What a Fire Pump Is For and When a Building Needs One

Why this matters

People ask whether a building needs a fire pump as though it were a property of the building. It is not. A fire pump exists to close a measured gap between two curves: what the water supply can deliver at the point of connection, and what the most hydraulically demanding part of the system requires. Height, floor area and occupancy all push the demand point around, but none of them answer the question. A flow test laid against a calculated demand point answers it, and the same building can need a pump on one side of the street and not on the other.

Getting this backwards has a specific cost. A pump installed where the supply was already adequate adds a machine tested weekly, flow tested annually, and taken out of service for each of those events. A pump omitted where the gap was real leaves remote heads discharging below the density the calculation promised, and nobody finds out until the fire.

What a fire pump is, and what it is not

A fire pump is a centrifugal pump, a driver, a controller and a set of appurtenances, arranged so it starts automatically on a pressure drop and runs until somebody stops it by hand. The machine itself is an ordinary centrifugal pump: impeller behaviour, curve shape, suction conditions and parallel or series operation belong to the pump cards elsewhere in this library. This card owns the part specific to fire service.

The most important thing it is not: a fire pump adds pressure, it does not add water. It raises the pressure of whatever flow the supply can actually deliver at the suction flange. If the main cannot pass 800 gallons per minute, no pump will make it pass 800 gallons per minute. That is why a supply problem and a pressure problem have completely different remedies, and why a pump proposed as the fix for an undersized main is the wrong machine for the wrong fault.

The two curves

The supply curve comes from a hydrant flow test at or near the point of connection. Three readings produce it: static pressure with nothing flowing, residual pressure with a hydrant flowing, and the flow itself, measured at the outlet. Those points define a curve that can be projected to other flows, because pressure drop through a piping network varies with flow raised to a power near 1.85 for the materials and turbulent conditions typical of water mains. That exponent holds at constant pipe geometry and roughness, and it describes the main you tested on the day you tested it, not the main after somebody closes a sectional valve three streets away.

The demand point comes from the hydraulic calculation: a single flow and a single pressure at a defined reference, usually the base of the riser or the system side of the backflow device. It bundles the density required over the design area of operation, the friction and fitting losses back to that reference, the elevation lift, and the hose allowance the adopted edition adds.

A pump is needed when the demand point sits above the supply curve at the demand flow. That is the whole test.

Constants, with the conditions they were derived under

• Elevation head: 0.433 psi per vertical foot, for water near ambient temperature at ordinary gravity. It is a property of the fluid column, not of the pipe, so it applies at any pipe size and does not care about flow.
• The 1.85 exponent on the flow versus friction relationship, at constant pipe geometry and roughness in turbulent flow. It is a projection tool for the supply curve, not a design method.
• Gauge error has a basis and a character, and both change the arithmetic. A test gauge specified at 1 percent of full scale on a 0 to 200 psi face is a bound of 2 psi, and because the basis is full scale rather than reading, that 2 psi does not shrink when the reading is low. Its character is a fixed systematic offset from one instrument, so it cancels in a difference read on that same gauge and survives in full on an absolute reading. Static minus residual is a difference; static alone is not.

What the installation standard fixes about the pump itself

Three numbers travel with any listed centrifugal fire pump, from NFPA 20, which binds only in the edition the authority having jurisdiction has adopted and amended. The AHJ is a named role with authority over that adoption and its amendments, not a synonym for whoever performs the acceptance test.

• Churn, meaning shutoff, no flow. Net pressure at churn is not more than 140 percent of rated net pressure, because churn pressure plus the highest expected suction pressure has to stay inside the pressure rating of the components downstream.
• The 150 percent point. The pump delivers not less than 65 percent of rated net pressure at 150 percent of rated flow. The pump has to hold up past its rating, not just at it.
• Suction floor. Suction pressure must not fall below the value the adopted edition sets at 150 percent of rated flow, and many local water purveyors impose a higher floor to protect the main. Both apply; the stricter governs.

The case: an addition that triggered the question

A single-story building with an added four-story wing. The design team assumed the existing city supply would carry the new system, because it always had. The permit review asked for a current flow test, which is the moment the question actually gets answered.

The flow test. At the point of connection: static 68 psi, residual 52 psi with a measured flow of 1,180 gallons per minute, so a 16 psi drop at that flow. The gauge is a 0 to 200 psi test gauge at 1 percent of full scale. Correction, printed: the 2 psi bound is a fixed systematic offset on one instrument with a full-scale basis, so it cancels in the 16 psi difference and does not cancel on the static, which is therefore not less than 66 psi. Written with one inequality sign, not as an interval, because a one-sided bound is not a plus-or-minus.

The demand point. The hydraulic calculation returns 520 gallons per minute at the base of the riser for the design area, plus a 250 gallon per minute hose allowance at that same reference, giving 770 gallons per minute. The required pressure there is 84 psi. Inside that 84 psi, the elevation term alone is the 46 foot lift to the design area: 46 ft x 0.433 psi/ft = 19.9 psi, using the constant at ambient water temperature as stated above.

Projecting the supply to the demand flow. 770 divided by 1,180 is 0.653, raised to the 1.85 power is 0.454, so the drop at 770 gallons per minute is 16 psi x 0.454 = 7.3 psi. Nominal residual at demand flow is 68 - 7.3 = 60.7 psi. On the one-sided static bound: not less than 66 - 7.3 = 58.7 psi.

The gap. Demand 84 psi against a nominal 60.7 psi is a shortfall of 23.3 psi at 770 gallons per minute; against the bound of 58.7 psi it is not less than 25.3 psi. The pump is sized against the conservative figure, because rounding in the flattering direction here means a remote head running under its design density.

Selection. A pump rated 750 gallons per minute at 40 psi net is proposed. The demand flow of 770 is 103 percent of rated, between churn and the 150 percent point, so net pressure there is a little under rated: call it 38 psi off the manufacturer's curve. Available at the demand point becomes 58.7 + 38 = 96.7 psi against an 84 psi demand, a margin of 12.7 psi on the conservative basis and 14.7 psi on the nominal one. Both comparators carry the same correction, so the margin is not flattered by mixing a corrected demand against an uncorrected supply.

Checking the suction floor. 150 percent of 750 is 1,125 gallons per minute. The drop there is 16 psi x (1,125/1,180) raised to 1.85, which is 16 x 0.916 = 14.7 psi, so the residual is 68 - 14.7 = 53.3 psi nominal and not less than 51.3 psi on the bound. That clears both zero gauge and a 20 psi purveyor floor.

Checking churn against the downstream rating. Net at churn is not more than 140 percent of 40 psi, or 56 psi. On the highest expected suction of 68 psi that is 124 psi at the system side, inside the ordinary component rating.

Sibling-rule check, run against this scenario before it ships:

• The control valve card's rule that a shut valve removes the whole downstream area with no outward sign: applied. The pump cut-in closes the riser control valve, so it opens an impairment rather than a work ticket, closed on the acceptance test result and the valve proven open under supervision rather than on the pipe being reconnected.
• The obstruction card's statement that density is delivered over an area of operation: used unchanged, since the demand flow is that density over that area plus hose allowance, not a peak flow at one head.
• The example introduces two figures the general section routed rather than stated: the adopted edition's suction floor, taken here as zero gauge, and an illustrative 20 psi purveyor floor. Both are named as what they are, and neither is offered as the answer for another building.

When the answer flips to no pump

Three conditions genuinely invert the method rather than adjusting it.

A tank supply. With a stored-water tank rather than a pressurised main there is no supply curve to project. The tank gives a static head that falls as it drains, and the question stops being whether a pump is needed and becomes what its suction condition is at the lowest expected level. That is the pump cards' territory.

A repiped or re-zoned main. A flow test is one moment. A purveyor who has closed a sectional valve, reduced a zone pressure, or connected a new large user changes the curve without telling you, which is why a pump decision made on an old test is not a decision.

A demand point that can be lowered instead. Larger feed mains, a shorter run to the remote area, a lower design area, or a different sprinkler type can move the demand point below the supply curve. Adding pipe often costs fewer maintenance hours over a decade than a machine needing a weekly churn run, an annual flow test, and an impairment for each.

How to verify you got this right

Confirm the demand point and the supply curve are stated at the same reference. A demand at the base of the riser compared against a supply at the hydrant, with the underground and backflow losses left out of one of them, is the most common way this comparison quietly passes when it should fail.

Confirm the hose allowance is in the demand and is not also subtracted from the supply. It goes in once.

Re-read the flow test date and the gauge basis. A test that does not name its gauge accuracy or its full-scale range cannot support a bound, and without a bound you are designing to the nominal.

Hazard control for the work this section requires. Flow testing discharges a large volume at high velocity: route the discharge where it cannot undercut a foundation, flood an occupancy, cross a roadway or ice a walking surface, and open the outlet slowly, because a sudden opening on a main produces a pressure surge that damages the main. With a diesel driver, run only with the exhaust routed outdoors and stand clear of the plume, since engine exhaust carries carbon monoxide, and use hearing protection under a program meeting 29 CFR 1910.95, because a pump room with a diesel at rated speed is a high noise environment. Do not run any fire pump at churn beyond the brief interval its circulation relief valve is designed for, since a pump churning with no flow puts its input power into the water in the casing and overheats it. Work inside the controller is electrical work: de-energize and lock out under 29 CFR 1910.333(b)(2) and prove dead with the live-dead-live sequence of NFPA 70E-2021, 120.5, rather than 29 CFR 1910.147, which expressly excludes exposure to electrical hazards from work on electric utilization equipment.

References

• NFPA 20, Standard for the Installation of Stationary Pumps for Fire Protection, in the edition the authority having jurisdiction has adopted and amended, for pump curve limits, suction conditions and controller requirements.
• NFPA 13, in the adopted edition, for the hydraulic calculation, design area and hose allowance that form the demand point.
• NFPA 25, in the adopted edition, for the weekly and annual fire pump test regime referenced above.
• 29 CFR 1910.95 for the occupational noise program, and 29 CFR 1910.333(b)(2) with NFPA 70E-2021, 120.5, for work inside the pump controller.
• See related: the centrifugal pump cards for the machine itself, and the control valve and impairment cards in this category.