What a Backflow Preventer Does to a Fire Line

Why this matters

A backflow assembly is the only component anybody deliberately adds to a fire service main in order to make it worse hydraulically. It is there for a good reason, it is required by somebody other than the fire code, and it takes away pressure the sprinkler system was designed to have. The mistake that follows is almost never the physics; it is the bookkeeping. Crews subtract the whole device loss from the available pressure and conclude a working system has failed, or they subtract nothing and hand over a system that will not make its design density. The device loss is a re-basing of a calculation that already contains a device, not a subtraction from the water supply.

Who actually requires it, and why the fire code is not the answer

Cross-connection control on a fire service is a drinking water requirement, administered by the water purveyor under the state drinking water program and the adopted plumbing code, and enforced through the service agreement. It is not a federal fire requirement and it is not an NFPA requirement, although the adopted NFPA 13 and NFPA 25 editions govern what the assembly does to the fire system once it is there. Your authority having jurisdiction for the fire side and the purveyor for the water side are two different named roles with two different sets of authority, and they routinely require different things about the same pipe.

The reason is the water in the pipe. Sprinkler piping is stagnant for years, often black steel, and it may hold antifreeze, foam concentrate, corrosion inhibitor, or water from an auxiliary supply such as a tank, a well, or a fire department connection. The hazard classification the purveyor assigns decides the family of device:

• Double check detector assembly, for a pollutant-level, non-health-hazard cross connection: a plain wet system with no additives and no unapproved auxiliary supply.
• Reduced pressure detector assembly, for a contaminant-level health hazard: additives, or an auxiliary supply of unapproved quality.

The detector variant on both is a metered bypass sized to register small flows, so a leak or an unauthorized tap shows up on a meter instead of hiding inside a large main.

What it costs you, and how that cost behaves

Two components, and they behave differently.

There is a nearly fixed component from the check springs, which is what a device loses at almost no flow. Then there is a velocity-dependent component that grows roughly with the square of flow, holding the device size and geometry constant with the check discs fully lifted. Near the low end the fixed component dominates and a square-law estimate under-predicts the loss; well above the minimum the square law is a reasonable screening tool for the same device.

A reduced pressure assembly costs more than a double check because its relief zone actively maintains a differential between the two checks. As a family, double check assemblies commonly sit on the order of 5 to 10 psi at design flow and reduced pressure assemblies commonly 10 to 15 psi or higher, but neither range is a number you may use. The only figure that governs is the manufacturer's listed head loss curve, for that model, that size, at YOUR design flow, and you read whether the listing includes the assembly's shutoff valves or excludes them.

   street main
       |
     purveyor meter
       |
   [ shutoff ] [ check ] [ check ] [ shutoff ]   the assembly
       |                                          (both shutoffs are
       |                                           fire system control
       |                                           valves and must be
       |                                           supervised)
       +---- fire department connection ties in HERE,
       |     on the system side, so pumper pressure
       |     never passes back through the assembly
       |
     sprinkler riser

The bookkeeping rule

A hydraulic calculation states a required pressure at a required flow at a defined point and elevation. Everything the designer modeled between that point and the sprinklers is already inside the required figure. So before you charge a device loss, write down what the number you are correcting already contains.

• If the calculation modeled a backflow assembly and you are replacing it with a different one, the correction is a re-basing. You charge the difference between the installed listed loss and the modeled loss, not the full installed loss.
• If the calculation modeled no device and the purveyor added the requirement afterward, the correction is an addition. You charge the full listed loss.

Say which of the two you are doing, in the same breath as the number. The common error runs one way, a re-basing written as an addition, and every one of them lands in a direction that makes somebody redesign a system that was fine.

A narrative case: the upgrade that looked like a failure

An existing sprinklered building adds a static tank on a well as an auxiliary suction supply. The purveyor reclassifies the fire service from a pollutant hazard to a health hazard and requires the existing double check detector assembly be replaced with a reduced pressure detector assembly.

What was on file. A sealed hydraulic calculation: system demand 480 gpm at 68 psi required at the point of connection, and the calculation explicitly modeled a double check detector assembly at 7 psi at 480 gpm. The design documents state that the 68 psi figure includes a 10 psi cushion.

The supply curve came from a flow test with static 82 psi and residual 71 psi at 940 gpm, so the drop at the test flow was 11 psi.

Available pressure at the system's flow. Available drop scales with flow raised to about the 1.85 power, holding the same supply and the same test point constant. At 480 gpm the drop is 11 times (480 divided by 940) to the 1.85, which is 11 times 0.288, about 3.2 psi. Available at 480 gpm: 82 minus 3.2, or 78.8 psi. Margin over the 68 psi requirement: 10.8 psi.

The arithmetic the crew did. The new assembly's listed loss at 480 gpm is 13 psi. Ten point eight minus thirteen is negative 2.2 psi, so the job stopped and a fire pump got quoted.

The arithmetic the calculation supports. The 68 psi already contains a 7 psi device. Replacing the device re-bases that term rather than adding to it: charge 13 minus 7, which is 6 psi. New required pressure 68 plus 6, which is 74 psi. Margin 78.8 minus 74, which is 4.8 psi. Positive, thin, and real.

Corrections and qualifiers, each printed.

Cushion, not double-counted. The design documents put a 10 psi cushion inside the 68 psi. It stays inside the 74 psi. It is not added again on top of the 4.8 psi margin, and it is not removed to make the margin look healthier. The margin of 4.8 psi sits above a requirement that already carries the cushion.

Elevation, net zero and printed anyway. The new assembly sits 4 ft lower in the riser room than the point the calculation referenced. Water is 0.433 psi per foot, so 4 ft is 1.73 psi. Moving down gains 1.73 psi of static head at the assembly and pays the same 1.73 psi back climbing to the reference point. A relocation within the same room is a wash, and the reason to print it is that half the field corrections of this shape claim the gain and skip the payback.

Loss curve basis, stated. The 13 psi is read from the listed head loss curve for that exact model and size at 480 gpm, not from a catalog typical figure and not from the value at the assembly's rated flow. The listing was checked for whether the assembly's shutoff valves are included; if they are excluded, their loss is a separate addition, not part of the 13.

Supply curve basis, stated. Static 82 psi and residual 71 psi at 940 gpm are a dated snapshot of a live municipal supply. The 1.85 exponent holds the same supply and the same test point constant; a new test after a main change re-bases the whole curve, not just one point.

Who is allowed to make this change. Nobody on the crew. The calculation is a sealed engineering document, so the engineer of record revises it and the AHJ accepts the revision. The value of doing the arithmetic in the field is knowing whether you are looking at a 6 psi conversation or a 13 psi one before you quote a fire pump the building may not need.

Sibling-rule check. The 4.8 psi result is stated with the flow it applies at and the point it is referenced to, so no pressure travels here without its basis. The shutoffs added by the assembly are named as fire system control valves requiring supervision, so the "shut valve nobody notices" failure is not created by this work. The purveyor's annual test and the adopted NFPA 25 forward flow test are both treated as planned impairments with a start, a compensating measure, and an end, which the records card requires. No test discharge is routed across an exit discharge path, which the egress card requires and which an iced walkway would violate.

The parts that bite after the install

The shutoffs are fire system control valves. Two of them, on the fire service, and a closed one is the most common way a sprinkler system is out of service without anybody knowing. The adopted NFPA 13 and NFPA 25 editions require control valves to be supervised, by lock, seal, or electronic supervision, and the assembly's valves are not exempt because a plumber installed them.

The tests are recurring impairments. The purveyor's annual backflow test and the adopted NFPA 25 edition's forward flow test at system demand are two separate tests, on two separate schedules, and both take the fire system down while they run. That is a planned impairment with an impairment coordinator, a tag, and compensating measures, not a maintenance visit.

A reduced pressure assembly discharges. Under a supply fluctuation or a fouled check, the relief port dumps water at a rate that can exceed a floor drain. An interior installation needs a drain sized for the relief discharge, and an assembly in a vault turns a nuisance discharge into a drowning hazard.

Opening it is stored energy. Before any test cock is opened or any flange broken, the control valve is closed and tagged, which starts the impairment process, then the line is drained and the gauge is read at zero before the first bolt moves. A large flange at service pressure holds a serious load. Where the assembly is in a vault or pit, treat it as a permit-required confined space under 29 CFR 1910.146 until an evaluation says otherwise, with atmospheric testing before entry and a plan for water entry, not just for atmosphere.

References

• NFPA 13, Standard for the Installation of Sprinkler Systems, and NFPA 25 for inspection, testing and maintenance, in the editions adopted and amended by your authority having jurisdiction, for control valve supervision, forward flow testing, and the impairment program
• The state drinking water program's cross-connection control requirements and the adopted plumbing code, administered by the water purveyor, for device selection by hazard classification
• Manufacturer listed head loss curves for the specific assembly model and size at the design flow
• 29 CFR 1910.146 for permit-required confined spaces where the assembly is in a vault or pit
• See related: What a Fire Protection Record Has to Show a Year Later; What a Means of Egress Cannot Pass Through