Steam Separator

A Steam Separator is a pressure vessel fitted in a steam main that mechanically removes entrained water droplets from wet steam before the steam reaches the engine, turbine, or process load. It works by abruptly changing steam direction across baffles or a cyclone vane so that water droplets, being roughly 1,500 times denser than steam, lose momentum and fall out into a drain pocket. The purpose is to deliver steam at a dryness fraction of 0.98 or better to the working cylinder. The result is no slugging, no water hammer, and a measurable gain in indicated power.

Steam Separator Baffle Type Cross-Section Animated cross-section diagram showing how a baffle-type steam separator works: wet steam enters, flows around a baffle plate where water droplets impact and drain, while dry steam exits. Steam Separator (Baffle Type) Wet Steam In Dry Steam Out Baffle Plate Drain Pocket Steam Trap Droplets impact here Steam follows curve Collected water Key Physics Water: ~950 kg/m³ Steam: ~3.6 kg/m³ Ratio: ~1500× denser
Steam Separator Baffle Type Cross-Section.

How the Steam Separator Works

Wet steam carries water in two forms — visible droplets suspended in the flow and a thin film clinging to the pipe wall. Both kill performance. Droplets entering a cylinder slam against the piston crown and ring belt, causing the dull thud you hear as water hammer, and they wash oil off the bore. The Steam Separator forces the flow to do something the droplets cannot follow — typically a sharp 90° or 180° turn across a baffle stack, or a tangential entry into a cyclone body. Steam, with a density around 3.6 kg/m³ at 8 bar absolute, follows the bend easily. Water droplets, at roughly 950 kg/m³ at the same temperature, carry far more momentum per unit volume and impact the baffle plate, where they coalesce, run down the wall, and collect in the drain pocket at the bottom of the vessel.

The geometry has to be right or the separator becomes a restriction with no benefit. Inlet velocity should sit between 15 and 25 m/s for a baffle-type unit. Below 10 m/s the droplets do not have enough momentum to leave the streamline and they sail straight through. Above 30 m/s droplets re-entrain off the baffle face and get carried out the outlet — you get carryover even though the separator looks correctly sized. The drain pocket needs a steam trap that actually keeps up with the condensate load. If the trap fails closed, the pocket fills, the water level rises into the separation chamber, and the unit becomes a humidifier feeding the engine wet steam at full flow rate.

The usual symptom of a failing separator is the engine starting to thump on every revolution within 30 seconds of opening the regulator wide. That is the cylinder taking water. Pull the trap, check the float or bucket, and confirm the inlet strainer is not blanked off with scale before you blame the boiler.

Key Components

  • Separator Body: The pressure vessel itself, typically cast iron or fabricated steel rated to the line pressure plus a margin. Internal volume is sized so superficial velocity drops to 3-5 m/s in the separation chamber, giving droplets time to settle out before the outlet.
  • Baffle Stack or Cyclone Vane: The active separating element. Baffle plates force a 90-180° flow reversal; cyclone vanes spin the flow at 8-15 g of radial acceleration. Either way, the droplets cannot follow the streamline and impact a wetted surface.
  • Drain Pocket: The collection sump at the base of the vessel. Volume should hold at least 60 seconds of expected condensate flow at full load so a slow-acting trap does not let the water column rise into the separation zone.
  • Steam Trap: Float or inverted-bucket type, sized for 2-3× the expected condensate rate. The trap must discharge continuously under load — a thermostatic trap is the wrong choice here because it cycles and lets the pocket flood between discharges.
  • Inlet and Outlet Flanges: Sized to match the steam main, usually with the outlet one pipe size larger than the inlet on baffle units to drop velocity into the chamber. Mismatched sizing is the most common cause of poor separator performance on retrofits.

Who Uses the Steam Separator

You find Steam Separators anywhere wet steam would damage the downstream equipment or rob it of power. They sit immediately upstream of reciprocating engines, turbines, sterilisers, and process heaters, and they sit at the bottom of long vertical risers where condensate naturally accumulates.

  • Marine Steam: Fitted ahead of the throttle valve on the SS Shieldhall's triple-expansion engine to dry steam off the Scotch boiler before it reaches the HP cylinder.
  • Heritage Railway: Used on the Ffestiniog Railway's double Fairlies to strip moisture from the saturated steam line ahead of the regulator, reducing cylinder cock blowdown on cold starts.
  • Brewing: Installed before the wort copper jacket inlet at Hook Norton Brewery to ensure steam-quality dryness fraction stays above 0.97 during boil-up.
  • Power Generation: Cyclone moisture separators on the LP turbine crossover of CANDU reactors at Bruce Power, dropping steam wetness from around 12% back to under 1%.
  • Hospital Sterilisation: Spirax Sarco S series separators ahead of autoclave inlets in NHS sterile services departments, where HTM 2031 mandates dryness fraction ≥ 0.95 at the load.
  • Steam Wagon Operation: Fitted into the steam main between superheater and engine on preserved Sentinel DG and Foden wagons to handle priming when the boiler is rocked on uneven ground.

The Formula Behind the Steam Separator

The core question for a separator is how much water droplets actually leave the flow. The separation efficiency depends on droplet diameter, steam velocity, and the residence time inside the vessel. At the low end of the typical 15-25 m/s inlet velocity range, you get clean separation of droplets down to about 10 µm but the vessel is oversized for the flow. At nominal velocity around 20 m/s the unit hits its design sweet spot — droplets above 5 µm consistently leave the flow and the pressure drop stays under 0.2 bar. Push velocity past 30 m/s and re-entrainment kicks in: droplets that hit the baffle smear into a film, the film shears off the trailing edge, and you carry water out the outlet despite the separator nominally being in service.

η = 1 − exp(−(ρw × dp2 × v) / (18 × μs × L))

Variables

Symbol Meaning Unit (SI) Unit (Imperial)
η Separation efficiency (fraction of droplets removed) dimensionless dimensionless
ρw Water droplet density at saturation temperature kg/m³ lb/ft³
dp Droplet diameter (target cut size) m in
v Steam velocity through the separation zone m/s ft/s
μs Steam dynamic viscosity at line conditions Pa·s lb/(ft·s)
L Effective separation path length m ft

Worked Example: Steam Separator in a recommissioned heritage textile mill steam main

Sizing the droplet-removal performance of a recommissioned Spirax Sarco S2 baffle-type Steam Separator being fitted to the saturated steam main of a 1903 Lancashire boiler at the Queen Street Mill heritage textile site in Burnley, where the main feeds a 500 IHP Roberts cross-compound mill engine through a 200 mm bore steam pipe at 7 bar absolute, and the trustees want droplet-removal performance verified at slow Sunday demonstration running, nominal weaving-cadence load, and a brisk full-load showpiece burst before the public open day.

Given

  • ρw = 740 kg/m³
  • dp = 10 × 10<sup>−6</sup> m
  • μs = 1.6 × 10<sup>−5</sup> Pa·s
  • L = 0.30 m
  • vnom = 20 m/s
  • vlow = 10 m/s
  • vhigh = 30 m/s

Solution

Step 1 — compute the exponent argument at nominal 20 m/s, the design point for the S2 unit:

xnom = (740 × (10 × 10−6)2 × 20) / (18 × 1.6 × 10−5 × 0.30) = 1.48 / 8.64 × 10−5 ≈ 1.71

Step 2 — compute nominal separation efficiency for 10 µm droplets:

ηnom = 1 − exp(−1.71) = 1 − 0.181 = 0.819 → 81.9%

That is the design sweet spot. About 82% of the 10 µm droplet population leaves the flow on a single pass, and droplets above 20 µm are removed essentially completely. Dryness fraction at the engine throttle climbs from a typical wet-boiler value of 0.94 to roughly 0.985 — clean enough that you stop hearing the cylinder cocks spit water 10 seconds into the run.

Step 3 — at the low end of the operating range, slow Sunday demonstration running at 10 m/s:

ηlow = 1 − exp(−0.855) = 1 − 0.425 = 0.575 → 57.5%

Only 58% of the 10 µm droplets get caught. The mill engine still runs without thumping because total moisture flow is low at slow throttle, but the dryness fraction sits closer to 0.96 and you will see an oily mist at the chimney that you do not see at full load.

Step 4 — at the high end, the brisk showpiece burst at 30 m/s:

ηhigh,theoretical = 1 − exp(−2.57) = 1 − 0.077 = 0.923 → 92.3%

The formula predicts 92% removal, but in practice you do not get it. Above roughly 25 m/s the captured film shears off the trailing edge of the baffle and re-entrains. Measured efficiency on a Spirax S2 at 30 m/s typically drops back to about 75%, with pressure loss across the unit climbing past 0.3 bar — you can hear it in the change of tone at the regulator.

Result

Nominal separation efficiency works out at 82% for 10 µm droplets at the 20 m/s design velocity, lifting steam quality at the engine to a dryness fraction of about 0. 985. At slow Sunday running (10 m/s) efficiency drops to 58% but absolute moisture is low so the engine runs cleanly; at the 30 m/s burst the formula promises 92% but real-world re-entrainment pulls it back to around 75%, and the sweet spot clearly sits at the 18-22 m/s band the manufacturer rates for. If you measure carryover worse than predicted, three suspects come up first: a stuck or undersized float trap on the drain pocket letting the sump flood into the chamber, a partially blocked inlet strainer raising local velocity above the re-entrainment threshold, and a cracked or warped baffle plate that has opened a bypass path between the inlet and outlet flanges.

Steam Separator vs Alternatives

A Steam Separator is one of three common ways to deal with wet steam. The choice between a baffle separator, a cyclone separator, and a drip-leg-with-trap arrangement depends on flow rate, available pressure drop, droplet size you need to catch, and how much vertical space you have above the steam main.

Property Baffle Steam Separator Cyclone Steam Separator Drip Leg with Trap
Droplet cut size (smallest droplet reliably removed) ~10 µm at design velocity ~5 µm down to 2 µm at high g-loading ~50 µm — only catches gross slugs
Typical separation efficiency (10 µm droplets) 80-90% 95-99% 20-30%
Pressure drop at design flow 0.1-0.2 bar 0.2-0.5 bar <0.02 bar
Installed cost (DN100 line, indicative) £600-1,200 £1,500-3,500 £80-200
Service life before internal wear matters 20+ years 10-15 years (vane erosion) 30+ years (no internals)
Best application fit Engines, sterilisers, process heaters Turbine inlets, high-purity steam Long pipe runs, end-of-main drips
Sensitivity to off-design velocity Moderate — re-entrainment above 25 m/s High — needs tangential velocity in band Low — passive device

Frequently Asked Questions About Steam Separator

This is a velocity-dependent capture problem and it surprises a lot of operators. At low throttle the steam velocity through the separator drops below the threshold where droplet momentum is enough to leave the streamline — typically below 8-10 m/s for a baffle unit. The droplets simply follow the flow around the bend and out the outlet.

The fix is either to size the separator for the lowest expected flow rather than the highest, or to fit a smaller-bore separator on a bypass for low-load running. On heritage engines, the simpler answer is to crack the cylinder cocks for the first minute every time you start from cold or run at a creep, accepting that the separator is only really earning its keep above half throttle.

Three questions decide it. First, what droplet size do you need to catch? If 10 µm is fine — and it is for almost any reciprocating engine or process heater — a baffle unit is cheaper, simpler, and has no vanes to erode. If you are feeding a turbine and need to get below 5 µm, a cyclone is the only realistic choice.

Second, what pressure drop can you afford? A cyclone takes 0.3-0.5 bar at design flow. On a 7 bar line that is recoverable; on a 1.5 bar low-pressure heating line it kills you. Third, is the line flow steady or swinging? Cyclones lose efficiency hard when tangential velocity falls outside their design band, so a baffle handles wide-turndown applications better.

If the separator itself is healthy, the moisture is coming from downstream of it. The most common cause is an uninsulated or poorly-insulated steam main between the separator and the engine throttle. Steam re-condenses against the cold pipe wall and you generate fresh wet steam in the 10-20 metres after the separator outlet. Lag the line properly with at least 50 mm of mineral wool and recheck.

The second cause is a long horizontal run with no drip leg at the engine end. Condensate accumulates in the bottom of the pipe and gets picked up as the engine takes a heavy stroke. Fit a drip pocket and trap immediately upstream of the throttle valve.

Most baffle separators are designed for horizontal flow with a vertical drain — that is the standard orientation and the one the manufacturer's efficiency curves are based on. What you cannot do is invert the drain or mount the unit so the drain points sideways. The separated water has to fall under gravity into the trap, and if the drain is not at the lowest point the pocket fills and the chamber floods.

If you have a vertical riser and need to fit a separator in it, buy a unit specifically rated for vertical-flow service. Spirax Sarco, Armstrong, and TLV all make versions with internal geometry tuned for vertical mounting — do not try to repurpose a horizontal unit.

That is a thermostatic trap or a wrongly-sized float trap. Thermostatic traps cycle by design — they open when the condensate cools below saturation by 10-20°C and close again when hot condensate arrives. On a separator drain you want continuous discharge because the condensate is always at saturation temperature. Cycling lets the pocket fill between discharges and you carry water past the baffles on every cycle peak.

Replace it with a float-and-thermostatic trap or an inverted-bucket trap sized for at least 2× the worst-case condensate flow, and the discharge will go continuous. You will hear the difference at the trap outlet within minutes.

Yes — and the gain is bigger than most people expect. Wet steam loses two ways: latent heat is wasted re-evaporating the entrained water inside the cylinder, and the water itself does no expansion work but takes up volume. Going from a dryness fraction of 0.92 to 0.98 typically returns 4-6% on indicated power at the same throttle setting and steam rate.

On the Roberts cross-compound at Queen Street Mill, fitting a separator ahead of the throttle was worth roughly 25 IHP on a 500 IHP engine — measurable on the indicator card as a cleaner expansion line and a higher MEP in the HP cylinder.

References & Further Reading

  • Wikipedia contributors. Steam separator. Wikipedia

Building or designing a mechanism like this?

Explore the precision-engineered motion control hardware used by mechanical engineers, makers, and product designers.

← Back to Mechanisms Index
Share This Article
Tags: