What a Flash Tank Is Recovering

Why this matters

A flash tank looks like a small receiver and gets treated like one, which is how they end up flooding, carrying water into a low-pressure header, and being blamed for hammer that started at their own drain trap. It is not a receiver. It is a separator, its whole job is to catch vapor that was going to form anyway, and the quantity that sizes the vessel is a vapor rate rather than a condensate rate. Size one on the condensate flowing through it and you will build a vessel that is correct for a fluid that occupies less than two percent of what is moving through it.

Isolating one before any work

A flash tank is a pressure vessel holding saturated liquid at its operating pressure, which means the liquid in it flashes the instant that pressure is released.

To open one: close the condensate inlet and the vapor outlet, lock and tag both under your written energy control procedure per 29 CFR 1910.147, drain through the vessel drain to a piped discharge directed away from every walkway, confirm zero at a gauge on the vessel itself rather than assuming the drain finished the job, and let the shell cool before breaking any joint. Where the tank sits in a below-grade room or pit, that space is a permit-required confined space until evaluated under 29 CFR 1910.146.

Never valve, plug or gag the relief device, and never close the vapor outlet while condensate is still being admitted. Closing that outlet with flow coming in is an instruction that puts the vessel on a path toward the source pressure, which is the one thing the tank is not built for. The vessel itself is built and inspected to the ASME Boiler and Pressure Vessel Code, Section VIII, in the edition your state or local boiler authority has adopted, and it binds you through that jurisdiction's registration and inspection requirements rather than on its own authority.

What the tank is, and what it is not

Condensate leaving a high-pressure trap is saturated liquid. Drop its pressure and a fraction of it re-boils, because the excess sensible heat it carried at the higher boiling point has nowhere else to go. That mechanism and the enthalpy balance behind it belong to the article on what flash steam is, and are used here rather than re-derived.

Vent that flash to atmosphere and it is gone, mass and heat together. A flash tank does one thing: it drops the pressure to an intermediate value instead of all the way to the return, holds the mixture still long enough for the vapor to separate from the liquid, and pipes the vapor to a header that can use it. The liquid then continues to the return through a drain trap.

So it is not storage, and it is not a receiver with a fancy name. It is a controlled pressure step with a separation volume around it.

The recovery, in one relationship

Flash fraction = (liquid enthalpy at the high pressure - liquid enthalpy at the low pressure) / latent heat at the low pressure, with all three values from steam tables for the actual pressures.

Representative table values for the range these systems run:

Saturated liquid runs about 0.017 ft3/lb across all of it.

Two directions follow from those numbers and they pull against each other. Staging at a higher intermediate pressure recovers a smaller fraction, because the numerator of the balance shrinks. But it recovers it at a more useful pressure and in half the volume, because vapor specific volume falls as pressure rises. The tank pressure is therefore chosen by what the low-pressure header actually needs, not by maximizing the fraction.

The vessel is sized by the vapor

The separation happens because the vapor rises through the tank slowly enough that entrained droplets fall out of it. That makes the governing quantity the upward vapor velocity, which is the vapor's volumetric rate at tank pressure divided by the tank's free cross-sectional area. Diameter sets that velocity. Height and inlet position give the liquid a quiet zone below it.

The acceptable velocity comes from the vessel manufacturer's separation criteria for that design, not from a rule of thumb, because it depends on droplet size and on whether the tank carries internals.

      vapor out, to low pressure header
                  |
            +-----------+
 high       |  vapor    |
 pressure   |  space    |
 condensate |           |
 in ------->|~~~~~~~~~~~|  liquid level
            |  liquid   |
            +-----+-----+
                  |
            drain trap sized on tank
            minus return pressure

The inlet enters tangentially and above the liquid level so the incoming mixture spins against the shell and throws liquid outward rather than firing it straight up into the vapor space. An inlet piped into the liquid or aimed at the outlet defeats the separation no matter how the diameter was calculated.

The drain trap is sized on the tank's differential, not the source's

This is the single most common installed fault, and it is an arithmetic mistake rather than a hardware one.

The trap on the tank drain sees tank pressure minus return pressure. It does not see source pressure. A trap selected from a capacity table read at the source differential appears to have far more capacity than it will actually deliver.

For liquid through a fixed orifice, rate scales roughly with the square root of the differential, so an error in the differential shows up as roughly its square root in capacity. That is the incompressible approximation and it is a floor on the error, not the whole of it, because flash forming inside the orifice reduces real capacity further.

Get it wrong in the undersizing direction and the tank floods, liquid carries over into the low-pressure header, and you get hammer in a main that has no obvious water source. Get the drain trap failing open instead and the tank's vapor blows straight into the condensate return, pressurizing it and stalling every trap that drains into that return. Both ends of that range are ways one small trap takes down a system it is not obviously connected to.

Worked example: 1,000 lb/hr staged at 15 psig

The service is 1,000 lb/hr of condensate leaving 100 psig traps. The plant has a 15 psig header with a continuous load. The return runs at 2 psig.

The recovery. Flash fraction from 100 psig to 15 psig is (309 minus 218) divided by 946, which is 9.6 percent. That is 96 lb/hr of 15 psig steam, and 904 lb/hr of liquid continuing to the drain.

The volume, which is what sizes the vessel. The vapor is 96 lb/hr at about 13.7 ft3/lb, which is about 1,315 ft3/hr, or 21.9 ft3 per minute. The liquid is 904 lb/hr at about 0.017 ft3/lb, which is about 15.4 ft3/hr, or 0.26 ft3 per minute.

The vapor occupies about 85 times the volume of the liquid, and it is the vapor rate that has to be divided by the tank's free area to land under the manufacturer's separation velocity. A tank sized on 0.26 ft3 per minute of water would be a pipe.

What the liquid still carries. That 904 lb/hr is saturated at 15 psig, so it flashes again on the way to a 2 psig return. Against atmospheric, the second-stage fraction is (218 minus 180) divided by 970, which is 3.9 percent, or about 35 lb/hr.

Check the staging closes. Recovered plus second-stage flash is 96 plus 35, which is 131 lb/hr, or 13.1 percent of the original 1,000. Dropping the same condensate straight from 100 psig to atmospheric gives 13.3 percent. Those agree to within 0.2 of a percentage point, and that gap is rounding in the two fractions rather than a physical difference. The tank did not create vapor. It captured about three quarters of the vapor that was going to form anyway, at a pressure where something can use it, and let the remaining quarter go to the return as before.

Now size the drain trap correctly. Tank at 15 psig, return at 2 psig, so the differential is 13 psi. Sizing it against the source instead gives 100 minus 2, which is 98 psi, and 98 divided by 13 is about 7.5 times the real differential. Square root of 7.5 is about 2.7, so a trap picked that way is overstated in capacity by at least a factor of about 2.7 and will not pass 904 lb/hr at the 13 psi it actually sees.

What that failure looks like in the field. The tank level climbs, the liquid reaches the tangential inlet region, separation stops working, and the 15 psig header starts receiving water. The complaint arrives as hammer in the low-pressure main, which sends a tech to the low-pressure main, which is not where the fault is. The tell is the tank sight glass, and it is the first thing to look at on any hammer complaint downstream of a flash tank.

When a flash tank is the wrong answer

• No continuous low-pressure load. If the header's demand is intermittent and smaller than 96 lb/hr in the example above, the recovered vapor relieves or vents anyway and you have built a vent with a vessel registration attached to it.
• The two pressures are close together. A small drop gives a small numerator and a small fraction. Run the balance for your actual pair before assuming there is anything to recover.
• The condensate load is spiky rather than steady. Separation velocity is set by peak vapor rate, not average, so a tank sized on average load floods on every peak.
• The return already runs at elevated pressure. Staging at 15 psig into a return at 12 psig leaves almost no second-stage drop and very little differential for the drain trap, which is the undersizing case above arriving by a different route.

How to verify an installed one is doing its job

• Watch the sight glass under load, from the side and at arm's length, with eye and face protection and the glass guard in place. A gauge glass on a saturated vessel fails by releasing flashing water at eye level. A stable level with the drain trap cycling is the working condition. A level that climbs and does not come back is the drain trap, not the tank.
• Read the tank gauge against the header it feeds. A tank sitting at the return pressure rather than at its design pressure is not staging anything, and the usual cause is the vapor line being undersized or valved.
• Check the drain trap differential on paper against the capacity table, at tank pressure minus return pressure. This is a desk check and it finds the most common installed fault without touching anything.
• Confirm the relief device is unobstructed and in its inspection interval under the boiler authority that adopted the vessel code for your jurisdiction.
• Trace the vapor line's destination and confirm it has continuous load. A recovery line into a header that is dark half the year recovers nothing for that half.

References

• ASME Boiler and Pressure Vessel Code, Section VIII, in the edition adopted by your state or local boiler authority, which binds through that jurisdiction's vessel registration and inspection
• 29 CFR 1910.147, control of hazardous energy, for isolating and verifying a vessel before opening it
• 29 CFR 1910.146, permit-required confined spaces, where the tank is in a below-grade room or pit
• Steam tables for liquid enthalpy, latent heat and vapor specific volume at the actual pressures, and vessel manufacturer separation criteria for the acceptable upward vapor velocity
• See related: What Flash Steam Is and Why It Looks Like a Failure; Why Condensate Return Lines Fail From the Inside