Why Feedwater Is Treated Differently From Loop Water

Why this matters

Two systems in the same mechanical room can carry the same water and need almost nothing in common from a treatment program. The difference is not the temperature and not the metal. It is whether the water is consumed and continuously replaced, or is a fixed inventory that only gets loaded once. In a replaced system, chemical demand tracks throughput. In a fixed inventory, demand tracks leakage, which means a closed loop's treatment program is really a leak-monitoring program wearing a chemistry label, and the meter on the makeup line is its primary instrument. Shops that import cooling tower habits into a closed loop spend money on the wrong thing and, in two specific cases below, do active harm.

Before you sample or add anything to a closed loop

• A closed loop holds stored pressure and, on the heating side, water hot enough to scald. Isolate, relieve to a vented point, and verify zero at a gauge before breaking any fitting. Water above 212 F flashes to steam when released to atmosphere; below that it is still a scald hazard, and the difference decides whether you are managing a spray or a jet.
• Never open a chemical pot feeder under pressure. Isolate both valves, open the vent, confirm flow has stopped and pressure is gone, then open the fill.
• Inhibitors and glycols are chemicals with different classes. Take glove class and any respiratory control from each product's safety data sheet. Ethylene glycol is acutely toxic if swallowed and both glycol families are a serious slip hazard on a plant room floor; contain spills rather than washing them to a drain, and dispose of drained glycol through the route your local requirements set rather than to a storm drain.
• Do not heat glycol beyond the manufacturer's stated film temperature limit while flushing or during any temporary bypass; the degradation products are acids and you will have converted a fluid change into a corrosion event.

Two regimes, one distinction

Continuously replaced water describes boiler feedwater, tower makeup, a domestic hot water system, a pressure washer drawing from a tap. Water enters, does its job, and leaves as steam, vapour, drift, blowdown or down a drain. Whatever treatment it needs must be supplied continuously and proportionally to the volume passing through, because the treated water is leaving as fast as you treat it.

Fixed inventory describes a closed hydronic heating loop, a chilled water loop, a glycol loop, a sealed radiant system. In principle you fill it once, treat it once, and it runs. Nothing evaporates, so nothing concentrates.

That last clause is where most of the useful consequences come from, and it is worth stating precisely: because there is no evaporation, a closed loop does not cycle up. Water leaking out takes its dissolved minerals with it at loop concentration, and makeup brings the same concentration back in, so the loop's concentration sits where the makeup sits and does not climb. Cycles of concentration, the number that governs a tower, is meaningless here.

What a closed loop does not need

The negative space is the payload of this card, because each of these is a habit that travels in from open-system work.

• Continuous blowdown. There is nothing to remove. Bleeding a closed loop discharges inhibitor you paid for and pulls in fresh makeup, which is the one thing the loop is trying to avoid.
• A cycles target. No evaporation means no concentration, so any target is a number without a mechanism behind it.
• Routine oxidising biocide. A tower is continuously re-inoculated because it washes outside air; a closed loop is not, so the recurring dose has no recurring source to answer. Worse, an oxidiser consumes many corrosion inhibitors and attacks the metals directly. This is one of the two habits that does harm rather than nothing.
• Scale inhibitor sized to throughput. Sizing chemistry to circulating flow rather than to system volume is the other one: it over-doses the loop, and some inhibitor packages have an upper limit above which they stop helping.
• Softening the makeup as a strategy. Softening the makeup on a leaking loop treats the symptom of a leak with a piece of equipment. Fix the leak.

What a closed loop does need

Three things are consumed or deposited in proportion to makeup volume, even though nothing concentrates. This is the mechanism that makes the makeup meter the instrument.

Oxygen is consumed, not conserved. The oxygen in the first fill reacts with the metal, and a properly sealed loop then sits at very low dissolved oxygen indefinitely. Every gallon of makeup delivers a fresh charge of dissolved oxygen at saturation for its temperature, near 12 mg/L for cold makeup around 45 F at sea level and one atmosphere, falling as temperature rises and falling with altitude. The corrosion is not caused by the water being there. It is caused by the water arriving.

Inhibitor is diluted and carried out. Whatever leaks out takes inhibitor with it and whatever comes in has none, so reserve decays with makeup volume.

Hardness is deposited even though it does not concentrate. Calcium carbonate solubility falls as temperature rises, so hardness arriving with makeup precipitates at the hottest surface in the loop. Loop concentration stays flat while deposit accumulates, at a rate set by makeup volume. That is why a loop can test at ordinary hardness for years and still be laying down deposit on a heat exchanger.

Worked example: a loop that is not closed

A chilled water loop with a system volume of 4,000 gallons, taken from the drawings and confirmed by a dilution test, which is the honest way to get it when the drawings are missing: add a known mass of a conservative tracer, circulate for a full turnover, then measure.

The makeup meter reads 1,200 gallons over 90 days, about 13.3 gallons a day. That is 30 percent of system volume in a quarter, so at that rate the loop replaces its entire inventory in roughly 300 days and takes on about 1.2 complete fills a year.

Set that against a tightness target. A commonly used one for a sound closed loop is makeup under roughly 5 percent of system volume per year; take your own target from the treatment supplier and the system's age, and log the meter monthly so the trend rather than the absolute is what triggers you. At 120 percent a year against a 5 percent target, this loop is running about 24 times its target, and no reasonable version of that target changes the conclusion.

Follow the three consequences through with those numbers.

Oxygen. The loop receives the dissolved oxygen content of 1.2 complete fills a year. A sealed loop receives it once, at commissioning.

Inhibitor. Reserve is being diluted and carried out at roughly 30 percent a quarter, which is exactly what the site's history showed: a quarterly top-up dose that never held, read for three years as normal consumption.

Hardness. At makeup hardness of 120 mg/L as calcium carbonate, the loop receives the hardness content of 1.2 fills a year and its measured hardness has not risen. That is not reassuring, it is the finding: the arithmetic says the calcium entered and the sample says it is not in the water, so it is on a surface, and the hottest surface in a chilled water loop is the one you care least about losing and the one nobody inspects.

The failure mode here is treating the quarterly inhibitor dose as the service. Three years of correct chemical additions kept every water test in band while the loop consumed its own heat exchanger, and the one instrument that would have named it cost nothing to read because it was already installed.

When a closed loop does show rising conductivity

Because a closed loop cannot concentrate on its own, a conductivity trend that climbs is a positive finding and has only two explanations, which fail in opposite directions.

Something is evaporating. An open expansion tank, a loop that has been boiling locally, or a vent passing steam. Water leaves without its minerals and the loop cycles up like a miniature tower.

Something is being added. A failed makeup valve or backflow preventer letting domestic water through continuously, a heat exchanger leaking process fluid or domestic water into the loop, or a second fill connection nobody documented. A continuously passing makeup connection is also a cross-connection question: a chemically treated loop connected to potable water is classified as a high hazard by the model plumbing codes and requires a tested reduced-pressure principle backflow assembly, so a makeup valve that will not seat is a public-health item and not only a water bill item. Confirm the required assembly and its test interval with the authority having jurisdiction.

Both show as rising conductivity and both are found the same way: read the makeup meter, then isolate the makeup connection for a shift and see whether the loop holds pressure.

Glycol changes the maintenance, not the regime

A glycol loop is still a fixed inventory, so everything above holds. What changes is that the fluid itself has a condition as well as a concentration, and only one of those is easy to measure.

Freeze point, by refractometer or hydrometer, tells you concentration. It does not tell you condition. Glycol degrades to organic acids when it is overheated or when its inhibitor package depletes, and the pH crash that follows turns a protective fluid into an aggressive one while the freeze point still reads fine. Test pH and inhibitor reserve on the schedule the fluid manufacturer sets, not just freeze point, and treat a falling pH as a fluid replacement question rather than a top-up question.

Never top up a glycol loop with water without recalculating concentration, and take the concentration for a given freeze or burst protection from the fluid manufacturer's own table for that product, since inhibited glycols for hydronic service and automotive coolants are not interchangeable and their tables are not the same.

How to verify which regime you are actually in

• Find the makeup meter, or install one. If a loop has no way to measure its own makeup, it has no way to know whether it is closed, and everything else on the service sheet is an opinion.
• Isolate the makeup for a measured period and watch pressure. A closed loop holds. This is the single test that settles the argument.
• Get system volume before you dose anything. From drawings if they are trustworthy, by dilution test if they are not. Dosing a fixed inventory without knowing the inventory is guessing.
• Sample the loop and the makeup on the same visit. A loop concentration equal to makeup is what a sealed loop looks like; higher means evaporation or addition, lower means something is depositing.
• Log the meter monthly and read the slope. A step change in makeup rate dates the leak, and dating it is usually what finds it.

References

• Fluid manufacturer documentation for glycol concentration, freeze and burst tables, film temperature limits and test intervals
• Water treatment supplier documentation for inhibitor type, target reserve and testing on your specific loop metallurgy
• Model plumbing code and the authority having jurisdiction for backflow protection on makeup to chemically treated systems
• See related: What a Boiler Does to the Water in It; Cycles of Concentration in Plain Terms; How Scale Forms and What It Actually Costs