Water Hammer Diagnosis and Arrestor Install

Purpose

A water hammer arrestor absorbs the pressure spike from a sudden change in water velocity. That is all it does, which is why so many get installed and fail to fix the complaint. Half the bangs a shop is called out for are not shock at all but pipe moving against framing, a chattering cartridge, or thermal expansion in a closed system. Of the half that are shock, most devices go in wrong one of two ways, too far from the valve or on a system whose static pressure was never corrected. This procedure makes the diagnosis explicit before a device is bought, and the placement verifiable afterward.

Scope

Covers diagnosis of banging, knocking and hammering on potable water distribution in residential and light commercial buildings, and the selection, placement and verification of mechanical arrestors conforming to ASSE 1010.

Does not cover pressure regulation or thermal expansion control; the pressure reducing valve replacement SOP owns both, and that work goes first for reasons this procedure explains. Does not cover drain, waste or vent noise, or new system design, where velocity is chosen at the drawing stage; this works on buildings whose pipe sizes are not changing.

Roles and responsibilities

Procedure

1. Reproduce the noise before you diagnose it, and note when it occurs in the cycle. Operate each candidate fixture and appliance in turn. Acceptance: the noise reproduced on demand, trigger named, moment recorded as at-opening, during-flow, or at-closing. Wrong looks like a diagnosis built on the customer's description, which uses one word for three noises. Stop rule: if you cannot reproduce it, install nothing and leave a listening assignment; an arrestor fitted on a guess is one you will be asked to justify. Hazard: fill cycles mean live supply lines and possible discharge, so check the floor around the machine and never leave one mid-cycle.

2. Classify the noise, because only one of four is hammer. A single sharp bang the instant a valve closes is shock. Ticking or creaking after a hot draw that continues once flow stops is pipe expanding against a tight hanger or an undersized hole through framing. A buzz at partial opening that stops at full open is a worn cartridge. A thud or relief discharge in the evening after hot water use is thermal expansion in a closed system. Acceptance: the noise placed in one of the four, in writing. Wrong looks like calling everything hammer. Stop rule: only the first proceeds here; the second is a hanger fix, the third a cartridge, the fourth routes to the regulator SOP. Hazard: none, but misclassifying puts a device on a wall for a noise it cannot touch.

3. Take a static pressure reading before you consider any device. Gauge on a hose bib, every fixture closed. Acceptance: a stable reading held two minutes, location named. Wrong looks like skipping this because the complaint is noise, not pressure. Stop rule: above 80 psi, IPC 604.8 and UPC 608.2, in the edition your authority having jurisdiction has adopted, require a pressure reducing valve and that goes first, not because it fixes the bang, which it barely touches, but because verifying an arrestor at a pressure the building should not run at is meaningless; a customer who declines the regulator gets that and the reading on the ticket. Hazard: threading a gauge onto an old hose bib can shear the nipple, so support the bib and stand clear of the spray line.

4. Identify the quick-closing device and confirm it closes fast enough to matter. Appliance solenoid fill valves, single-lever faucets snapped shut, and modern toilet fill valves are the usual sources. Acceptance: the valve named. Wrong looks like blaming "the plumbing" when one appliance is responsible. Stop rule: closure counts as instantaneous when it beats the pipe's reflection time, 2L divided by wave speed, L being the run back to the nearest large-volume junction; 30 ft of rigid copper near 4,000 ft/s gives about 0.015 seconds, and a solenoid beats that. Where no quick-closing valve exists, return to step 2: a bang at a two-handle faucet is usually the occupant snapping the handle. Hazard: do not open an appliance valve to inspect it, since that moves the manufacturer's warranty problem onto your shop.

5. Compute the branch velocity and the spike it produces, because velocity is the term the rise is proportional to. Take the flow rate from the appliance's literature and the bore from the pipe type and size actually installed. The Joukowsky relation gives pressure rise as fluid density times wave speed times velocity change; for water in rigid copper at ordinary temperature, wave speed near 4,000 ft/s and instantaneous closure, that is near 54 psi per 1 ft/s stopped. Acceptance: a velocity in ft/s and a peak written as static plus spike, computed rather than guessed. Wrong looks like assuming nominal size equals bore, or carrying that 54 psi per ft/s onto PEX, whose much lower wave speed gives a proportionally smaller spike, so a real hammer complaint on PEX points at a rigid section or metal transition. Stop rule: well above the common guidance of about 8 ft/s cold and 5 ft/s hot for copper, the branch is undersized, a re-pipe conversation for the lead, since an arrestor there quiets the noise and leaves the erosion in place; and where the peak exceeds the rating of any fitting or connector on that branch, say so on the ticket. Hazard: none, this is arithmetic, and skipping it is how a shop sells a third arrestor on one branch.

6. Check for and empty any air chambers before you condemn them. Older buildings rely on capped vertical stubs above fixtures, which work until they fill with water. Acceptance: a stated finding of whether air chambers exist and whether draining down and refilling changes the noise. Wrong looks like fitting an arrestor beside a waterlogged chamber and calling the noise fixed, when the drain-and-refill is what silenced it for a week. Stop rule: if draining and refilling silences it and it returns within days, the building relies on air chambers and the fix is mechanical arrestors, which do not waterlog. Hazard: draining the system drains the water heater, so on an electric unit open the breaker and prove it dead before the tank level drops; a dry element energized is destroyed in seconds.

7. Size the arrestor to the fixture unit load on that branch, and place it close to the valve. Acceptance: an ASSE 1010 rated arrestor whose PDI-WH201 size letter covers the branch's fixture unit load, on that branch near the quick-closing valve rather than at the trunk, connection accessible. Wrong looks like an oversized unit at a manifold twenty-odd feet upstream, leaving most of the branch to build its own spike before the device sees it. Stop rule: PDI-WH201, in the edition adopted through the plumbing code your authority having jurisdiction enforces, places the device within about 6 ft of the valve; if the only accessible spot is farther, open access rather than accept the distance. Hazard: on a soldered connection shield combustibles behind the joint, keep a charged extinguisher in reach, and hold a fire watch after the last heat per NFPA 51B in the edition your authority having jurisdiction or insurer requires; ventilate rather than working over heated flux.

8. Fit it accessible, right way up, and on the correct branch. Acceptance: arrestor in the orientation its manufacturer specifies, on the branch serving the identified valve, reachable without cutting finished surfaces. Wrong looks like a device buried behind tile. Stop rule: an arrestor is serviceable, with a bellows or piston and a seal, so if the only way to fit it is to bury it, use an access panel; an inaccessible one becomes an unexplained noise for the next tech. Hazard: press connections close with several tons of jaw force, so hands stay outside the jaw envelope and the fitting is confirmed seated before the tool cycles.

9. Verify by reproducing the original trigger, not by listening to a faucet. Run the identified appliance through its full cycle three times. Acceptance: no audible bang on any of the three, and where a peak-indicating gauge is available, a peak well below the pre-install figure at the same test point. Wrong looks like one quiet cycle and a signature. Stop rule: if the noise persists, do not add a second arrestor, go back to step 2, because a persisting noise after a correctly sized and placed device means the classification was wrong. Hazard: leave the system pressurized and watched a few minutes after the last cycle, since a joint that will weep does it then.

The record this produces

Eight fields: the reproduced trigger and the moment in the cycle, the classification from step 2, the static pressure with its test point, the computed branch velocity with the pipe type and size behind it, the estimated peak, whether air chambers were found, the arrestor's PDI size and ASSE rating, and its location described well enough to find without a search.

They land on the property record. The next tech reads the classification and location before diagnosing a returning noise, which prevents the second visit where another arrestor goes in beside the first. The office reads the static pressure to know whether the regulator work is still outstanding. The lead reads the velocity when the customer asks why the noise came back after a higher-flow appliance went in, because a branch already at the edge of its guidance has no headroom.

One worked pass, including a failure

Two-story house, complaint logged as a bang in the laundry wall when the washer runs. Step 1 reproduces it on demand: a single sharp bang the instant the cold fill solenoid closes, at no other fixture. Step 2 classifies it as shock, since it occurs at closing and does not continue after flow stops.

Step 3 fails. Static at the laundry hose bib reads 92 psi, above the 80 psi threshold in IPC 604.8 as adopted locally. Stop rule taken: the regulator work goes first, routed to the pressure reducing valve SOP, which finds a failed regulator and sets the building to 60 psi.

Be precise about what that did for the noise. Step 5: the washer's literature gives about 5 gpm, the branch is 1/2 in type L copper with a bore near 0.545 in, so about 6.9 ft/s. At roughly 54 psi per ft/s, 6.9 ft/s stopped is about 372 psi of rise, putting the peak near 464 psi at the original 92 psi static and near 432 psi at the corrected 60 psi. Reducing static took about 32 psi off a peak of roughly 464, about 7 percent. The regulator was required and was the right first move, and it is not the fix for the bang; saying so on the ticket stops the customer concluding the second visit was unnecessary.

Step 4 confirms instantaneous closure: about 30 ft back to the cold trunk, so reflection time is 60 divided by a wave speed near 4,000 ft/s, about 0.015 seconds. The 6.9 ft/s sits under the 8 ft/s cold guidance, so no re-pipe. Step 6 finds a capped stub above the laundry box, waterlogged, explaining why the house was quiet for a decade. Step 7 sizes to that branch's fixture unit load, which the smallest PDI-WH201 size covers, and places it about 3 ft from the solenoid inside the laundry box. Step 9 runs three complete fill cycles: no bang on any, and the peak reads far below the pre-install figure.

References

• ASSE 1010, performance requirements for water hammer arresters, and PDI-WH201, arrester sizing, consensus standards reaching your installation through the plumbing code your authority having jurisdiction has adopted, in the edition it names.
• IPC 604.8 and UPC 608.2 for the 80 psi static threshold, and IPC 604.9 for fitting arresters where quick-closing valves are used, in your adopted edition.
• NFPA 51B, fire prevention during welding, cutting and other hot work, in the edition your authority having jurisdiction or insurer requires.
• Appliance literature for fill rate, and pipe manufacturer's data for actual internal diameter by type and size.
• See related: the Plumbing pressure reducing valve replacement SOP, which owns static pressure and thermal expansion.