Why a New System Needs Cleaning Before It Needs Treating

Why this matters

A new system is not a clean system. It is a system full of the residue of its own manufacture and installation: oil on the tube bore, flux at the joints, mill scale on the steel, swarf in the low points, and often several weeks of untreated hydrostatic test water that has been standing at room temperature growing things. Dose inhibitor into that and the chemistry goes to work on the debris instead of the metal, the residual falls, somebody adds more, and a year later the system has a corrosion history that started the week it was commissioned. The order is not a preference. Cleaning is what makes the treatment able to reach the surface it was bought to protect.

Before any cleaning chemical goes in

Read the safety data sheet for the specific product and take your glove class, eye and face protection and any respiratory control from that sheet rather than from what is on the truck; a caustic degreaser and an inhibited acid call for different glove materials and neither is covered by a general-purpose glove. Never add one product to a system or a drum that holds another, and never combine an acid product with a chlorine-bearing product, because that combination releases chlorine gas, an inhalation hazard that eye protection and gloves do nothing to address, and the control for it is not being there rather than a mask chosen on the spot.

Circulating a warm chemical means a hot pressurized system with a corrosive fluid in it. Relieve pressure through a drain valve to a receptor and confirm zero on a gauge before opening any joint, confirm the fluid is below scald temperature since a few seconds of contact at 140 F causes a serious burn, and isolate and lock the circulating pump under a procedure meeting 29 CFR 1910.147 before working on it. Protect the fill connection with the backflow assembly class the adopted plumbing code requires for a system holding a chemical, and confirm with the local sewer authority what the spent cleaner may be discharged as, and at what pH, before you open a drain valve.

What a new system is actually full of

Cutting, drawing and forming oil. Copper tube and steel pipe carry a film from manufacture, and a coil arrives with whatever was used to form its fins and tubes. It is thin, invisible, and everywhere.

Thread-cutting oil, pipe dope and jointing compound. Field-applied, generous, and mostly on the inside where nobody looks.

Flux residue. Soldering and brazing fluxes are chemically active by design, several are acidic and some carry chlorides or zinc, and residue left inside a joint keeps being active after the joint is made.

Mill scale. Hot-rolled black steel carries an iron oxide layer from the mill. It is not a coating in any protective sense; it is a discontinuous layer that is electrically conductive and cathodic to the bare steel exposed wherever it is cracked or missing, which produces exactly the small-anode, large-cathode geometry that drives deep local pitting rather than general thinning. The library's galvanic corrosion article carries that mechanism in full.

Swarf, slag, tape and debris. Cuttings from threading and reaming, weld slag, shreds of thread tape, plastic caps, insulation fibres and cardboard.

Standing test water. Hydrostatic test water is usually raw, often taken from a yard hydrant, and frequently left in the system for weeks while the job finishes. It carries silt, oxygen and a microbial population, and every day it stands is a day of unprotected exposure.

Why inhibitor applied to a dirty surface is spent, not stored

Corrosion inhibitors work at the metal surface. Filming inhibitors adsorb onto it; passivating inhibitors maintain an oxide layer on it. Both require access to the metal, and both are consumed at a rate set by the surface area they have to cover and by whatever else in the water reacts with them.

An oil film is a barrier between the inhibitor and the metal, so the chemistry is present in the water and absent where it matters. Debris and oxide give the inhibitor a large, reactive area to be consumed on. Organic residue feeds microbes, which consume treatment chemistry and colonise the very surfaces the inhibitor is meant to hold. The result is a residual that falls fast and keeps falling, and a system that is chemically treated and physically unprotected at the same time.

The failure mode this produces in the field is a specific and recognisable conversation. The residual comes back low, the shop tops it up, it comes back low again, and somebody concludes the test is wrong or the product is weak. Both readings were correct. The system was eating the chemistry.

The gate

Clean before treating whenever the wetted surfaces have not been documented as cleaned since fabrication.

Two words in that sentence carry the load. Documented means a cleaning record or a manufacturer's statement that the component ships cleaned, not an assurance that it is new; new is the condition that guarantees the residue, not the condition that excludes it. Wetted surfaces means the whole circuit, so a system with one uncleaned branch is an uncleaned system, since the loop mixes.

Run against real jobs, that one rule resolves in opposite directions.

Outcome one: the loop where the gate fires

A 900 gallon hydronic loop, black steel headers and copper branches, field-brazed, hydro-tested six weeks before startup and left full. No cleaning record exists, so the gate fires.

Take the case where it was ignored, because the numbers are the argument. The loop was filled and dosed to 100 percent of the treatment supplier's target residual. Two weeks later it read 35 percent of target. The makeup meter showed 18 gallons over that period, which on 900 gallons is 2 percent dilution, so dilution accounts for 2 of the 65 points lost and the other 63 points went into the system. The loop was topped back to 100 percent. Four weeks after that it read 55 percent.

Read the rate rather than the readings. The first charge lost about 33 points a week; the second lost about 11 points a week. The demand was falling, which tells you the surface was slowly getting satisfied, and the loop had still lost 110 percentage points of residual over the six weeks, of which about 108 went into the system rather than out of the drain, a little more than one full charge, to do a job a cleaning product does in a day. Every one of those points was inhibitor reacting with oil, oxide and debris rather than adsorbing onto metal, and during all six weeks the steel underneath was unprotected.

Now the cleaned comparison. The same loop chemically cleaned, flushed to a measured endpoint, refilled and dosed to 100 percent read 92 percent at 30 days and 88 percent at 90 days, on a metered 14 and then 44 gallons of makeup. The 92 to 88 step is what 5 percent dilution predicts on its own, so consumption is at or below test tolerance, and that is only legible because the makeup was metered. That is what a satisfied surface looks like, and it is the reading that tells you the film has formed.

A residual that keeps falling months into service is a different finding and worth separating: with a clean system and a stable makeup volume, continued loss means either an unmetered leak or active corrosion still consuming chemistry, and the makeup meter is what tells the two apart.

Outcome two: the system where the gate does not fire

A packaged chilled water skid arrives with the manufacturer's cleaning and passivation certificate for its wetted circuit, connected to distribution that is entirely polymer pipe, fusion-joined, with no ferrous content, no field brazing, no flux and no threading. The system was pressure tested with treated water that was not left standing.

The gate does not fire, and the correct answer is a flush to remove installation debris and displace the test water, a fill, and treatment to target. Running a full alkaline and acid clean here would be work with a real downside: an inhibited acid formulated for mill scale has nothing to remove, and both extremes of pH are hostile to some polymer and elastomer components that were never in a steel system's cleaning specification.

The instructive part is what would flip it back. Any field-brazed copper, any black steel, any threaded joint, any component without a cleaning statement, or test water left standing, and the gate fires again for the whole circuit rather than for the added part, because one uncleaned branch contaminates the loop it joins.

Choosing the cleaning chemistry without wrecking a metal

Alkaline detergent and dispersant handles oil, general debris and light deposits, circulated warm for the period the product specifies. It is the default first stage and it is what most new systems actually need.

Inhibited acid handles mill scale and hard mineral deposit. It removes metal as well as scale, so it is time-limited, inhibited, and monitored rather than left to run, and it is followed immediately by neutralisation and a flush because acid residue left in a system is a corrosion programme of its own.

Two exclusions belong in the same breath as the recommendation. Acid strips the zinc from galvanized steel, so an acid clean on a galvanized system removes the protection and is not a decision to make on site. Aluminium is amphoteric, meaning it is attacked at both high and low pH and is stable only in a mid-range window, so a strong alkaline cleaner is as damaging to it as an acid is; where a system contains any aluminium component, the pH limits come from that component's manufacturer and they govern the whole circuit.

Then a full flush to a measured endpoint, refill, and dose. Do not leave a cleaned system standing full of raw water, because freshly cleaned bare metal in aerated water corrodes fast and the film you are trying to establish forms under treatment, not before it.

Proving the system is clean before you treat it

Look at something removable. Pull a strainer basket, a flanged spool or a coupon and look at the bore. This is the only direct evidence anyone gets.

Run a patch test on the circulating water. A measured volume through a fine membrane patch, with a magnet run underneath it, separates iron oxide made inside the system from mineral and construction grit, and a patch that is still loading heavily says the flush is not finished.

Check the flush endpoint chemically, not visually. Conductivity within a stated margin of the makeup water is a measurement; clear water at the hose is an impression.

Then watch the residual. The residual holding flat over the first month at a known makeup volume is the proof that the surface is clean and the treatment is doing what it was bought for. Where a programme uses corrosion coupons, a rack in a side stream exposed for the supplier's stated period and returned for mass loss gives you a corrosion rate rather than an inference, and that rate is the number to hand the customer at handover.

References

• Water treatment supplier procedures and target residuals for pre-commission cleaning, passivation and inhibitor dosing
• OSHA 29 CFR 1910.1200, hazard communication, for safety data sheet availability and product-specific protection, and 29 CFR 1910.147, control of hazardous energy, for isolating circulating pumps and pressurised sections
• Equipment manufacturer limits on cleaning chemistry, particularly pH limits where aluminium or galvanized components are present
• The adopted plumbing code for backflow protection at a fill connection to a chemically treated system, and the local sewer authority for discharge limits on spent cleaner
• See related: How to Flush a System So the Flush Removes Something; What Passivation Is and When It Matters; Galvanic Corrosion and the Metals That Fight