Home Page
pictures of buildings I have successfully converted from steam heat to hydronic heat

Steam Pipes Barely Corrode

Old steam pipes cut in half are indistinguishable from new piping. The cross section of cut pipe shows no loss of material, even though the pipes are a century old.

This is typical. Steam system operation does not degrade steam piping. In most hydronic conversions, virtually all the pipes and radiators can be re-used. There are some exceptions: buried pipes, patched pipes, and undersized dry returns should be replaced. But virtually everything else is sound.

Re-using existing piping and radiators contributes greatly to the viability of these projects. But widespread misconceptions about these components' service lives must be countered before conversion projects will be seriously entertained.

It is certainly surprising that old steam piping shows such minimal wear. Wet carbon steel regularly flushed with fresh air would seem to be an ideal recipe for corrosion. And yet the pipes look like new. How?

For the same reason that steel shipwrecks in deep water do not dissolve into rust: there's insufficient oxygen available. Water doesn't rust steel, oxygen does. Water is just the catalyst.

Even a hundred years of air-flushing doesn't cycle enough oxygen through steam piping to generate significant corrosion. If the pipes stayed wet on the outside they'd deteriorate. But too little air travels through steam systems to cause appreciable interior corrosion.

Proof of this can be found at every valve in a steam system. It's common knowledge that you shouldn't mix brass valves with galvanized pipes. The pipes will quickly rot and leak due to galvanic corrosion. Even worse would be mixing brass valves with plain steel pipe, because without the protection of the zinc layer, the steel pipe would corrode even faster.

Virtually all steam radiators have brass hand-valves with black pipe screwed directly into them, with a film of water bridging the joints. Yet even after a hundred years these pipes are not leaking. How could this be?

It takes more than dissimilar metals for galvanic corrosion to occur. There also has to be oxygen to attack the anodic metal. Fresh water contains so much oxygen that even galvanized piping succumbs to it. The absence of galvanic corrosion next to brass hand-valves in steam piping proves that these systems do not admit sufficient oxygen to affect the piping. If they did, every pipe screwed into a hand-valve would have leaked decades ago.

Testing the Piping

This is not to say that a steam system can be filled with water without proper air-testing beforehand. In a system of any size, a few leaks will be found. But these leaks are not a result of deterioration: almost all of them occur at joints in fairly recent work, where components were changed. The original joints and components virtually never leak.

The crucial thing is to find and repair leaks before filling the system with water. The first step is to fill the system with pressurized air, then inspect all piping while listening for hissing air leaks. Almost all leaks will be detected this way. Once the needed repairs are performed, the test should be repeated to check the work.

The second step is to fill the system with pressurized air that has been scented with peppermint oil, and repeat the walk-through. The aroma of peppermint will be noticeable near leaks.

The final step is to fill the system with water, heat it to its maximum operating temperature, then repeat the walk-through. A couple slow, seeping leaks at joints may be discovered at this walk-through.

This testing protocol requires assiduosness from the installer, and unhindered access to the building. But if it is rigorously followed, no damage from water leaks will occur.

As it happens, the converted system will have probably no more leaks than a newly piped steel or copper system. Even brand-new piping contains leaks and requires testing. The complicating factor in a hydronic conversion is not the age of the piping, it's that the testing occurs in an occupied building. The challenge is logistical, not material.

Pipes to Be Replaced

As noted, a few sorts of piping should not be re-used. First and foremost, assume that all pipes embedded in dirt or concrete leak. Buried pipes are notorious for leaking, probably because of electrolysis (buried steel pipes form an excellent ground for a building's electrical wiring).

Patched pipes should be replaced. This issue is typically encountered only in very old loft buildings. These had very crude heating systems with undersized pipes run with minimal pitch. Such piping hammers, and thus leaks (hence the patches).

Pipes out of service should not be used. Steam system pipes are subjected to thermal expansion with every heat cycle, and so can be counted on to remain sound when heated. Out of service pipes have not been regularly stressed this way, and could fail under thermal expansion.

Systems supplied with district steam may be suspect. I have heard that district steam is corrosive, and I know of a district steam building in which the risers had all rotted. While these data points are anecdotal, they are enough to suggest that a system supplied with district steam should not be converted to hydronic until the effect of district steam on piping is understood.

Suitability of Heaters

Virtually every radiator, convector, and blower can be converted to hydronic operation. However, certain very old steam radiators will not operate at full output. The sections in these radiators connect to each other at the bottom but not at the top (the sections in more modern radiators connect at both top and bottom). Each section thus becomes an isolated chamber that traps air. Since it is not practical to install an air vent onto every radiator section, these radiators cannot be vented, so they stay mostly filled with air. But not completely filled. The pressure in a hydronic system is high enough to compress the trapped air, so these radiators get at least partly filled. At 15 PSI, a radiator will be half-filled with heating water. And heat flows from the heating water into the metal more rapidly than it flows from the radiator to the room air, so a half-full radiator can get more than halfway hot. On the other hand, the connections on these radiators allow the water to flow straight across, potentially reducing how much heat the water gives up to the radiator.

But: then again, most radiators are vastly larger than they need to be. They were sized to keep rooms warm even with the windows open. Now the windows will be closed, and they're much better windows. On top of that, with conversion you're going from a system that turns on and off repeatedly to a system in which heat is always available at every heater.

So if the radiator sections are not connected at the top, do some heat-loss calculations, and compare the heat-loss to the radiators' output. If the radiators' rated output is at least double the heat loss, you'll probably be fine. It's hard to put real numbers to this, but typically even an undersized radiator will be adequate in all but the coldest weather, and my experience is that typical heat loss calculations result in oversized radiators. Radiators that prove inadequate are easily changed out later. The degree of conservatism that's appropriate varies with the building and program, but ultimately these systems are flexible and forgiving. All that said, old radiators of this type are not a common problem.

Cast-Iron Baseboard

One system I converted had long runs of cast-iron baseboard. Several of them leaked at their internal joints (at the push-nipples). The system pressure was much less than the baseboard's rating, and I've seen plenty of cast-iron baseboard work perfectly well in hydronic systems, so it wasn't a material failure. My guess is that these baseboards had been exposed to a lot of steam hammer (unavoidable in long baseboard runs), and this kept the push-joints from fully sealing. Another possibility is that the baseboards had been put together poorly. Their length could have made them hard to assemble, with poor joints the result. Given how rare leaks are, even in pipes that have been hammering for a century, I don't think there's any wider lesson here, other than that long runs of cast-iron baseboard may be problematic.


The taller a building, the greater the pressure inside a hydronic system, due to static head. It's the same phenomenon as occurs when swimming underwater: the greater the depth of water overhead, the greater the pressure on your eardrums.

Steam pipes can withstand very high pressure, but cast iron's limit is much lower. How much lower? Interesting question. Conservative practice is to limit cast iron to 30 psi. As a practical matter that means nothing above six stories.

But a smart plumber once told me that you can go as high as 50 PSI so long as the system has constant circulation with outdoor reset (which minimizes thermal stress). I typically design for 50 maximum pressure, and I have had no problems. (Admittedly, I've only converted about a dozen systems.)

Even so, some additional reassurance might be wanted. I can offer two data points. One is that cast iron boilers are tested to double their pressure rating, and they never fail the test. The second data point is a building I'm aware of where several dozen cast-iron baseboard radiators were inadvertantly exposed to excessive air pressure. It wasn't until the pressure reached ~90 PSI that one of them failed.

Again, conservatism is a judgment call. Sticking to a 30 PSI limit is defensible. But in light of experience and the properties of cast iron, a 50 PSI limit is also defensible.

Expansion Tanks

Given the importance of accurately controlling pressure, proper tank sizing is crucial. In a conversion that size is BIG: these systems need far more expansion capacity than is usual, because the pipe volume is so great. As the water in the system expands while it's heated, the expansion tanks give that added volume of water someplace to go. The more water you start out with, the more expansion you have to accommodate, and the more tanks you need. Yes, "tanks" plural: in any building larger than a few apartments you'll need several tanks. With cast-iron components, accurate pressure control is essential, and adequate expansion capacity is part of that. Siegenthaler's methodology makes the calculation straightforward.

The tanks are more effective the higher they are, so in a building of any height, the size and/or number of tanks can be reduced if the tanks can be installed into a roof bulkhead or on the top floor.

Expansion Tank Piping

This may seem counterintuitive, but the connection from the tanks to the system can be 1/2" piping. The flow through this piping is miniscule: as the water heats up or cools down, water moves between the system and the tank. But this expansion and contraction happen slowly, so the flow is tiny. Minimal flow means minimal pressure loss, so small piping is fine.

The tanks should connect to the system upstream of the circulator, with minimal distance between the circulator and the connection. But "distance" in this case isn't physical distance, it's resistance, which you can also call pressure drop. For example, if a system has a common return that's made of 4" pipe, you could connect the expansion tanks anywhere along that return with virtually no reduction in expansion capacity. This can be very useful when shoe-horning large tanks into an old building.

I prefer not to have a valve between the tanks and the heating plant. It's too easy for someone to leave the valve closed, causing the system to become over-pressurized. Instead, rely on the heating-plant isolation valves for tank testing and replacement. To make life easier, offset the tank connection up to the ceiling then back down before connecting to the plant, and put a drain cock at the tank connection. This makes it possible to drain the tank without draining everything else at the same time.

If I'm Sizing a Boiler for the Converted System, I Have to Add Capacity to Account for the Large Steam Pipes, Right?


I said, If I'm Sizing a Boiler for the Converted System, I Have to Add Capacity to Account for the Large Steam Pipes, Right?

No, you don't.

But Someone Told Me That if I'm Sizing a Boiler for the Converted System, I Have to Add Capacity to Account for the Large Steam Pipes

They're wrong.

But They Seemed Smart

So what? I seemed nice.

I'm Sizing a Boiler for the Converted System, So I'm Going to Add Capacity to Account for the Large Steam Pipes

Don't. Just don't. Really. Don't.

This idea about boiler sizing seems to make intuitive sense, which is what makes it so dangerous. So let's go through it.

Steam pipes are really big. Which means they hold a lot of water. And water soaks up and holds a lot of heat. The argument is that before you can heat the building, you have to heat up all that water, so you need to add that heating capacity to the boiler.

Let's start out by accepting the argument's basic premise, which is that you have to heat up the water completely before you can heat the building. The thing to realize here is that the argument makes sense only if the heat that goes into the water doesn't come right back out to heat the building, that somehow the heat just stays in the water. The argument is that first you have to put in enough heat to get the water hot, and then after that you have to put in ADDITIONAL heat for the building, and that's why the boiler has to sized for building + pipes.

Let's accept the argument at face value. I'm going to accept that the boiler has to start out by pouring heat into the water in order to get it fully hot, and only that I put in after that will go towards heating the building. So, I'm going from a cold start, and my pathetic undersized boiler is huffing and puffing, and putting it's entire miserable output of heat into the water, and none of that heat is going to come out of the water to heat the building because the water itself needs all of that heat all for itself. Okay, fine, I get it, the building will get no heat until the water is completely hot. So, how long will that take? About twenty minutes. That might seem bad, until you realize that a hydronic system only goes from a cold start in fall. After that it operates continuously, staying warm. So even if the argument were right, the only effect would be a twenty-minute delay in fall, when the system first starts up.

But the argument isn't right, because it assumes that heat goes into the water and just stays there. So that's why you need an extra-large boiler, to supply enough heat to heat the building AND to heat all that water. Because the heat is going to go into the water and piping and just stay there forever.

This may be the most intuitive way to understand why the argument is wrong: if I need to supply enough btu's to heat both the water and the building, what happened to all those btu's I poured into the water? Did they just disappear? Does the heat just sit in the water, refusing to come out?

Of course not. All the heat that goes into the water comes right back out into the building. The system isn't a hole in the ground that you pour heat into. It's a conveyor belt that quickly carries heat from one spot to another. As soon as I put btu's into the water, the pipes warm up. And as soon as they start to warm up, they start giving up their heat to the building.

I've gone on at some length on this because the argument is so dangerous. It has an intuitive appeal, and sounds smart, but just results in yet more oversized boilers.

Hydronic Boiler Sizing Based on Historic Fuel Use

Heat loss calculations are a reasonable way to size a hydronic boiler, and often the only way. But the calculations are inherently conservative, mostly due to the unpredictability of air infiltration rates. A more accurate method of boiler sizing is to analyze the building's historic fuel use, comparing it against heating degree days. The math is straightforward.