The Heating System That Worked Until the Boiler Was Replaced

The complaint

A three-story building on low-pressure steam heat. The old sectional boiler was replaced over the summer by a competent crew, the changeover passed its inspection, and the first cold week produced four complaints at once: hammer in the basement mains on every morning start, the top floor at the far end of the building running cold all day, two air handling coils flooding, and the domestic hot water exchanger slow to recover.

Nothing had been replaced except the boiler. Every trap in the building was the trap that had worked the previous winter. Every radiator, every coil, every foot of pipe was untouched. The building had run acceptably for years and now did not, and the only change was upstream of all of it.

Stop the hammer before anybody diagnoses anything

Water hammer in a steam main is not a noise complaint. A slug of condensate driven by steam velocity into a fitting or an elbow arrives with enough force to rupture the fitting, and a rupture in an occupied basement discharges steam at main temperature into the space. That risk is present on every morning start until the cause is removed.

Containment first, on the same visit. Bring the system up slowly rather than letting it come up on its normal call: crack the main supply valve so the mains warm over a long ramp instead of admitting steam at velocity over cold standing water, and keep people out of the affected basement runs while it comes up. Where the site has no way to slow the start, the honest containment is to keep the mains that hammer out of service until the cause is fixed, and to say so in writing.

Nobody stands next to a hammering main to listen for the location. Locate it from the ends of the run and from the pattern across the building, not by walking beside it during a start. And nothing on that system is opened, drained, gauged or unbolted until the section is isolated and locked out as control of hazardous energy under 29 CFR 1910.147 in general industry, or under the construction energy control requirements of 29 CFR Part 1926 on a construction site, drained to a point piped away from any walkway, confirmed at zero on a gauge you have just watched read something else, and cooled below 120 F verified with a contact probe on a spot already established as representative, or with a non-contact reading derated for the surface, because an infrared reading on bare or jacketed pipe reads LOW and low is the dangerous direction for a safe-to-touch check.

What the building could tell us, and what it could not

Four faults, one date. That is the shape of the whole diagnosis and it is worth naming early, because four unrelated components do not fail over one summer in one building. Where a complaint set appears on a date rather than at a load, the change is upstream of everything complaining.

The reason this is worth saying is that it points away from the standard single-device reasoning. On one device with the system pressure unchanged, a failure that shows up at every load points at an obstruction, a shut valve or a genuinely failed-closed trap, which is how the sibling procedure on recognising stall separates stall from the alternatives. Here every device failed at every load simultaneously, and none of those three explanations survives being applied to four devices at once.

So the useful move was not to test anything. It was to read what the building had written down.

Four documents and one absence

The original commissioning sheet, found in a binder in the boiler room, listed the operating pressure control as cut-in 8 psig, cut-out 12 psig. The system had been designed and balanced around a 12 psig header.

The new boiler's start-up sheet listed cut-in 3 psig, cut-out 5 psig. That was a deliberate and defensible choice on its own terms: a lower operating pressure means a lower saturation temperature in the boiler, less standby loss and less flash at the receiver, and the new control could hold it comfortably.

The original trap submittal listed the coil trap capacities as selected at a 10 psi differential.

The as-built riser diagram showed the condensate return main running above the basement ceiling, with the air handling coil traps discharging upward into it.

The absence. Nobody had ever written down the vertical distance from those coil trap outlets up to the return main. It was not on the as-built, not on the commissioning sheet, and the new boiler's start-up sheet had no field anywhere on it that referred to the return system. The one number that decides whether the building drains was not in the building's records, which is why the operating pressure could be changed without anybody being wrong on any form.

The number nobody had written down

Measured with a tape, from the coil trap outlets to the return main: 14 feet.

Three things stack up on the outlet side of those traps, and the trap has to beat their sum before one drop leaves. The return main gauge read 1 psig. The 14 foot lift converts at 2.31 feet of water per psi, the value for water at 60 F, and hot condensate near 212 F is about 4 percent less dense so one psi supports nearer 2.4 feet of it; 2.31 is the conservative figure for a lift requirement because it makes the pressure needed come out larger, and it is the one used here. That is 6.06 psi. Add 1 psi as an allowance for friction in a long return run.

Required at the trap inlet: 1 plus 6.06 plus 1, which is 8.06 psig.

The same arithmetic at both pressures

At the old 12 psig header: 12 minus 8.06 leaves 3.94 psi of margin. The coils drained. Not generously, and nobody had ever known how little room they had, but they drained at every load.

At the new 5 psig header: 5 minus 8.06 is 3.06 psi short. There is no operating condition at which those coils drain. They fill on the first call and stay full, which is exactly what the flooding complaint described.

The drip traps on the basement mains are a separate calculation and they moved too. Those discharge into the low-level return at 1 psig with no lift, so their differential went from 12 minus 1, which is 11 psi, to 5 minus 1, which is 4 psi. For a trap passing subcooled condensate through a fixed opening, flow scales roughly with the square root of the differential, so the square root of 4 over 11 is 0.60: about 60 percent of their former capacity. Where the condensate arrives at saturation and flashes across the seat, the flashing vapour chokes the passage and real capacity falls below that, so 60 percent is an optimistic bound and the manufacturer's capacity table at the actual differential owns the real figure.

The condensate load did not fall by 40 percent. The building still needed the same heat, so the mains still made the same water. Undersized drip traps leave condensate standing in a main with steam moving above it, and that is the standing condition that generates hammer on every morning start.

The far end went cold for a second reason, not the same one

It would be tidy to attribute the cold top floor to the waterlogged mains alone, and that is only half of it.

Saturation temperature at 12 psig is about 244 F; at 5 psig it is about 227 F. Against a 70 F room that is a driving temperature difference of 174 F before and 157 F after, a ratio of 0.902, so on a straight linear read the radiation lost about 10 percent of its output. Free-standing cast-iron radiation in still room air does not scale linearly: standard practice takes the exponent near 1.3, which gives 0.902 raised to 1.3, or 0.875. About 13 percent of output, gone from every radiator in the building at once, before any water reaches anything. A forced-convection coil behaves differently and is not covered by that exponent.

Thirteen percent is survivable in mild weather and not survivable at the far end of the longest run in a cold week, which is exactly where the complaint came from. Both mechanisms point the same direction and they arrive together, which is why the top floor was the first to be noticed and the coils were the first to be measurable.

Why every trap in the building tested good

Because they were good. Three traps had already been replaced on the first two visits before this one, and all three were pulled in working condition.

A trap is a valve that distinguishes condensate from steam. It supplies no motive force of its own. Give it no differential and it does nothing, and it does nothing in precisely the way a failed-closed trap does nothing, which is why a temperature check at the trap inlet on a flooded coil returns cool and reads as a confident failed-closed diagnosis. The condensate standing in a partly flooded coil subcools against cooler surface, so the reading is real and the conclusion drawn from it is wrong.

Testing traps on this building was never going to find anything, and it would have kept finding nothing for as many visits as anyone was willing to fund.

The fix, and the limit on the obvious version of it

The direct fix is to put the operating pressure back where the distribution was designed for it. That is a control adjustment, and it is gated by two things that get checked before the setting is touched, in that order.

First, the relief valve setting and the pressure rating of every component on the system. A boiler in the low-pressure heating class under ASME Boiler and Pressure Vessel Code Section IV, in the edition your state's boiler law adopts, is limited to 15 psig, and its relief valve is set at or below that. An operating pressure is raised by adjusting the operating pressure control and never by adjusting, resetting or replacing the relief valve, and where the desired operating pressure does not sit comfortably below the relief setting with margin, the answer is not to move the relief valve.

Second, whether anything downstream was replaced or added at a lower rating during the boiler work, which is a question for whoever did the changeover rather than an assumption.

Where the pressure cannot go back up, the drainage has to change instead. Drop the coil traps to a vented receiver set below the coils and pump from there, or fit a device that supplies its own motive pressure, so the 14 foot lift stops being the trap's problem. The manufacturer owns which of those suits the load. That is real work and it is the honest answer for a building that has a genuine reason to run at the lower pressure.

What does not fix it: more traps, larger traps, a different trap family, or a second boiler. None of those creates a differential that is not there.

What this generalises to

Operating pressure is not a boiler setting. It is the one setting in the plant that fixes the saturation temperature at every heating surface and the differential across every trap simultaneously, so changing it at the boiler changes every terminal and every drainage path at once. The numbers that decide whether that change is survivable live at the far end of the building: the lift from each trap to its return, the return main pressure, and the differential each trap's capacity was selected at. Those belong on the new boiler's commissioning record, measured, before the operating control is set.

References

• ASME Steam Tables, or an equivalent saturated-steam table, for the saturation temperatures at the two operating pressures
• ASME Boiler and Pressure Vessel Code Section IV, in the edition adopted by your state's boiler law, which binds the installation and reaches the shop through the jurisdictional inspection, for the low-pressure heating class limit and relief valve setting
• Trap and pump-trap manufacturer capacity data at the actual differential, and their selection literature for a device that supplies its own motive pressure on a lifted return
• 29 CFR 1910.147 for control of hazardous energy in general industry, and the construction energy control requirements of 29 CFR Part 1926 on a construction site
• See related: How to Recognise Stall Before You Replace the Trap; What Water Hammer Actually Is in a Steam Line; What Makes Steam Different From Every Other Working Fluid