Views: 0 Author: J-VALVES Publish Time: 2026-11-06 Origin: Site
Every steam trap in a plant is also a small steam generator. When high-pressure condensate passes through the trap orifice, part of it boils instantly, and that vapor - flash steam - is either a usable low-pressure steam supply or a plume above the vent stack. A flash vessel is the piece of equipment that turns the first outcome into the second: it separates the vapor from the condensate at a controlled pressure, so the steam can be routed to a real user instead of the atmosphere. The equipment itself is simple; the engineering decisions around it are not. This guide walks through the decisions in the order a specifier should follow: how much flash steam you actually have, at what pressure you should flash it, how big the vessel must be, where the steam goes, and what the system does to every trap upstream when you raise the return-line pressure. The failure section at the end covers the signatures that show up on site when one of those decisions was skipped.
Condensate leaving a steam trap at working pressure is saturated liquid: it carries exactly as much sensible heat as it can hold at that pressure. Throttling across the trap seat is essentially an isenthalpic process - total enthalpy is conserved while pressure falls. At the lower pressure the saturation temperature is lower, so the condensate now holds more heat than saturated liquid can contain, and the surplus evaporates a fraction of the water on the spot. That vapor is flash steam, also called secondary steam.
Two properties matter for recovery planning. First, flash steam is chemically identical to boiler steam and, because it comes from condensate that has already been treated and boiled, it is clean - there is no reason to treat it as waste water vapor. Second, it is generated everywhere condensate meets a pressure drop: at every trap, at condensate control valves, at blowdown lines, and inside the condensate return main itself wherever a high-pressure branch dumps into a lower-pressure collector. A plant does not decide whether to produce flash steam; it only decides whether to catch it. Vented flash steam is a direct fuel loss, because every kilogram that escapes the vent must be re-raised in the boiler from cold makeup water.
The flash fraction is an enthalpy balance across the pressure drop:
x = (hf1 - hf2) / hfg2
where hf1 is the saturated-liquid enthalpy at the high-pressure side, hf2 the saturated-liquid enthalpy at the low-pressure side (the vessel), and hfg2 the latent heat of evaporation at that low pressure. All three come from saturated steam tables. The numerator is the surplus sensible heat the condensate must shed to reach the new saturation temperature; the denominator is the heat each kilogram of flash steam absorbs in evaporating, read at the LOW pressure because that is where the boiling happens. Reading hfg at the upstream pressure is the classic table error - it changes the answer by only a percent or two, which is exactly why it survives a casual check.
The formula also shows the two levers that reduce flash when flash is unwanted: sub-cooling the condensate before the drop (a thermostatic trap with a deliberate cooling leg subtracts cp x delta-T from hf1), and narrowing the pressure gap. Typical industrial flash fractions run 5 to 20 percent by mass for common pressure drops. That sounds small, and it is the most misleading number in condensate system design, as the volume arithmetic below shows.
The table below gives the flash fraction for the three most common trap discharge pressures flashing into a 2 bar g recovery vessel or to atmosphere. Values are computed from saturated steam tables (hf in kJ/kg at 10 bar g = 781, 7 bar g = 721, 4 bar g = 670; at 2 bar g hf = 561 and hfg = 2164; at atmospheric hf = 419 and hfg = 2257). Exact figures vary slightly between steam table editions; the pattern does not.
Condensate source pressure | Flash to 2 bar g vessel | Flash to atmospheric tank |
|---|---|---|
10 bar g (184 deg C saturated) | 10.2 percent | 16.1 percent |
7 bar g (170 deg C saturated) | 7.4 percent | 13.4 percent |
4 bar g (152 deg C saturated) | 5.0 percent | 11.1 percent |
Published vendor data agrees with this pattern: flash tables in standard steam handbooks show about 12 to 16 percent flashing to atmosphere from high-pressure condensate and correspondingly less into a pressurized vessel. The visible consequence: a plant returning 5 t/h of 10 bar g condensate to an unpressurized tank sends roughly 800 kg/h of steam out of the vent - steam the boiler must replace kilogram for kilogram.
Take a plant with 5,000 kg/h of saturated condensate returning from 10 bar g traps, and a flash vessel operating at 2 bar g. From the table, the flash fraction is 10.2 percent:
Quantity | Value | How it is obtained |
|---|---|---|
Flash steam generated | About 510 kg/h | 5,000 x 0.102, enthalpy balance |
Residual condensate to pump out | About 4,490 kg/h at 144 deg C | Mass balance; leaves at 2 bar g saturation temperature |
Flash steam volume flow | About 310 m3/h | 510 kg/h x 0.606 m3/kg at 2 bar g saturated |
Residual condensate volume flow | About 4.9 m3/h | 4,490 kg/h at roughly 926 kg/m3 |
Steam share of total volume | About 98.5 percent | 310 / (310 + 4.9) |
The last row is the point of the whole calculation: 10 percent by mass is 98.5 percent by volume. Low-pressure steam is roughly 600 times less dense than hot water, so a mass fraction that looks negligible dominates every volumetric decision - vessel diameter, vent line size, return-main capacity, and pump selection. This is also why a condensate return main sized on liquid flow alone hydraulically chokes once flash is present: the two-phase volume in the line can be two orders of magnitude larger than the design liquid flow. Sizing condensate lines for flash is its own discipline, covered in any condensate return reference; the flash vessel exists precisely to strip the vapor out before the remaining liquid is pumped.
As a cross-check against published data, a standard process steam reference works the same problem in imperial units: 3,000 lb/h of condensate from 160 psig traps flashing into a 20 psig vessel yields 12.4 percent flash, or 372 lb/h of flash steam - the same arithmetic at the same pressure ratio, from an independent source.
The most common design error is treating the flash pressure as something to minimize. It is the opposite: the flash pressure should be set by the low-pressure steam user the flash steam will feed, plus a small margin for line losses. Flashing to a higher pressure produces less flash steam by the enthalpy balance, but every kilogram produced is steam that genuinely displaces boiler output at a usable pressure. Flashing to atmosphere maximizes the flash fraction and then usually wastes all of it, because a plant with no low-pressure consumer cannot use atmospheric steam for anything except makeup water heating.
The practical sequence for choosing the flash pressure:
Identify the low-pressure users first: a deaerator, a feedwater tank preheating coil, a space-heating header, a low-pressure process user such as a dilution heater or tracer header.
Set the vessel operating pressure slightly above the pressure the user's steam space actually needs, so flow is always from vessel to user.
Check the trap differential consequence before committing (next section) - this is where most recovery projects create problems they did not anticipate.
If no pressurized user exists, flash to a vented tank and recover the heat in a vent condenser heating makeup water; that is still far better than a live steam plume.
Pressure control between vessel and header follows the same logic as any steam reducing station: when the flash vessel must feed a header whose pressure varies, a self-acting or pilot-controlled pressure regulating valve holds the vessel pressure steady, and a non-return valve prevents backflow from the header into the vessel when header pressure rises above vessel pressure. The valve sizing arithmetic for that duty follows the same critical-flow and Kv methods used for any valve Cv/Kv sizing calculation.
The vessel has two jobs that pull its diameter in different directions: separate flash steam from condensate without entraining droplets, and hold enough liquid to give the drain system a stable level. Vapor-side sizing starts from a droplet-settling criterion - the Souders-Brown relationship:
v_max = k x sqrt((rho_L - rho_V) / rho_V)
where rho_L and rho_V are liquid and vapor densities at vessel conditions and k is an empirical constant. Published practice for vertical flash tanks puts k around 0.08 to 0.15 m/s without a demister, rising toward 0.3 m/s-class values where a mesh pad is fitted and the service is clean; simple vendor charts land in the same place, corresponding to allowable vapor velocities of roughly 2 to 3 m/s for typical steam flash conditions. Continuing the worked example at 2 bar g (vapor density about 1.65 kg/m3, liquid about 926 kg/m3):
Step | Result |
|---|---|
Allowable velocity, k = 0.10 m/s: 0.10 x sqrt((926 - 1.65)/1.65) | About 2.4 m/s |
Cross-section for 310 m3/h (0.086 m3/s) of flash steam | About 0.036 m2, i.e. only about 215 mm diameter |
Liquid hold-up, 5 minutes of 4,490 kg/h at 926 kg/m3 | About 0.40 m3 between normal liquid level and bottom |
Vessel volume with level at half-full | About 0.8 m3 minimum |
Governing geometry at L/D about 2:1 | Roughly 750-800 mm diameter x 1.6-1.8 m high |
The teaching point is in the third row: at low flash rates the velocity criterion alone produces a vessel hardly bigger than the inlet pipe, and it is the liquid hold-up and the L/D proportion that govern the real diameter. Published flash-vessel guidance converges on: vertical orientation, normal liquid level at about half volume, liquid hold-up of roughly 3 to 5 minutes (some references use up to 10 for upset tolerance), height-to-diameter between about 2:1 and 3:1, and final size rounded up to a standard vessel diameter. Confirm the final selection against the vendor's own sizing chart for the actual pressure pair - chart practice differs between manufacturers, and a vessel supplied against the wrong flash pressure is undersized for carryover even when it passes a casual volume check.
Geometry inside the vessel does most of the separation work, and four details decide whether it does it cleanly:
Inlet above the liquid level, arranged to kill momentum: a tangential entry or an impingement baffle opposite the nozzle. Condensate arriving from trap discharges can enter as a two-phase jet at 15 to 45 m/s; let that jet hit the liquid surface and it entrains droplets straight into the vapor outlet.
A demister mesh pad at the vapor outlet for clean condensate service, removing entrained droplets down to the tens-of-microns class. In fouling or oily condensate, a vane pack is preferred because mesh pads blind.
A dip tube or internal elbow on the liquid outlet with a siphon-breaker hole, so the draining liquid cannot pull a siphon and drag vapor out with it.
Level control on the liquid outlet - a level transmitter and control valve holding the normal level at half volume, with high and low alarms. A flash vessel with no level control eventually passes liquid up the steam line or empties and passes steam down the condensate line; both failures surface somewhere far away from the vessel.
One more fitting belongs on the list: a vacuum breaker. The vessel sits at 144 deg C full of saturated steam. On a plant trip the steam space condenses, pressure collapses below atmospheric, and an unrated tank designed only for modest internal pressure sees a full vacuum load from the outside. Vacuum protection is a standard line item on flash vessel specifications for exactly this reason.
Flash steam is only worth what the receiving user is worth. The routing table below covers the standard destinations:
Destination | When it works | Watch-outs |
|---|---|---|
Low-pressure process header | A stable LP consumer exists near the condensate collection point and accepts steam at the vessel pressure | Needs pressure control and a non-return valve; vessel pressure must always exceed header pressure or flow reverses |
Deaerator steam space | The default destination - deaerators run at low pressure and absorb flash steam continuously; flash displacement cuts live LP steam taken by the deaerator one for one | When flash supply exceeds deaerator demand the surplus must vent; a pressure-maintaining arrangement keeps the vessel from drifting up |
Feedwater tank preheating / makeup water heater | No LP header available; atmospheric or low-pressure vessel with the steam condensed against cold makeup | Heat is recovered, steam is not; fine where no pressurized user exists |
Space heating / HVAC header | Seasonal match is acceptable | Summer demand may vanish - provide a fallback vent condenser or dump route |
Vent condenser on atmospheric tank | Last resort when nothing else can take the steam | Still recovers most of the latent heat into makeup water instead of the sky |
Wherever the destination is a header, the interface looks like a miniature of a full reducing station: regulating valve, non-return valve, and a defined pressure relationship. The station-architecture logic - parallel paths for wide turndown, sensing-line practice, and set-pressure discipline - carries over directly from steam pressure reducing valve station design.
Installing a pressurized flash vessel raises the pressure in everything upstream of it: the collection main, and then every trap discharge line connected to it. Each of those traps now works against a higher back pressure, and three consequences follow mechanically from that single change.
First, capacity: every trap passes less at the reduced differential, and traps that were sized with a thin margin may no longer clear their load. Second, stall: on any modulating heat exchanger draining through the collection system, the steam-space pressure at partial load may fall below the new return pressure, condensate stops flowing, and the exchanger floods at exactly the loads the process runs at most. No trap family cures stall - the remedies are pump-traps or pumped condensate, exactly as in steam trap selection and stall diagnosis, and the stall check must be redone with the flash vessel's operating pressure substituted for the old return pressure. Third, thermodynamic disc traps that tolerated the old back pressure may lock closed at the new one; their published minimum operating pressure is a differential, not a gauge reading.
The return main between traps and vessel also changes character. With flash present the line runs two-phase at much higher volume flow, and slugging condensate in that regime produces the same impact events covered in check valve water hammer and non-slam selection - risers immediately at trap outlets and undersized collection mains are where the hammer concentrates. The honest project sequence is therefore: model the new return pressure, re-verify every significant trap's differential and stall margin, and only then order the vessel.
A flash vessel operating above atmospheric pressure is a pressure vessel in every code sense - ASME Section VIII or the local equivalent governs its design, materials, welding, examination and registration, and it takes a code-stamped nameplate like any other item in the plant. Two boundary items deserve spec-level attention.
Overpressure protection: a vessel that can be isolated, or that receives flash steam through a control valve that can fail open, needs a safety relief valve sized for the largest credible vapor generation - maximum condensate flow at the minimum vessel pressure, plus any contribution from the failure case. The relief-valve selection method, back-pressure corrections and nameplate verification follow the same practice as any steam-side PSV, covered in pressure safety valve selection per API 520/526; the trap here is remembering that the relieving medium is low-pressure steam at high specific volume, so the required orifice is generous for the mass flow involved. Even a nominally atmospheric flash tank needs its vent sized for the maximum flash generation at a tolerable back pressure - published practice keeps vent velocities in the 15 to 20 m/s class so the tank never quietly pressurizes.
Instrumentation: at minimum a pressure gauge, a level indication (gauge glass or transmitter) visible at the vessel, temperature indication, and the high/low level alarms tied into the drain valve logic. A flash vessel without a level indication is a black box that fails silently, and the failure surfaces as wet steam or flooded equipment somewhere else.
Flash vessels fail in patterns, and the pattern points at the design step that was skipped:
Field signature | Most likely cause | Verification and fix |
|---|---|---|
Continuous steam plume from the vent | Vessel pressure set below a usable level, or flash generation exceeds the receiving user's demand | Check vessel pressure against design; add a pressure-maintaining route or a real LP user; a plume is unrecovered revenue |
Wet steam or water carryover into the LP header | Vapor velocity above the separation criterion - undersized vessel or wrong flash pressure | Recompute Souders-Brown at actual pressure and flow; check demister condition; retrofit a larger vessel or vane pack |
Upstream equipment runs cold; traps flood after the recovery project | Return pressure raised by the vessel; traps lost differential; stall on modulating duties | Measure trap inlet and discharge pressures; redo stall checks; fit pump-traps on stalled exchangers |
Hammering in the collection main near the vessel | Two-phase slugging from undersized collection lines or lift legs after traps | Inspect riser geometry; resize the worst main sections; drain by gravity to the vessel where possible |
Vessel level swings; drain valve hunts | Level control poorly tuned, or siphon breaker missing on the drain dip tube | Retune the loop; verify the dip-tube siphon hole; check pump NPSH if condensate is pumped out |
Vessel or pipework deformed after a plant trip | Vacuum collapse - no vacuum breaker, or vent line unable to admit air on shutdown | Fit a vacuum breaker sized for the condensation rate; inspect for damage before restart |
The first signature deserves emphasis in financial terms: a vent plume on a vessel receiving 5 t/h of 10 bar g condensate is on the order of 800 kg/h of live steam, and at typical industrial steam costs that runs to five figures per year per vent. The vessel pays for itself by eliminating that plume - provided the flash steam is routed somewhere real.
State the design duty explicitly: maximum and minimum condensate flow (kg/h), source pressure, and target flash pressure - not just a pipe size and "flash tank".
Require the flash fraction calculation and the resulting flash steam flow in the vendor offer, with the steam table state points shown.
Specify the vessel code (ASME VIII U-stamp, PED, or local equivalent), design pressure and temperature, and corrosion allowance.
Require the sizing basis: allowable vapor velocity or k value used, demister fitted or not, and the carryover claim behind it.
Specify liquid hold-up (minutes at normal level) and normal liquid level position - half volume is the default.
Specify internals: inlet arrangement (tangential or baffled), demister or vane pack material for the condensate chemistry, dip tube with siphon breaker.
Include level control scope: transmitter, control valve on the drain, high/low alarms - or state clearly that plant instrumentation covers it.
Include the safety relief valve sized for maximum vapor generation with its set pressure and code certification, plus the vacuum breaker.
Define the flash steam interface: connection size, a pressure control and non-return valve where the vessel feeds a variable header, and the routing philosophy.
Attach the system study: recomputed trap differentials and stall checks for every significant upstream drain, so the recovery project does not arrive with its own hidden list of corrective work.
In summary: flash steam recovery is arithmetic before it is hardware. Compute the fraction from the enthalpy balance, set the vessel pressure from the user that will receive the steam, size the vessel for droplet separation and liquid hold-up rather than the pipe it connects to, and re-verify every trap upstream against the new back pressure. Do those four things and the vessel is a quiet, permanent fuel saving; skip any of them and the plant will tell you - in plumes, wet steam or flooded heat exchangers - which one it was.
How much flash steam will my condensate produce?
Compute the flash fraction from the enthalpy balance across the pressure drop: x = (hf1 - hf2) / hfg2, with hf read as saturated liquid at the high and low pressures and hfg read at the LOW pressure, where the boiling actually happens. Typical fractions run 5 to 20 percent by mass for common industrial drops. Worked figures from saturated steam tables: 10 bar g condensate flashes 10.2 percent into a 2 bar g vessel and 16.1 percent to atmosphere; 7 bar g gives 7.4 and 13.4 percent; 4 bar g gives 5.0 and 11.1 percent. Sub-cooled condensate - for example from a thermostatic trap with a cooling leg - flashes less, because the surplus sensible heat has already been removed upstream. Remember the volume consequence: a 10 percent mass fraction is roughly 98 percent of the two-phase volume, because steam at 2 bar g occupies about 0.6 m3/kg against roughly 0.001 for hot water.
What pressure should the flash vessel operate at?
Set it from the low-pressure user the flash steam will feed, plus a small margin for line losses - never simply as low as possible. Flashing to a higher pressure yields less flash steam by the enthalpy balance, but every kilogram produced is boiler-quality steam at a pressure a real consumer can use, so it displaces live steam one for one. Flashing to atmosphere maximizes the fraction and then usually wastes it, unless a vent condenser recovers the latent heat into makeup water. The practical order: identify the deaerator, LP process header or feedwater heating duty first, set the vessel slightly above the pressure that user needs, then check what the new back pressure does to every upstream trap before committing.
Can flash steam be piped directly into a low-pressure steam header?
Yes, provided the vessel operating pressure always exceeds the header pressure, so flow can never reverse. The interface is a miniature reducing station: a self-acting or pilot-controlled pressure regulating valve holds the vessel pressure steady, and a non-return valve protects the vessel when header pressure rises above it. Watch turndown: when flash generation exceeds the header's demand, the surplus must have a defined path - a pressure-maintaining vent or a secondary user - otherwise the vessel pressure drifts up and raises the return-line back pressure on every trap connected to it, which is exactly the failure that floods upstream equipment.
Why did our steam traps start flooding after the flash recovery system was commissioned?
Because the flash vessel raised the pressure in the collection main, and every upstream trap now works against a higher back pressure at lower differential. Three mechanical consequences follow: reduced trap capacity, thermodynamic disc traps that can lock closed at the new differential, and stall on modulating heat exchangers - at partial load the steam-space pressure falls below the new return pressure and condensate stops draining. No trap family cures stall; the remedies are pump-traps or gravity drainage to a receiver below the equipment. The correct project sequence is to recompute each significant trap's differential and stall margin with the vessel operating pressure substituted for the old return pressure, before the vessel is ordered.
Does a flash vessel need a safety valve if it is just a vented tank?
A vessel that operates above atmospheric pressure and can be isolated, or that receives flash through a control valve that can fail open, needs a code-certified safety relief valve sized for the largest credible vapor generation - maximum condensate flow at minimum vessel pressure plus the failure case. The relieving medium is low-pressure steam at high specific volume, so the required orifice is generous for the mass flow involved. Even a nominally atmospheric tank still needs two items: a vent sized for maximum flash generation at a tolerable back pressure (published practice keeps vent velocities around 15-20 m/s so the tank never quietly pressurizes), and a vacuum breaker, because a steam-filled vessel at 140-plus deg C will pull a full vacuum on shutdown as the vapor space condenses.
