Backflow failed test repair field guide for plumbers and testers

A failed backflow test, and what it actually tells you

A failed backflow test means the assembly is no longer holding pressure the way its standard requires, so it can no longer be trusted to stop contaminated water from reversing into the potable supply. The number on the form is the symptom. The job is the diagnosis.

A relief valve that discharges, a check that leaks back, an air inlet that will not open, each one points to a specific part and a specific repair, and each is fixable on most assemblies without replacing the whole unit. Treat the failed test as the start of a repair path, not the end of a report.

The tester who writes failed and walks away has done half the work. The repairer flushes the line, finds the failed component, rebuilds it with the matching kit, and proves the fix with a passing retest before the assembly goes back in service. The difference between those two is the difference between a failed report and a closed job.

The four testable assemblies and how they fail differently

Four testable assemblies cover almost everything in the field, and they fail in different places because they are built differently. The reduced pressure principle assembly, the RP or RPZ, is the high-hazard workhorse: two independently acting check valves with a relief valve between them that dumps the middle zone to atmosphere the moment either check fails. The double check, DC or DCVA, is the same two checks in series without the relief valve, used on lower hazards. The pressure vacuum breaker, PVB, has a single check and a spring-loaded air inlet, and it protects against backsiphonage only, never backpressure. The spill-resistant version, the SVB or SPVB, is a PVB built so it does not spill water across the gauge during the test, which is what lets it go indoors.

The standards behind them, which you should verify against the edition your program adopts, map cleanly. ASSE 1013 covers the RP, ASSE 1015 the double check, ASSE 1020 the PVB, and ASSE 1056 the spill-resistant vacuum breaker. The fire-line detector versions, the reduced pressure detector and the double check detector, fall under ASSE 1047 and 1048. The common atmospheric vacuum breaker is a different animal and is generally not a testable assembly, so it does not belong in this workflow.

What does each gauge reading tell you?

Every backflow test reads pressure differential across the working parts with a calibrated differential gauge, and each reading isolates one component. The framework is the USC Foundation for Cross-Connection Control and Hydraulic Research field test procedure, the one most programs adopt, and it runs the same logic whether you carry a five, three, or two needle valve gauge.

On an RP, the relief valve opening point tells you whether the relief will dump the zone when it has to. It should start to discharge at a differential of at least 2.0 psid. The first check reading tells you whether CK1 holds a tight differential above that relief opening point. The second check tells you whether CK2 holds against backpressure. On a double check, you read both checks the same way, each holding tight on its own. On a PVB or SVB, you read whether the air inlet opens at its minimum differential and whether the single check holds in the flow direction.

The pass values vary by procedure and by the assembly standard, so confirm the exact criteria against the procedure your AHJ has adopted. The numbers below are the common minimums, not a universal law.

Why did my backflow assembly fail?

A backflow assembly fails at one of a short list of components, and the gauge reading tells you which one before you ever open it. Map the symptom to the part, flush, then repair the part the reading points at. Replacing rubber the gauge never accused is how you end up retesting a failure twice.

The table below is the field map from symptom to cause to repair. Read it as a starting diagnosis, not a guarantee, because debris can imitate almost any of these failures, which is why flushing comes before parts every time.

The number one field failure: debris on the seat

The most common reason an assembly fails is not a worn part. It is debris sitting on a check seat or under the relief, usually after the water main was disturbed. A main break, a hydrant flush, a meter change, a repair upstream, any of these kicks scale, sand, gasket shavings, and pipe dope downstream, and a single grain on a seat holds a check open just enough to fail the differential.

This is why flushing comes before condemning parts. Before you order a kit or pull a check, isolate the assembly, relieve the pressure, and flush the line and the assembly to wash debris off the seats. A real share of failures pass on retest after nothing more than a good flush, because the part was never bad. It was holding a piece of the main.

The rookie move is to read a failed check, order rubber, rebuild it, and retest, only to fail again because the debris is still in the line and lands on the fresh seat too. Flush first. If it still fails after a clean flush, the part is genuinely worn and the kit is the answer.

Why is the relief valve on my RPZ weeping or discharging?

A relief valve that weeps or runs is the most common RP complaint, and it is almost always telling you that the first check is passing pressure backward into the zone, not that the relief itself is broken. The relief valve sits below the first check and opens when zone pressure climbs to within about 2.0 psid of the supply. If CK1 leaks back, the zone pressure rises, and the relief does exactly what it is built to do. It dumps to protect you. The discharge is a symptom of the check, read correctly.

So diagnose in order. Flush first, because debris on CK1 is the usual culprit and a flush can clear it. If it still weeps, inspect and rebuild the first check with its rubber kit. Only after the checks hold do you suspect the relief valve itself, which fails from a swollen or torn diaphragm, a weak relief spring, or debris under the relief seat.

A relief that opens below 2.0 psid, or that will not seat, gets its own relief kit. One pattern points straight at the relief: continuous discharge under no-flow conditions while the checks test fine. That is the relief diaphragm or spring, not a check. Watch also for thermal expansion, covered later, which makes a healthy relief weep on hot afternoons for a reason that has nothing to do with the assembly.

Check valves that leak: CK1 and CK2

A check valve fails when it cannot hold its rated differential, and the cause is one of three things: debris on the seat, worn or swollen rubber on the disc, or a weak spring that no longer loads the check closed. On an RP, the first check carries the higher differential and sees the harder duty, so it wears first, and it is the one a weeping relief usually accuses.

On a double check, both checks have to hold independently, because the whole protection is the two of them in series. A DC with one good check and one leaker has lost its redundancy even if the assembly still passes overall, so each check is read on its own merit, not as a pair.

The repair is the same logic across both. Flush to rule out debris. If the check still will not hold, replace the rubber, the disc, or the whole check module per the kit the manufacturer lists for that model and size. Inspect the seat while it is open, because a pitted or scored seat chews up new rubber, and a rubber-only kit will not save a damaged seat. That is when the complete kit or a new module earns its cost.

Why won't the air inlet on my PVB open?

On a PVB or SVB, the air inlet is the part that admits air to break a siphon, and when it sticks shut the assembly fails, because it can no longer do its one job against backsiphonage. The air inlet should open at a differential of at least 1.0 psid as the supply pressure falls. When it does not, the usual causes are a fouled or corroded poppet, a stuck or broken air-inlet spring, or debris and mineral scale binding the moving parts. That last one is common on irrigation PVBs that sit outdoors and freeze.

The single check on a PVB also has to hold, at least 1.0 psid in the flow direction, and it fails the same way any check does. Because a PVB has fewer parts than an RP, the repair is usually a clean and a spring or poppet kit for the air inlet, or a check rubber, both available as model-specific kits.

The PVB and SVB differ in how the air inlet is loaded against the check, which is what lets the spill-resistant version stay dry on the test. Match the kit to the exact assembly, not just the brand, because the spill-resistant internals are not interchangeable with a standard PVB.

RPZ vs DCVA vs PVB: what fails most on each

The three common assemblies fail in characteristic ways, and knowing the pattern speeds the diagnosis. An RP fails most often at the relief valve discharge, which is really a first-check problem most of the time, because the relief is sensitive and reports any zone pressure rise immediately. The RP is the most diagnostic assembly precisely because the relief tells on the checks.

A double check has no relief valve, so it fails quietly. There is no discharge to warn you. A DC just stops holding a differential on one or both checks, and you only find it on the annual test, which is one reason the test matters more on a DC than people treat it. By the time a DC reads failed, it has often been failing for a while with nobody the wiser.

A PVB fails at the air inlet more than anywhere else, because the air inlet lives exposed, moves every time the system depressurizes, and is the first thing to seize from scale, corrosion, or a freeze. On irrigation PVBs the seasonal pattern is plain: they come up failed in spring after sitting through winter. Match your first suspicion to the assembly in front of you and you flush and open the right part first.

Matching the repair kit to the assembly

The repair kit has to match the make, model, and size of the assembly, and getting this wrong is a wasted trip and a failed retest. Every major manufacturer publishes two tiers of kit for most models: a rubber kit, which is the discs, o-rings, seals, and diaphragm for routine wear, and a complete or total rebuild kit, which adds springs, seats, or full check modules for an assembly worn or damaged past the rubber.

Read the assembly's nameplate or stamped model and size before you pull a kit. The same brand sells different internals across model generations and revisions, so the number on the tag drives the part, not the year or the look of the body. Do not guess part numbers from memory. Use the manufacturer's repair-parts lookup for the exact model, size, and revision, because they change kit contents between revisions and a close-but-wrong kit will fight you at the seat.

The table lists common testable lines by manufacturer as a starting point for the lookup, not as a substitute for it. Verify the model and size on the assembly itself, every time.

Isolating and depressurizing the assembly safely

Before any part comes out, the assembly has to be isolated and the pressure relieved, and this is the step where people get soaked or get hurt. Close the downstream shutoff first, then the upstream shutoff, so you are working on a section closed at both ends. Then bleed the trapped pressure through the test cocks before you crack any fitting, because an assembly closed at both ends is still holding line pressure, and a check or relief under pressure will let go when you loosen it.

Open a test cock and watch for flow to stop, confirming the upstream shutoff is actually holding and not passing. A shutoff that leaks by means you never get the assembly to zero, and you do not want to discover that with a check half out. If the line cannot be reliably isolated at the assembly, isolate further upstream or coordinate a shutdown, because working a backflow assembly on a live line is how you flood a mechanical room.

Plan for the water sitting in the body even after the pressure is off. Have a bucket and rags ready. On a high-hazard line, treat whatever is in the assembly as suspect and protect yourself accordingly, because the assembly is there precisely because the downstream side is not clean.

The rebuild: seats, rubber, and seating

With the assembly isolated and drained, the rebuild is methodical. Open the cover or the check retainer, lay the parts out in the order they come, and note orientation, because a check disc or a relief diaphragm installed backward fails the retest and sends you right back in. Inspect every seat as it is exposed. Rubber is the easy part. A scored, pitted, or corroded seat is what quietly ruins a rubber-only repair, and that is when you step up to the module or the complete kit.

Clean the seats and the bore before the new rubber goes in. Lubricate o-rings and seals only with a lubricant approved for potable water, never petroleum grease, which swells the rubber and contaminates the line you are trying to protect.

Seat the new check and relief components without forcing them, and tighten covers and retainers to the manufacturer's torque and sequence, which lives in the model's repair instructions and varies by size. Do not improvise torque on a backflow cover. Over-torque distorts the seat you just cleaned, and under-torque weeps. The torque value belongs to the manufacturer, not to feel.

Do I need to retest after a backflow repair?

Yes. A repaired backflow assembly has to be retested and pass before it goes back in service, no exception, because the repair is not proven until the gauge says so. A rebuild can fail its own retest from a backward diaphragm, a pinched o-ring, debris that landed on the fresh seat, or a part that was the wrong kit. The retest is how you catch your own work before the customer does.

Run the full field test again, all readings, the same procedure as the original test, and record every value, not just the word pass. The retest values are what the program accepts as proof the assembly works.

Whether the retest can happen the same day, and whether the same person can perform and certify both the repair and the retest, varies by jurisdiction. Some programs allow it, some restrict who signs, some want a separate report or a waiting period. Confirm the same-day and same-tester rules with your AHJ or water purveyor before you assume the closeout is done. The rebuild is your work. The passing retest is the proof, and the proof is what the program is buying.

AHJ and water purveyor submittal

The compliance side of a backflow repair is a paper trail to a specific authority, usually the water purveyor or a cross-connection control program, and it runs on deadlines. A failed test typically has to be reported, and the assembly returned to passing condition and documented, within a window the program sets, often counted in days from the failed test. Miss the window and the customer can face a notice, a fine, or a shutoff even after the assembly is fixed, because the program tracks the dates, not just the outcome.

What the program wants is the failed report, the record of the repair, and the passing retest report, tied to the same assembly by serial number and signed by a certified tester or repairer. Many programs require the certification number and a current gauge calibration date on the form, and they reject reports missing either one.

The deadlines, the forms, who may sign, and where it gets submitted are all local. Confirm them with the AHJ or purveyor for the jurisdiction the assembly sits in, because two towns on the same water system can run different programs with different clocks.

Freeze damage, thermal expansion, and seasonal failures

Backflow assemblies fail on a seasonal clock, and freeze is the biggest reason. Water left in an assembly through a hard freeze expands as it turns to ice and cracks bodies, splits covers, distorts seats, and tears rubber. Irrigation PVBs and RPs above grade are the classic casualties. They come up failed or visibly cracked in spring because nobody winterized them. A cracked body is not a rubber-kit repair. It is a replacement.

Thermal expansion is the quieter seasonal failure. A check valve closes against backflow, and on a closed system with no expansion path, water heating up has nowhere to go, so pressure climbs behind the check. On an RP that shows up as the relief valve weeping on hot afternoons even though the checks test fine, because the relief is dumping thermal pressure the system has no other way to release. The fix there is not the assembly. It is a thermal expansion tank or relief on the system, and chasing it as a backflow repair wastes a kit.

Read the season. A spring batch of failures after a cold winter is freeze. A relief weeping only when the sun is on the building is thermal expansion. Neither is solved by blindly rebuilding checks.

Parts and tools you carry

• Calibrated differential test gauge with a current calibration date and the needle-valve hose set.
• Manufacturer rubber kits and complete or module kits for the assemblies you service, by model and size.
• Potable-water-safe lubricant for o-rings and seals, never petroleum grease.
• Wrenches and the cover or retainer tools for the assemblies you work, plus a torque wrench for cover bolts.
• A bucket, rags, and a way to flush the line and the assembly before condemning parts.
• The manufacturer's repair-parts lookup and torque specs for the model in front of you.
• Your tester or repairer certification number and the forms the AHJ accepts.

What to document

A backflow repair that is not documented to the program is not a closed job, even if the assembly works. The record proves the same assembly was tested, failed, repaired, and retested, and it is what the purveyor accepts and what defends you if the assembly is questioned later.

Capture the assembly identity by serial, the failure, what you did, and the passing retest, tied together on one trail. Record the make, model, size, and serial number, the location and the service it protects, the failed readings and the failure mode, the flush you performed before parts, the kit and parts used, every retest reading, your certification number, the gauge calibration date, and the submittal to the AHJ with its deadline.

Common mistakes

• Condemning and replacing parts before flushing, when debris on the seat was the whole problem.
• Rebuilding the part the gauge never accused, then failing the retest on the part that actually leaked.
• Pulling the wrong kit by guessing the model instead of reading the nameplate and looking it up.
• Returning a repaired assembly to service without a passing retest with all readings recorded.
• Using petroleum grease on the seals, which swells the rubber and contaminates the line.
• Not recording the serial number, so the failed report and the passing retest cannot be tied to the same assembly.
• Rebuilding checks on an RP whose relief only weeps from thermal expansion, when the fix is an expansion tank on the system.
• Missing the AHJ submittal deadline, so the customer is non-compliant even though the assembly is fixed.

Field checklist

Standards and references

The assembly standards come from ASSE International, and they map to the assemblies you test. ASSE 1013 covers the reduced pressure principle assembly, ASSE 1015 the double check, ASSE 1020 the pressure vacuum breaker, and ASSE 1056 the spill-resistant vacuum breaker. The fire-line detector assemblies fall under ASSE 1047 and ASSE 1048. Confirm the standard and its edition against what your jurisdiction and the approval list adopt, because the standards are revised on a cycle.

The field test procedure most programs adopt is the USC Foundation for Cross-Connection Control and Hydraulic Research procedure, published in its Manual of Cross-Connection Control, with the field test steps for RP, DC, PVB, and SVB. The common pass values, relief opening at least 2.0 psid and checks holding at least 1.0 psid, come from that framework, but the exact criteria and the gauge format your AHJ accepts control the test. AWWA's M14 manual, Backflow Prevention and Cross-Connection Control, is the water-industry reference for program practice.

Tester and repairer qualification is its own standard, the ASSE Series 5000 professional qualification standards. The plumbing code that adopts all of this is local, the IPC or UPC as amended, and the water purveyor's cross-connection program sets the reporting and the deadlines. The standard that controls any given call is the one your AHJ has actually adopted, so verify before you cite a number on a submittal.

Units and terms

Backflow work has its own shorthand, and the same part goes by a few names across manufacturers and forms.

Pressure differential is read in psid, pounds per square inch differential, which is the difference the gauge sees across a check or the relief, not the line pressure. The assemblies are abbreviated by type: RP or RPZ, DC or DCVA, PVB, and SVB or SPVB. The internal parts are CK1 and CK2 for the checks and RV for the relief valve. The hazard the assembly guards against is the cross-connection, and reverse flow arrives two ways, backpressure and backsiphonage.

RPZ / RP

Reduced pressure principle assembly, the highest-protection testable backflow assembly (ASSE 1013)

DC / DCVA

Double check valve assembly, two checks in series for low-hazard protection (ASSE 1015)

PVB

Pressure vacuum breaker, a single check plus an air inlet for backsiphonage-only protection (ASSE 1020)

SVB / SPVB

Spill-resistant pressure vacuum breaker, a PVB built not to spill on test (ASSE 1056)

CK1 / CK2

Check valve number one and number two, the independently acting checks inside the assembly

RV

Relief valve, the part on an RP that dumps the zone to atmosphere when a check fails

psid

Pounds per square inch differential, the pressure difference the gauge reads across a check or the relief

Cross-connection

Any point where the potable supply can connect to a non-potable source

Backpressure / backsiphonage

The two ways flow reverses: downstream pressure pushing back, or upstream vacuum pulling back

Related tools

Calculators and readiness checks for this work

Compare your options

FAQ

People also ask

• How do you test a backflow preventer?
• What is the difference between an RPZ and a double check?
• What is graywater?
• How do you water-test a DWV system?
• Why is my water pressure low?