How to Tell Whether a Pump Is Running Off Its Curve

Why this matters

"The pump is weak" is the most common wrong diagnosis in a mechanical room, and it costs a pump replacement that changes nothing. A pump either matches its published curve or it does not, and you can settle that question in about twenty minutes with two gauges, a clamp meter and a tachometer, without a flow meter anywhere in the building. What makes it work is that the three measurements disagree in a readable way: head and power move in opposite directions when the system changed, and in the same direction when the pump itself degraded. One reading alone cannot separate those two cases. Three readings can, and this article is the procedure for taking them and reading the disagreement.

The curve itself, and how a pump curve and a system curve set an operating point, are covered elsewhere in this library. This is the field method for locating yourself on those curves when nobody installed a flow meter.

Before you put a gauge or a meter on a running machine

The coupling guard stays on, and you stand outside the plane of the coupling. You are working beside a rotating shaft with a hose or a lead in your hand. No loose sleeves, no lanyards, no gloves that can be drawn in; route a gauge hose so its free end cannot reach the coupling.

Before opening any gauge port, isolate that port and confirm the isolated side reads zero on a gauge. On a heating system, shut the heat source and let the fluid drop below 120 F first, because water held above 212 F flashes to steam the instant the port opens; if you cannot wait for it to cool, discharge through a hose to a floor drain while standing to the side with a face shield and heat-rated gloves.

Clamping a meter on motor leads means working inside an energized enclosure, which is 29 CFR 1910.333(b)(2) in general industry and 29 CFR 1926.417 for the construction counterpart, with shock and arc-flash boundaries and PPE determined before the cover comes off. Prove the meter on a known live source before and after the reading per NFPA 70E-2021, 120.5, in the edition your employer or your customer's facility has adopted. If there is a permanently mounted meter or a drive display reading kW, use it and leave the cover on.

Apply reflective tape for an optical tachometer with the machine stopped and locked out at the disconnect under 29 CFR 1910.147, not by reaching at a turning shaft. Then restart and read the tape from outside the guard.

Step 1: establish the speed before anything else

Every published curve is drawn at a stated speed. If the machine is not at that speed, nothing else you measure means anything, and this is the leg most often skipped because the nameplate is right there and reads like a measurement. It is not. A 1750 rpm nameplate is the speed at rated load; a lightly loaded induction motor runs faster than nameplate because slip falls as load falls, so 1770 rpm is normal and is not evidence of anything wrong.

If the pump is on a variable speed drive, the published curve does not apply at all until you correct it, and the correction is not optional. A sibling article covers what a drive changes about the whole pumped system, including why the correction is bounded rather than exact.

Read the actual shaft speed optically, and write it down next to the curve's stated speed. Those two numbers travel together through everything below.

Step 2: measure differential head, then apply three corrections

Take a suction gauge reading and a discharge gauge reading on the same instrument if you can, moving one gauge between two ports rather than trusting two gauges to agree with each other. Then convert, and there are three corrections that most techs skip. Each one has a sign and each one gets written on its own line.

• Specific gravity. The 2.31 conversion from psi to feet is feet of cold water. Hot water, glycol or any other fluid needs the differential divided by its specific gravity to get feet of that fluid, which is what the curve is drawn in.
• Elevation between the two gauge taps. Total dynamic head is measured between the two tap elevations, so add the height of the discharge tap above the suction tap. On a close-coupled pump this is inches and disappears; on a vertical or split-case arrangement it is feet and does not.
• Velocity head difference. Discharge piping is usually smaller than suction piping, so the discharge carries more velocity head. Include it when the suction and discharge are two pipe sizes apart or when the difference exceeds about one percent of the total head; below that it is noise.

A suction reading in vacuum is negative and is subtracted algebraically, which means the differential gets larger, not smaller. Reading a 4 in. Hg vacuum as if it were zero understates the pump's head by about 4.5 ft.

Step 3: get a power leg, and know what kind of number it is

Power is the leg that breaks the tie, and it is also the leg most likely to be taken badly.

If you have a true-power reading, in kW at the drive display or a power meter on the leads, use it. Motor current is not proportional to shaft power at part load, because the magnetizing component of the current is roughly constant regardless of load. A motor pulling 70 percent of nameplate current can easily be at 55 to 60 percent of rated shaft power. So a clamp meter gives you an ordinal reading, not a power measurement: it tells you reliably whether power went up or down against a recorded baseline on the same machine at the same voltage, and it does not tell you the horsepower.

That is still enough, because what you need from this leg is a direction. Record the voltage with the current, because current rises when voltage sags for the same shaft load, and a reading taken at a different supply voltage is not comparable.

Step 4: land the point and read which pair disagrees

Now put the corrected head on the published curve at the measured speed and read across to a flow. That flow is only true if the pump still matches its curve, and the power leg is what tests that assumption. Five combinations cover almost everything you will meet.

Reversed rotation deserves its own line because it hides so well. A three-phase centrifugal running backwards still pumps, usually at substantially reduced head and flow rather than at zero, which is exactly why nobody notices after a motor swap or a panel change. Confirm rotation against the arrow on the casing, with the guard in place, before you conclude the impeller is worn.

Worked example: a branch running short after a piping change

An end-suction pump on a hydronic loop. Published curve at 1750 rpm with a full-diameter impeller: shutoff 104 ft, 100 gpm at 100 ft, 150 gpm at 96 ft, 200 gpm at 90 ft, 250 gpm at 82 ft. Published shaft power on the same curve sheet: 4.1 bhp at 100 gpm, 5.0 bhp at 150 gpm, 5.8 bhp at 200 gpm. Design point is 200 gpm at 90 ft. Motor is 7.5 hp, 460 V three-phase. Commissioning record shows 8.4 A at 462 V.

Speed. Optical tach reads 1768 rpm against the curve's 1750 rpm. That is 1.0 percent high, which shifts head by roughly two percent under the affinity relationship, so about 2 ft at this part of the curve. Small enough to note and carry, not large enough to change the conclusion. Recorded, not ignored.

Head, with all three corrections printed. Fluid is water at 140 F, specific gravity 0.985. Discharge gauge 47 psi, suction gauge 6 psi.

• Raw differential: 47 - 6 = 41 psi
• Converted: 41 x 2.31 = 94.7 ft of cold water
• Specific gravity correction: 94.7 / 0.985 = 96.2 ft of this fluid
• Elevation correction: discharge tap sits 1.5 ft above the suction tap, so +1.5 = 97.7 ft
• Velocity head correction: 4 in. suction, 3 in. discharge, at the implied flow this is +0.4 ft = 98.1 ft total dynamic head

Where that lands. Between the curve's 150 gpm at 96 ft and 100 gpm at 100 ft, 98.1 ft interpolates to roughly 125 gpm. Design was 200 gpm. If the pump is on its curve, flow has fallen by about 38 percent.

Power. Clamp reads 6.6 A at 461 V against the commissioning 8.4 A at 462 V. Same machine, same meter, voltage within a volt, so the comparison is legitimate as a direction. Current fell by about 21 percent. Because this is a current reading and not a power measurement, the honest statement is "shaft power fell", not "shaft power fell 21 percent". The curve says power at 125 gpm should be roughly 4.5 bhp against 5.8 bhp at design, a fall of about 22 percent in shaft power, which is consistent with a current fall of that order for a motor in this load band.

The read. Head is above design and power is below baseline. That is row one: a restriction added downstream. It is not a worn pump, because a worn pump makes less head, not more, and the head reading would have come in below 90 ft rather than above it. It is not runout for the same reason in reverse. The suction gauge sat steady at 6 psi throughout, which takes the suction row off the table.

What was found. A triple-duty valve at the pump discharge, closed to about a third of travel by a contractor balancing a new branch two months earlier. Opening it restored 88 ft at 8.3 A, which lands at roughly 205 gpm on the curve at a power draw slightly above the commissioning figure, exactly where a design-flow point should sit.

Failure mode of getting this wrong. The shop that reads only the discharge gauge sees 47 psi, calls it strong, and blames the branch. The shop that reads only flow-by-symptom replaces the pump with a larger one, which rides its own steeper system curve to a slightly higher flow at much higher power and considerably more valve wear, and the complaint half-returns in a season. The restriction is still there in both cases.

How to verify you got this right

Change one thing and predict the result before you make the change. That is the only real proof that you located the operating point rather than guessed it.

In the example above, the prediction was "open that valve and head falls toward 88 to 90 ft while current rises toward the commissioning 8.4 A." Both moved as predicted and in the right proportion. If head had fallen and current had not risen, the operating point was not where the reading implied and the pump's own curve is suspect.

Two more checks worth the minute they cost. Read shutoff head by closing the discharge valve for no more than a few seconds with the pump running and the suction valve fully open, and only on a centrifugal, never on a positive displacement pump, and never on a hot system where the trapped water can flash: a pump making less than roughly 90 percent of its published shutoff head has genuinely lost something internally. And re-read the suction gauge with the strainer basket pulled and cleaned, because a loading strainer produces the same falling-flow complaint from the other end of the machine.

References

• Hydraulic Institute standards for centrifugal pump field testing, in the edition adopted by your customer's specification or your service contract, which is what defines an acceptable field-test tolerance
• 29 CFR 1910.147 for lockout of the energy isolating device before applying tachometer tape, and 29 CFR 1910.333(b)(2) with 29 CFR 1926.417 as the construction counterpart for taking readings inside an energized enclosure
• NFPA 70E-2021, 120.5, in the edition adopted by your employer, for the live-dead-live instrument proving sequence
• See related: Reading Pump Curves Reference; What a Pump Curve and a System Curve Do Together; Why a Bigger Pump Does Not Fix a Restriction; What a Variable Speed Drive Changes About a Pumped System