How Do You Build a Pump Curve From 1105 GPM at 130 Ft?

A single stamped duty point — 1105 gpm at 130 ft — plus 50 hp at 1775 rpm is enough to bound a curve, not enough to draw one. Use it to check the hydraulics, calculate specific speed, and narrow the pump to a family; then go get the real curve from the manufacturer or the building's submittal set. Do not run a piping redesign off a sketched parabola. And before you touch the model, reconcile that 40 psig differential across the pump with the 130 ft stamp, because those two numbers do not agree and the disagreement is the whole story.

Read the Nameplate Data Before You Read the Model

Start here. You have four hard numbers and one soft one. The hard ones are the stamped duty (1105 gpm, 130 ft), the motor rating (50 hp), the motor speed (1775 rpm), and the mechanical room pipe size (8 in). The soft one is 40 psig across the pump, which converts to 92.4 ft of head on water at SG ~1.0:

Head (ft) = psi x 2.31 / SG
40 psi x 2.31 / 1.0 = 92.4 ft

The casing says the pump was selected for 130 ft. The gauges say it is making 92 ft. That is a 38 ft gap and it has a short list of causes.

Check Hydraulic Power Against the 50 HP Motor

Run the power balance before anything else. It tells you whether the stamped point is even real.

WHP = Q x H x SG / 3960
WHP = 1105 x 130 / 3960 = 36.3 hp
BHP = WHP / efficiency

The 50 hp nameplate requires at least 72.6% efficiency at the duty point, or 63% if you are willing to lean on a 1.15 service factor. A single-stage centrifugal moving 1105 gpm at 1775 rpm should land in the 80-85% band, so 50 hp is a legitimate selection — not evidence of a mismatch. If you assume 70% efficiency out of habit you will conclude the machine needs 55 bhp and a 75 hp frame, and you will start chasing a problem that does not exist. That is not the fault. Read the motor's actual service factor and FLA off the nameplate instead of assuming.

Use Specific Speed to Bound the Curve Shape

Specific speed fixes the impeller geometry and therefore the shape of the head-capacity line you are trying to reproduce:

Ns = N x sqrt(Q) / H^0.75
Ns = 1775 x sqrt(1105) / 130^0.75
Ns = 1775 x 33.24 / 38.50 = 1530   (single suction)

If the pump is double suction — common at this flow in a chilled water plant — split the flow per eye and recompute: sqrt(552.5) = 23.5, giving Ns ~1085. Either value lands solidly in radial-flow territory. Practical consequences:

• Shutoff head runs roughly 110-125% of the duty head, so 143-163 ft.
• The curve is comparatively flat near BEP, so small head errors translate into large flow errors. Hold that thought for the 40 psi reading.
• Brake power rises continuously with flow. A radial pump pushed right of BEP overloads the motor before it runs out of head.

Cross-check the duty against the plant load while you are here. At 1105 gpm and a 10 F delta-T, that pump serves about 460 tons (gpm x delta-T / 24); at 12 F, about 552 tons. If the chiller nameplates do not add up to that, you are looking at a distribution pump, not a chiller-dedicated primary pump, and 130 ft of head makes sense.

Find the Real Pump, in This Order

1. Cut a window in the insulation over the pump nameplate. Six inches square is enough. You do not need a camera — transcribe every line by hand: manufacturer, size designation, casing or pattern number, serial number, impeller diameter, and the stamped duty. The casing number alone is usually enough for a rep to pull the selection.
2. Record the full motor nameplate. HP, FLA, service factor, frame, voltage, RPM. FLA is what makes the amp test in the verification step meaningful.
3. Pull the mechanical submittals. The building has an O&M set, a shop drawing package, or a permit file with the pump schedule and the certified curve on it. The design engineer of record keeps job files. This is the fastest path and it is almost always skipped.
4. Ask maintenance where they buy parts. Whoever supplies the mechanical seal, the wear rings, or the coupling knows the pump model. Check the CMMS spare parts list and past purchase orders.
5. Call the local manufacturer's rep. Give them the casing number, suction and discharge flange sizes, bolt count, base plate dimensions, and the stamped duty. Reps identify castings by sight routinely and can supply the factory curve at the correct impeller trim.
6. Substitute an equivalent published curve. If identification fails, take a curve from a major manufacturer's chilled water family that hits 1105 gpm at 130 ft at 1750 rpm with the same suction arrangement and Ns. Correct it to 1775 rpm with the affinity laws and label it as a substitution in your report.

Construct a Placeholder Curve Only as a Last Resort

If you must model something today, fit a parabola through the known point:

H(Q) = H0 - k*Q^2
H0 = 1.25 x 130 = 162.5 ft (shutoff assumption)
k  = (162.5 - 130) / 1105^2 = 2.66e-5 ft/gpm^2

Correct for speed with the affinity laws where needed: Q scales with N, H with N squared, P with N cubed.

Now see how fragile that is. Solve the placeholder curve for the 92.4 ft you measured and it predicts about 1620 gpm. Assume a flatter pump — shutoff at 1.15 x 130 = 149.5 ft — and the same 92.4 ft predicts about 1890 gpm. A 10-point swing in your shutoff assumption moves the answer 270 gpm. That is why you do not size pipe off a sketched curve, and it is why a real shutoff measurement is worth more than any assumption.

Verify the Operating Point With Three Field Measurements

1. Read the drive first. If there is a VFD, get the output frequency. Head falls with the square of speed: 92.4/130 = 0.711, and the square root of that is 0.843, which is about 50.6 Hz on a 60 Hz machine. A distribution pump loafing at 50 Hz explains the entire 38 ft discrepancy with no hydraulic problem at all. Confirm the drive speed before you do anything else.
2. Clamp the motor leads. Average the three phases and compare to nameplate FLA. Predict first, then measure. If the pump really is at ~1620 gpm and 92 ft, brake power is about 1620 x 92.4 / 3960 / 0.70 = 54 bhp — at or above a 50 hp nameplate, inside a 1.15 service factor. So: amps at or above FLA confirms runout. Amps at 60-70% of FLA rules runout out and points at reduced speed, worn internals, or parallel operation sharing the header.
3. Confirm with a transit-time flow meter. Strip insulation on a straight run of the 8 in header with at least 10 pipe diameters upstream and 5 downstream of the transducers. This is the only measurement that settles the flow argument outright.

Take the differential with one calibrated gauge and a three-valve manifold, valved to suction then discharge. Two separate gauges at two elevations with two calibration histories will manufacture 5-10 ft of error on their own.

Pitfalls That Send Chilled Water Analyses Off the Rails

• Static head in a closed loop. Chilled water distribution is a closed circuit. The down-leg returns whatever the up-leg gains, so the pump sees friction only and the system curve passes through the origin. If the piping program was given building height as static head, it will report a head requirement that no installed pump could meet and hand you a false "redesign the piping" conclusion. Zero that term and rerun.
• Parallel pumps. With multiple chillers, pumps commonly share a header. Parallel machines add flow at equal head, so each unit delivers far less than its solo curve suggests. Modeling one pump against the whole building load will always over-predict required head.
• Primary/secondary confusion. Chiller-dedicated primary pumps typically develop modest head; 130 ft points to a secondary or distribution duty. Assigning a distribution pump the evaporator-only circuit — or the reverse — wrecks the balance.
• Modeled valve positions. Balancing valves set at 40% in the field and modeled wide open will make the real system look like it has far less resistance than the analysis predicts, or vice versa. Get actual handle positions and coil control valve authority before trusting the pressure requirement.
• The bogus part number. Numbers pulled by field personnel come off the motor, the seal, or the coupling as often as the pump. Verify any number against the manufacturer's nomenclature before you build an analysis on it.
• Assuming the building is broken. If the loop holds design supply temperature at the far coils and the pump is drawing sensible amps, the hydraulics work. The model is the suspect.

Stop reconstructing the moment you have the casing or pattern number and a rep who will answer the phone — the factory curve at the as-built impeller trim ends the exercise, and no synthesized parabola comes close to it. If the pump cannot be identified after the insulation is opened and the submittal set has been searched, hand the manufacturer's application group your measured shutoff head, measured duty point, motor FLA and amps, and flange dimensions and let them do the selection. Do not issue a piping redesign on an assumed curve; a wrong curve produces a right-looking answer and an expensive rebuild.

Frequently Asked Questions