Expansion Tank Verification on a Closed System
Purpose
Water heated in a system it cannot back out of has nowhere to go, so the pressure climbs until something gives. On most houses the something is the relief valve, which does exactly what it was built to do and gets blamed for it; on the unlucky ones it is a supply line, a toilet fill valve or the tank seam. An expansion tank is the release, and it is the single most commonly mis-verified device in a residential system because the check most techs perform, tapping it and listening, cannot distinguish a healthy tank from a dead one. This procedure verifies pre-charge the only way it can be verified and ties it to a measured number the house actually runs at.
Scope
Covers verifying, setting or replacing a bladder-type thermal expansion tank on a residential or small-commercial domestic hot water system, and establishing whether the system is closed in the first place.
It does not cover the relief valve or its discharge, which the T&P valve and discharge piping verification SOP owns and which routes drip complaints here; setting or replacing a pressure reducing valve, which is its own scope; or hydronic heating expansion tanks, which are sized and charged against a different design condition entirely.
Roles and the handoff
| Role | Owns | Hands off |
|---|---|---|
| Office | Books a relief-drip or banging complaint with time for pressure work, not as a 30 minute call | Asks whether the meter or the service was recently changed, since that is how a house goes closed |
| Service tech | Every step below, including the isolate-and-drain before any pre-charge reading | Returns cold static, recovery peak, and pre-charge as three separate numbers |
| Owner or lead | Any scope beyond the tank, such as a pressure reducing valve above 80 psi static | Quotes it as its own line so the customer sees two problems, not one |
The handoff that breaks: a tech puts a gauge on the tank's air valve with the system still pressurized, reads system pressure, calls the pre-charge correct, and leaves. That reading is guaranteed to look right on a tank that is completely waterlogged.
Procedure
Establish whether the system is actually closed, and name the device that closes it. Acceptance: a pressure reducing valve, a backflow preventer or a check in the meter identified and recorded, or the system positively confirmed open. Wrong looks like assuming closed because there is an expansion tank fitted: plenty of tanks are fitted to open systems by habit, and plenty of houses went closed the day the utility swapped a meter, with nobody told. Hazard: none, this is a look at the service entry, but do not force a meter pit lid or enter a pit, which is a separate confined-space question.
Measure cold static at a hose bib with nothing running in the house. Acceptance: static in psi recorded, taken with no fixture, no irrigation and no appliance drawing. Wrong looks like a reading taken while a toilet is refilling, which reads low and then sets every downstream number wrong. A static above 80 psi is separately its own scope under International Plumbing Code 604.8 in the edition your jurisdiction has adopted. Hazard: thread the gauge fully home before opening the bib, because a gauge blown off a fitting at high static is a projectile at head height.
Measure what the system actually does on a recovery, with a lazy-hand gauge. Leave a gauge with a maximum-reading pointer on the bib through a full heat cycle after a decent draw. Acceptance: the peak in psi recorded, and the shop gate is a peak no more than 10 psi above measured cold static and comfortably below the relief valve's marked set pressure. Wrong looks like a peak that reaches the valve's marked set: the valve has been relieving on every cycle, which is why it drips and why it will not reseat forever. Hazard: none at this step beyond the gauge already fitted, but do not leave a gauge on a bib the customer might open.
Determine whether a tank exists and whether its bladder is intact, using the air valve rather than your knuckles. Press the Schrader core briefly with the system still pressurized. Acceptance: air comes out, which means the bladder is holding water on the other side of it. Wrong looks like water at the air valve: the bladder has failed, the tank is full of water on both sides, it has been doing nothing for some time, and it is replaced rather than recharged. Tapping the shell for a hollow ring is not a test and does not go on the record as one. Hazard: expect a short spit of water or a jet of air at the valve, so keep your face out of line and wear eye protection.
Isolate the tank and drain its water side before you read or set pre-charge, because a pressurized system makes the gauge lie. With system pressure on the water side, the bladder transmits it and the air gauge reads system pressure regardless of what the charge actually is. Acceptance: tank isolated at its service valve, water side drained to a bucket or hose until it stops, and only then a gauge on the Schrader. Wrong looks like a pre-charge reading taken hot and live: that number is worthless and is the reason so many tanks are recorded as correct while sitting dead on the pipe. Hazard: a tank on the hot side holds stored water at 120 to 140 F, which scalds bare skin in seconds, so route the drain before opening it and keep hands off the outlet.
Set the pre-charge to match the measured cold static, not to what the tank shipped with. Acceptance: air pressure set within about 2 psi of the static recorded at step 2, using a low-range gauge rather than a tire gauge, and never above the tank's marked maximum working pressure. Where a pressure reducing valve is being set on the same visit, set the valve first and charge the tank to the new static. Wrong looks like leaving the factory charge, commonly 40 psi, on a 65 psi system: the tank is partly compressed before the heater even fires and much of its acceptance volume is already gone. Hazard: charge with a hand pump or a regulated line, because an unregulated compressor will pass the tank's rating in a second and rupture a bladder you just paid for.
Check size and support against the manufacturer's table and the fitting it hangs on. Acceptance: tank size read from the manufacturer's sizing table using stored volume, the cold-to-setpoint temperature rise, the static, and the relief set pressure, with the size read recorded; and the tank independently supported rather than cantilevered off a threaded nipple. Water expands on the order of 2 percent by volume across a typical mains-to-setpoint rise, which is a rule of thumb for sanity-checking the table's answer and not a substitute for it, since the true figure depends on both end temperatures. Wrong looks like a tank hanging on its own nipple: a waterlogged tank carries its full volume in water, and the joint that fails puts that water on the floor. Hazard: support the tank before you let go of it.
Return to service and prove the gate with the gauge still on. Open the isolation valve, restore the appliance, and run a full recovery after a real draw. Acceptance: the recovery peak now within 10 psi of cold static, and the relief valve dry at 24 hours. Wrong looks like a peak still climbing to the relief set with a new tank fitted: the tank is undersized, mis-charged, or isolated by a valve somebody left shut, so recheck the isolation valve first. Hazard: this step restores stored energy and pressure to a system you opened, so on a gas appliance repeat the spillage test at the draft hood about a minute after the burner fires with a personal carbon monoxide monitor on you, and on an electric one confirm the tank is full before the breaker closes.
When the tank is not the answer
System is open: an expansion tank is not required and is not the fix, so keep looking, because a drip on an open system is a valve or a temperature problem. Static above 80 psi: the pressure reducing valve is the primary scope and the tank is charged to whatever static the new valve is set to, in that order. No isolation valve on the tank: fitting one is part of the job, because a tank that cannot be isolated cannot be verified on any future visit. Customer declines the tank: record the measured recovery peak beside the relief valve's marked set in their own file, because that pairing is the whole argument and it will be needed again.
The record this produces
- The device that closes the system, or a positive statement that the system is open
- Cold static in psi, with the time and the confirmation nothing was running
- Recovery peak in psi from the lazy-hand gauge, and the relief valve's marked set it was compared against
- Schrader result as air or water, and the tank's age if known
- Pre-charge as found, measured after isolate-and-drain, and pre-charge as left
- Tank size read from the manufacturer's table, with the four inputs used
- Post-work recovery peak against the 10 psi gate, and the 24 hour relief check
Three numbers on one line, static, peak and pre-charge, tell any later tech in five seconds what took this visit an hour to establish.
Worked pass: 50 gallon electric, relief valve dripping every evening
Single family house, pressure reducing valve at the service, expansion tank about four years old fitted on the cold line above the heater. Relief valve marked 150 psi and 210 F.
- Step 1: pressure reducing valve present at the service, so the system is closed. Recorded.
- Step 2: cold static 55 psi at the hose bib with nothing running, taken at 8 am before anyone drew water.
- Step 3: lazy-hand gauge left through a full recovery after a bath draw. Peak 150 psi, which is exactly the relief valve's marked set, so the valve has been relieving on every cycle. Against the gate, 150 minus 55 is a 95 psi rise where the gate allows 10.
- Step 4: FAILED. Schrader core pressed and water came out, not air. The bladder is gone, the tank is solid water, and it has been contributing nothing. Stop rule taken: no attempt to recharge it, tank condemned and replaced.
- Step 5: with the replacement fitted, isolation valve closed and the water side drained to a bucket before a gauge went anywhere near the air valve. Pre-charge as shipped read 40 psi.
- Step 6: charged with a hand pump to 55 psi, matching the measured static, checked at 56 psi which is inside the 2 psi band. Tank's marked maximum well above that.
- Step 7: manufacturer's table entered with 50 gallons stored, a rise from 55 F mains to a 130 F setpoint, 55 psi static and a 150 psi relief set, and it returned a 2 gallon nominal tank from the 40 to 60 gallon row. Sanity check: 2 percent of 50 gallons is 1.0 gallon of expansion, and the table's answer is larger than that because a tank's acceptance volume is only part of its shell volume. Fitted on a bracket to the framing, not hanging on the nipple.
- Step 8: isolation valve opened, breaker closed with the tank confirmed full, full recovery run after another draw. Peak 61 psi, so 61 minus 55 is a 6 psi rise, inside the 10 psi gate and far below the 150 psi relief set. Relief valve dry at the 24 hour call-back.
The number that closes the ticket is the pair: 150 psi peak before, 61 psi after, against the same 55 psi static. A tech who had only recharged the old tank would have measured 55 psi on a gauge, written "pre-charge correct", and left the customer with the identical drip.
References
- Manufacturer sizing table and installation instructions for the expansion tank, for size against stored volume, temperature rise, static and relief set pressure, and for the tank's maximum working pressure
- International Plumbing Code 604.8, in the edition your authority having jurisdiction has adopted, for the pressure reducing valve requirement above 80 psi static
- International Plumbing Code 607.3 or Uniform Plumbing Code 608.3, in your adopted edition, for the thermal expansion control requirement on a closed system, which the two codes word differently
- See related: the T&P valve and discharge piping verification SOP, the water heater annual service SOP, and the water hammer diagnosis and arrestor install SOP