What Suspended Solids Do Once They Are Inside a System

Why this matters

Dirt in a system reads as a housekeeping issue and behaves as a corrosion variable. Solids in motion are almost harmless; solids that stop moving become an insulating blanket, a shelter that inhibitor cannot get under, and a habitat for the organisms that pit steel from the inside. That is why a loop can carry visible dirt for years with no complaint and then produce a pinhole, and why the pinhole shows up where the flow was slowest rather than where the water was dirtiest. If you take one thing from this card, take that solids do their damage after they settle, which makes velocity and deposition, not turbidity, the numbers to manage.

Before opening a loop that has run dirty, isolate the section and relieve pressure through a drain valve to a receptor, confirming zero on a gauge, because a strainer cap on a live loop releases with force. Where deposits smell of rotten eggs, treat the space as containing hydrogen sulfide, which is an inhalation hazard released when sulfide-bearing sludge is disturbed: ventilate, keep your face out of the opening, and where the work is inside a pit, tank or basin, do not enter without the atmospheric testing and entry controls required by 29 CFR 1910.146, since a sulfide atmosphere in a confined space kills before it smells worse. Take glove class and any respiratory control from the safety data sheet for whatever cleaner or biocide the system holds.

Where solids come from, and the source changes the answer

Three origins, and telling them apart is the first real diagnostic step.

Brought in at construction. Mill scale off black steel, weld slag, swarf and cutting chips, thread compound, tape shreds, insulation and cardboard fibres, and the sand that comes with a hydrostatic test taken from a yard hydrant. This load is finite. It comes out with a proper clean and flush and it never comes back.

Brought in with makeup. Silt and sand from a well or a distribution main, plus everything an open cooling tower scrubs out of the air it pulls through. This load is continuous and proportional to makeup volume, which is why a system with a leak has a dirt problem as well as a chemistry problem.

Generated inside. Corrosion product, mostly iron oxides, plus biological mass and precipitated scale that has spalled off a surface. This load is self-sustaining and rises with the very deposits it forms, so it is the one that gets worse on its own.

A shop that treats an internally generated load as a filtration problem installs filters forever. A shop that treats a construction load as a corrosion problem doses inhibitor into a system that needed a flush.

Where solids stop moving

Deposition is a velocity problem. Fine particles stay in suspension as long as the flow keeps them there, and drop out wherever the local velocity falls below the transport threshold for that particle. The design convention for keeping fine silt and corrosion product moving in a horizontal service water line sits on the order of 2 feet per second, derived for typical service water carrying light silt in a full pipe; a coarser or denser solid needs more, a vertical run needs less to keep material moving up, and the figure is a transport threshold rather than a code limit.

So the map of where solids collect is the map of where velocity drops:

• the bottom of horizontal runs, especially oversized pipe left from a load that was reduced
• inside heat exchanger tubes and the lower tubes of a coil, where flow divides among many parallel paths and the parallel paths are not equal
• a tower basin, a boiler mud drum, a buffer tank, any deliberate low-velocity volume
• immediately downstream of a sudden enlargement or a partly closed valve, where the flow separates
• dead legs, capped stubs, isolated branches and bypasses, where velocity is zero by design

Note the feedback that makes this progressive. A deposit narrows the passage, which raises resistance, which reduces flow through that path in a parallel network, which lowers the velocity further, which deposits more. Nothing about the water changed; the geometry changed, and it changed in the direction that makes the next hour worse than the last one.

What a deposit does that suspended dirt does not

It insulates. A porous sediment layer holds stagnant water in its pores and conducts heat poorly, so it costs approach temperature the way mineral scale does. The scale article in this library carries the heat transfer arithmetic; the point here is that the mechanism does not require the deposit to be hard or mineral.

It hides the metal from the inhibitor. Corrosion inhibitors work at the surface, by filming or by maintaining a passive layer, and they can only do that where the treated water reaches the metal. Under a deposit, the bulk water residual you measure at the sample point is irrelevant to the square inch of steel that is corroding. This is the reason a system can test perfectly on paper and pit anyway, and it is the single most useful idea on this card.

It sets up an oxygen concentration cell. Water under a deposit is depleted of oxygen relative to the water beside it, which makes the covered metal anodic to the clean metal around it. The area ratio does the damage: a small covered anode against a large exposed cathode drives deep, local metal loss rather than general thinning. The library's articles on pitting, crevice and galvanic corrosion carry that mechanism in full; what belongs here is that a deposit manufactures the geometry those articles describe.

It shelters organisms. Sulfate reducing bacteria need the low-oxygen conditions that exist under a deposit and nowhere else in an aerated system, and their by-product attacks steel and produces the black sulfide sludge and rotten-egg smell that identifies microbiologically influenced corrosion in the field. Biocide dosed into the bulk water does not reach them either, for the same reason the inhibitor does not.

It plugs the smallest passage first. Control valve seats, balancing valve ports, spray orifices, plate exchanger channels and instrument taps go before the pipe does, and the symptom is a control problem rather than a water problem, which is why the plumbing and controls trades often chase the same fault from two directions.

Reading the dirt

The patch test. Pass a measured volume of system water, one litre is plenty, through a fine membrane patch and look at what it collects. This tells you the concentration and the character in one shot, and it takes minutes.

The magnet. Run a magnet under the patch. Black magnetic solids are iron oxide made inside the system. Non-magnetic tan, grey or gritty solids are mineral or brought in from outside. That distinction chooses between "install filtration" and "stop the corrosion," and getting it backwards means filtering forever while the metal keeps thinning.

The iron trend. Total iron in a closed loop should settle to a low, stable value once a system is clean and treated, with the ceiling set by the treatment supplier's own programme. Watch the trend rather than the number: iron that rose during initial passivation and then flattened is a film forming; iron that keeps climbing sample after sample is metal leaving the system.

The strainer basket. Useful for mass, useless for count. A satisfying handful of debris tells you nothing about the fine fraction still circulating, because mass scales with the cube of particle size and the count is dominated by particles a basket never sees.

Worked case: one coil, two numbers moving opposite ways

A chilled water loop of about 900 gallons serves three air handlers at roughly 55 gallons per minute each. The complaint is that the second unit stopped holding its supply air temperature on warm afternoons, and the building had already replaced its control valve actuator.

The measurements were flow and differential pressure across the coil, taken at the same time. Flow through that coil read 38 gallons per minute against 55 design, which is 69 percent. The differential pressure across the coil read higher than its commissioning value, not lower.

That pair is the whole diagnosis. In a parallel network the header sets the differential across each branch, so a restriction inside the coil drops the flow through it while the differential across the coil holds or rises, the rise coming from the pump riding back up its curve as total flow falls. On a header with many branches that rise can be small, so read the pair rather than the sign: flow down with differential holding or rising is a restriction inside the measured element. Had the restriction been upstream, a plugged strainer or a valve closing, the flow would still have fallen but the differential across the coil itself would have fallen with it, because a coil passing less water drops less pressure. Two candidate faults, opposite signatures, one pair of gauges to tell them apart.

The patch test on the loop water came back heavily loaded and strongly magnetic, so the solids were internally generated iron oxide rather than construction sand. Opening the coil header confirmed sediment banked in the lower tubes, which is exactly where the parallel paths run slowest.

The correction had three parts, and the reasoning matters more than the parts list. Clean the coil, because the deposit is what is throttling the flow. Fit side-stream filtration, because the loop is making solids faster than it can carry them, and specify the element change as part of the recommendation rather than as a consequence of it. A side stream on a loop making iron oxide and sulfide sludge concentrates both into a bag somebody opens monthly, so the housing needs isolation and a vent to a receptor, the change needs ventilation and the protection the safety data sheet gives for what the loop holds, and the spent elements are a waste stream with a disposal route decided before the first one is fitted. Find the oxygen source, because iron oxide at that concentration is a corrosion rate, and filtering corrosion product without stopping corrosion is a permanent subscription.

Sizing the side stream: the convention is to run a side-stream filter at roughly 5 to 10 percent of system flow, which at 165 gallons per minute total gives a range of about 8 to 17 gallons per minute. At 12 gallons per minute a 900 gallon loop turns over in 75 minutes, which is about 19 turnovers a day, enough that the loop's solids concentration falls on a scale of days rather than seasons. Push the side stream lower to save on element cost and the turnover stretches out until the filter is a decoration; push it higher and you are moving system flow through a filter for diminishing returns.

Confirmation came from three trends read together over the following weeks: the coil's flow returned toward design at a differential matching its clean curve, the patch loading fell sample over sample, and the total iron flattened instead of climbing. The first number alone would only have proved the coil was clean today.

What changes the answer

A once-through system does not accumulate, so its answer is protective straining sized on the largest particle that matters, not on the total load. An open evaporative system takes in solids continuously with makeup and with the air it scrubs, so it needs basin sweeping or separation as a permanent feature rather than a corrective one. And a system whose solids are non-magnetic and whose iron trend is flat has a genuine filtration problem with no corrosion behind it, which is the one case where filtration alone is the complete answer.

References

• OSHA 29 CFR 1910.146, permit-required confined spaces, for entry into a tower basin, pit or tank where sulfide-bearing deposits may be disturbed
• OSHA 29 CFR 1910.1200, hazard communication, for the safety data sheet governing glove class and respiratory control for the cleaner or biocide in use
• Water treatment supplier programme limits for total iron and suspended solids, and filter manufacturer data for element differential and holding capacity
• See related: How Scale Forms and What It Actually Costs; Pitting, Crevice and Uniform Corrosion; How to Flush a System So the Flush Removes Something