Why You Never Test a Relief Valve to Prove a Point

Why this matters

The try lever on a steam safety valve looks like a test button and it is not one. It is a mechanical assist designed to work at one specific condition, and pulled outside that condition it drags a lapped metal seat across a lapped metal nozzle under near-full spring load. That is not a test. That is machining.

The shop cost is not the valve. It is that the damage is silent for a while, gets written into the log as a passing test, and then presents weeks later as an unrelated complaint that somebody else has to chase.

If you ever find a steam safety valve discharging, the response is to leave the immediate area, keep everyone away from the discharge outlet and the pipe run, and stop the heat input at the remote burner disconnect or emergency switch outside the boiler room rather than reaching past a discharging valve to touch anything. The plume is invisible for the first stretch, sonic, and hot enough to strip skin, and it is loud enough at close range to damage hearing without protection.

The call

Low-pressure steam heating system in a three-story office building. Boiler maximum allowable working pressure 15 psig, one ASME-stamped safety valve set at 15 psig, system operated at 5 psig. Call was a no-heat complaint on the top floor on a cold morning, with the boiler running essentially without cycling.

Nothing about that presentation points at a safety valve. It points at distribution, and that is where the first visit went.

What got eliminated, and on what evidence

Boiler capacity. The boiler held 5 psig easily in mild weather and could not hold it below roughly 20 F outdoor. A genuinely undersized boiler is short at design conditions and fine in the shoulder season, so the shape fit. It was eliminated by the fact that the building had run the same boiler through eight winters colder than this one with the same load, and nothing on the load side had been added.

Traps blowing through. A group of failed-open traps dumps live steam into the return and can absolutely starve the far end of a system. The return main was checked from a distance with an infrared thermometer, never by hand, and it ran at temperatures consistent with condensate and flash rather than the near-supply temperatures a group of blown traps produces. Two suspect traps were tagged for follow-up but neither was passing enough to explain the shortfall.

A leak in the mains. The makeup water meter had been read weekly for two years. The last three weeks ran about seven times the two-year weekly average, which confirmed the system was losing water somewhere and was the strongest lead of the visit. The crawlspace and the buried section were checked and dry.

Seven times the normal makeup with dry piping means the water is leaving as steam, at a point where nobody is calling it a leak.

The line in the log

The maintenance log for the walkthrough three weeks earlier had a single entry: tested safety valve, OK.

The safety valve was weeping steadily out of its discharge pipe, to a drip pan elbow that drained to the floor drain, in a boiler room nobody had walked since. It had been doing it since the day of that entry.

The walkthrough had gone like this. The facility manager asked the tech to show that the safety valve worked. The tech pulled the try lever, the valve chattered, he let it go, and both of them heard it stop. Then he wrote the line.

Why the lift is what did the damage

Two mechanisms, and they compound.

The lifting gear on an ASME-stamped steam safety valve is designed to positively lift the disc when the vessel is at or above roughly 75 percent of set pressure. That design condition, stated in the ASME Boiler and Pressure Vessel Code section the valve was stamped to, in the edition your state's boiler law adopts, exists because the lever is an assist, not an opener. At 75 percent of set, steam force under the disc has already taken most of the spring load, so the lever only has to supply the rest and the disc lifts more or less straight up.

This system was operating at 5 psig against a 15 psig set, which is 33 percent of set. At that pressure the spring is carrying nearly its full load, and the lever does not lift the disc off the seat, it skids it across the nozzle. Steam safety valve seats seal metal to metal on a lapped, mirror-finish contact band. One drag under load leaves a scratch across the band, steam finds the scratch, and the flow wire-draws it wider every hour after that.

A valve that has sat for years is often lightly bonded to its seat by scale and corrosion product. Breaking that bond with a lever is exactly as likely to leave the valve unable to reseat as it is to leave it working, and there is no way to know which one you got by listening.

Both mechanisms end at the same place: the valve is now weeping, the weep is destroying the seat further, and the boiler is making steam to replace steam that is going out the roof.

The test that was never performed

Here is the part worth sitting with. Even if that lift had been done correctly, at or above 75 percent of set on a hot system with the discharge clear and nobody near the outlet, it would not have told the facility manager what he asked for.

A lever lift demonstrates that the disc is not seized. It says nothing about set pressure. A valve whose spring has taken a permanent set, or whose corroded spindle binds under load, or whose seat is already leaking will still lift on the lever. The question "does the safety valve work" means "will it open at 15 psig and pass its rated capacity," and there is exactly one way to answer that: a set-pressure test with a calibrated source.

That test happens on a bench, at an organization holding the National Board VR certificate for pressure relief valve repair, under the National Board Inspection Code in the edition your jurisdiction has adopted, which binds the owner and reaches you through the jurisdictional boiler inspector. On many low-pressure heating valves the practical answer is not testing at all but scheduled replacement, because the valve costs less than the labor to send it out and get it back stamped.

So the honest answer to the customer is that the only way to know is to remove the valve, and the cheapest way to know is to replace it on a schedule. Neither of those involves the lever.

What it cost, in hours

Set the parts aside and count labor.

• The lever pull: about ten seconds, producing no information.
• First no-heat visit: about 2.5 hours, spent eliminating boiler capacity, traps, and buried piping.
• Second visit, including finding the weep and confirming it: about 1.5 hours.
• Replacement: the boiler off, cooled below any pressure, drained, confirmed at zero on the gauge, locked and tagged under the shop's written energy control procedure required by 29 CFR 1910.147 for stored pressure energy, then the valve changed and the system brought back up and proved for leaks from a distance. About 3 hours, plus coordinating the jurisdictional inspector's sign-off where the state's boiler law requires it.

That is roughly 7 hours of recovery labor, three weeks of a boiler running near continuously, and three weeks of makeup water at about seven times normal, all traceable to a ten-second act that produced nothing. There is no ratio to compute against the value gained, because the value gained was zero.

The other cost has no hours attached. A second tech read "tested safety valve, OK" and, on the first visit, moved past the boiler room because of it. What that log line does operationally is put the shop's name next to a claim the act could not support, and the practical fact you can bank is that the next person believed it. Whether such an entry carries any weight in a dispute is a question for the shop's own attorney, and that is a separate question from the operational one.

What to say when a customer asks you to prove it works

Say the mechanism, not the policy. "Policy says no" invites an argument. The mechanism ends it.

Something close to this works: the lever on that valve is designed to help it open when the boiler is already near its trip pressure. Your system runs at a third of that, so pulling the lever would scrape the sealing surface instead of lifting it, and the valve would leak afterward. It would also not tell us the one thing you want to know, which is what pressure it actually opens at, because that takes a calibrated test on a bench.

Then give them the thing they can actually have, which is a record. Note the valve's set pressure, its stamped capacity, its installation or replacement date, the condition of the discharge piping, and whether the operating pressure sits far enough below set that the valve is not simmering. The section on relief valve protection in this group covers what those checks look for. That record is a real answer to a real question, and it is one you can produce without touching the valve.

Two situations change this. Where the jurisdiction's boiler law or the facility's own written operating procedure requires a periodic try lever test, that requirement governs and it will specify the pressure at which the lift is done, which will be at or above the design condition and never at idle. And where a valve is already weeping, there is nothing to preserve; that valve is coming off regardless, so the question stops being whether to test it and becomes how soon it can be changed.

References

• ASME Boiler and Pressure Vessel Code, in the edition adopted by your state's boiler law, which binds the owner and reaches the service contractor through the jurisdictional inspector, for the lifting device design condition and for stamping and capacity
• National Board Inspection Code, in the edition adopted by your jurisdiction, for pressure relief valve repair and the VR certificate held by organizations permitted to reset and re-stamp a valve
• 29 CFR 1910.147, the OSHA general industry energy control standard, for isolating stored pressure energy before a valve is removed
• Valve manufacturer instructions, which govern the conditions under which any in-place lift is performed
• See related: What a Safety Relief Valve on a Steam System Protects