loses pressure to friction against the walls — that's head loss. It depends on the flow velocity, the pipe's diameter and length, the fluid's viscosity and the wall's roughness — and it behaves differently depending on whether the flow is laminar or turbulent, which is what the Reynolds number tells you.

Enter the flow rate, the pipe's diameter and length, pick the fluid and the pipe material: the calculator gives the flow regime, the friction factor, the head loss and the resulting pressure drop.

DQP1P2L

Flow, fluid and pipe

Flow

Optional: sum of the K coefficients of elbows, valves, fittings (0 to skip).

Fluid

ν and ρ auto-fill (except "Custom").

Pipe

ε auto-fills (except "Custom").

Pipe-flow terms marked * are set out in the glossary.

The Colebrook-White equation, and why Swamee-Jain approximates it without iterating

Straight-pipe head loss* is the Darcy–Weisbach relation. The same loss is written two ways: as metres of fluid, or as a hydrostatic pressure in pascals.

Δh = f · (L/D) · v² / (2g)  ·  Δp = f · (L/D) · ρ v² / 2

What remains is the friction factor* f. Laminar flow, Reynolds number* Re below 2,300: f = 64 / Re, the Poiseuille relation, exact. Turbulent flow: f depends on both Re and the relative roughness* ε/D. Colebrook–White puts f on both sides, so the value is found by iteration — repeated substitution by hand, a loop in a spreadsheet. Swamee–Jain writes f in one explicit expression.

1 f = − 2 log10 ( ε 3.7D + 2.51 Ref )

f ≈ 0.25 / [ log10( ε/(3.7 D) + 5.74 / Re0.9 ) ]²

From Re = 5,000 to 108, for everyday ε/D, the explicit value usually lies within one percent of Colebrook. This page switches to Swamee–Jain the moment Re ≥ 2,300: the same family, and no Newton loop. A Colebrook iteration for a published figure lands a few thousandths away in f.

The transitional zone (2,300 < Re < 4,000): an uncertainty worth knowing

Re = v D / ν. Below 2,300 the filaments stay parallel, f = 64/Re, and the loss is linear with speed. Above 4,000 the flow mixes, f settles, and Darcy–Weisbach holds. Between those thresholds the regime is not signed off. The flow flips on a vibration or a weld bead, and f is no longer a single-valued function of Re and ε/D.

A line is not sized to run in that band. The scatter on Δh is too large, and so is the scatter on pump power. When the duty still falls there — a night set-back, oil warming through the morning — the turbulent f is taken, the higher of the two. The loss is overstated slightly. That costs less than a pump that cavitates every other day.

The calculator cuts at 2,300: Poiseuille below, Swamee–Jain above. The grey band is treated as turbulent. Cautious, not an error. The printed Re is what gets read. At 2,800, D or Q is changed rather than a pressure drop trusted to a tenth of a bar.

Equivalent lengths of fittings (elbows, tees, valves) and typical K coefficients

An elbow is not two metres of pipe, yet it dissipates head as though it were. The minor-loss coefficient* K in Δhs = K · v² / (2g) is dimensionless and fixed by the geometry. Equivalent length*, Leq = K · D / f, is the straight length with the same loss. Convenient when losses are stacked; misleading once f moves with Re.

Δhs = K · v² / (2g)  ·  Leq = K · D / f

  • Long-radius 90° bend: K ≈ 0.3 to 0.4. Tight mitre, shop-welded: nearer 0.9.
  • Full-bore ball valve, open: 0.05 to 0.1. Butterfly: 0.2 to 0.5. Globe valve: 4 to 10.
  • Sharp inlet: 0.5. Bellmouth: 0.04. Sudden enlargement: (1 − A1/A2)².

The ΣK field adds those K values, all referred to the velocity entered for the pipe. A short machine circuit, full of fittings, can lose more in the singularities than in the hose. A 400 m irrigation main does the opposite. Both are counted. A flat addition of 20% is not a substitute.

The Moody chart: reading f visually from Re and relative roughness

The Moody chart is Colebrook drawn. Re on a log axis, f vertical, one curve per ε/D. Lower left, the laminar line f = 64/Re. Far right the curves flatten: fully rough, f depends on ε/D only. The transitional region in the middle is left indistinct on many prints — no line is drawn where the physics supplies none.

Still useful at a whiteboard, or to catch a spreadsheet whose f has become implausible. Not for reading f to 0.001 with a pencil: for telling smooth-turbulent flow (tiny ε/D, f still falling with Re) from rough flow (f flat). The arithmetic is done on the page; the chart exposes an arbitrary f.

Sizing a pipe diameter to keep head loss under a target value

A cap is usually set first: 5% of working pressure, 0.5 bar on the discharge, 0.05 m/m on an irrigation lateral. Too small a D and v rises, Δh with it, and the pump grows. Too large a D and the pipe costs, weighs and takes space. The diameter kept is a compromise, not a unique formula.

Mean velocity is the first filter. Cold water in a distribution main: 0.8 to 2 m/s. Oil on a machine discharge: 3 to 6 m/s, sometimes more on a short hose. Suction is much slower; that limit is the cavitation section below. Once a target v is set, D = √(4Q / (π v)). Darcy–Weisbach is run again. If Δh overshoots, the next commercial DN is taken, not the millimetre sketched beforehand.

Swamee–Jain can be rearranged for D at a given Δh, with a little algebra. Trying a D, reading Δh and stepping up a size remains sounder: pipe is sold in steps, not as a continuum.

Pump-suction cavitation: the role of head loss on the suction side

Every metre lost between the sump and the suction flange is a metre taken from the available NPSH* (NPSHa). The pump cannot restore it. A strainer, a foot valve, three elbows and an undersized hose do more damage than a discharge head that is a little high. Discharge head loads the motor. Suction loss cavitates the impeller.

  • Suction bore larger than the discharge bore.
  • As few fittings as the layout allows before the pump. A filter on the suction side is included in the NPSHa.
  • The pump is placed as low as the layout allows, on a flooded suction where that is possible.
  • NPSHa > NPSHr, with margin — the rest of that comparison sits with pump power.

Suction losses are worked on their own, even when the discharge run is long. ΣK = 5 on 3 m of suction at 2 m/s is already serious. The same 5 on 40 m of discharge remains acceptable.

Relative roughness ε/D: why a small pipe “sees” its wall as rougher

ε is the material and its age. New commercial steel ≈ 0.045 mm. Drawn stainless, copper, PVC ≈ 0.0015 mm. Cast iron ≈ 0.26 mm. Braided hose, nearer 0.03 mm. The flow does not respond to ε. It responds to ε/D.

Same steel, 15 mm bore and 80 mm bore. The asperities take a much larger share of the small section. ε/D rises, f rises, and because v is already higher at the same Q, Δh rises on both counts. A small pipe is not a large one scaled down. It is rougher, faster, and more costly in power.

Catalogue ε is new pipe. Ten years of scale, rust or biofilm, and ε has moved into another range. On raw-water mains ε is often doubled as a precaution. On clean irrigation PVC, 0.0015 mm still holds.

Worked example: earthmoving-equipment hydraulic circuit, farm irrigation line

Machine discharge, VG46 oil. Steel, ε = 0.045 mm. L = 8 m, D = 25 mm, v = 3 m/s, ν = 46 cSt = 4.6 × 10−5 m²/s, ρ = 880 kg/m³. Q = v · π D² / 4 ≈ 1.5 L/s.

Re = (3 × 0.025) / (4.6 × 10−5) ≈ 1,630 → laminar, f = 64 / 1,630 ≈ 0.039

Δh = 0.039 × (8 / 0.025) × 3² / (2 × 9.81) ≈ 5.7 m  ·  Δp ≈ 0.50 bar

Half a bar in 8 m, at 3 m/s, laminar. Against a 200 bar circuit that looks negligible. It is not, once six such runs are added, plus the fittings. At 07:00, ν is no longer 46 cSt.

Irrigation lateral, 20 °C water, PVC. 10 m³/h, L = 150 m, DN80 (D = 80 mm), ε = 0.0015 mm. v ≈ 0.55 m/s. Re ≈ 44,000, turbulent. ε/D = 0.0015 / 80 = 1.9 × 10−5.

f ≈ 0.021 (Swamee–Jain)  ·  Δh / L ≈ 0.004 m/m  ·  Δh ≈ 0.6 m over 150 m

Target 0.05 m/m: the loss lies well under. DN65 would still pass. DN50 needs v and f afresh, not a division by the diameter. The head-loss calculation is run again rather than scaled by eye.

Typical velocities by fluid

Velocity is a choice, not a result. Too slow, and the pipe is oversized, solids settle, and the oil does not carry heat. Too fast, and Δh rises, noise with it, and the suction cavitates. Workshop figures, not laws.

  • Water suction: 0.5 to 1 m/s. Water discharge: 1 to 2.5 m/s.
  • Oil suction: 0.6 to 1.2 m/s. High-pressure oil discharge: 3 to 6 m/s, occasionally 8 on a short hose.
  • Return to tank: 2 to 3 m/s — enough to reject heat, not enough to entrain air.

A fire main will take 4 m/s for a few minutes. A potable borehole will not. The band is then checked against Δh.

Why Hazen–Williams is the wrong tool for oil

Water practice still reaches for Hazen–Williams: an empirical fit for water in the usual municipal range, with a C-factor that folds roughness and viscosity into one coefficient. No Reynolds number. No ν. VG46 put through it does not yield a usable head loss.

Darcy–Weisbach with a proper f (Colebrook or Swamee–Jain) does not depend on whether the fluid is water, diesel or hydraulic oil, provided ρ and ν are supplied. That is why the calculation is built that way. Hazen–Williams stays on the clean-water main already known. It is not the relation for an excavator circuit.

Winter start-up: viscosity, Reynolds and a pump that looks undersized

ISO VG46 is 46 cSt at 40 °C. At 0 °C the viscosity can sit in the hundreds of cSt. Re collapses, the flow drops into the laminar regime, f = 64/Re rises sharply, and so does Δp. The positive-displacement pump still delivers the same Q. The diesel sees a line pressure that is absent once the oil is hot; the relief lifts and the oil heats in the valve. Ten minutes later the condition has cleared. On the first crowd of the ram, the breaker has already tripped.

The discharge is sized for oil at running temperature. Suction and start-up are checked on cold oil. A line that passes at 40 °C can still leave the NPSHa inadequate at 5 °C. What changes in the form is ν. The relation stays Darcy–Weisbach.

Glossary

Head loss
Energy dissipated by the flow, as a height of fluid Δh or as the pressure drop Δp = ρ g Δh. The straight-pipe part is the Darcy–Weisbach major loss; fittings add a minor loss.
Reynolds number
Re = v D / ν. Laminar below 2,300, where f = 64/Re. Turbulent above 4,000. Between the two the regime is transitional and f is not single-valued.
Relative roughness
ε/D, absolute wall roughness divided by internal diameter. Turbulent f responds to this ratio, not to ε alone, so the same wall is effectively rougher in a smaller bore.
Friction factor
Dimensionless f in Darcy–Weisbach. Laminar flow uses the exact Poiseuille value 64/Re. Turbulent flow takes f from Re and ε/D, by iterating Colebrook–White or from Swamee–Jain.
Minor-loss coefficient
Dimensionless K in Δhs = K · v² / (2g). Fitting geometry fixes it — bend, valve, inlet — and it refers to the velocity in the pipe.
Equivalent length
Leq = K · D / f, the straight pipe that would match the fitting loss at that f. It changes when f changes with Re.
NPSH
Net positive suction head. NPSHa is what remains at the suction flange once suction losses are subtracted; NPSHr is what the pump requires. NPSHa has to exceed NPSHr, with a margin.

Related calculators