Why an Inhibitor Runs Out and What Happens After It Does

Why this matters

Inhibitor leaves a closed loop by exactly two routes: it gets diluted out with water, or it gets consumed by something. Those two look identical on a single reserve reading and they have nothing in common as problems. One is a plumbing job and one is a chemistry job, and a shop that cannot tell them apart ends up on a top up treadmill, adding product every quarter to a system that will never hold it, while the actual cause runs untouched for years.

Separating them costs one extra number on the same sample you are already drawing.

Before you draw anything or add anything

Read the loop's pressure and temperature at the gauge before opening a sample valve. A heating loop above about 140 F will flash out of a cracked valve as scalding water and steam; either isolate the section and let it cool or draw through a sample cooler, into a container positioned so the discharge path is not toward anybody.

Read Section 8 of the safety data sheet for every product you are about to handle, including the biocide, and use the glove class and eye protection it names for that product. Biocides in particular are a different hazard class from inhibitors and several are skin sensitisers, meaning the reaction gets worse with each exposure rather than better. 29 CFR 1910.1200 is the standard that puts that data sheet in your hands.

Never combine an oxidising product with an acid product, in the loop, in a feeder, or in a bucket. That generates chlorine gas in seconds, which is an inhalation event; the control is separation and a thorough flush between steps, and if a respirator ever enters the answer it is under a written program with fit testing per 29 CFR 1910.134.

The flushed and drained volume goes somewhere. Treated loop water carrying a biocide is not automatically a legal drain discharge; the local sewer authority sets what may go down and in what concentration, and that is a call you make before you open the valve, not after the bucket is full.

The loop that would not hold its charge

A chilled water loop in a small commercial building, measured at 900 gallons. Nitrite based inhibitor package. The shop had topped it up once already the previous quarter and it had come back low again, which is the point at which someone finally pulled the three sample records and put them next to each other.

Five columns, and only one of them was ever being read.

Separating dilution from consumption

Dilution is indiscriminate. Water going out and fresh water coming in removes every dissolved species in the same proportion, including the ones nothing reacts with. Consumption is selective: it removes the reactive species and leaves the rest.

That gives you the test. Pick a species in the loop that is not consumed, read it alongside the inhibitor, and compare the two falls. Total conductivity works and is a meter reading rather than a lab result. Chloride works and is more specific but needs a test.

Run it on this loop. Makeup over the two quarters totalled 19 gallons against a system volume of 900, which is 2.1% of the loop's water replaced. Conductivity fell from 2,400 to 2,350, a drop of 50, which against 2,400 is 2.1%. Those two agree to the decimal, so dilution is fully accounted for and there is nothing unexplained on that side.

Now the inhibitor. Nitrite fell from 1,100 to 520, a drop of 580, which against 1,100 is 53%. Dilution at 2.1% predicts nitrite would have landed near 1,076. It landed at 520. Something removed roughly 556 ppm that water loss cannot explain, and that is 96% of the total fall.

The verdict falls out cleanly and it inverts the shop's working theory. This was never a leak. Chasing one would have consumed a day of pressure testing and found nothing, which is exactly what "no leak found" usually means on a loop like this.

What was eating it

Nitrite is consumed by oxidation, and there are two common oxidisers in a closed loop. Dissolved oxygen arriving from outside will do it. So will nitrifying organisms, which use nitrite as a nitrogen source and oxidise it to nitrate. Both leave nitrate behind, so nitrate alone does not separate them.

The pH column separates them. Nitrification is acid producing, and a loop losing pH while its inhibitor disappears is a biological finding, not an oxygen one. This loop dropped 0.8 pH units across two quarters through a buffered package, which means the acid load had already consumed a meaningful part of the buffer before pH moved at all. Dip slides confirmed it, and the fluid had the faint sulphurous smell that goes with a heavy population.

The system that produced this is the ordinary one for chilled water. A loop running in the low forties to mid fifties Fahrenheit never gets hot enough to suppress anything, and cold water holds substantially more dissolved oxygen in solution than hot water can, so the growth conditions are better than in a heating loop in both respects at once. The same finding in a heating loop that runs 160 to 180 F would point somewhere else entirely, because that temperature suppresses most of these populations; a hot loop losing nitrite with stable conductivity is far more likely to be oxygen ingress or a large area of freshly exposed metal from recent pipe work.

Why 520 ppm was not "a bit low"

The iron column is the one that makes this urgent rather than scheduled. Dissolved iron rose from 0.3 to 2.1 mg/L, a factor of seven, while nitrite was still comfortably detectable on a strip. A nitrite package protects by maintaining the passive film on the anodic sites, and below the supplier's published minimum it covers some of those sites and not others. The corrosion current does not stop; it now has far less bare area to come out of, so it comes out concentrated. That is why detecting nitrite tells you nothing useful on its own and why the supplier's minimum, not the detection limit, is the number that matters.

This behaviour belongs to the anodic passivating class specifically. On a cathodic or all organic package, protection fades roughly in proportion to concentration and a reading at half of target is genuinely half protected. The class the product belongs to is on the data sheet and it changes the urgency of every low reading you will ever take.

What the top up alone had already cost

The previous quarter's response was to add product and move on. The arithmetic says why it failed. Restoring 900 gallons from 780 back to 1,100 ppm means closing a 320 ppm deficit; 900 gallons is about 3,406 litres, so that is roughly 1.09 kg of nitrite ion, and at a 10% active product about 10.9 kg of product. That chemical went into a loop with an active population still consuming it, so it bought one quarter and then the reading was lower than before the top up.

The recurring cost is not the product, it is the visit: an hour on site plus lab turnaround, every quarter, forever, with a reading that never improves. Against that, the diagnostic that resolved it was one extra conductivity reading on a sample already being drawn, which adds a minute. That is the whole trade.

The real fix was a non oxidising biocide on the supplier's program, then re inhibition, then a shortened test interval until two consecutive samples held. The choice of a non oxidising biocide is not incidental: an oxidising biocide dosed into a nitrite loop will oxidise the nitrite along with the organisms, so you would be paying for the treatment and destroying the inhibitor in the same drum.

When both are running at once

The loop above was a clean case because dilution accounted for essentially nothing. Most loops are not that tidy, and the split is worth doing properly rather than declaring one winner.

Take a 600 gallon loop whose makeup meter shows 72 gallons over a year, which is 12% of its volume. Conductivity fell from 2,000 to 1,760, a drop of 12%, so dilution is again fully explained. Reserve fell from 1,200 to 720 ppm, a drop of 480. Dilution at 12% predicts 1,056, so 144 ppm of that fall is dilution and 336 ppm is consumption. Dilution owns 30% of the loss, consumption 70%, and both numbers are large enough to be real jobs.

The tempting next sentence is that the makeup is causing the consumption, since fresh water carries dissolved oxygen and oxygen oxidises nitrite. Check it before you say it. Nitrite oxidation consumes roughly one gram of oxygen for every 2.9 g of nitrite, so removing 336 ppm from 600 gallons, about 2,271 litres, would take on the order of 265 g of oxygen. Cold makeup carries around 10 mg per litre, so that much oxygen arrives in roughly 7,000 gallons of water. The meter says 72.

Off by a factor near a hundred, so the makeup water cannot be driving the consumption here and the two findings are independent. Fixing the leak will not recover the reserve, and topping up the reserve will not stop the water loss. Two work orders, and the consumption one points at the routes that admit oxygen without losing water, or at fresh metal, or at organisms.

Confirming it took

Two samples at a shortened interval, read as a set of three numbers rather than one.

Nitrite holding at target with conductivity stable means nothing is being consumed and nothing is being diluted. Iron falling across those two samples is the surface re passivating, and that is the reading that actually says the metal is protected; iron is the outcome, nitrite is only the input. If iron holds or keeps climbing while nitrite sits at target, the chemical is in the water and not on the metal, which points at a deposit shielding the surface rather than at the dose.

And put the makeup meter reading in the record beside them every time. The whole separation above depends on knowing how much water moved, and a loop with no meter cannot run this test at all; it can only guess, top up, and come back.

References

• 29 CFR 1910.1200, Hazard Communication, for the safety data sheet whose Section 8 sets the glove class and eye protection for each inhibitor and biocide handled
• 29 CFR 1910.134, Respiratory Protection, for the written program and fit testing required before a respirator is any part of a chemical control
• Local sewer authority discharge limits for treated loop water containing biocide or inhibitor
• Fluid and inhibitor supplier documentation for the published minimum concentration, the inhibitor class, and biocide compatibility with the package in use
• See related: What a Corrosion Inhibitor Actually Does, which owns the mechanism and class distinctions used here