What a Boiler Does to the Water in It

Why this matters

A boiler does something no other piece of equipment on a service route does: it splits one water into two waters with opposite chemistry. What leaves as steam is nearly pure. What stays in the drum gets more concentrated every minute the burner runs. Almost every chronic steam-side complaint, a blowdown rate that keeps creeping up, returns that keep pinholing, carryover that wets the steam, is one of those two streams doing exactly what physics says it will. The case below is a building where the two streams were feeding each other's problem for two heating seasons and every individual repair was correct.

Before you go near a steam boiler

• Never open a boiler that is under pressure. Verify zero at the gauge and confirm it by cracking a vent to atmosphere before breaking any joint. A gauge can lie; a vent that produces nothing is a second opinion.
• If the boiler has lost its water level and is hot, secure the fuel and let it cool. Do not add water. Introducing water onto overheated metal is the mechanism behind catastrophic failures of this equipment, and there is no version of that call that is worth making quickly.
• A low-water cutoff or a pressure control that tripped is evidence, not the fault. Establish why it opened before you replace it. Replacing a correctly operating protective device reaches the same end state as jumpering it, one step slower and with a part number on the invoice.
• Manual blowdown has a valve sequence. Open the quick-opening valve first, then throttle with the slow-opening valve, and close in the reverse order, so the quick-opening valve never does the throttling and is never left as the only thing holding back the drum. Confirm the discharge is piped to a blowdown separator or an approved receiver before you touch either one, and confirm the discharge temperature limit your adopted plumbing code sets for a sanitary sewer with the authority having jurisdiction.
• Treatment chemicals here are usually a caustic-class product and an amine. Take the glove class and any respiratory control from each product's safety data sheet, keep them physically separate from any oxidising product, and dilute into water rather than adding water to concentrate.

The signal

A low-pressure steam boiler serving a school. The service contractor's log showed continuous blowdown increased twice over two heating seasons, both times correctly, to hold boiler water inside its dissolved solids limit. The makeup water meter, which nobody had been reading, showed makeup roughly double what it had been. Two sections of buried return line had been repaired for pinholes, both correctly. Nobody had connected the three.

The boiler as a still

Steam carries almost nothing but water. Everything dissolved in the feedwater, hardness, alkalinity, silica, chloride, sulfate, stays in the drum, so the drum concentrates continuously and blowdown is the only way anything leaves. That gives the boiler form of cycles of concentration: boiler water dissolved solids divided by feedwater dissolved solids. Rearranged for the number you actually control, blowdown as a fraction of feedwater is one divided by cycles.

Two things set feedwater quality, and only one of them is the water utility. Feedwater is condensate return plus makeup, and condensate is essentially distilled, so feedwater concentration is makeup concentration multiplied by the makeup fraction. Lose return and you have raised feedwater concentration without anything changing in the street.

The gases leave with the steam

The second stream is where the returns come from. Bicarbonate alkalinity in the feedwater does not survive boiler temperature. It breaks down to carbonate and then releases carbon dioxide, and carbon dioxide is volatile, so it leaves the drum with the steam, travels the distribution, and dissolves into the condensate as it forms. Dissolved carbon dioxide in water is carbonic acid, and condensate has essentially no hardness and no buffering to resist it, so the pH of untreated condensate falls and the return line corrodes from the inside.

The classic conversion used in boiler work puts the carbon dioxide released somewhere between roughly 0.4 and 0.8 mg/L per mg/L of feedwater alkalinity as calcium carbonate, depending on how completely the bicarbonate breaks down at that drum pressure and temperature. Take the factor your treatment supplier uses for your pressure rather than picking one, because the spread between those two ends is a factor of two.

Dissolved oxygen is the other gas, and it moves the opposite way from what people expect. Oxygen solubility in water falls as temperature rises, which is exactly what a deaerator exploits: heat the feedwater and scrub it with steam and most of the oxygen comes out before it ever reaches the drum. State the other end so the direction is established rather than asserted: cold makeup fed straight to a hot boiler arrives carrying its full dissolved oxygen load, and that oxygen finds the hottest metal in the building.

Eliminating the obvious explanations

The blowdown increases were wrong. They were not. Boiler water was measured, it was at the limit, and continuous blowdown was raised to hold it. Every one of those decisions was correct on the information available at the time.

The utility's water changed. Makeup dissolved solids measured 400 ppm, matching the sample from two seasons earlier. Killed.

The return line was simply old. Plausible, and it is the reading that produced two correct repairs and no change in trajectory. The problem with it is that it predicts pinholes at a steady rate; what the log showed was an accelerating one, and age does not accelerate.

The loop that was closing on itself

Condensate return had fallen from about 70 percent to about 45 percent, mostly through the buried line and a receiver vent nobody had looked at. Follow what that does to both streams.

Drum side. At 70 percent return, the makeup fraction is 30 percent, so feedwater dissolved solids are 0.30 times 400, which is 120 ppm. Against this boiler's own dissolved solids limit of 3,000 ppm, taken from its manual and the ASME guidance table for its drum pressure and not from any general figure, cycles are 3,000 divided by 120, which is 25, and blowdown is one divided by 25, or 4.0 percent of feedwater.

At 45 percent return, the makeup fraction is 55 percent, so feedwater is 0.55 times 400, which is 220 ppm. Cycles are 3,000 divided by 220, which is 13.6, and blowdown is 7.3 percent of feedwater. Blowdown rose by a factor of 7.3 divided by 4.0, which is 1.83.

That is the mass of water heated to saturation temperature and then discharged rising by 83 percent. It is a share of fuel input, not all of it, and how large a share depends on how much of the boiler's output comes back as condensate, which is the same variable that caused the problem.

Steam side. Makeup alkalinity measured 180 mg/L as calcium carbonate. At 70 percent return, feedwater alkalinity is 0.30 times 180, or 54 mg/L. At 45 percent return it is 0.55 times 180, or 99 mg/L. The ratio is 99 divided by 54, which is 1.83, identical to the blowdown ratio, and that is not a coincidence: both scale with the makeup fraction, because condensate contributes neither solids nor alkalinity.

Convert that to carbon dioxide with the range above. At 54 mg/L of feedwater alkalinity, the condensate sees roughly 22 to 43 mg/L of carbon dioxide. At 99 mg/L, roughly 40 to 79. Whichever end of the factor is right for this pressure, the returns got 83 percent more acid gas.

So the return line leaks, which lowers return, which raises feedwater alkalinity, which raises carbon dioxide in the steam, which attacks the return line faster, which produces the next leak sooner. Both repairs were correct and neither one touched the mechanism. Name the other end of it too: a tight system with high return runs low feedwater alkalinity and low carbon dioxide, so its returns corrode slowly, which keeps return high. This mechanism is self-reinforcing in whichever direction it is already going, which is why the middle of it is unstable and why a return-rate trend matters more than any single reading.

What would have flipped the conclusion

• Return holding steady while blowdown rose. That moves the case to makeup quality or to a boiler water limit that was set too low for the pressure.
• Carryover symptoms rather than corrosion symptoms. High drum solids, high alkalinity or oil contamination produce foaming and wet steam, which sends solids into the distribution and hammers traps. Same root cause, entirely different presentation, and the tell is water in places steam should be rather than metal loss.
• Oxygen pitting rather than general thinning. Carbonic acid attack thins the return along the path the liquid takes; localised deep pits point at air in-leakage instead, usually through a vacuum drawn on a cooling system at shutdown, and that is a different repair.
• A makeup meter that had not moved. Without that reading, the accelerating pinhole rate is just bad luck. With it, the case makes itself.

How to verify the return fraction you are reasoning from

• Meter makeup. It is the single most useful instrument on a steam system and most buildings do not have one. Log it weekly and reason from the trend, not the absolute.
• Compare feedwater conductivity against makeup conductivity. The ratio between them gives the makeup fraction directly and does not require a meter, provided you sample both on the same visit from fixed points.
• Measure condensate pH and iron at the receiver. Iron in the condensate is metal that used to be pipe, and it quantifies the attack rather than describing it.
• Recompute cycles from the boiler water sample rather than trusting a setpoint. Boiler water solids divided by feedwater solids is two tests and it tells you whether the continuous blowdown valve is passing what someone thinks it is.
• Sample hot boiler water through a sample cooler, never into an open container, and give it time to reach the cooler's steady temperature before you draw the sample.

References

• The boiler manufacturer's water quality limits and the ASME guidance table for the drum pressure in service
• Water treatment supplier documentation for the carbon dioxide conversion factor and condensate treatment appropriate to the return metallurgy
• 29 CFR 1910.147 for isolation of mechanical and stored energy; local plumbing code and authority having jurisdiction for blowdown discharge
• See related: Why Feedwater Is Treated Differently From Loop Water; Why a Condensate Line Is a Chemistry Problem Too; Cycles of Concentration in Plain Terms