What a Closed Loop Does That an Open One Does Not

Why this matters

Techs use "closed loop" to mean "the pipes are all connected up." That is not the property that matters. The property that matters is that new water only enters through the make-up line, which means the oxygen and the minerals that attack the system arrive as a one-time charge rather than as a continuous supply. Get that straight and a lot of field decisions become obvious: why an inhibitor lasts years in one system and vanishes in another, why one loop rots out in six seasons while an identical one lasts thirty, and why a variable-speed retrofit pays back beautifully on one system and barely at all on the next.

Before you sample, dose or drain

• Treatment chemicals are selected by exposure route, not by how nasty they look. Read the safety data sheet for the actual product and use the protection it specifies. Liquid inhibitors and glycols are a skin and eye route, so chemical-resistant gloves and goggles. Powdered inhibitors and biocides that you scoop, pour or mix are also an inhalation route, so dose them in a ventilated space with a respirator selected under a program meeting 29 CFR 1910.134, not with a dust mask. Never combine two treatment products in one container.
• Do not open a diaphragm expansion tank on a pressurized loop. It stores energy. Isolate it, drain the system side to a bucket, and confirm zero on a gauge before a fitting comes loose (29 CFR 1910.147).
• Never plug, gag or crank up a weeping relief valve. A weeping relief is telling you something about pressure or expansion capacity, and disabling it removes the last barrier on a system that is already misbehaving.
• On an open evaporative system, basin water aerosolizes. Shut the fan and stay out of the drift before sampling, and use a fitted respirator: the hazard here is inhaled aerosol, and gloves do not address that route. Name what you are protecting against rather than leaving it as dust, because it changes how seriously the step is taken: the organism of concern is Legionella, an open evaporative system is one of its recognized amplification sites, and a building operating one should have a water management program along the lines of ASHRAE Standard 188. Do not re-enter the drift zone until the fan has been off long enough for aerosol to settle and the site's own program says it is clear.

The make-up audit

The single most useful number on a closed loop is how much water it swallows per year, and almost nobody measures it. Two ways to get it:

1. Fit a small totalizing meter on the make-up line. It costs almost nothing, needs no power, and turns a guess into a trend you can read on every visit.
2. Close the fill valve and watch system pressure for a week, allowing for the swing that temperature and the expansion tank normally produce. A loop that cannot hold pressure with the fill shut is feeding, and the fill valve was hiding it.

Work an example. A loop holding 400 gallons.

• A tight loop takes make-up only for occasional service work. Call a reasonable target under 5 percent of loop volume per year, so under 20 gallons a year for this system. Confirm the target against the treatment supplier's guidance for your chemistry, because inhibited systems often specify tighter.
• A loop with a weeping relief and a packing leak takes 2 gallons a day. That is 730 gallons a year, which is 1.8 times the entire loop volume every year.

The leaking loop is being handed about 36 times the fresh-water load of the tight one (730 divided by 20). That is 36 times the dissolved oxygen and 36 times the hardness, arriving continuously, and it is why the second loop's inhibitor test comes back low every visit while the first one holds for years.

The one-time charge, and why it makes a closed loop survivable

Fill a steel loop with fresh water and you have added a fixed quantity of dissolved oxygen. That oxygen attacks the steel, gets consumed doing it, and then it is gone. The loop's oxygen supply is exhausted, and corrosion slows to a crawl. Same story with hardness: the minerals in the fill water precipitate once, coat what they are going to coat, and there is no more where that came from.

That is the whole reason a closed loop can run for decades on plain treated water. It is not that the water is special. It is that the supply of attacker is bounded.

Which is exactly why make-up is the number to watch. Every gallon of make-up is a fresh delivery of oxygen and hardness into a system whose defence was that it did not get any. A loop feeding 1.8 volumes a year is not a closed loop that leaks a bit. Chemically, it is an open system in closed-loop clothing, and it will corrode and scale like one no matter what the drawing calls it.

The practical consequence for treatment: on a tight loop, chemistry is a periodic test and a top-up after service work. On a feeding loop, dosing chemistry before you find the leak is buying inhibitor to send down a drain, and the shop that keeps selling the top-up without fixing the feed is selling a subscription to its own failure.

Why an open loop concentrates and a closed one cannot

An evaporative open system, a cooling tower being the common case, has a mechanism a closed loop does not: water leaves as vapour and the minerals stay behind. Make-up replaces the water but brings its own minerals in with it, so dissolved solids climb until something removes them. That something is blowdown, deliberately dumping concentrated water. The ratio of make-up to blowdown is the cycles of concentration, and it is the knob that decides whether the system scales.

A closed loop has no evaporation and no blowdown, so it has no concentrating mechanism at all. Its solids sit where they were when it was filled. This is why open-system treatment is a continuous dosing and bleed program while closed-system treatment is a test-and-adjust program, and why applying one system's service routine to the other wastes money in one direction and destroys equipment in the other.

The other difference: the system curve has an intercept

A closed loop's up-leg and down-leg cancel. Whatever height the water is lifted on the way out, it gives back on the way home, so the pump overcomes friction only and the system curve passes through the origin:

head = k x flow squared

An open system that lifts water from one level to another does not get that back. The pump must supply the lift whether or not anything is flowing:

head = static lift + k x flow squared

Work it on a system lifting 30 ft with 12 ft of friction at its design 40 GPM, so 42 ft total.

At 20 GPM, friction falls to 12 x 0.25 = 3 ft, and total head is 33 ft. Head dropped only 21 percent, from 42 ft to 33 ft, because the 30 ft of lift never moves. Hydraulic work goes as flow times head, so the system at 20 GPM takes (20 x 33) against (40 x 42), which is 660 against 1680: about a 61 percent reduction.

Run the same comparison on a closed loop whose 42 ft at 40 GPM is all friction. At 20 GPM its head is 42 x 0.25 = 10.5 ft, and hydraulic work is (20 x 10.5) against (40 x 42), which is 210 against 1680: an 87 percent reduction. That is the cube law people quote for variable-speed savings, and it is a closed-loop result. Quoting it on a high-lift open system overstates the saving badly.

The minimum speed nobody plans for

The intercept has a second consequence that catches people on open-system speed retrofits. Pump head falls with the square of speed, so at half speed a pump makes a quarter of its full-speed head at every point on its curve, including shutoff.

Say that same pump's full-speed shutoff head is 55 ft. At half speed its shutoff head is 0.25 x 55 = 13.75 ft. The static lift is 30 ft. The pump cannot reach the top of the lift, so flow is zero, not half.

Solve for the speed where it just barely delivers nothing: you need shutoff head to equal the 30 ft lift, so the speed ratio is the square root of 30 divided by 55, which is 0.74. Below about 74 percent speed this pump moves no water at all. A closed loop has no such floor, because at any speed above zero the pump makes some head and the system needs only friction.

If you are commissioning a drive on an open system, find that floor and set the minimum speed above it. Otherwise the first time the control calls for low flow, the pump spins, draws current, produces nothing, and gets condemned as failed.

Telling which one you actually have

Systems get relabelled by renovation, not by decision. Run these on any loop somebody called closed:

• Make-up trend. Meter it or hold the fill shut for a week. Feeding regularly means treat it, and diagnose it, as open.
• Does the pump head match a curve through the origin? Take head at two different flows. If the two points fit k x flow squared with no intercept, it is closed. A stubborn intercept means there is a free surface or a vent somewhere that you have not been told about.
• Is there a free surface anywhere in the circuit? An open expansion tank, a gravity break at a high point, a hot water tank vented to atmosphere. Any one of them makes the loop open in every sense that matters.
• Where does the expansion tank connect? On a properly arranged closed loop the tank connection is the point of no pressure change, and the pump is arranged to push away from it. A tank connected on the discharge side moves the whole loop's pressure profile down and pulls air in at every vent, which then presents as a chronic air complaint that no amount of purging fixes.

References

• Manufacturer and treatment-supplier guidance for closed-system inhibitor levels, test intervals and acceptable make-up rates
• ASHRAE Handbook, HVAC Systems and Equipment, for hydronic system arrangement, expansion tank connection and open-circuit design
• OSHA 29 CFR 1910.134 for respiratory protection when handling powdered treatment chemicals or sampling aerosolizing basin water; 29 CFR 1910.147 for stored energy before opening an expansion tank
• See related: What a Pump Curve and a System Curve Do Together; How Air and Water Behave Differently in the Same Run; Cavitation and What It Tells You