Surface Steam Condensers

Because the steam keeps expanding and doing work until it reaches the condenser pressure, so every inch of mercury the condenser takes off the back pressure is extra enthalpy drop through the last stages. On a condensing turbine the usual field rule is that each 1 in HgA increase in back pressure costs somewhere around 2 to 3 percent of output, with the exact figure depending on the exhaust annulus and the steam conditions.

That means a unit designed for 2.0 in HgA that drifts to 3.5 in HgA in August is giving up several percent of shaft power for the same fuel, which on a 5,000 hp turbine-driven compressor is a cost you can put a dollar figure on by the hour. Cooling water temperature sets the floor, and air blanketing is what usually lifts the pressure above it.

The ejector or vacuum pump does the removing, but the condenser has to deliver the air to it in a form the equipment can handle, which is the part that gets overlooked. Air leaves the bundle mixed with steam, and the ratio matters: a well designed air cooling section at the vent subcools that mixture 5 to 10 F below the shell saturation temperature, which condenses most of the steam out of it and cuts the volume the ejector must move by half or more.

Skip that section and the ejector is sized for mostly steam, costing motive steam every hour the plant runs. A vent nozzle at the wrong end of the bundle defeats the air cooling section entirely, and shows up as an ejector that runs hot and a vacuum that never reaches design.

Mostly from the dozens of small connections below atmospheric pressure, and the way to find which ones is to measure before you hunt. A rate-of-rise test takes an hour: isolate the air removal equipment with the condenser under vacuum and time the shell pressure climb. Multiply the rise in inches of mercury per minute by the shell plus piping volume and you have the leak as a dry air flow, which tells you whether you are looking for one bad gland or a general case of loose flanges.

Then a helium sniffer or an ultrasonic detector walks the system while it runs. The glands on valves that are normally open are the usual suspects because nobody looks at them, and a turbine gland seal steam pressure that has drifted low will admit air faster than every flange combined.

The hotwell gives the condensate pump a stable suction and gives the control system a buffer, and it does both with a surprisingly small volume. A typical design holds somewhere between one and three minutes of full load condensate flow between the low and high level alarms, so a condenser handling 40,000 lb/hr of steam needs a hotwell of only about 80 to 240 gallons of working volume.

Two details are worth specifying. Condensate pumps in this service run with very little net positive suction head, since the liquid is at its saturation pressure, so the hotwell elevation above the pump centerline is a design number, not an afterthought. And condensate should leave the hotwell at or close to saturation; a drop of more than about 2 F below shell saturation, called condensate depression, means the condensate has been subcooled and has picked up dissolved oxygen on the way.

Yes, but the achievable back pressure rises by a noticeable margin, because an air-cooled condenser works against the dry bulb temperature rather than the wet bulb. On a humid summer afternoon the wet bulb can sit 15 to 20 F below the dry bulb, and that is exactly the margin a cooling tower loop enjoys. A water-cooled unit on tower water at 85 F will typically hold 2 to 2.5 in HgA; an air-cooled condenser on a 95 F design day is more likely to run at 4 to 6 in HgA, and the turbine gives up the output that difference represents.

The decision therefore depends on the value of the lost power against the cost of the water, its treatment and its tower. On a small turbine-driven pump the air-cooled answer is often right; on a generating set it is usually only right where there is no water to be had.

Design for the standard in-leakage allowance for the shell size, and then confirm it against a measured figure from any similar unit on the site, because the two frequently disagree. The standard allowances scale with condenser volume and the number of turbine exhaust connections; for a small industrial condenser serving a single turbine the design figure usually falls between 5 and 20 lb/hr of dry air, and a unit that actually measures 50 lb/hr after a few years of service is not unusual.

The point of the allowance is to size the running ejector stages with margin, not to describe a new tight system. A plant with a history of poor vacuum should design the air removal for twice the standard figure, since the cost of an oversized ejector is a little motive steam while the cost of an undersized one is turbine output every day.