Why Makeup Water Is the Variable That Decides Everything

Why this matters

A water report gives you a concentration. Damage is done by a mass. The bridge between them is volume, and for a closed system the volume that matters is not the one printed on the tag - it is the total quantity of fresh water the system has swallowed since it was filled.

That number is almost never recorded, is trivially cheap to record, and it decides outcomes that the water report alone cannot explain. A tight loop on mediocre water outlasts a leaking loop on excellent water, every time, because the tight one received its impurity load once and the leaking one keeps receiving it. Say it at the other end too, since half of it gets quoted alone: a loop taking no makeup goes anoxic within weeks and stops corroding even on aggressive water, while a loop on continuously renewed makeup keeps a fresh oxygen supply at every heat-transfer surface no matter how good the supply is.

Before you go looking for the loss

Do not isolate a makeup connection on a system whose low-water protection, pump seal cooling or expansion control depends on that feed, and never leave a makeup valve closed unattended on an operating system. Closing it is a legitimate diagnostic step and it removes a protective function while it is closed, so it is a supervised, time-boxed test with the equipment watched, not a valve you shut on the way out.

A loop that has been closed for years may hold hydrogen sulfide from sulfate-reducing bacteria in its stagnant legs. That is an inhalation hazard, not a smell to note - the rotten-egg odor deadens at higher concentrations, so the nose is not a detector. Opening or draining that loop is done with ventilation and, where the space is confined or poorly ventilated, with atmospheric monitoring and respiratory protection issued under a written program (29 CFR 1910.134).

Draining or flushing a loop puts hot, dirty, treated water somewhere it was not before. Confirm the receiving drain can take the volume and the temperature before opening the valve, isolate the section and relieve pressure to a gauge reading zero before cracking any fitting, and stand clear of the discharge while you open it.

Isolating a pump to inspect a seal means locking out its stored energy under 29 CFR 1910.147, with the electrical disconnect handled separately under 29 CFR 1910.333(b)(2), since 1910.147 excludes electric utilization installations at 1910.147(a)(1)(ii)(C).

A fill connection between a treated loop and the potable supply is a cross-connection. Backflow protection to what the local plumbing code requires for that chemical goes in before anything else happens, and a missing one is a public-health finding you report rather than a detail you note.

The case: a plate exchanger that plugged twice

A six-year-old hydronic heating loop, about 600 gallons, in a small commercial building. The plate heat exchanger had been cleaned two years earlier and was plugging again. System water ran black, strainers filled with fine magnetite sludge between visits, and the building's water was unremarkable municipal supply at hardness 220 mg/L as CaCO3.

The three explanations on the table. Bad fill water. A failed air separator letting the loop breathe. An inhibitor that was never charged properly after the last cleaning.

Bad fill water went first. The supply report showed a hard but well-buffered water, nothing aggressive, nothing unusual. If 600 gallons of that water were the entire exposure, the loop would carry a modest one-time mineral load and no ongoing supply of anything. The report could not explain a second plugging in two years, and reading it as though it could was the trap: a concentration cannot explain a rate.

Air ingress went second, and it survived longer than it deserved. Black water and magnetite are an oxygen story, so the theory fit the evidence. It was killed by a properly drawn dissolved-oxygen sample - tube to the bottom of the bottle, overflowed, capped with no bubble, read immediately - which came back at a few tenths of a milligram per liter. The loop was anoxic at the moment of sampling. That does not mean oxygen never enters; it means oxygen entering is being consumed close to where it enters, which is exactly what happens in a system taking makeup.

The inhibitor test is what broke it open. The loop had been charged to its target concentration at the cleaning two years earlier. The reserve now read about 5 percent of the charged concentration. Nothing in a closed loop destroys inhibitor at that rate. What removes it at that rate is water leaving and being replaced.

Turning a depleted reserve into a volume

Treat the inhibitor as a conservative tracer, which requires the assumption that it is diluted rather than consumed or precipitated - reasonable for the common closed-loop reserve chemistries, and worth confirming with the supplier for the specific product.

For fresh water entering a well-mixed loop more or less continuously, the fraction of the original charge remaining is e raised to minus the ratio of makeup volume to system volume. Working backwards from 5 percent remaining:

• Negative natural log of 0.05 is about 3.0 system volumes of makeup over the two years since the charge.
• That is 1.5 system volumes per year, or 1.5 x 600 = 900 gallons a year.
• Assuming that rate held across the loop's whole life, which is the load-bearing assumption here and the reason the next step is a meter rather than more arithmetic, six years at 1.5 volumes a year is 9 system volumes, or 9 x 600 = 5,400 gallons total.

What 9 system volumes actually delivered

Convert to liters for the mass arithmetic: 5,400 gallons is about 20,439 liters.

Hardness delivered. 20,439 L x 220 mg/L = 4,496,580 mg, or about 4.5 kilograms expressed as calcium carbonate. A tight loop filled once would have received 600 gallons, about 2,271 liters, carrying about 0.5 kilograms. Nine times the mineral, into the same 600 gallons of circulating volume, and none of it left as vapor because a closed loop has no vapor path. It came out of solution where the surface was hottest, which is the plate exchanger.

Oxygen delivered. Cold makeup arrives near air saturation; call it 8 mg/L, a deliberately conservative figure for cold municipal water. 20,439 L x 8 mg/L = 163,512 mg, or about 164 grams of dissolved oxygen.

Convert that to iron consumed. Iron reacting with oxygen consumes roughly 2.3 grams of iron per gram of oxygen if the product is ferric oxide and about 2.6 if it is magnetite, so 164 grams of oxygen accounts for roughly 380 to 430 grams of iron turned into oxide. That is a plausible quantity of black sludge to find in the strainers of a system this size, and it is the number that finally matched the symptom.

Compare the tight loop again: one fill delivers about 18 grams of oxygen, worth roughly 42 to 48 grams of iron, once, after which the loop is anoxic and the reaction stops. The difference is not nine times more corrosion spread evenly. It is a one-time event against a process that never ends.

What the numbers changed

The exchanger was not the problem and cleaning it again would have bought two more years. The loop was losing about 900 gallons a year through a weeping pump seal and a buried section, and an automatic fill valve was replacing it silently. An automatic fill valve does not cause this; it hides it. That is its design intent - keep the system full - and it means a leak large enough to destroy a heat exchanger produces no low-pressure alarm, no complaint, and no visible symptom.

What would have changed the conclusion. If the inhibitor had come back near its charged concentration, the dilution model collapses and with it the volume estimate, and the investigation moves to a one-time contamination or to a genuine air path. If the loop had been drained and refilled once during those two years for unrelated work, that single event contributes its own volume as a batch rather than a drip, the exponential model overstates the remainder, and the honest answer is to stop calculating and install a meter.

The failure mode, and it is common: the exchanger gets cleaned, the loop gets a fresh inhibitor charge, the readings on the way out are excellent, and nobody touches the fill. The next reserve test two years later reads 5 percent again. Every number on every visit is correct and the system is on its third exchanger.

The instrument that ends the argument

A totalizing meter on the makeup line converts all of the above into two readings and a subtraction. It is the cheapest permanent instrument on a hydronic system and the least frequently installed. Record the totalizer at every service visit alongside the date, and within two visits the system has a makeup rate; within a year it has a trend.

Express the result as system volumes per year, not gallons, because that is the unit that travels between a 200-gallon loop and a 5,000-gallon one and the unit that maps to inhibitor dilution. Set your own action threshold on that basis and hold it: a reasonable starting point is to flag any closed loop above about 10 percent of system volume per year, measured over a window of at least 30 days at steady operating temperature, and to stop treating it as a chemistry problem and start treating it as a leak to find above roughly 50 percent per year. There is no published universal limit for this - tune the two numbers to what your own records show for tight systems in your area, and write them down so the whole crew uses the same ones.

References

• 29 CFR 1910.134 for respiratory protection where opening a long-closed system may release hydrogen sulfide, particularly in a confined or poorly ventilated space
• 29 CFR 1910.147 for stored energy isolation on pumps and pressure vessels, and 29 CFR 1910.333(b)(2) for the electrical disconnect, which 1910.147 excludes at 1910.147(a)(1)(ii)(C)
• 29 CFR 1910.1200 (Hazard Communication) and Section 8 of the product safety data sheet for handling any inhibitor or cleaning chemical
• Treatment supplier documentation for whether a specific inhibitor behaves as a conservative tracer, which the dilution calculation above depends on
• See related: How to Find Out How Much Makeup Water a System Is Really Taking; The Difference Between an Open Loop and a Closed Loop Chemically; How to Take a Water Sample That Means Something