Separator Sizing — Two/Three-Phase, Horizontal/Vertical
Give it the gas, oil and water rates and the separator’s pressure and temperature, and it works out how big a vessel has to be to do two jobs: hold the gas velocity low enough that the liquid droplets fall out of it, and hold the liquid long enough for the phases to separate. You get a recommended standard diameter and seam-to-seam length, the whole table of sizes it chose between, and — if you switch the optional modules on — the inlet-nozzle check, how much liquid actually leaves with the gas, and what a mist extractor does about it. It is a first-pass process sizing tool, not a mechanical design and not a vendor guarantee.
Preliminary sizing only — this is not a design calculation
Preliminary sizing only. Souders-Brown, droplet-settling and retention-time methods per GPSA, API 12J, Arnold & Stewart and Bothamley (SPE OGF 2013) with the assumptions listed on this page. Excludes slug/surge volumes, level-alarm spacing, re-entrainment, foaming, degassing, mechanical design and nozzle reinforcement. Verify against the vendor's sizing and the current codes before procurement.
What to buy
Marginal - the closest workable size, not a clean pass
Recommended 30 in x 2.13 m (7.0 ft) horizontal three-phase separator, L/D 2.8, gas at 69 % of allowable, oil retention 5.0 min. Why it is marginal: L/D 2.8 is outside the 3.0-5.0 target window.
Marginal — see the verdict above.
0.804 m/s in the gas space · 48 % of the Souders-Brown velocity limit · 69 % of the settling length the design droplet needs
design basis 3.0 min
design basis 3.0 min
Outside the target window — a shipping and support question, not a process one.
design target 150 µm
The vessel
Every standard diameter, side by side
The highlighted row is the recommendation: the smallest diameter that passes every check and lands inside the L/D window. The rows around it show what the next size up or down buys you.
| ID (in) | Lss req. (m) | Lss chosen (m) | L/D | Gas % of allow. | Oil ret. (min) | Water ret. (min) | Settles (µm) | Status |
|---|---|---|---|---|---|---|---|---|
| 16 | 6.97 | 7.01 | 17.3 | 170 % | 4.7 | 3.0 | 47 | fail |
| 20 | 4.46 | 4.57 | 9.0 | 109 % | 4.8 | 3.1 | 60 | fail |
| 24 | 3.10 | 3.20 | 5.3 | 76 % | 4.8 | 3.1 | 75 | marginal |
| 30pick | 1.98 | 2.13 | 2.8 | 69 % | 5.0 | 3.2 | 108 | marginal |
| 36 | 1.70 | 1.83 | 2.0 | 86 % | 6.2 | 4.0 | 132 | marginal |
| 42 | 1.74 | 1.83 | 1.7 | 89 % | 8.5 | 5.4 | 135 | marginal |
| 48 | 1.81 | 1.83 | 1.5 | 97 % | 11.0 | 7.1 | 146 | marginal |
| 54 | 1.90 | 1.98 | 1.4 | 86 % | 15.1 | 9.7 | 132 | marginal |
| 60 | 2.00 | 2.13 | 1.4 | 78 % | 20.1 | 12.9 | 120 | marginal |
| 66 | 2.11 | 2.13 | 1.3 | 94 % | 24.4 | 15.6 | 142 | marginal |
| 72 | 2.22 | 2.29 | 1.3 | 86 % | 31.0 | 19.9 | 132 | marginal |
| 84 | 2.47 | 2.59 | 1.2 | 74 % | 47.9 | 30.8 | 115 | marginal |
| 96 | 2.73 | 2.74 | 1.1 | 97 % | 66.2 | 42.5 | 146 | marginal |
| 108 | 3.01 | 3.05 | 1.1 | 86 % | 93.1 | 59.8 | 132 | marginal |
| 120 | 3.28 | 3.35 | 1.1 | 71 % | 126.5 | 81.2 | 111 | marginal |
| 132 | 3.57 | 3.66 | 1.1 | 59 % | 167.0 | 107.2 | 94 | marginal |
| 144 | 3.85 | 3.96 | 1.1 | 50 % | 215.3 | 138.2 | 81 | marginal |
“Gas % of allow.” is how close the gas section runs to whichever criterion binds first — the Souders-Brown velocity limit, or (in a horizontal vessel) the settling length the design droplet needs. Over 100 % it fails.
About the recommended vessel: marginal: L/D 2.8 is outside the 3.0-5.0 target window
Warnings
Marginal - the closest workable size, not a clean passfor information
No standard diameter satisfies every check outright, so the closest workable one is shown. For the 30 in vessel above: L/D 2.8 is outside the 3.0-5.0 target window. That is the normal outcome for a wet three-phase duty - the liquid sets the length, so the vessel comes out shorter and fatter than the usual length/diameter window rather than failing anything. Look down the sizing table before settling on one: a larger diameter will usually sit inside the window, for more steel.
Length/diameter outside the target windowfor information
L/D is 2.8 against a target window of 3.0 to 5.0. The vessel still works, but it is an unusual shape to buy.
Liquid/liquid settling is outside the Stokes rangefor information
The Reynolds number of the design liquid droplet is above 2, so Stokes' law is being used beyond its strict range. Industry practice (Arnold & Stewart) keeps Stokes here, and so does this calculator, but the settling velocity is optimistic. For the 500 µm water droplet, Stokes gives a 0.09 min fall time through the oil pad; the intermediate law that its Reynolds number of 13 actually calls for gives 0.16 min. So treat the oil-pad check as optimistic by roughly 1.8x. Use a smaller design droplet, or a thinner oil pad, if the margin matters.
How this was calculated
This tool sizes a two- or three-phase separator, horizontal or vertical, or rates one you already have. It follows the standard published methods — API Spec 12J, GPSA Section 7, Arnold & Stewart’s Surface Production Operations, and Mark Bothamley’s 2013 SPE separator series. Every number it uses is listed with its source in the full method document.
The gas side is checked two ways, and the tighter answer wins. The first way is the traditional velocity factor (K): the gas can only move so fast before it drags liquid out with it. The second way follows a single droplet of a chosen size — 150 microns unless you change it — and asks whether that droplet has time to fall out of the gas before the gas reaches the outlet. In a vertical vessel that sets the diameter. In a horizontal vessel it sets the length.
The liquid side is sized on holding time: how many minutes the oil and the water each sit in the vessel. In a three-phase vessel the tool also checks the two settling jobs directly — whether a water droplet has time to fall through the oil pad, and whether an oil droplet has time to rise through the water layer.
It then works through every standard vessel diameter, adds the usual allowances for the inlet and the mist extractor to get a seam-to-seam length, and recommends the smallest vessel that passes everything and has a sensible length-to-diameter ratio.
What it does not do
It sizes for capacity and holding time only. It does not set level-alarm spacing or slug volume, check whether gas velocity is ripping liquid back off the surface, allow for foaming or waxy crude, check gas carried out in the liquid, or do any mechanical design. It assumes a clean, steady feed. Foaming service typically needs up to four times the holding time. Cyclone bundle capacities are vendor-specific — published values vary by a factor of two, so confirm with the supplier before issuing.
This is a screening tool. It does not replace a vendor’s design or your own judgement.
The numbers behind this case
1 · The gas
- Absolute pressure 551 kPa, molecular weight 22.0, z 0.983 → density 5.1 kg/m³.
- Standard flow re-expressed at operating conditions = 660 m³/h — the flow the vessel actually has to handle.
2 · How fast the gas is allowed to go
- Souders-Brown: V = K·√((ρliquid − ρgas)/ρgas) = 1.663 m/s.
- Droplet settling: a 150 µm oil droplet falls at 0.323 m/s (intermediate law); after the maldistribution factor, 0.323 m/s.
- In a horizontal vessel the K figure is the velocity ceiling and the droplet figure becomes a length the gas space must have — the two are not compared as velocities.
3 · The vessel that came out of it
- Gas space 0.228 m² of 0.456 m², gas velocity 0.804 m/s.
- Liquid volume 0.36 m³ → 5.0 min of oil retention, 3.2 min of water.
- Seam-to-seam 1.98 m required, rounded up to 2.13 m (next half foot) → L/D 2.8.
- The smallest droplet this geometry drops is 108 µm, settling by the intermediate law.
References
- M. Bothamley, “Gas/Liquids Separators — Quantifying Separation Performance”, SPE Oil & Gas Facilities, Parts 1–3, 2013.
- GPSA Engineering Data Book, Section 7 — Separators and Filters.
- API Spec 12J — Specification for Oil and Gas Separators.
- K. Arnold & M. Stewart, Surface Production Operations, Vol. 1.
- API RP 14E — erosional velocity for the inlet piping.
Read before you rely on these numbers
- The K-factor, the maldistribution factor and the retention time are judgement calls, not measurements. Change one and the answer moves — try it and see how much.
- Standard diameters are treated as inside diameters. A real vessel’s ID is the nominal OD minus two wall thicknesses, so a nominal 36 in vessel holds slightly less.
- No slug or surge volume is included. If the upstream line slugs, the liquid section has to be sized for the slug, not the steady rate.
- Foaming crude, high-viscosity oil and tight emulsions all beat gravity separation and need residence times well beyond the table defaults.
- Preliminary sizing only. Souders-Brown, droplet-settling and retention-time methods per GPSA, API 12J, Arnold & Stewart and Bothamley (SPE OGF 2013) with the assumptions listed on this page. Excludes slug/surge volumes, level-alarm spacing, re-entrainment, foaming, degassing, mechanical design and nozzle reinforcement. Verify against the vendor's sizing and the current codes before procurement.