Boiler kettling: a six-test diagnostic tree

Open the installation manual of any copper-tube boiler and look for the word “warranty.” You will find a sentence very few building owners have ever read: lime build-up caused by excessive water hardness is not a fault of the equipment and is not covered. Lochinvar goes further and labels an exchanger destroyed by a leak that let the system drink fresh water for months a non-warrantable failure. Which means that when boiler kettling ends in a punctured exchanger, the invoice belongs to the owner — and that the noise had been giving notice for months.

What boiler kettling is actually telling you

The kettle-like rumble has a single mechanism behind it: water boils locally against a surface of the heat exchanger, and the steam bubbles collapse a few centimetres downstream in cooler water. Those collapses are what you hear. Two conditions make it possible, and either one is enough — the wall is too hot, or water is not moving fast enough to carry away the heat the burner deposits into it.

Scale delivers both at once. It insulates the heat transfer surface — Health Canada notes that hard water scales hot water pipes and water heaters — and it narrows the passage, cutting flow. That is why an honest diagnosis never starts with “it needs descaling”: fixing the scale without fixing the flow just rebuilds the same deposit inside three years.

Tools for this job: two contact thermometers or a calibrated infrared gun, a differential gauge or the circulator’s pressure ports, the pump curve for the circulator actually installed, the boiler manufacturer’s manual, an inspection light, and the equipment’s service history.

⚠ Safety — Kettling accompanied by high-limit or relief-valve trips is no longer something to schedule; it is a shutdown. Any work on the gas train, the burner or its settings belongs to a contractor holding the appropriate licence under CSA B149.1. Every test below is on the water side, with the appliance running normally.

Test 1: when exactly does the noise appear?

A noise has diagnostic value only once it is timed. Record the phase precisely: during temperature rise, in steady state, or just after the burner shuts off. A rumble that builds as supply temperature climbs and then settles points to an overheated wall — scale or flow. A sharp bang when the pump stops is far more likely water hammer or badly supported piping, and has nothing to do with the exchanger. Irregular gurgling that ignores temperature means air is still in the loop.

Note the supply temperature at the instant the noise starts. That number is your reference for the five tests that follow.

Test 2: what is the real supply-to-return delta T?

Read both temperatures at the same moment, burner firing and circulator running, as close to the boiler flanges as you can get. Then compare the spread to the design delta T. North American manuals commonly size loops on a 20 °F (about 11 °C) drop — Weil-McLain works its bypass-circulator examples on exactly that basis, where a 1,000,000 Btu/h boiler at a 20 °F drop calls for roughly 100 US gpm.

A delta T well above design means one thing: water is spending too long inside the exchanger and has time to boil. An unusually low delta T points the other way, to excessive flow, which is a different failure altogether — the same manuals warn that too much velocity through copper tubes causes pitting and erosion, also outside warranty.

Test 3: does flow meet the manufacturer’s minimum?

Before you blame scale, prove that water is moving. Published minimums are higher than most people assume: on the Copper-Fin II line, the 300,000 Btu/h model needs 19 US gpm just to close the flow switch, and the 990,000 to 2,070,000 Btu/h models need 26. Below that, the burner should not be lighting at all.

Work the list in order: isolation valves fully open, balancing valve back at its commissioned position, circulator strainer clean, check valve free, circulator turning the right way and not seized, air purged at the high points. In Montreal’s building stock the most common cause is not mechanical but human — a balancing valve closed down during a tenant fit-out and never reopened.

Test 4: how much fresh water is this loop drinking?

This is the test that separates a diagnosis from a hunch. A healthy closed loop consumes almost nothing. Weil-McLain is blunt about the alternative: continual make-up water reduces boiler life, minerals build up in the sections, heat transfer drops, the cast iron overheats and sections fail.

Read the make-up meter if there is one. If there isn’t, close the fill valve for a week and watch system pressure. A measurable drop means a leak somewhere — a pump seal, an air vent, a coil. If pressure climbs instead of falling, the next suspect is the expansion tank and its pre-charge, whose failure pushes the relief valve open and triggers still more make-up. Island of Montreal water is moderately hard and stays under the 7-grain threshold at which Weil-McLain requires a softener — but a loop taking in a few dozen litres a week still imports, winter after winter, more than enough to line an exchanger.

Test 5: what does the exchanger show once it is open?

The first four tests let you open the appliance knowing what you are looking for. When opening it up means half a day of teardown, a borescope inspection through the sight port often settles the question before a wrench comes out. Otherwise, pull an inspection plug, a section plate or the bundle cover, and read three things: deposit thickness, colour and location. White to beige, hard, concentrated on the hottest side, is lime — the make-up water story. Black, soft and spread everywhere is magnetite: a corrosion and dissolved-oxygen problem, not a hardness problem, and the remedy is different.

Take a water sample while the unit is open. pH is the highest-value single reading: Weil-McLain recommends keeping boiler water between 7.0 and 8.5, and a value outside that band redirects the whole correction plan toward a boiler water treatment program.

Test 6: correct in order, then close the loop on paper

Sequence matters more than technique. Restore flow first — it is nearly free, and it protects the exchanger through everything that follows. Eliminate the fresh-water intake second; without that step, descaling is a recurring expense. Treat or replace the exchanger last, once the cause that destroyed it is no longer running.

Then close the loop in writing. Log delta T before and after, verified flow, measured make-up volume and the nature of the deposit in the equipment file. That page is what makes the next noise readable in ten minutes rather than half a day, and it is the same page that shows documented boiler service was in fact performed if the question ever becomes contractual.

Field case: the Saint-Laurent boiler that sang every evening

Three-storey office building in Saint-Laurent, boiler room in the basement, two hot water boilers in cascade. The caretaker reported a rumble in the evening, never in the morning. The previous vendor’s first proposal was a full descaling of both units.

The measurements said otherwise. On the noisy boiler, the supply-to-return spread was far wider than on its silent twin at the same setpoint. The circulator ran, the strainer was clean. The secondary loop’s balancing valve, however, had been closed three-quarters of the way during a floor renovation a year earlier. In the evening, as zone valves closed, remaining flow dropped below threshold and water started boiling against the wall. Second finding: an automatic air vent on the roof was weeping, and the loop was taking in water every week.

Valve reopened and tagged, vent replaced — the noise was gone the same day. The exchanger, inspected three months later as a precaution, carried only an early deposit. The original quote covered two appliances; the actual fix took half a day.

Listen to your boiler rooms before the first cold night

Do a listening round this week, while the boilers are still barely firing, and have two numbers written into each equipment file: the measured delta T, and the make-up water volume over the last twelve months. Those two figures are enough to classify a noise the moment it shows up in November, and to know whether the call that follows is about a valve or an exchanger.

That distinction is not a technician’s detail. It is the difference between half a day of labour and an exchanger replacement the manufacturer will decline to cover — which is why service reports from Montréal Combustion always carry a measured delta T beside the noise being described.