Troubleshooting Pump Overload Against Published Curve Data

A pump model that ran inside motor service factor for years and now overloads by about 10% has either changed physically or had its operating point moved. Field cases on record include a casting-process change with no change to impeller shape that flattened the head curve, cut shut-off head by 18%, and raised power demand by 40% at one performance point. Another case: a 30 hp motor delivering about 38 hp output (about 127% of rating) on a pump that had only a non-witnessed performance test. Work the checks below in order. Each one tells you which check to run next.

Leave the overload setting alone until the cause is known

The quick fixes all hide the problem and fail in predictable ways:

• Raising the overload relay or drive current limit. It stops the trips and pushes the motor into the service factor or beyond at the far end of the curve. Winding life drops and nobody has fixed anything.
• Swapping in a bigger motor. It moves the failure to the coupling, shaft, and starter, and the pump still does not match its published curve.
• Trimming the impeller by feel. Trim tables assume the published curve is right. If the curve is wrong, the trim is wrong too.
• Trusting the datasheet curve over your own readings. A curve printed on a datasheet describes the pump that was tested, not necessarily the one on your skid.

Take measurements first. If the motor is tripping now, use the temporary restore in the section on holding production and come back to the checks.

Check 1: Rule out site conditions with four readings

Take these readings at the overloaded operating point before you suspect the pump. Pump power scales with fluid specific gravity and, by the affinity laws, roughly with the cube of speed, so a small error in either one produces a large power error.

1. Shaft speed. Read it with a tachometer, or the drive output frequency if a VFD is fitted. Compare to nameplate speed.
2. Supply voltage and motor current. Log all phases. A voltage imbalance or low voltage raises current without raising shaft load.
3. Fluid. Confirm specific gravity, viscosity, and solids or slurry content against the design basis.
4. Operating point. Read discharge and suction pressure and flow. If the system head dropped (a valve opened, a pipe section changed, a zone added), the pump slid to the right on its curve where power is higher. Convert pressure to head: H (ft) = 2.31 x psi / SG.

Convert motor readings to shaft power. If the motor is three-phase (confirm on the nameplate): input kW = sqrt(3) x V_LL x I_line x PF / 1000. Then shaft bhp = input kW x motor efficiency / 0.746. Use the motor efficiency at the actual load from the motor data sheet, not the nameplate full-load figure.

Outcome: If speed, voltage, fluid, or operating point explains the extra power, fix that and stop. If all four match design and the overload persists, go to Check 2.

Check 2: Compare the pump against your own baseline records

Repeat-order pumps give you the best diagnostic available: your own historical flow, head, and power for the same model. Pull the commissioning test or the earliest site readings for the same model, impeller diameter, and speed.

Finding	Meaning	Next
Same flow and head as baseline, power up	Hydraulic efficiency changed or friction increased (impeller finish, clearances, geometry)	Check 3, then Check 4
Same head at shut-off, power up only at high flow	System curve moved or wear-ring clearance changed	Return to Check 1, then inspect clearances
Lower shut-off head and flatter curve than baseline	Impeller or casting geometry differs from the pump you originally tested	Check 3 and Check 4
No baseline exists	You have nothing to compare against	Test to the published curve now (Check 3) and start the record

A flatter head curve is the mechanism behind the high-power reports. With the system curve unchanged, a flatter pump curve intersects it further to the right at higher flow. On a pump whose power curve rises with flow, that means more power than the trim was selected for. Trims chosen so the motor never overloads at any point on the published curve fail as soon as the delivered curve differs from the published one.

Check 3: Read shut-off head and one more point against the curve

Shut-off head is the fastest field test for a changed impeller. It needs only a calibrated pressure gauge and a short run against a closed discharge valve.

1. Confirm the pump is safe to run at shut-off for a short time. Closed-valve operation heats the fluid and loads the bearings and seal. Check the manufacturer's minimum-flow guidance before you start. If in doubt, skip this step and use a low-flow point instead.
2. Record suction and discharge pressure with calibrated gauges and convert to head with the specific gravity from Check 1.
3. Record head at one or two flows you can measure, plus motor power at each.
4. Overlay all points on the published curve and on your baseline.

Head at shut-off 18% below published is not a measurement error. A gauge error of that size is visible against a second calibrated gauge. Impeller trim gives a first approximation: head scales roughly with diameter squared and power with diameter cubed. Use that only to check whether a diameter difference could explain what you see. If the measured impeller diameter matches the published one and head is still low, the geometry differs (vane angle, vane thickness, passage shape, surface finish).

Outcome: Head and power match baseline: return to Check 1, the fault is in the system. Head or power differs from both baseline and published: go to Check 4.

Check 4: Inspect the impeller and casting when the curve has shifted

A manufacturer can change the casting process without changing the pattern drawing and without knowing it changes hydraulic performance. Shape on paper stays the same while the as-cast passages, vane edges, and surface finish do not. Pull the pump and check:

1. Impeller outside diameter after trim, against the trim you ordered.
2. Vane thickness, vane tip shape, and passage cross-section against a known-good impeller from the same model if you kept one.
3. Surface finish inside the passages against the older impeller.
4. Wear-ring and running clearances.
5. Casting marks, pattern numbers, and date codes on the new impeller against the old one.

Photograph and measure both. Keep the old impeller and the new one side by side. Ask the manufacturer in writing whether the casting process, foundry, or material changed for the lot. Their answer tells you whether the change is deliberate. If they confirm a process change, they need to re-test the affected pump, because the published curve no longer describes it.

Outcome: Geometry or process differs: go to the permanent repair section. Everything matches and the pump still misses its curve: go to Check 5.

Check 5: Judge the test that produced the curve or the acceptance data

A test result is only as good as the bench behind it. Three problems recur:

• Uncalibrated or in-house-calibrated instruments. Ask for the date of the last calibration and the third party that performed it. Reject in-house calibration as evidence.
• Cross-contamination on the bench. A bench that runs used and new equipment, or that shares fluid or stored parts with a repair shop, contaminates test fluids and parts and voids the comparison. Ask whether used equipment is tested on the same bench.
• Numbers that are physically implausible. A small mixed-flow impeller of about 4 in. OD with a BEP near 120 GPM has an expected new-pump efficiency near 68%. A test bench that reported 82% for a four-year-old pump from a sour-gas well (over 5% H2S) produced a number that used equipment cannot reach. Check any claimed efficiency with efficiency = Q(gpm) x H(ft) x SG / (3960 x bhp) against what the size of the impeller can plausibly deliver.

A non-witnessed test carries no assurance. The 30 hp motor at about 38 hp output came from a pump with a non-witnessed performance test. Only a witnessed test on the actual pump verifies performance.

Hold production while the cause is open

These are temporary restores. Log each one and remove it when the permanent repair is in.

• Throttle the discharge to move the operating point left on the curve until motor current falls under nameplate. Accept the lower flow.
• Limit VFD speed if a drive is fitted. Reduced speed cuts power at roughly the cube of the speed change. Set the limit from measured power, not from a calculation alone.
• Swap in the old impeller from the same model if you kept one and it matches the baseline data.
• Leave the overload relay at the motor's rated setting. A trip is protecting the motor. Do not defeat it.

Fix the pump line permanently with witnessed testing and spec language

A pump that passed a test once and was then reordered for years with only casual checks has no protection against a supplier's process change. The permanent fix is a check on every batch, not only the first.

1. Re-test every pump model in the line-up against its published curve, not just the one that failed. Field experience here is that a supplier-side process change can affect multiple models.
2. Require a witnessed performance test in the purchase specification for pumps where power draw against the motor matters. State the acceptance tolerance and the test standard your spec names.
3. Require calibration certificates from a third party, with the calibration date, for the test instruments.
4. Require that the test bench does not test used equipment or share fluid and parts with a repair operation.
5. Require notification of any change to casting process, foundry, or material, and re-test after any such change.
6. Record flow, head, and power for each delivered pump against its serial number. Your own baseline is the fastest detector of a change.

Spec language shifts the burden. If the supplier claims performance that a witnessed test disproves, the contract gives you a remedy. A consulting client that lacks such a clause has to chase the supplier on its own for compensation.

Confirm the repaired pump runs inside nameplate at every curve point

1. Run the pump at shut-off (if permitted), at design flow, and at the highest flow the system can reach. Record head, flow, and motor current or kW at each point.
2. Convert to shaft power with the formulas in Check 1 and confirm it stays below motor nameplate at every point. If you sized the trim to stay out of the service factor, confirm that too.
3. Overlay the points on the published curve and your baseline. Shut-off head, curve slope, and power should match within the tolerance in your spec.
4. Log ambient and fluid conditions with the readings so you can distinguish site changes from pump changes on the next test.
5. Remove the temporary restores (throttling, speed limit) and repeat the run at normal operating conditions.

FAQ