A new condensing boiler bolted onto forty years of unloved pipework is one of the most common installations in British housing, and one of the most likely to fail early. The reason is not the boiler. It is the water it inherits, and the fact that modern heat exchangers are far less forgiving than the cast iron lumps they replaced.
Ask a heating engineer what kills new boilers and you will not hear much about electronics. You will hear about black water.
The material in question is magnetite, an iron oxide that forms when oxygen, water and steel radiators are left to get on with each other over a decade or two. It is magnetic, dense and mildly abrasive, and once it is circulating there are only a few sensible responses: clean the system properly, capture the debris continuously, or take the new appliance out of contact with the old water altogether by separating the circuits with a plate heat exchanger. Which of those three is right depends almost entirely on how much you know about the pipework you are inheriting.
What follows is the reasoning behind that choice, because the sales conversation about power flushing tends to skip it.
Why the old boiler put up with it and the new one will not
Corrosion in a sealed or open vented heating system is an electrochemical process. Oxygen enters through the feed and expansion tank, through permeable plastic pipe, or simply every time the system is drained and refilled. It reacts with steel radiators and steel pipework, and the product is a fine black sludge that settles at the bottom of emitters and circulates through everything else.
A 1980s cast iron boiler had waterways you could put your fist through and water velocities low enough that debris passed straight through. A modern condensing heat exchanger is a completely different object. The waterways are narrow, the surface area is deliberately large, and the water moves fast to get the return temperature low enough to condense. Put a spoonful of magnetite into that and it does not pass through. It lodges.
Once a passage is partly blocked, the water behind the deposit stops moving properly and the metal on the other side keeps receiving heat. Local boiling follows. That is kettling, and the banging noise homeowners describe as the boiler sounding like a kettle is the sound of steam bubbles collapsing inside a component that cost several hundred pounds to make.
The other failure route is mechanical. Abrasive particles wear pump bearings, jam diverter valve mechanisms and clog the small orifices in automatic bypass and pressure differential valves. None of it happens dramatically. It happens over two or three winters, which is usually just long enough to be outside the warranty conversation.
The symptoms worth taking seriously
You rarely get a clean diagnosis, but the pattern is recognisable:
Radiators hot at the top and cold across the bottom third, especially the ones furthest from the boiler
Water at the drain point coming out grey or black rather than tea-coloured
Banging or ticking from the boiler shortly after it fires on a cold start
A circulating pump that is louder than it was, or a second pump failure in five years
A magnetic filter that is full again three months after being cleaned out
That last one is the most useful signal on the list. A filter that keeps refilling is telling you the system is still generating debris, not that the filter is doing a great job.
Cleaning: what a flush can and cannot fix
There are three broad approaches and they are not interchangeable.
A chemical clean means dosing a cleanser, running the system hot for a period, then draining and flushing with fresh water. It is cheap, it is low risk, and on a system in reasonable condition it is often enough.
Power flushing adds a pump that reverses and increases flow through each radiator in turn, usually combined with a magnetite capture unit on the flow and some percussive encouragement of the emitters. It shifts far more material. It also takes most of a day, costs several hundred pounds, and carries a genuine risk that is rarely mentioned in the quote: on very old pipework, particularly microbore and buried steel, dislodging decades of deposit can open up pinhole leaks that the sludge itself was plugging.
Mains pressure flushing sits between the two and works well on systems with good access and no obvious fragility.
BS 7593:2019, since amended, is the code of practice the whole argument sits on. At summary level it expects a chemical clean and a fresh water flush before inhibitor is added, it expects a permanent in-line filter fitted to all systems rather than treated as an optional extra, it expects an on-site water test annually to check inhibitor level and system cleanliness, and it expects inhibitor to be re-dosed at five year intervals or the water sent for laboratory analysis instead. Approved Document L in England refers directly to that standard, and most boiler manufacturers make documented cleaning and inhibitor a warranty condition. Which is why the Benchmark checklist has water treatment entries on it at all.
Worth knowing: in hard water areas, above roughly 200ppm, a scale reducer is recommended on top of all of the above. Corrosion and scale are separate problems with separate solutions.
Filters and inhibitor are maintenance, not a cure
A magnetic filter is a capture device. It removes what is already circulating and it does that well, but it does not stop the reaction that keeps producing more. That job belongs to the inhibitor, and inhibitor depletes: every partial drain-down to move a radiator, every top-up after a leak, every bit of remedial work dilutes it.
The annual test that the standard asks for takes about two minutes with a test kit and almost nobody does it. If you take one practical thing from this article, make it that one.
On larger systems, and on anything sealed where draining a radiator to pour chemicals in is either awkward or actively unhelpful, inhibitor goes in through a dosing pot instead. It is a small vessel plumbed across the circuit with isolation valves, filled, then flushed through by the system flow. Same chemical, no guesswork about dilution, and it gives you somewhere to draw a genuine water sample from.
When separation is the honest answer
Sometimes the correct engineering answer is that the existing water is not going to be brought up to standard, and pretending otherwise transfers a risk onto whoever signs the commissioning sheet.
The situations recur:
Communal pipework in a converted building, where nobody can tell you what the risers are made of or when they were last touched. Systems with buried or inaccessible steel. Mixed metal installations where aluminium components sit alongside old steel. Listed buildings where lifting floors is not on the table. And any commercial retrofit where the client wants a new appliance this month and a system refurbishment never.
In those cases you build two circuits. The primary contains the new appliance and water you filled and treated yourself. The secondary is the existing system, as it is. Between them sits a plate heat exchanger: a stack of thin corrugated stainless steel plates forming alternating channels, with the two fluids running counterflow past each other through the plate wall. Heat crosses. Water does not. Whatever is circulating in the old system stays in the old system.
Stainless steel matters here, both for corrosion resistance and because these units are frequently brazed and cannot be opened for cleaning, so what goes into them has to be filtered.
The trade-offs are real and worth stating. You lose a few degrees to the approach temperature, which means the primary has to run slightly hotter to deliver the same secondary flow temperature. You need a second circulator. There is more kit in the plant room and a higher parts bill. Separation is not the default answer, and anyone presenting it as one is selling rather than advising.
What it buys you is a boundary. The appliance warranty stops depending on the archaeology of somebody else’s pipework, and that is often worth more than the components cost.
Questions installers get asked
Does a new boiler really need a power flush?
Not always. It needs a documented clean appropriate to its condition, plus a filter and inhibitor. On a reasonable system a chemical clean and flush meets that. On a heavily contaminated one it will not.
Will a magnetic filter clean an existing system?
Slowly and partially. It captures circulating debris but does not remove what has settled in radiator bottoms, and it does nothing about the corrosion still going on.
Does hydraulic separation cost efficiency?
Slightly, through the approach temperature and the extra pump. On a system that would otherwise foul a new heat exchanger, that loss is trivial by comparison.
How often should inhibitor be checked?
Annually, at the boiler service, with a full re-dose or a laboratory water test every five years.
Before the new appliance goes in, get three things in writing: what cleaning method was used, which inhibitor and at what concentration, and whether a permanent filter has been fitted. Those three lines on the commissioning record are what a warranty department will ask for, and they are considerably easier to obtain on the day than eighteen months later.
