Pressure Drop in Pipes 101: Darcy-Weisbach, Reynolds, and Minor Losses

The Darcy-Weisbach Equation

The standard relationship for friction loss in a full, straight run of pipe is Δp = f · (L/D) · (ρ·V²/2), where f is the dimensionless Darcy friction factor, L is the pipe length, D the internal diameter, ρ the fluid density, and V the mean velocity, meaning volumetric flow divided by internal area. The same statement in head form, h(f) = f · (L/D) · (V²/2g), is how the app's fluids formula entry expresses it; the two are related by Δp = ρ·g·h(f).

Read the structure of that equation and most of the practical guidance follows. Loss is proportional to length. It is inversely proportional to diameter, and velocity itself depends on diameter squared at fixed flow, so sensitivity to bore is severe. And loss goes with velocity squared, which is why velocity limits rather than pressure calculations are the usual first-pass design rule in piping.

Δp = f · (L/D) · (ρ·V²/2). The units resolve to kg/(m·s²), which is a pascal. Check that before trusting a number.

Reynolds Number Picks the Friction Factor

The friction factor is not a constant; it depends on the flow regime and, in turbulent flow, on the pipe's roughness. The Reynolds number, Re = ρ·V·D/μ, where μ is the dynamic viscosity, tells you which regime you are in. It is the ratio of inertial to viscous effects, and being dimensionless it comes out the same in any consistent unit set.

For internal flow in circular pipes, flow is laminar below roughly Re = 2300, turbulent above roughly Re = 4000, and unpredictably transitional in between. In the laminar regime the friction factor has an exact analytical solution, f = 64/Re, with no dependence on roughness at all. Real water systems are almost always turbulent; laminar pipe flow shows up mostly with viscous fluids such as oils.

In the turbulent regime f depends on both Re and the relative roughness ε/D, where ε is the absolute roughness height of the wall, typically around 0.045 mm for commercial steel and much lower for drawn tubing or plastic. The implicit Colebrook equation is the reference correlation and must be iterated. The explicit Swamee-Jain approximation, f = 0.25 / [log₁₀(ε/(3.7D) + 5.74/Re⁰·⁹)]², typically agrees with it to within about one percent, comfortably inside the uncertainty on ε itself.

• Laminar, Re below about 2300: f = 64/Re exactly, independent of roughness.
• Turbulent, Re above about 4000: f from Colebrook or the explicit Swamee-Jain approximation.
• Transitional, roughly 2300 to 4000: treat predictions as uncertain and design away from this band.

A Worked Turbulent Example

Take water at about 20 °C flowing at 2.0 m/s through a 50 m straight run of DN 50 Schedule 40 commercial steel pipe. The pipe schedule tables give that size an internal diameter of 52.48 mm, so D = 0.05248 m. Use ρ = 998 kg/m³ and μ = 1.0 × 10⁻³ Pa·s as typical properties at that temperature; the flow is about 4.3 L/s.

First the regime: Re = 998 × 2.0 × 0.05248 / 1.0 × 10⁻³ ≈ 1.05 × 10⁵, comfortably turbulent. Then relative roughness: ε/D = 0.045 mm / 52.48 mm ≈ 8.6 × 10⁻⁴. Feeding both into Swamee-Jain gives f ≈ 0.0217, and iterating Colebrook on the same inputs gives 0.0216, a difference well under one percent.

Now the loss: Δp = 0.0217 × (50 / 0.05248) × (998 × 2.0² / 2) ≈ 41 kPa, which converts to Δp/(ρg) ≈ 4.2 m of water over the 50 m run. Every term is in SI base units and the result carries units of pascals, as it should.

Push the velocity to 4.0 m/s and rerun. Re doubles to about 2.1 × 10⁵, f falls slightly to about 0.0205 because friction factor decreases weakly with Reynolds number, and Δp rises to roughly 156 kPa. That is a factor of about 3.8, slightly below the four the V² term alone would give, with the shortfall explained entirely by the drop in f.

Doubling velocity roughly quadruples pressure drop. At fixed flow rate, halving the bore raises the loss by roughly a factor of 32, because velocity scales with 1/D² and loss scales with V²/D.

A Laminar Example, and Why It Differs

Change the fluid to a viscous oil, ρ = 880 kg/m³ and μ = 0.1 Pa·s, running at 0.5 m/s through the same 50 m of DN 50 pipe. Now Re = 880 × 0.5 × 0.05248 / 0.1 ≈ 231, firmly laminar. The friction factor comes straight from f = 64/Re ≈ 0.277, more than ten times the turbulent value above, and roughness plays no part in it.

The pressure drop is Δp = 0.277 × (50 / 0.05248) × (880 × 0.5² / 2) ≈ 29 kPa. A quarter of the velocity in the same pipe still produces a loss of the same order as the water case, purely because the fluid is a hundred times more viscous. Viscosity, not velocity, dominates here.

The scaling laws differ too. Substituting f = 64/Re into Darcy-Weisbach gives a pressure drop proportional to velocity rather than its square, and inversely proportional to the fourth power of diameter, which is the Hagen-Poiseuille result. That is why intuition built on turbulent water systems misleads badly when applied to hydraulic oil.

Minor Losses Are Often Not Minor

Straight pipe is only part of a real system. Every elbow, tee, reducer, valve, entrance, and exit disturbs the flow and dissipates extra energy. These are handled with a loss coefficient K applied to the same velocity head: Δp = K · ρ·V²/2. Each K is empirical, so it should come from the manufacturer or a recognised reference rather than from memory.

That form makes fittings easy to add: total the K values, evaluate once at the relevant velocity, and add to the straight-pipe loss. In the water example, four elbows with a typical K near 0.3 each give a combined K of 1.2 and about 2.4 kPa on top of the 41 kPa from the straight run. But in a compact skid or a pump suction the fittings can easily exceed the pipe loss.

The alternative bookkeeping is equivalent length, expressing each fitting as the straight pipe that would produce the same loss, L(eq) = K·D/f; those four elbows work out to roughly 2.9 m here. One category deserves care: a partly closed throttling valve can have a K one or two orders of magnitude above an elbow, so if a system behaves nothing like the calculation, a valve position is a likelier explanation than a wrong friction factor.

Δp = K · ρ·V²/2 for each fitting. Sum the K values, evaluate at the relevant velocity, and add to the straight-pipe loss.

Turning the Numbers Into a Design

In practice the calculation runs backwards from a sizing question. Start from the required flow rate and a target velocity, since piping practice sets velocity bands by service to limit noise, erosion, and pumping cost. That gives a required bore, which you round up to the nearest real size from a schedule table. Compute the actual velocity in that real size, then work the loss.

Add the straight-run loss, the fitting losses, and any static elevation change to get the total the pump must overcome. Elevation is not a friction term and does not scale with velocity, so keep it separate. Then check the result across the operating envelope, not just at the design point, because a system comfortable at design flow can starve at maximum flow or run a pump off its curve at minimum flow.

Finally, keep the uncertainty in perspective. Roughness increases over a pipe's life with scaling and corrosion, published K values approximate real geometries, and viscosity moves with temperature. A careful calculation is a good estimate, not a precise one.