When should a pump test be considered a failure?
If you don't want to read all the way through the details, the simple answer is that at initial acceptance you should be perfect within the accuracy of your test equipment. At your annual pump tests, not more than 5% drop in from the original curve/data points.
Note that the initial acceptance testing per NFPA 20 is different than the the annual fire pump testing of requirements NFPA 25.
NFPA 20 Initial acceptance testing
You bought a new pump and it should perform exactly per specs is the simple answer. The pump manufacturers take this seriously before it leaves the factory in order to maintain their UL listings and FM approvals. As such, the pump impeller was factory trimmed to be exactly the rated pressure at 100% flow. There should be no questions about it and you should always be provided with a certified factory curve for your backup documentation.
Our goal in the field after installation is to make sure there was no shipping damage or issues with the driver. NFPA 20 states this requirement as follows:
14.2.4.2 - At all flow conditions, including those required to be tested in 14.2.6.2, the fire pump as installed shall equal the performance as indicated on the manufacturer's certified shop test curve within the accuracy limits of the test equipment.
This should be verified at minimum at no-flow (churn), 100% (rated flow), and 150% (max flow).
Variable-speed fire pumps have additional requirements at the 25%, 50%, 75%, and 125%, points (seven in total) with the speed limiting features enabled. A second test with the speed limiting featured turned off must also be run at the standard three points.
Simple enough, but lots of potential issues with accurately taking those measurements. But first lets briefly talk about pass criteria after the pump has been initially accepted.
NFPA 25 Annual Fire Pump Test
Things wear-out. Small pieces of debris (i.e. sand, dirt, and other particles) in the water are slowly wearing down the impeller and changing it's shape. Plus bearings can wear out causing friction. But how much change in performance before you need to take action?
A 5% reduction is the simple rule. NFPA 25-2023 states this requirement as follows:
8.3.2* No-Flow Test.
8.3.2.1.2.1* The pressure readings on the discharge and suction gauges shall be recorded, and a pressure difference that is greater than 95 percent of the rated pump pressure shall be investigated and corrected.
8.3.7.2 Evaluation of Fire Pump Test Results.
8.3.7.2.3 The fire pump test results shall be considered acceptable if both of the following conditions are satisfied:
The fire pump test results shall be considered acceptable if all of the following conditions are satisfied:
(1) Fire pump meets the flow and pressure requirements of the most demanding system(s) supplied by the fire pump based on owner-provided system design information
(2)* Fire pump supplies 100 percent of rated flow
(3)* Net pressure at each flow point is at least 95 percent of one of the following:
(a) Original manufacturer's pump curve
(b) Original unadjusted field test curve
(c) Test curve generated from the fire pump nameplate
Note that the 5% reduction apply to all three test points; Churn, Rated, and 150%.
Error Sources
Now let's discuss real-world conditions and why your field curve might not perfectly match the factory-certified curve. Discrepancies often arise from several standard factors that can be accounted for during testing.
Equipment Accuracy / Calibration
Per NFPA 25 (Sections 8.3.3.2.2.2 and 8.3.3.2.2.3), testing devices must be calibrated annually. Standard gauges and transducers must maintain an accuracy of ±1%, while flow meters require an accuracy of ±3%. These small variances can accumulate, creating a noticeable difference between field results and the factory's high-precision lab environment.
RPM Variation
Factories test pumps using precision-calibrated electric motors to control RPM exactly. In the field, your electric motor or diesel engine will have slight variations in speed. Because pump performance is tied to the affinity laws—where pressure varies by the square of the speed—even a small deviation in RPM leads to a significant change in discharge pressure.
Circulation Relief Valve (1-20 gpm flow potentially)
On smaller pumps, it is common to see lower-than-expected churn pressure due to the required flow of the circulation relief valve. This valve is designed to flow water (typically 1–20 GPM) during no-flow conditions to prevent the pump from overheating. Additionally, diesel engines utilize a cooling loop that can pull another 10–100 GPM from the pump discharge.
These "hidden" flows mean your "no-flow" test isn't truly at zero GPM. For a small 100 GPM pump, a 20 GPM relief flow represents a 20% shift on the curve. To get the most accurate churn reading, you can temporarily close these valves, but only for a very brief period and while closely monitoring the pump temperature to ensure it does not overheat.
Field Application Guide
Step-by-Step Calculation Guide
To determine if the fire pump is within 95% of required performance:
Subtract the suction pressure from the discharge pressure to get the differential (net) pressure for each flow condition (churn, rated, and peak flow).
For vertical turbine pumps, add the total suction lift in psi to the discharge pressure to calculate net pressure. Lift in height can be converted to psi by multiplying by 0.433 or dividing by 2.31.
Divide the result by the PSIG value on the data plate for that specific flow condition.
Compare the result to 95% of the data plate value to see if the pump is meeting the required performance.
Applying Affinity Laws
If results fall below the 95% threshold, apply Affinity Laws to adjust net pressure and flow:
Use the formulas:
Q2 = Q1 * (N2 / N1)
Pn2 = Pn1 * (N2 / N1)^2
Where N1 is the field speed (RPM) and N2 is the data plate speed (RPM). This recalibrates results to verify if they meet the 95% requirement.
Testing Methods
There are three accepted methods for annual flow testing:
Hose streams (using diffusers, play pipes, or nozzles on the test header).
Flow meters (discharging to drain, atmosphere, or suction reservoir).
Closed-loop metering (piping back to the suction side, permitted for only two consecutive years; methods one or two must be used at least every third year).