Water Supply · Calculation
How Do You Read a Fire Pump Test Curve (Speed Corrected)?
Reviewed by a licensed Professional Engineer · NY, NJ, PA, MD, DE · Last reviewed · Code basis: NFPA 20 Chapter 14 and Chapter 6 (DE: 2025 edition; elsewhere the edition referenced by the adopted building code); NFPA 25 (2023) Chapter 8, Section 8.3.3; SFPE Handbook, 5th ed.
General guidance, not a code determination for any project. Confirm the adopted code and local amendments with the authority having jurisdiction.
A fire pump test curve is read by converting each gauge reading to net pressure (discharge minus suction, with velocity head and elevation corrections), correcting each test point to the pump’s rated speed with the affinity laws (flow scales with speed, pressure with speed squared), and then plotting the churn, 100 percent and 150 percent points against the manufacturer’s certified curve and the National Fire Protection Association (NFPA) 20 envelope. A pump that was running 1 percent slow during the test reads about 2 percent low on pressure, which is enough to turn a passing pump into a failing one on paper, so the correction is not optional.
This article covers the arithmetic: the inputs to record, the formulas, a worked example with real numbers for a 1,000 gpm at 100 psi pump, the errors that spoil most field curves and the cases that need an engineer rather than a spreadsheet. The test day itself, who attends and what is witnessed, is covered in DGA’s article on what happens at a diesel fire pump acceptance test. Pump selection is covered in how a fire pump is sized for rated flow and pressure.
At a Glance: Reading a Speed-Corrected Fire Pump Curve
- What is calculated: net pump pressure at churn, 100 percent and 150 percent of rated flow, each corrected to the rated speed on the pump nameplate, then compared with the certified curve and the NFPA 20 limits.
- Method in one line: net pressure = discharge gauge minus suction gauge (plus velocity head and elevation corrections); corrected flow = measured flow × (rated rpm ÷ test rpm); corrected pressure = net pressure × (rated rpm ÷ test rpm)².
- Inputs needed: nameplate rated flow, pressure and speed; suction and discharge gauge readings at each point; flow from the test header pitot readings or a flowmeter; tachometer speed at each point; gauge elevation difference; suction and discharge pipe sizes.
- Code and section: NFPA 20 Chapter 14 for the acceptance test and the three test points; NFPA 20 Chapter 6 for the pump performance envelope; NFPA 25 (2023) Chapter 8, Section 8.3.3, for the annual flow test; the affinity laws are standard pump hydraulics, as covered in the Society of Fire Protection Engineers (SFPE) Handbook, 5th ed.
- Typical result range (units): speed ratio 0.98 to 1.02; pressure correction within about 4 psi on a 100 psi pump; corrected churn 120 to 140 percent of rated pressure; corrected 150 percent point 65 to 75 percent of rated pressure.
- Most common input error: plotting discharge gauge pressure instead of net pressure, which overstates the pump by the full suction pressure (often 20 to 60 psi).
What Inputs Does a Fire Pump Test Curve Need?
Every number on the curve traces back to a reading taken during the test. The table lists each input, its units and where it comes from. A test record that is missing the tachometer readings or the gauge elevations cannot be corrected afterward, which is the main reason acceptance curves get rejected on review.
| Input | Units | Source |
|---|---|---|
| Rated flow | gpm | Pump nameplate and the certified curve (for example 1,000 gpm) |
| Rated pressure | psi (net) | Pump nameplate and the certified curve (for example 100 psi) |
| Rated speed | rpm | Pump nameplate; electric motors are nominally 1,750 or 3,550 rpm and the nameplate full-load speed is usually a little higher (for example 1,770 rpm); a diesel runs at the engine’s governed speed |
| Measured speed at each point | rpm | Tachometer on the shaft or coupling during each flow point, not the governor setting or the motor nameplate |
| Suction gauge reading | psi | Calibrated gauge on the suction flange at each point; may read below 0 psi on a lift or a weak supply |
| Discharge gauge reading | psi | Calibrated gauge on the discharge flange at each point |
| Flow | gpm | Sum of the hose stream flows from pitot readings at the test header, or a calibrated flowmeter in the test loop |
| Pitot reading per hose stream | psi | Pitot tube at each nozzle, with the nozzle diameter and discharge coefficient |
| Elevation between gauges | ft | Measured from the suction gauge centerline to the discharge gauge centerline; 2.31 ft of water equals 1 psi |
| Velocity head correction | psi | Calculated from the flow and the inside diameter of the suction and discharge pipes where the two gauges sit on different pipe sizes |
The rated values define the three test points. NFPA 20 Chapter 14 calls for readings at churn (no flow), at rated flow and at 150 percent of rated flow, and most test crews add points in between to shape the curve. The 150 percent point is the one that strains the supply, so the suction gauge reading there decides whether the test was valid.
How Is Each Test Point Corrected to Rated Speed?
The method has four steps, applied to every test point before anything is plotted: convert the gauge readings to net pressure, read the tachometer and form the speed ratio, scale the flow by the ratio and the net pressure by the ratio squared, and plot the corrected points against the certified curve and the NFPA 20 limits. The formula card gives each step.
Figure 1: Net Pressure and the Speed Correction
Net pressure: Pnet = Pdischarge − Psuction + hv,d − hv,s + (zd − zs) ÷ 2.31, where hv is the velocity head at each gauge in psi and z is the gauge elevation in ft.
Velocity head at a gauge: hv = 0.001122 × Q2 ÷ d4, in psi, with Q in gpm and d the inside pipe diameter in inches.
Corrected flow: Qc = Q × (Nr ÷ N), where Nr is the rated speed and N the tachometer reading.
Corrected pressure: Pc = Pnet × (Nr ÷ N)2. Power scales with the cube of the same ratio.
Net pressure is the pressure the pump adds. A discharge gauge alone includes whatever the suction supply contributed, so a pump on a 50 psi city main looks 50 psi better than the same pump on a tank. The velocity head term matters when the suction gauge sits on a larger pipe than the discharge gauge, which is the usual arrangement.
At 1,520 gpm with Schedule 40 pipe at both gauges (an 8 inch suction with a 7.981 inch inside diameter and a 6 inch discharge with a 6.065 inch inside diameter), the velocity head is about 0.6 psi at the suction gauge and about 1.9 psi at the discharge gauge. The pump is therefore credited with roughly 1.3 psi that the discharge gauge does not show.
The speed ratio is the step most often skipped. An electric motor runs below synchronous speed by its slip, and the slip grows with load, so the 150 percent point is usually the slowest reading of the test. A diesel engine drifts with governor response and fuel temperature. Both effects are small, usually within 2 percent, and both are large enough to move a marginal pump across the 65 percent or 140 percent line.
Once every point is corrected, the plot has three comparisons. The corrected curve should sit on or above the manufacturer’s certified curve for that serial number. The corrected churn must not exceed 140 percent of rated pressure, and the corrected 150 percent point must deliver at least 65 percent of rated pressure; both limits come from the NFPA 20 Chapter 6 performance envelope for centrifugal fire pumps. For an annual test under NFPA 25 (2023) Chapter 8, the comparison is with the original unadjusted field acceptance curve, the original manufacturer’s curve or the nameplate curve, rather than with the envelope alone.
Worked Example: A 1,000 gpm at 100 psi Pump Rated at 1,770 rpm
The pump in this example carries a nameplate of 1,000 gpm at 100 psi net at 1,770 rpm, driven by an electric motor. The test crew recorded net pressures, already corrected for velocity head and the 2 ft gauge elevation difference, and read a tachometer at each point. The three raw points were 142 psi at churn and 1,790 rpm; 1,015 gpm at 98 psi and 1,760 rpm; and 1,520 gpm at 66 psi and 1,745 rpm.
Taken raw, the churn reading of 142 psi exceeds the 140 psi ceiling, and the 150 percent reading of 66 psi barely clears the 65 psi floor. Both conclusions change once the speed is accounted for. The table shows the arithmetic for each point.
| Test point | Raw reading | Speed ratio Nr ÷ N | Corrected flow | Corrected net pressure |
|---|---|---|---|---|
| Churn | 0 gpm, 142 psi at 1,790 rpm | 1,770 ÷ 1,790 = 0.9888 | 0 gpm | 142 × 0.98882 = 142 × 0.9778 = 138.8 psi |
| 100 percent | 1,015 gpm, 98 psi at 1,760 rpm | 1,770 ÷ 1,760 = 1.0057 | 1,015 × 1.0057 = 1,021 gpm | 98 × 1.00572 = 98 × 1.0114 = 99.1 psi |
| 150 percent | 1,520 gpm, 66 psi at 1,745 rpm | 1,770 ÷ 1,745 = 1.0143 | 1,520 × 1.0143 = 1,542 gpm | 66 × 1.01432 = 66 × 1.0289 = 67.9 psi |
The same arithmetic can be run in Python. The snippet below reproduces the table:
N_r = 1770
points = [(0, 142, 1790), (1015, 98, 1760), (1520, 66, 1745)] # gpm, net psi, rpm
for Q, P, N in points:
r = N_r / N
print(f"{Q*r:7.0f} gpm {P*r*r:6.1f} psi (ratio {r:.4f})")
# 0 gpm 138.8 psi (ratio 0.9888)
# 1021 gpm 99.1 psi (ratio 1.0057)
# 1542 gpm 67.9 psi (ratio 1.0143)
The corrected churn is 138.8 psi, which is 138.8 percent of rated pressure and inside the 140 percent ceiling (140 psi). The corrected 150 percent point is 67.9 psi at 1,542 gpm, which is 67.9 percent of rated pressure at 154 percent of rated flow, so the pump clears the 65 psi floor at 1,500 gpm with margin. The corrected rated point, 99.1 psi at 1,021 gpm, sits within 1 psi of the nameplate, and on a plot the three corrected points follow the certified curve. The pump passes.
Figure 2: Raw and Corrected Test Points Against the NFPA 20 Envelope
The same points can be plotted and corrected in DGA’s fire pump test curve plotter, which takes the gauge readings, tachometer speeds and nameplate values and draws the corrected curve against the envelope. The tool applies the same affinity correction shown in the formula card.
DGA designs fire pump installations and reviews acceptance and annual test results across New York, New Jersey, Pennsylvania, Maryland and Delaware. Project briefs get confirmation and preliminary code feedback within one business day. See where DGA works →
Which Errors Spoil Most Field Test Curves?
The arithmetic is short, so most bad curves come from a bad input rather than a bad formula. The table lists the errors DGA’s engineers see most often when reviewing contractor test reports, how each one shows up on the plot and how to fix it.
| Error | How it shows on the curve | Fix |
|---|---|---|
| No speed correction | A pump that ran 1 to 2 percent slow at 150 percent flow plots 2 to 4 percent low on pressure and can fail the 65 percent floor on paper | Record a tachometer reading at every point and correct flow by the ratio and pressure by the ratio squared |
| Discharge gauge used as net pressure | The whole curve sits above the certified curve by the suction pressure, and churn appears to exceed 140 percent | Subtract the suction gauge reading at each point; a suction reading below 0 psi adds to the net |
| Velocity head ignored on different pipe sizes | Net pressure reads 1 to 3 psi low at 150 percent flow with an 8 in. suction and 6 in. discharge | Add the discharge velocity head and subtract the suction velocity head using the inside diameters |
| Wrong pitot coefficient | Every flow point shifts left or right by the coefficient error, often 3 to 10 percent, so the curve looks steep or flat | Use the coefficient for the nozzle actually used and confirm the nozzle diameter with a caliper |
| Diesel speed taken from the governor setting | Points are corrected to a speed the engine never ran, usually overstating pressure at 150 percent | Read a tachometer on the shaft at each point; the governor setting is a target, not a measurement |
| Compared with the wrong curve | A pass or fail call against a catalog curve, a different impeller trim or a sister pump’s certified curve | Compare with the certified curve for the serial number at acceptance, and on the annual test with the original field acceptance curve, the manufacturer’s curve or the nameplate curve named in the test record |
Two of these errors pull in opposite directions, which is why an uncorrected curve can look fine. A discharge gauge used as net pressure flatters the pump, and a missing speed correction at 150 percent penalizes it; a reviewer who sees a curve with neither correction cannot tell which effect won. Both corrections should be shown on the test sheet, point by point.
Who Witnesses the Acceptance Test in Each State?
The arithmetic is identical everywhere, because it comes from NFPA 20 and pump hydraulics rather than from a state code. What changes is who witnesses the acceptance test and who receives the curve. The table lists the usual reviewer in each jurisdiction DGA serves; local practice varies, and the installing contractor’s permit should name the witness.
| Jurisdiction | Adopted code basis for the pump | Who witnesses the acceptance test | Filing or record |
|---|---|---|---|
| New York City | 2022 NYC Building Code with NFPA 20 (2016) as modified by Appendix Q; 2022 NYC Fire Code | A Department of Buildings (DOB) inspector or a certified professional, at the permit holder’s option, with the pump manufacturer’s authorized representative on site | Fire pump test result recorded in DOB NOW |
| New York State (outside NYC) | 2025 Uniform Code (2024 I-Codes), NFPA 20 as referenced | Local code enforcement official or the fire official the municipality designates | Acceptance test report with the permit closeout |
| New Jersey | Uniform Construction Code, 2024 I-Codes, NFPA 20 as referenced | Fire protection subcode official of the municipality | Test report with the fire protection subcode final inspection |
| Pennsylvania | Uniform Construction Code, 2021 I-Codes, NFPA 20 as referenced | Municipal or third-party code official; Philadelphia Department of Licenses and Inspections (L&I) in the city | Test report with the final fire protection inspection |
| Maryland | 2021 International Building Code (IBC) through the county building code; NFPA 1 (2024) under the State Fire Prevention Code | County fire marshal or the Office of the State Fire Marshal, with the county building inspector | Test report with the fire marshal’s final |
| Delaware | 1 Del. Admin. Code 701 (2026) with the NFPA 20 and NFPA 25 editions it adopts | Office of the State Fire Marshal (Wilmington Fire Marshal inside the city) | Test report with State Fire Marshal plan review closeout |
Delaware adopts its NFPA editions directly by regulation, so the edition there can differ from the neighboring states. Elsewhere the NFPA 20 edition is the one referenced by the adopted building code, and the NFPA 25 edition is the one referenced by the adopted fire code, so the test record should name the edition the witness is applying.
When Does the Simple Correction Stop Working?
The affinity correction assumes a constant-speed centrifugal pump drawing from a flooded suction, with the relief valve closed and the discharge gauge reading the pump alone. Four common situations break one of those assumptions, and each needs a licensed Professional Engineer (PE) to interpret the result rather than a spreadsheet.
Variable speed pressure limiting control. A pump with variable speed control deliberately runs below rated speed at low flow to hold the discharge pressure down, so the churn and low-flow points are not measurements of the pump at rated speed. NFPA 20 Chapter 14 expects the test to also demonstrate performance at rated speed with the control bypassed, and the two curves have to be read separately.
Vertical turbine pumps with lift. A vertical turbine pump drawing from a wet pit or tank below the discharge head has no suction gauge; the suction side is the water level, which drops as the test runs. Net pressure is then the discharge gauge reading plus the vertical distance from the gauge down to the pumping water level (2.31 ft per psi) plus the discharge velocity head, so the water level at each point has to be recorded alongside the tachometer reading.
Relief valve open during the test. If the main relief valve on a diesel pump lifts at churn or at rated flow, part of the discharge returns to suction or to drain without passing the flowmeter or the hose streams. The measured flow is then low by the relief valve flow, and the point cannot be corrected without knowing that flow; the test is usually repeated with the valve set properly.
Suction pressure below 0 psi at 150 percent. A suction gauge that drops into vacuum at the 150 percent point is telling the reviewer that the supply, not the pump, limited the test. The affinity correction still applies to the pump, but the supply has to be evaluated against the demand on its own; DGA’s article on how engineers decide whether a fire pump is needed explains how the supply curve enters that decision. The pump room arrangement that produces a sound suction, including the pipe sizes the velocity head term depends on, belongs to the article on what NFPA 20 requires of a fire pump room.
Key Takeaway: Read a fire pump test curve in three moves: net pressure (discharge minus suction, with velocity head and elevation), speed correction (flow by the rpm ratio, pressure by the ratio squared) and comparison (certified curve, 140 percent churn ceiling, 65 percent floor at 150 percent flow). In the worked example the raw churn of 142 psi corrected to 138.8 psi and the raw 150 percent point of 66 psi corrected to 67.9 psi at 1,542 gpm, turning two marginal readings into a clear pass under NFPA 20 Chapter 14.
Frequently Asked Questions
Does every fire pump test point have to be corrected for speed?
Yes. NFPA 20 Chapter 14 judges the churn, 100 percent and 150 percent points at the rated speed on the nameplate, and a motor or engine rarely holds that speed exactly under load. Flow is multiplied by the ratio of rated speed to measured speed, and net pressure by that ratio squared.
Can a pump pass if the raw churn reading is above 140 percent of rated pressure?
Conditional: the raw reading only matters once it is corrected. In the worked example a raw churn of 142 psi on a 100 psi pump corrected to 138.8 psi at 1,770 rpm and passed; a raw churn that is still above 140 psi after correction fails the NFPA 20 Chapter 6 envelope.
Is the annual NFPA 25 flow test compared with the same limits as the acceptance test?
No. NFPA 25 (2023) Chapter 8, Section 8.3.3, requires an annual flow test, and Chapter 8 treats the result as acceptable when the pump still meets the system demand and the net pressure at each flow point is within the NFPA 25 margin, commonly read as at least 95 percent of the original unadjusted field acceptance curve, the original manufacturer’s curve or the nameplate curve. The 140 percent and 65 percent limits from NFPA 20 Chapter 6 are the acceptance envelope, and a pump can sit inside the envelope and still show degradation against its own original curve.
Does the speed correction apply to a diesel engine running at its governed speed?
Yes. The governed speed is a setting, not a measurement, and the engine drifts under load, so a tachometer reading is taken at each point and the correction is applied in the same way as for an electric motor. A diesel test report without tachometer readings cannot be corrected and is usually sent back.
More in Water Supply
- Do You Need a Fire Pump? How Engineers Decide (NFPA 20)
- Water Supply for Fire Protection: The Complete Guide
- What Is a Fire Pump Controller and Transfer Switch?
- When Can a Well or Private Source Supply Fire Protection?
- What Is Backflow Preventer Pressure Loss in Sprinkler Calcs?
- Split-Case vs In-Line vs Vertical Turbine: Which Fire Pump?
This article is general information, not a substitute for a project-specific engineering evaluation. Code editions, local amendments and reviewer practice change; confirm the adopted code and the authority having jurisdiction for each project.
Need a Fire Pump Test Curve Reviewed?
Send the test sheet, the certified curve and the nameplate data. DGA corrects the points to rated speed, plots them against the envelope and the original curve, and reports whether the pump passes and what to retest. Project briefs get confirmation and preliminary code feedback within one business day.