How do you calculate coolant pressure drop in a liquid cooling loop?

With the Darcy–Weisbach equation: ΔP = f × (L/D) × (ρv²/2), where f is the friction factor (about 0.03-0.04 for smooth hose), L the length, D the inner diameter, ρ the fluid density (about 1,020 kg/m³ for PG25) and v the velocity from flow ÷ pipe area. For example, 90 L/min in 25 mm hose over 3 m loses roughly 20 kPa before fittings.

What happens if pressure drop exceeds the CDU pump head?

Flow drops until the loop’s resistance curve intersects the pump curve lower down. Less flow means a wider supply-return temperature rise and less heat moved — the rack runs hotter than the design, typically failing first at high ambient. Fix it with larger diameters and shorter runs, not a bigger pump.

Formula · 3 min read

Pressure drop (Darcy–Weisbach)

Every meter of hose and every coupling eats pump head: 90 L/min through 25 mm hose loses ≈ 20 kPa over 3 m (f = 0.035). If the loop's total drop exceeds the CDU's available head, flow falls below the formula on paper — and the rack overheats on a Friday afternoon.

Formula

ΔP = f × (L / D) × (ρ × v² / 2) · v = Q / A

Darcy–Weisbach with friction factor f (≈ 0.03-0.04 for smooth hose at these flows), ρ ≈ 1,020 kg/m³ for PG25, and v the flow velocity. Velocity is the killer — it enters squared, which is why stepping down one hose size hurts so much.

Worked example

90 L/min · 25 mm ID · 3 m · f 0.035 → v 3.06 m/s → ΔP ≈ 20.0 kPa (2.9 psi)

That is just one straight hose. Add the rack manifold, two UQD couplings, elbows and the CDU heat exchanger, and a loop can easily demand 100-150 kPa — which is exactly what CDU pump curves are sized against.

Try it

Flow (L/min)Inner diameter (mm)Length (m)Friction factor fCDU available head (kPa)

20.0 kPa ≈ 2.9 psi · velocity 3.06 m/s

⚠️ Velocity 3.1 m/s — above the ~3 m/s comfort zone; erosion and noise risk, step up the diameter

Common mistakes

Undersized hose. Halving diameter multiplies velocity fourfold and drop roughly sixteen-fold (v² and D both bite). 25 mm+ supply lines are normal at rack flows for a reason.
Ignoring fittings and the UQD. Each elbow adds an equivalent length; a quick disconnect is often the single highest-drop element in a sub-loop. Count them, don't wish them away.
Reading the pump curve wrong. A CDU's "available head" is at a flow rate; as loop drop rises, operating flow slides down the curve. Check the intersection, not the brochure peak.

Coolant flow rate (what Q should be) · What is a CDU? · Commissioning (verifying real ΔP and flow)

Sources

Darcy–Weisbach with Darcy friction factor; PG25 density ≈ 1,020 kg/m³ at 25 °C. Friction factor from Moody/Colebrook smooth-pipe turbulence — engineering approximation for first-pass sizing, not a substitute for vendor loop data.