How Fouling Changes Heat Transfer

Why this matters

Fouling is the only equipment fault that arrives on a schedule and still surprises everybody. Nothing breaks, nothing trips, and the customer's first complaint is usually about comfort or recovery time rather than about the equipment. By then a tech has often already cleaned the wrong surface, because the visible face is the one you can reach and the controlling one is frequently the one you cannot. This card follows a single job where that mistake was avoidable, and the number that avoided it.

The load-bearing idea: a deposit is a resistance added in series, and what it costs you depends entirely on how large a share of the existing total resistance it becomes. The same layer thickness can be invisible on one side of the same exchanger and a third of the capacity on the other.

Before any of this: the controls for what you are about to do

Every step below involves either a descaling chemical or a hot pressurized circuit, and the protection differs by route of exposure.

• Acid descaling solution: inhalation is the route that gets skipped. Circulating warm acid off-gasses mist and vapor, so run it with the space ventilated and with respiratory protection selected and fit-tested under 29 CFR 1910.134 (fit testing at least annually per 1910.134(f)(2)), not with gloves alone. Chemical splash goggles and a face shield under 29 CFR 1910.133 handle the contact route, and the specific product's SDS, which you are entitled to under the hazard communication standard at 29 CFR 1910.1200, names the concentration limits and the neutralizer.
• Never introduce an acid into a system that has held a chlorine-bearing cleaner or bleach, and never store the two on the same shelf on the truck. The mixture releases chlorine gas, which injures at concentrations well below what feels alarming.
• Opening a plate pack, a strainer, or any joint on a hot circuit: water above 140 F scalds on contact, and above about 180 F a line opened under pressure flashes to steam and sprays. Isolate the section, drain it to an open point, and confirm a gauge on that section reads zero before you loosen a fastener. Let the metal come down in temperature before handling it or use insulated gloves rated for the contact temperature.
• Any pump or fan on the circuit: lock and tag the disconnect for the mechanical and stored-energy hazard under 29 CFR 1910.147 before you break into the loop, so a pressure-drop switch or a time clock cannot restart it against your open joint.
• Cleaning a fouled air-side coil: alkaline and acidic coil cleaners aerosolize under a pressure sprayer, so the control is respiratory plus eye protection, and the wash should run outdoors or with the space ventilated. Dry-brushing an unidentified deposit puts it airborne; wet the surface first.

The signal

Domestic hot water off a brazed plate exchanger fed from a boiler loop. The complaint was drift, not failure: outlet temperature used to hold through a long draw and now sags after about a minute. No alarms, no codes, boiler cycling normally.

Commissioning readings were in the job file, taken two heating seasons earlier at the same flows and the same entering temperatures. That record is what made this job a measurement instead of a guess.

Duty then: 500 x 8.0 x 30 = 120,000 Btu/h on the boiler side, and 500 x 4.0 x 60 = 120,000 Btu/h on the domestic side. The two sides agreeing is what lets you trust either.

Duty now: 500 x 8.0 x 22 = 88,000 Btu/h, and 500 x 4.0 x 44 = 88,000 Btu/h. Again balanced, which rules out a measurement problem on one side.

Delivered duty is at 73% of the commissioning figure (88,000 / 120,000).

Convert the loss into resistance, because that is what fouls

A percentage of capacity is the wrong unit for this fault, because capacity also depends on the temperature difference and that difference moved. Convert to UA, the exchanger's own conductance, using the counterflow log mean temperature difference on both sets of readings so the comparator gets the same treatment as the corrected figure.

Then: gaps of 180 - 115 = 65 F and 150 - 55 = 95 F. LMTD = (95 - 65) / ln(95 / 65) = 30 / 0.3794 = 79.1 F. UA = 120,000 / 79.1 = about 1,517 Btu/h per F.

Now: gaps of 180 - 99 = 81 F and 158 - 55 = 103 F. LMTD = (103 - 81) / ln(103 / 81) = 22 / 0.2402 = 91.6 F. UA = 88,000 / 91.6 = about 961 Btu/h per F.

UA is at 63% of commissioning while capacity is at 73%, and the difference between those two numbers is the whole lesson. As the exchanger transfers less, the streams stay further apart, the mean temperature difference rises from 79 F to 92 F, and that larger driving force hides part of the damage. Judge fouling by UA, never by capacity.

Total resistance is the reciprocal of UA. It went from 1 / 1,517 = 0.000659 to 1 / 961 = 0.001041 in those units, so something added 0.000381, which is 58% on top of the original total and 37% of what the total is today.

Ruling out the cheap explanations first

Three things imitate fouling and all three are quicker to check than a teardown.

Flow. Both flows were read at the device, not taken from the schedule, and both matched commissioning. Had domestic flow drifted up, the outlet would sag for a reason that has nothing to do with the surface. Flow enters the duty calculation on each side independently, so an assumed flow figure corrupts that side's duty by exactly its own error and nothing else flags it.

Entering temperatures. Boiler supply was still 180 F. A boiler that had quietly been turned down would produce the identical complaint with a clean exchanger, and the LMTD math would then correctly show UA unchanged.

Air in the pack. Trapped air is a fouling impostor that costs nothing to eliminate: it removes wetted surface rather than adding a deposit, so it also reads as low UA. It was purged at the high point before the second set of readings. If purging restores UA, you were never dealing with fouling.

Locating the side, without opening anything

Here is the test that decided the job. For fully developed turbulent flow, the film coefficient on a surface rises roughly with velocity to the 0.8 power. A deposit's resistance does not care about velocity at all. So stepping the flow on one side and watching UA tells you how much of the total resistance lives in that side's film.

Domestic flow was stepped from 4.0 to 6.0 gpm, a factor of 1.5. That predicts the domestic film conductance rising by 1.5 to the 0.8 power, about 1.38, so that film's own resistance falls to about 0.72 of what it was.

Observed UA at the higher flow: about 995 Btu/h per F, up 3.5% from 961.

If the domestic film held a fraction f of the total resistance, cutting it to 0.72 would raise UA by a factor of 1 / (1 - 0.28 x f). Solving against the observed 1.035 gives f of about 0.12. The domestic film is roughly an eighth of today's resistance. It is not the problem, and increasing domestic flow permanently would buy back about 3%, which is not a repair.

Working backwards with that same absolute film resistance against the commissioning total, the domestic film was about 19% of the resistance two seasons ago. It did not get worse. It got smaller as a share because something else grew underneath it. That is the series-resistance behaviour that makes fouling hard to see: the healthy components look like they are shrinking.

Today's resistance, apportioned: roughly 37% new deposit, 12% domestic film, and the remaining 51% split between the boiler-side film, the plate metal, and whatever was on the boiler side at commissioning. The new deposit is the single largest identified item and it is on a surface nobody had opened.

Confirming it, and what the confirmation is worth

The pack was isolated, drained, and circulated with an inhibited descaling solution under the controls listed above, then flushed and refilled. Repeat readings at the original 4.0 gpm and the original entering temperatures gave a domestic outlet of 114 F and a boiler-side drop to 151 F, so 500 x 4.0 x 59 = 118,000 Btu/h and UA back to about 1,480, which is 98% of commissioning.

That last figure is the verification, and it is stronger than "hot water is better now" for a reason worth naming: it is taken at the same flows and entering temperatures as the baseline, so it is a like-for-like comparison rather than a comparison against a mild-day condition that would have passed anyway.

The deposit was carbonate hardness scale, which is what most well and municipal water lays down on a hot surface because calcium carbonate becomes less soluble as temperature rises, so it comes out of solution exactly where the metal is hottest. Not every deposit behaves that way: sulfate scale, silica, and biological films each have their own solubility behaviour and some are not acid-soluble at all, so identify the deposit or have the water analysed before selecting a chemistry rather than reaching for the acid by default.

Turning two data points into an interval

The exchanger lost about 37% of its conductance over two seasons of service. Deposit resistance grows with accumulated thickness, and thickness grows with hours of service and with hardness delivered, so over a short history a straight line is a defensible first approximation and a conservative one, since many deposits accelerate once the surface is roughened.

Set the trigger on UA, expressed against the commissioning number, and state all three parts of the rule so a second tech applies it the same way. Unit of analysis: the exchanger, at commissioning flows and entering temperatures. Trigger: UA at or below 80% of commissioning. Action when triggered: clean, then re-baseline. On a straight line from 100% to 63% over two seasons, 80% arrives around the thirteenth month, which sets an annual check of UA rather than an annual clean. The check is a ten minute reading; the clean is a half-day with chemicals in an occupied building, and the point of the trigger is to stop paying for the second one on a calendar.

The failure mode this replaces is the one that produced the call: no baseline, no interval, and a first inspection that happens only after a customer notices. A shop that records commissioning UA on every exchanger it installs converts every future fouling complaint from a teardown into a ten minute measurement.

References

• ASHRAE Handbook, Fundamentals volume, chapters on heat transfer and on forced-convection correlations for in-tube flow
• TEMA standards for tabulated fouling resistances used in shell-and-tube design allowances
• 29 CFR 1910.134 (respiratory protection, with fit testing at 1910.134(f)(2)), 29 CFR 1910.133 (eye and face protection), and 29 CFR 1910.1200 (hazard communication and SDS access)
• 29 CFR 1910.147 for mechanical isolation and stored energy before breaking into a pumped circuit
• See related: What a Heat Exchanger Is Actually Doing; Why Approach Temperature Matters