In the design of duct networks, both in HVAC applications and in process plants, bends represent one of the most significant sources of concentrated pressure loss. Among the various geometries available, miter bends are often preferred for cost reasons and because standard (smooth) bends cannot be manufactured for large diameters, but they result in noticeably higher pressure losses than the latter.
This article presents the classic calculation method for estimating the concentrated loss coefficient ζ of a miter bend as a function of the deflection angle δ and the ratio between the equivalent radius of curvature and the duct diameter (referred to here as r/D, or equivalently r/R), and closes with a numerical example.
1. Theoretical Background on Concentrated Pressure Losses
The total pressure loss along a duct section is given by the sum of the distributed losses (pipe/duct friction, calculable using the well-known Darcy-Weisbach formula) and the concentrated losses localized at geometric singularities such as elbows, branches, contractions, and valves.
The concentrated loss associated with a singularity is expressed as:
Δp = ζ · (ρ · v²) / 2
where:
- Δp is the concentrated pressure loss [Pa]
- ζ (zeta) is the concentrated loss coefficient, dimensionless, characteristic of the geometry
- ρ is the fluid density [kg/m³]
- v is the average fluid velocity in the duct [m/s]
The term ρ·v²/2 is the dynamic pressure of the flow. Characterizing a bend therefore comes down to determining the appropriate ζ coefficient for the geometry under consideration.
2. Miter Bend Geometry: Single Miter vs. Segmented
It is useful to distinguish between two recurring configurations:
Single miter bend (sharp-edged). The duct is cut along a single inclined plane and the two sections are welded directly together. The equivalent radius of curvature is zero (r/D → 0): the flow undergoes an abrupt deflection, with significant flow separation on the inner wall of the bend and a resulting high pressure loss.
Segmented miter bend (or “gored bend”). The total deflection angle is divided across several segments (typically 3 or 5 gores), each rotated by a partial angle relative to the next. As the number of segments increases, the polygonal path increasingly approximates a smooth, constant-radius fitting, and an equivalent r/D ratio can be associated with the bend (the radius of curvature of the elbow’s centerline relative to the duct radius). As the number of segments and the r/D ratio increase, ζ asymptotically approaches the value of a smooth bend of the same angle and radius.
This distinction is the same one adopted in the classic hydraulic resistance literature, notably by Idelchik in his monumental Handbook of Hydraulic Resistance, and carried over into the ASHRAE fitting loss databases and the Crane TP-410 manual: the “sharp-edged” and “segmented/smooth” geometries are treated as two families of bends, characterized respectively by the absence and the presence of a significant radius of curvature.
3. Correlations for Calculating ζ
3.1 Single Miter (r/D = 0)
For a sharp-edged miter bend, ζ depends essentially on the deflection angle δ and, as a second-order effect, on the Reynolds number (the effect of Re is significant only at relatively low Re; under fully turbulent conditions, typical of industrial and HVAC applications, ζ is essentially independent of Re). A simplified correlation, widely used in the technical literature for order-of-magnitude estimates during preliminary design, is:
ζ = sin²(δ/2) + 2 · sin⁴(δ/2)
with δ expressed in degrees (or radians, as long as it is used consistently in the sine function), and valid indicatively for δ ≤ 90°.
This relationship should be regarded as an estimating tool: for detailed design it is always advisable to verify the coefficients against a validated technical reference (Idelchik, ASHRAE Fundamentals, Crane TP-410, or dedicated calculation software), since the numerical coefficients published in different sources may vary slightly depending on the reference experimental conditions (roughness, length-to-diameter ratio of the straight sections, inlet conditions).
3.2 Segmented Miter (r/D > 0)
When the total angle is divided across several segments, the ζ coefficient progressively decreases relative to the single-miter case, approaching the value of a smooth bend. The qualitative trend, consistent with the data reported in industry reference manuals, is as follows:
- for a given angle, ζ decreases as the number of segments and the r/D ratio increase;
- the effect is more pronounced going from 1 to 3 segments than from 3 to 5: beyond 3-4 segments, the marginal benefit decreases significantly;
- for r/D ≥ 1.5, the behavior approaches that of a constant-radius bend with the same r/D.
The table below gives indicative ζ values for a circular duct with a diameter of 1000 mm, under fully developed turbulent flow conditions (f_T ≈ 0.0115), useful for preliminary estimates during rough sizing. For detailed design it is necessary to refer to the complete diagrams in Idelchik or to the ASHRAE/Crane manuals, which also account for the number of segments, aspect ratio, and inlet conditions.
ζ values for a 1000 mm duct diameter
| No. of segments | r/D=0.75 | r/D=1 | r/D=1.5 |
|---|---|---|---|
| 3 | 0.222 | 0.173 | 0.140 |
| 4 | 0.118 | 0.092 | 0.074 |
| 5 | 0.116 | 0.090 | 0.073 |
3.3 Expressing the ζ Coefficient as an Equivalent Length L/D
An alternative approach, widespread especially in the methodology adopted by Crane (Technical Paper 410), consists of expressing the concentrated loss not as a standalone ζ coefficient, but as an equivalent length of straight duct, so that it can be added directly to the distributed losses within the same Darcy-Weisbach calculation. The conversion relationship is:
ζ = f_T · (L/D)
where:
- f_T is the reference friction factor under fully turbulent flow conditions, a function of the nominal diameter and duct roughness (typically tabulated for commercial steel piping; for ducts made of different materials or with different roughness it must be recalculated using the Colebrook or Moody relationships);
- L/D is the equivalent length of the singularity, expressed as the number of duct diameters of straight pipe that would produce the same pressure loss.
This approach is convenient because it allows bends, fittings, and straight sections to be summed into a single total equivalent length, which is then multiplied only by the duct’s distributed friction factor.
The table below shows typical L/D values as a function of the r/D ratio. Since f_T can vary, ζ must be updated accordingly, following the same logic adopted by Crane in its own reference tables. For r/D values other than 1, the values are estimated from those for r/D = 1, starting from the ζ coefficient values obtained by Locklin for r/D equal to 0.75 and 1.5.
| Angle ° | L/D r/R=0.75 | L/D r/R=1 | L/D r/R=1.5 |
|---|---|---|---|
| 15 | 5.1 | 4 | 3.2 |
| 30 | 10.3 | 8 | 6.5 |
| 45 | 19.3 | 15 | 12.1 |
| 60 | 32.1 | 25 | 20.2 |
| 90 | 77.1 | 60 | 48.6 |
4. Numerical Example
Let’s compare the pressure loss introduced by a 90° bend in two configurations, on a circular ventilation duct with the following data:
- Duct diameter: D = 1000 mm
- Air flow rate: Q = 50000 m³/h
- Dynamic viscosity of air: 1.85 · 10^-5 Pa·s
- Air density: ρ = 1.2 kg/m³
Velocity in the duct
Cross-sectional area: A = π·D²/4 = π·(1)²/4 = 0.78 m²
v = Q / A = 50000 / 0.78/3600 ≈ 17.8 m/s
Dynamic pressure calculation
pv = ρ·v² / 2 = 1.2 × (17.8)² / 2 ≈ 190.1 Pa
Case A: single sharp-edged miter bend, δ = 90°
From the correlation in §3.1: ζ_A = sin²(45°) + 2·sin⁴(45°) = 0.5 + 0.5 = 1.00
Δp_A = ζ_A × pv = 1.00 × 190.1 ≈ 190.1 Pa
Case B: 3-gore segmented miter bend, r/D ≈ 1, δ = 90°
From the table in §3.3, with three gores we have 2 changes of direction, so δ = 45° and r/D ≈ 1:
ζ_B = f_T · (L/D) = f_T · 15
Re ≈ 1154000, the flow is turbulent, although not fully developed.
From the Moody diagram, f_T ≈ 0.0123.
ζ_B = f_T · (L/D) = f_T · 15 = 0.0123 · 15 = 0.185
Δp_B = ζ_B × pv = 0.185 × 190.1 ≈ 35.2 Pa
The difference in pressure loss between the two configurations is about 155 Pa per single bend. The difference between the two cases is considerable. In a system with multiple 90° bends along the main run, the cumulative energy savings from the lower pressure losses can become significant on an annual basis, while the higher fabrication cost of miter bends (more cuts, more welds) should be weighed through a CAPEX/OPEX analysis of the component, no differently than for other plant design choices.
5. Final Considerations
- For small deflection angles (≤ 30°-45°) the difference between a single and a segmented miter is often negligible from an energy standpoint, and the single-miter solution may be preferable for construction simplicity.
- For angles close to 90° and for ducts with high flow rates or continuous operation, adopting a segmented miter bend (or a constant-radius bend, where space allows and it is physically feasible) is generally justified by the energy savings over the plant’s life cycle.
- The optimal number of segments is often 3 for angles of 90°.
- The coefficients presented in this article are intended for illustrative purposes and preliminary estimates: for detailed design, it is recommended to use the complete diagrams from Idelchik (Handbook of Hydraulic Resistance), the ASHRAE manuals (Fitting Loss Coefficients, or better yet the Duct Fittings Database), or the values given in the Crane Technical Paper 410, which include corrections for aspect ratio, inlet conditions, and Reynolds number.