Why Elbows, Tees and Take-offs Drive Pressure Drop
On most commercial runs the fittings cost more pressure than the duct they connect. This guide explains the flow physics behind that, quantifies each fitting family in feet of equivalent straight duct, and shows before/after swaps that recover static pressure without upsizing anything.
Last updated September 2026
Advertisement
Straight duct is the cheap part
Straight duct loses pressure to wall shear alone — a thin boundary layer rubbing along the metal. A fitting loses pressure to something far more expensive: the air separates from the wall, forms a recirculating eddy, and the kinetic energy fed into that eddy never comes back as pressure.
A 16 in trunk at 2,500 CFM runs at 1,790 FPM, with a velocity pressure of 0.199 in. w.g. and a friction rate of 0.251 in. w.g. per 100 ft. In that duct, each fitting is worth the following length of straight duct:
| Fitting | C | Loss (in. w.g.) | Equivalent straight duct |
|---|---|---|---|
| 90° smooth elbow, r/D 1.5 | 0.15 | 0.030 | 12 ft |
| 90° elbow, 3-piece segmented | 0.34 | 0.068 | 27 ft |
| 90° mitred elbow, no vanes | 1.20 | 0.239 | 95 ft |
| Conical take-off | 0.50 | 0.100 | 40 ft |
| Tee, flow into the branch | 1.00 | 0.199 | 79 ft |
| Abrupt duct exit | 1.00 | 0.199 | 79 ft |
16 in spiral galvanized duct, 2,500 CFM, 70 °F standard air.
One un-vaned mitred elbow in that trunk costs the same static pressure as roughly 95 feet of duct. Systems rarely fail because the duct was a shade too small; they fail because nobody counted the fittings.
Why each family behaves the way it does
- Elbows. The tighter the centreline radius, the more violently the flow separates at the inner wall. Going from r/D 1.5 to r/D 1.0 roughly doubles the loss; going to a sharp mitre multiplies it by eight.
- Tees and branch take-offs. Air has to be turned and accelerated or decelerated as the flow splits. The branch path pays for both, which is why branch coefficients near 1.0 are normal while the straight-through path pays almost nothing.
- Transitions. Contractions are cheap; expansions are not. A gradual 20° expander is about 0.25, an abrupt one approaches 1.0, because diffusing air without separation requires a shallow angle.
- Dampers and terminal connections. Even fully open, a blade sits in the airstream. Throttled for balance, it can dominate the branch.
Before and after: what a fitting swap is worth
| Change | Before | After | Saved (in. w.g.) |
|---|---|---|---|
| Mitred elbow → smooth r/D 1.5 | 0.239 | 0.030 | 0.209 |
| Mitred elbow → mitred with turning vanes (C 0.30) | 0.239 | 0.060 | 0.180 |
| Square branch tee → conical take-off | 0.199 | 0.100 | 0.100 |
| Abrupt expander → gradual 20° expander | 0.199 | 0.050 | 0.150 |
| All four, on one critical path | 0.878 | 0.239 | 0.638 |
Same 16 in trunk at 2,500 CFM. Recovered static pressure is available for the rest of the path — or for a smaller fan.
Price the swap on your own section in the fitting loss calculator, and check what the velocity is doing to it in the velocity calculator.
Method and assumptions
- Method
- Loss coefficient method, Δp = C × Pv; equivalent lengths derived from the friction rate of the reference section.
- Reference section
- 16 in spiral galvanized round, 2,500 CFM, 70 °F, sea level.
- Coefficients
- Representative published values for isolated fittings in preliminary design.
Limitations of this method
What the calculation on this page does not account for. Read these before using a number on a drawing or a submittal.
- Equivalent lengths shown are valid only for the reference section. Change the diameter or the airflow and every figure changes with the square of velocity.
- Coefficients assume isolated fittings with several diameters of straight duct either side. Back-to-back fittings lose more than the table predicts.
- Turning-vane performance depends entirely on installation. Badly spaced or loose vanes add noise and can perform worse than no vanes.
- Branch coefficients vary with the flow split between branch and main; a single value is a simplification.
- Lower pressure drop does not automatically mean lower noise. Evaluate acoustics separately, especially at take-offs serving quiet spaces.
Frequently asked questions
Why do elbows cause so much pressure drop?
Turning the air separates the flow from the inside of the bend. The separated region contracts the effective flow area, and the energy spent re-expanding and re-mixing downstream is lost as pressure. A mitred elbow with no vanes can cost eight times what a smooth long-radius elbow costs.
How much does a take-off add?
A conical take-off is typically around 0.5 velocity pressures and a square-edged branch tee around 1.0. On a short branch that is often more than the entire straight duct on the run.
Do turning vanes help?
Yes, substantially. Vanes in a mitred elbow bring the coefficient down from roughly 1.2 toward 0.2 to 0.35, provided they are correctly spaced and fixed. Poorly installed vanes generate noise and can make the loss worse.