Duct Fitting Loss Calculator
Straight duct is the easy part. On a congested run the elbows, tees, take-offs and dampers routinely lose more pressure than every metre of straight duct combined, and a fan selected on friction alone will be short. Count the fittings here, add the straight run, and get the total the fan actually has to overcome.
Total pressure loss on this run
19.4 Pa
0.078 in.wg · fittings 12.0 Pa · straight duct 7.4 Pa
| Fitting | C each | Qty | ΣC | Loss |
|---|---|---|---|---|
| 90° elbow, smooth radius, r/D 1.5 | 0.15 | 2 | 0.30 | 2.4 Pa |
| Tee or cross, flow into the branch | 1.00 | 1 | 1.00 | 8.0 Pa |
| Volume control damper, fully open | 0.20 | 1 | 0.20 | 1.6 Pa |
| All fittings | — | — | 1.50 | 12.0 Pa |
| Duct | 16.0 in ø · area 0.1297 m² |
| Velocity | 3.64 m/s · 716 fpm |
| Velocity pressure ρV²/2 | 7.97 Pa |
| Sum of loss coefficients | 1.50 |
| Fitting loss ΣC × pv | 12.0 Pa |
| Straight duct friction rate | 0.375 Pa/m over 65.0 ft |
| Air density used (sea level) | 1.2043 kg/m³ |
Fittings dominate Fittings account for 62% of the loss on this run — more than the straight duct. Radiused bends instead of mitred ones, and a shorter route with fewer junctions, will buy more here than increasing the duct size.
Coils, filters, attenuators, terminal units and diffusers are not included — those come from the manufacturer as a pressure drop at a stated flow and are added to this figure separately.
Why the fan comes up short
A duct schedule is produced by sizing straight runs to a friction rate. It is a complete answer to the question it was asked, and an incomplete answer to the question that matters, because the fan does not see straight duct — it sees the whole path.
Put numbers on it. A 400 mm duct carrying 500 L/s runs at 4.0 m/s, so its velocity pressure is about 9.5 Pa and its friction rate about 0.6 Pa per metre. Twenty metres of that duct loses 12 Pa. One unvaned mitred elbow in the same duct, at a coefficient of 1.2, loses 11 Pa on its own. Four bends, two tees and a fire damper on the same run add roughly 45 Pa — nearly four times the straight duct.
That ratio gets worse as velocity rises, because fitting loss scales with velocity squared while straight-duct friction rises more slowly. It also gets worse in exactly the places where routing is tightest and bends are most numerous: plant rooms, risers, and the last few metres to a terminal. Those are the runs where a fan selected on friction alone is not slightly optimistic, it is wrong.
How to use the fitting loss calculator
- Enter the airflow and the duct size. These set the velocity, and velocity pressure is what every fitting loss is multiplied by. Get this wrong and every number below is wrong by the square of the error.
- Count the fittings on the run. Work along the index run — the path from the fan to the most remote terminal — and count what is actually on it. Fittings on other branches do not belong in this total.
- Add the straight duct. Enter the developed length and the material, and the straight-duct friction is added on the same basis the ductulator uses.
- Check the split. The result shows what share of the total comes from fittings. On a congested plant room run it is often well over half, and that is the number worth showing a client who wants the route shortened.
- Override a coefficient where it matters. For a critical junction, look the coefficient up in the ASHRAE tables for your actual geometry and flow ratio, and enter it in the custom row instead of relying on a representative figure.
Formulas and a worked example
- Velocity pressure: pv = ρV² ÷ 2
- One fitting: Δp = C × pv
- All fittings: Δp = (ΣC) × pv
- Straight duct: Darcy-Weisbach with the Altshul-Tsal friction factor, as used by the ductulator
- Total: fittings + straight duct, before plant and terminal resistances
Worked example
0.5 m³/s through a 300 mm round duct at 20 °C. Area is 0.070686 m², so V = 7.074 m/s, and the velocity pressure is 1.2043 × 7.074² ÷ 2 = 30.13 Pa. Two radiused elbows at r/D 1.5 give ΣC = 0.30, so the fitting loss is 0.30 × 30.13 = 9.04 Pa. Twenty metres of the same duct at 1.848 Pa per metre adds 36.95 Pa, for a total of 46.0 Pa.
Now swap those two elbows for unvaned mitred ones. ΣC becomes 2.40, the fitting loss becomes 72.3 Pa, and the total goes from 46 to 109 Pa — the run more than doubles because of two fittings. Nothing about the duct size changed.
Loss coefficients used
| Fitting | C | Notes |
|---|---|---|
| 90° elbow, smooth radius, r/D 1.5 | 0.15 | The cheapest 90° turn available. Specify this where space allows. |
| 90° elbow, smooth radius, r/D 1.0 | 0.22 | Tighter radius, roughly half again the loss of r/D 1.5. |
| 90° elbow, 5-piece segmented | 0.24 | Standard fabricated round bend. |
| 90° elbow, 3-piece segmented | 0.34 | Coarser segmentation, noticeably worse. |
| 45° elbow, smooth radius | 0.10 | Two 45s on a long offset beat one 90 wherever the route allows. |
| 90° mitred elbow with turning vanes | 0.20 | Vanes are what make a square elbow acceptable. |
| 90° radiused elbow, r/W 1.5 | 0.15 | Best rectangular turn if the void gives you the radius. |
| 90° mitred elbow, no vanes | 1.20 | Six to eight times a radiused bend. The single most expensive fitting people leave on a drawing. |
| Tee or cross, flow into the branch | 1.00 | Highly dependent on the branch-to-main flow ratio — check the tables for a critical run. |
| Tee or cross, straight through | 0.10 | The through path is cheap; the branch is not. |
| 45° conical take-off | 0.50 | A conical or bellmouth tap is worth specifying over a straight spigot. |
| Gradual reducer, 30° included | 0.05 | Contractions are nearly free if they are gradual. |
| Gradual expander, 20° included | 0.25 | Expansions are not — keep the included angle small. |
| Sharp-edged duct entry | 0.50 | A bellmouth entry drops this to about 0.05. |
| Abrupt duct exit | 1.00 | The full velocity pressure is lost at a plain open end. |
| Volume control damper, fully open | 0.20 | Rises steeply as the damper closes — a throttled damper is a major loss and a noise source. |
| Fire damper, curtain type | 0.50 | Check the manufacturer figure; frame and blade design vary widely. |
| Flexible connector at the plant | 0.30 | Only if fitted taut. Slack or offset connectors are far worse. |
These are representative values. The published data in ASHRAE Fundamentals Chapter 21 and the SMACNA tables is a set of tables parameterised by geometry, and for junctions also by the ratio of branch flow to main flow — a tee can run from about 0.2 to over 2.0 depending on how the air divides. For a critical run, take the coefficient from the tables for your actual case and enter it in the custom row above.
Where the cheap savings are
- Radius instead of mitre. 0.15 against 1.20 for the same change of direction — a factor of eight, for a fitting that often costs no more.
- Vanes in any square elbow. They take an unvaned mitre from 1.20 to around 0.20. On a rectangular system this is the single highest-return specification note you can write.
- Two 45s instead of one 90. Where an offset allows it, 0.10 twice beats 0.15 once only marginally — but it beats a mitre comprehensively, and it usually routes better.
- Conical take-offs. A shaped tap instead of a straight spigot cuts the branch loss substantially, and branch losses are where balancing problems start.
- Do not balance with dampers you could have designed out. A throttled damper is a large coefficient and a noise source in the same place.
Frequently asked questions
How do you calculate the pressure loss of a duct fitting?
Multiply the fitting loss coefficient by the velocity pressure: Δp = C × ρV²/2. The coefficient is a dimensionless number for the geometry, and the velocity pressure carries the air density and the speed. A 90° mitred elbow with a coefficient of 1.2 in a duct running at 30 Pa velocity pressure loses 36 Pa — on its own, more than twenty metres of straight duct at a typical friction rate.
Why do fittings matter more than straight duct?
Because loss scales with velocity squared and fittings concentrate the disturbance. A straight metre of 400 mm duct at 5 m/s loses well under a pascal. A single unvaned mitred elbow in the same duct loses around eighteen. On a plant room run with four bends, two tees, a fire damper and a volume control damper, the fittings can exceed the entire straight-duct friction of the system — which is exactly why a fan sized from a duct schedule alone comes up short.
How accurate are these loss coefficients?
They are representative values in the range published by ASHRAE Fundamentals Chapter 21 and the SMACNA tables, and for simple fittings — radiused elbows, gradual transitions, straight-through paths — they are close. Junctions are the weak point: the coefficient for a tee depends strongly on how the air divides between branch and main, and it can range from around 0.2 to over 2.0. For a critical run, take the value from the tables for your actual geometry and flow ratio and enter it in the custom row.
What is the cheapest way to reduce fitting losses?
Specify radiused elbows instead of mitred ones, and put turning vanes in any square elbow that survives. A radiused bend at r/D 1.5 has a coefficient around 0.15 against 1.2 for an unvaned mitre — a factor of eight for the same change of direction. After that: use two 45° bends instead of a 90° where an offset allows it, use conical rather than straight take-offs, and keep transitions gradual.
Should I include the fan connection and the plant?
The flexible connector is in the list because it belongs to the duct system. Coils, filters and attenuators are not, because their resistance comes from the manufacturer as a pressure drop at a stated flow rather than as a loss coefficient — add those figures to this total separately. The same goes for terminal units and diffusers, which are selected on their own published data.
Does a damper belong in this calculation?
Yes, but the figure here is for a damper fully open, which is the design condition. A volume control damper throttled to balance the system is a different object entirely — its coefficient rises steeply as it closes, and it becomes a significant noise source at the same time. If a system needs heavy throttling to balance, the answer is usually to resize rather than to add the loss.
Last updated: 27 July 2026