How to Verify a Pump Against Its System Curve

Back to Blog

Take Field Points to Build the Curve

The system curve is the head the piping demands as a function of flow, and you can recover its real shape from the plant itself rather than a design document. Take at least two clean operating points at different flows, changing the flow by throttling a discharge valve, and at each point record the flow from the meter and the pump head from the discharge and suction pressures. Two points define the two unknowns in the system curve; a third point is worth taking as a check that the shape is really a static-plus-square-law resistance and not something anomalous.

Get the head right at each point. Pump head is the discharge pressure minus the suction pressure, both converted to head of the actual fluid, so read both gauges or transmitters simultaneously and convert with the fluid's specific gravity. Sloppy head numbers propagate into a wrong system curve, so verify the pressure instruments are trustworthy first, using the routine in the note on how to field-check a pressure gauge if the gauges have never been proven.

Understand what you are building. The system curve is the rising line whose fixed part is the static head and whose growing part is friction, described in the note on what a pump system curve is. By taking points at two flows you are measuring that line directly in the installed plant, capturing whatever real fouling and fitting losses exist, which is exactly the information a design curve cannot give you once the system has been in service.

Fit the Static Head and Resistance Coefficient

With two points (Q1, H1) and (Q2, H2), solve for the two unknowns in H = H_static + k*Q^2. Subtract the two equations to eliminate the static term: H2 - H1 = k*(Q2^2 - Q1^2), so k = (H2 - H1) / (Q2^2 - Q1^2). Then back-substitute into either equation to get H_static = H1 - k*Q1^2. You now have the fitted system curve for the plant as it actually is, static head and all, recovered from field data rather than assumed.

Sanity-check the fitted numbers against physical reality. The static head you recover should roughly match the known elevation difference plus any vessel back-pressure; if the fitted static head is far from what the plant geometry says, either a reading was wrong or something you assumed was static is actually changing. The resistance coefficient k should be positive and, if it is far higher than design, that extra resistance is the signature of a fouled line, a partly closed valve, or a plugged strainer raising friction across the board.

A third point turns the fit into a verification of the model itself. Predict the head at the third flow from your fitted curve and compare it to the measured head; a close match confirms the system really does follow the static-plus-square-law shape and your two-point fit is trustworthy. A poor match warns that something non-standard is in the line, a partly stuck check valve whose resistance jumps, or entrained gas, and that the simple curve does not fully capture the system.

Overlay the Pump Curve and Confirm the Duty Point

Now overlay the fitted system curve on the manufacturer pump curve and find where they cross. That intersection is where this pump must run in this system, and its predicted flow and head are what you verify against normal operation. Return the throttle valve to its normal position, read the flow, head, and motor amps at the real operating condition, and confirm they land on the intersection. When measured flow, measured head, and the power implied by the pump curve all agree at the crossing, the pump and system are verified together and behaving as the physics says they should.

Disagreement localizes the fault to one side or the other. If the field point sits on the fitted system curve but below the published pump curve, the pump has degraded and its real curve has dropped, pointing at impeller or wear-ring wear per the note on what a pump wear ring is. If the field point sits on the pump curve but the system curve came out far steeper than design, the system resistance is the problem, and you go hunting for the fouling or throttling that raised k. Separating the two is the whole value of building both curves.

This verification also sets a baseline you can trend against. Once you know the true duty point, continuous monitoring of flow, discharge pressure, and motor load on a platform such as Merobix lets you watch the operating point walk over time: a drift left with rising head means growing system resistance, while a drop in head at the same flow means the pump is wearing. Because you built the curves from real data, later deviations are measured against ground truth rather than an optimistic design sheet, which makes the trend genuinely diagnostic.

Common Mistakes

The most common mistake is taking both operating points too close together in flow, which makes the fit numerically unstable: a small error in either head reading swings the recovered k and static head wildly. Spread the two points well apart in flow, and take a third, so the square-law shape is well constrained and the fit is robust against normal reading scatter.

The second mistake is forgetting that the pump curve you overlay may no longer be true. The published pump curve is for a new machine, and a worn pump runs below it, so a field point that misses the intersection is not necessarily a bad measurement; it may be an honest sign the pump has degraded. Interpreting the gap correctly, wear on the pump side versus fouling on the system side, is the point of doing both curves rather than trusting either alone.

Frequently Asked Questions