What an Evaporative Tower Does to the Water in It

Why this matters

A cooling tower is usually described as a heat rejection device, which is true and is why it gets sized and never thought about again. From the water's point of view it is three machines running at once: a concentrator that removes pure water and leaves everything else behind, an aerator that keeps the loop saturated with air, and an extremely effective air washer that collects whatever is blowing past the building. Those three do independent things to the water, and the single instrument most towers have, a conductivity controller, only sees one of them. The case below is what that blind spot looks like on a service ticket.

Before you open a tower

• Lock and tag the fan motor disconnect under 29 CFR 1910.147 and confirm by eye that the fan has stopped before any part of you crosses the plane of an access door. A tower fan windmills on stack draft with the drive off, and the blades are inside arm's reach of the opening on most units.
• Work inside the tower's electrical panel is a different standard. 1910.147(a)(1)(ii)(C) excludes electrical work on utilization equipment, so panel work runs under 29 CFR 1910.333(b)(2), with the live-dead-live proving sequence per NFPA 70E-2021, 120.5.
• Do not pressure-wash fouled fill or a silted basin. That converts a contact exposure into an inhalation one and can throw a biologically loaded aerosol across a roof that people work on. Remove sludge wet, with a wet vacuum, and never dry-sweep a dried basin.
• Tower access is usually roof access. The fall hazard is the one that kills people on this equipment far more often than the water does, and it is governed by 29 CFR 1910 Subpart D for general industry work or 29 CFR 1926 Subpart M where the job is construction.
• Test reagents and biocides each have their own class. Take glove class and any respiratory control from that product's safety data sheet, and never let an oxidising product share a pump, jug or funnel with an acid or an amine.

The signal

A 600 gpm tower on a mid-rise, running a 10 F range, on a contract service with three residual checks a week. Through May, June and July the oxidant residual sat inside the supplier's band on almost every check. Through the first half of August, six consecutive checks came back at or near zero within an hour of the feed cycle. Conductivity was holding at 1,580 microsiemens against a 1,600 setpoint. Nothing on the controller said anything was wrong.

The obvious reading is that the feed system failed. It had not, and working out why took understanding what the tower was doing to the water in three separate ways.

Mechanism one: it removes pure water and leaves the minerals

Evaporation is the cooling. Water leaves as vapour and takes its latent heat with it, and it takes nothing else: every dissolved mineral that came in with it stays in the basin. The standard approximation is that roughly 1 percent of the circulating flow evaporates per 10 F of range, and that figure is derived assuming essentially all of the heat rejection is latent, so on a cold dry day when a larger share leaves as sensible heat the true evaporation is less than the rule gives.

For this tower at 600 gpm and a 10 F range, that is about 6 gpm evaporating. Over a 10-hour operating day, 3,600 gallons of pure water leaves the building as vapour. At the 5 cycles this loop was running, the non-evaporative loss needed to hold that concentration is the evaporation divided by cycles minus one, or 3,600 divided by 4, which is 900 gallons a day. Makeup is the sum: 3,600 plus 900 is 4,500 gallons a day, and 4,500 divided by 900 is 5, which is the cycles it started from.

That arithmetic is the concentrating machine stated plainly. It is also the reason a tower loop is nothing like the water in the street: at 5 cycles, every dissolved species that does not precipitate or volatilise sits at five times its makeup concentration all season.

Mechanism two: it saturates the water with air

Water falling through fill in a forced air stream leaves in equilibrium with that air. Oxygen solubility in water at sea level and one atmosphere runs near 13 mg/L around 40 F and near 7 mg/L around 90 F, falling as temperature rises and falling again with altitude. The warm end of that range is the lower number, which reads like good news and is not, because the tower re-saturates the water on every pass. A tower loop is continuously replenished with dissolved oxygen no matter what the solubility ceiling is at that temperature.

The practical consequence is that none of the reasoning that applies to a closed loop applies here. A closed system consumes its oxygen once and then sits low; a tower loop never gets that chance, which is why open-loop corrosion control is a continuous chemical program rather than a one-time dose.

Mechanism three: it washes the air

This is the mechanism nobody accounts for. Consider the air side of the same tower. At a mass ratio of water to air near 1.2, which is a common design region, 600 gpm at 8.34 pounds per gallon is about 5,000 pounds of water a minute against roughly 4,200 pounds of air a minute. At standard air density of 0.075 pounds per cubic foot at sea level, that is on the order of 55,000 cubic feet of air a minute, and over a 10-hour day something like 33 million cubic feet of air passing through a falling water curtain. At altitude the same mass of air is a larger volume, so the volumetric figure rises while the scrubbing does not change much.

Nothing about that geometry is different from a piece of air pollution control equipment. Dust, pollen, spores, combustion particulate and soluble gases go into the basin. Insoluble solids settle out or lodge in the fill. Organic material becomes food.

Working the case

Three hypotheses, killed in order on evidence.

The feed system failed. Checked first because it is cheapest to check and because a dead pump and a high demand read identically on a test kit. The metering pump was primed, the injection quill was clear, the drum was two-thirds full and the controller log showed feed cycles firing on schedule. Killed.

Cycles drifted up and something changed the water's chemistry. Conductivity was 1,580 against a 1,600 setpoint, and makeup conductivity measured 320, unchanged from the May baseline. Five cycles then, five cycles now. Killed, and it is the reading the controller was reporting all along, which is why the controller never flagged anything.

Demand rose. Tested directly: dose a measured sample of basin water and a measured sample of makeup water at the same rate, hold both for the same contact time, read residual on both. Makeup held. Basin water did not. That isolates the consumption to the loop rather than to the incoming supply or the product.

The basin had a silt bed at the low end and the fill was visibly loaded. Two blocks away, a demolition had been running since the last week of July. The tower had spent three weeks scrubbing that air.

The mass balance that closed, and what that proved

Worth running, because it rules out the other common explanation. Makeup hardness measured 120 mg/L as calcium carbonate. Per day, minerals entering with 4,500 gallons of makeup at 120 mg/L, against minerals leaving with 900 gallons of blowdown at the basin's measured 600 mg/L. Four thousand five hundred times 120 is 540,000; 900 times 600 is also 540,000. The balance closes.

That is a negative finding and it is the useful one: if calcium were dropping out as scale, the basin concentration would be lower than five times the makeup and the balance would not close. It closed, so the loop was not depositing hardness, and the fouling in the fill was not scale. It was what the air brought in. A shop that had assumed scale and gone in with an acid clean would have handled a hazardous product for no reason and left the actual load in place.

What would have flipped the call

• A makeup conductivity that had moved. A utility switching source water mid-season changes everything downstream of it, and the makeup reading is the cheapest way to see it.
• A basin calcium below five times makeup. That would have moved the case from air load to precipitation, and the corrective action from cleaning and increased blowdown to a scale-control problem.
• Residual holding on the basin sample in the demand test. That would have sent the case back to the feed system with a harder look at contact time and injection point mixing.
• Cold, clean, low-load conditions. State the other end of the demand mechanism or you have asserted direction rather than established it: on a mild, clean day the same feed rate against a low demand overshoots and residual runs high, which attacks steel and consumes inhibitor. High demand and low demand are both control failures, in opposite directions, off the same fixed feed rate. That is the argument for feeding to a measured residual rather than to a clock.

How to verify a tower reading before you act on it

• Sample from the same point every time, and say which point. Basin, sump strainer and the return header do not read the same on a dirty loop.
• Read residual within the window the test method specifies, on a fresh sample, in the shade. Oxidant residual falls in a sample sitting on a warm roof and you will chase a decay you created.
• Confirm cycles two ways before you trust a single instrument. A conductivity ratio and a chloride ratio disagreeing is information, not noise, and the sibling article on cycles of concentration covers what each disagreement means.
• Check drift eliminators are seated with no bypass gap while you are in there. Drift is water leaving at full concentration, so a bypass gap quietly changes your blowdown arithmetic and puts a full-strength aerosol into the air around the tower.
• Log what was happening outside the building. Construction, a nearby stack, harvest season, a cottonwood. The air side of a tower is part of its water chemistry and nothing on the controller records it.

References

• Manufacturer data for the specific tower: drift rate, fill type, design flow and range
• Water treatment supplier documentation for product residual bands, contact times and test methods
• 29 CFR 1910.147 for fan motor isolation; 29 CFR 1910.333(b)(2) for electrical work; NFPA 70E-2021, 120.5 for the proving sequence; 29 CFR 1910 Subpart D or 29 CFR 1926 Subpart M for the fall hazard depending on the work's scope
• See related: Cycles of Concentration in Plain Terms; How Scale Forms and What It Actually Costs; The Water Chemistry That Attacks a System