Thermal Expansion and Pressure Complaint Investigation

Purpose

A complaint that only happens at a certain time of day is a complaint with a cycle behind it, and on a domestic water system that cycle is almost always the water heater recovering. This procedure reproduces the cycle deliberately, watches pressure across it at one gauge point, and produces a paired figure: how far pressure rose, and how long after the burner or element started. That pair is what separates thermal expansion from a failing regulator, and it is what stops the single most common wrong repair in this trade, which is replacing a relief valve that was working perfectly and watching the new one weep the same evening.

Scope

Covers investigation of a pressure complaint whose symptom follows a cycle: a relief valve that weeps in the evening, banging after a shower, toilet fill valves that run at night, or a supply line that failed with no impact and no corrosion, on a residential or light-commercial domestic hot water system.

Does not cover verifying, setting or replacing the expansion tank itself, which the expansion tank verification on a closed system SOP owns and which this procedure hands to once the mechanism is established. Does not cover the relief valve or its discharge piping, owned by the T and P valve and discharge piping verification SOP, or the regulator's own condition, which is the house pressure survey and PRV assessment SOP, whose creep test deliberately removes heat input so the two do not contaminate each other. Does not cover hydronic heating loops.

Roles and the handoff

Procedure

1. Treat the weeping relief valve as evidence, not as the fault. Acceptance: a written statement that the valve discharged, with the discharge piping confirmed open, unthreaded at its end and terminating as installed. Stop rule: nothing is capped, plugged, restrained, shimmed or replaced at this step, because a relief valve that opens on overpressure is doing exactly what it was built to do and swapping it reaches the same failure one week later with a part number on the invoice. Hazard: relief discharge is water at heater temperature, commonly 120 to 140 F and higher during an overpressure event, which scalds in a couple of seconds, so nobody puts a hand or a face under the discharge line.

2. Fix one gauge point and prove whether the system is closed. Acceptance: the gauge location named, marked in increments no coarser than 2 psi, plus a yes or no on a check valve, backflow assembly or regulator with an integral check between the meter and the house. Stop rule: an open system cannot build pressure this way, so a genuinely open system sends the investigation back to the house pressure survey and PRV assessment SOP rather than forward. Hazard: threading a gauge onto a corroded bib at street pressure makes a projectile, so back the bib with a second wrench and stand off its axis.

3. Record the arrival state before you change anything. Acceptance: static at the gauge point, the heater's thermostat setting, the measured outlet temperature at the nearest hot tap, and the relief valve's stamped set pressure read off its own tag. Stop rule: an illegible or missing relief valve tag means that valve cannot be verified against any number, so it is replaced on its own merits and the record says plainly that the replacement is not the fix for the complaint. Hazard: drawing hot water to measure outlet temperature puts 140 F water at hand height, so run it into the tub rather than a lavatory and keep hands clear of the stream.

4. Establish a no-draw baseline with the heater satisfied. Lock out the ice maker, the softener and the irrigation controller, close every fixture, and wait until the burner or element has been off for a stated 10 minutes. Acceptance: a stable baseline pressure with two readings five minutes apart agreeing within 1 psi. Stop rule: a baseline that will not settle means something is still drawing or the regulator is passing, either of which is a different investigation. Hazard: a locked-out consumer is one somebody must put back, so tag each one now rather than trusting memory.

5. Force a recovery on purpose and watch the whole cycle. Draw a measured volume of hot water, close everything, and read pressure at 5 minute intervals from the moment heat starts. Acceptance: a series of readings with clock times, the volume drawn stated, and the peak identified. Stop rule: any draw during the watch voids the series, so the run is discarded and repeated after the offending consumer is locked out, rather than the dip being explained away in the notes. Hazard: standing beside a firing gas heater for a full recovery means standing beside a combustion appliance, so a personal carbon monoxide monitor is on before it fires and the room is left if it alarms.

6. Compare the peak with the stamped set pressure, in the same units. Acceptance: peak pressure in psi against the relief valve's stamped set pressure in psi, both written, and the elapsed minutes from heat start to peak. Stop rule: a peak at or above the stamped set pressure with the valve discharging confirms the valve operated correctly and the pressure is real, which closes the door on replacing it as the fix. Hazard: pressure is climbing toward a relief setting during this step, so no fitting, gauge or tap on that segment is disturbed while the series is running.

7. Correlate the cycle with the complaint's clock time. Acceptance: the customer's stated symptom time set beside the household's largest hot water draw and the recovery window that follows it, with the match or mismatch written. Stop rule: a symptom that does not follow any recovery window is not thermal expansion, so return to the regulator survey rather than fitting a tank and hoping. Hazard: none here, a comparison made at the table; discharge it rather than leaving it blank.

8. Apply the gate and write the disposition. The gate is a rise over 15 psi across one recovery cycle at one gauge point with no draw. Acceptance: a disposition of expansion control absent, expansion control present but not working, or not thermal expansion, with the rise and the elapsed minutes beside it. Stop rule: where a tank is already fitted, this procedure does not diagnose its pre-charge; that goes to the expansion tank verification on a closed system SOP, which owns the only check that distinguishes a healthy tank from a dead one. Hazard: none here; discharge it rather than leaving it blank.

9. Put every consumer and the heat back, and prove the protection still works. Restore the ice maker, softener and irrigation controller, remove the gauge, and bring the heater through one full cycle. Acceptance: static within about 5 psi of the arrival reading at the same gauge point, every locked-out consumer restored and confirmed running, the heater proved to fire and satisfy, and the relief valve confirmed free with its discharge piping open and terminating as found. Stop rule: any item on the step 4 lockout list you cannot restore is called out in writing before you leave. Hazard: this step puts fuel or power and pressure back into an occupied building, so the relief valve and its discharge are verified unobstructed before the heater fires, and the personal carbon monoxide monitor stays on for the confirming cycle.

The record this produces

Seven fields: the discharge confirmation from step 1; the gauge point and the closed-system answer; arrival static, thermostat setting, outlet temperature and the stamped relief set pressure; the settled baseline; the pressure series with clock times and the volume drawn; the peak against the set pressure with elapsed minutes; and the disposition with the rise figure.

The series is the field that survives. Any tech can say the pressure was high; a shop that can produce eighty-odd psi of rise across a sixteen minute recovery, at a named bib, with the ice maker locked out, has an argument nobody takes apart.

One worked pass, including a step that fails

A 50 gal gas water heater in a 1996 house. Complaint: the relief valve drips into a bucket most evenings between eight and nine, and the customer has already had one relief valve replaced by another shop.

Step 1: discharge line open, terminating downward as installed, valve weeping when found. Nothing capped. Recorded.

Step 2: gauge on the laundry bib, 1 psi increments. A dual check at the meter, so the system is closed.

Step 3: arrival static 68 psi, thermostat at the middle mark, outlet at the tub reads 138 F, relief valve tag legible and stamped 150 psi.

Step 4: ice maker, softener and irrigation locked out and tagged. Burner off ten minutes. Baseline reads 68 psi twice, five minutes apart.

Step 5 FAILS on the first run. The tub is run at 4.0 gpm for a timed 5 minutes, so 4.0 times 5, or 20 gal drawn, and the burner lights. Readings run 68, then 94, then 121, and then fall to 108 as the ice maker fills, which nobody caught in the lockout because it is plumbed off the laundry branch. Stop rule taken: the series is discarded rather than annotated, the ice maker is valved off, and the whole run repeats after the heater satisfies and cools back to baseline.

Valid run: burner starts 18:42 at 68 psi. 18:47 reads 96 psi. 18:52 reads 128 psi. 18:57 reads 149 psi. At 18:58 the gauge reads 152 psi and the relief valve begins to weep.

Step 6: peak 152 psi against a stamped 150 psi, so the valve opened above its set pressure and did its job. Elapsed from 18:42 to 18:58 is 16 minutes. The rise is 152 minus 68, or 84 psi.

Step 7: the household's biggest hot draw is three showers ending about eight in the evening, with recovery running roughly twenty minutes after that. That lands squarely inside the customer's stated eight to nine window, and it explains why the previous shop's new relief valve behaved exactly like the old one.

Step 8: 84 psi of rise across one recovery cycle, against a 15 psi gate, on a closed system with no expansion tank anywhere on the cold inlet. Disposition is expansion control absent. Sizing, pre-charge and installation go to the expansion tank verification on a closed system SOP rather than being improvised here.

Step 9: ice maker, softener and irrigation restored and each confirmed running. Gauge off, bib proved to shut with no drip. Static reads 67 psi at the laundry bib against 68 psi on arrival, a 1 psi difference and inside the 5 psi acceptance. The heater is watched through one full cycle with the carbon monoxide monitor on, and the relief valve and its discharge are confirmed free and terminating as found.

When the site does not match this procedure

An expansion tank is already fitted and the rise still clears the gate. The disposition is expansion control present but not working, and it goes to the sibling SOP for the pre-charge check. Do not add a second tank to a system whose first one is dead.

The rise is under the gate but the customer's complaint is real. Look at the regulator instead. A valve passing street pressure through raises static without any heater involvement, and the house pressure survey SOP separates the two by removing heat input from its creep test.

Electric heater with a long, low-output recovery. The rise is the same but arrives slowly, so the 5 minute interval stays and the watch runs longer. Do not shorten the draw to speed it up; a smaller draw gives a smaller rise and can drop a real fault under the gate. On a tankless heater there is no stored volume recovering at all, so the recovery series does not exist in this form and the investigation moves to the recirculation loop and its isolation valves.

References

• The relief valve's own stamped set pressure and its listing, read from the tag on the valve rather than assumed from the size or the fitting.
• The model codes' thermal expansion control requirement for a closed water supply system, at IPC Section 607.3 and the corresponding UPC section, in the edition your authority having jurisdiction has adopted.
• See related: the expansion tank verification on a closed system SOP, the T and P valve and discharge piping verification SOP, and the house pressure survey and PRV assessment SOP.