The naive expectation
If one pump delivers 500 gpm at 100 ft TDH, two identical pumps in parallel should deliver 1,000 gpm at 100 ft TDH. That's the manufacturer's catalog math, and it's almost always wrong in practice.
The catalog assumes the system curve is also at 100 ft of head when flow doubles. It isn't. As soon as the second pump turns on, flow rises — and so does friction loss, which rises with the square of flow. The system curve climbs steeply; the combined pump curve climbs less steeply. They cross somewhere between the single-pump operating point and the would-be doubled flow.
A useful working number: in a typical municipal force main with 50-70% friction at design flow, two identical parallel pumps deliver 1.5 to 1.8× the single-pump flow, not 2.0×.
How to find the actual operating point
Step 1 — plot the single-pump curve.
Step 2 — plot the system curve. Static head at Q=0, parabolic rise with Q² for friction.
Step 3 — construct the parallel-pump curve. For each head value H, the parallel flow is 2 × Q_single(H). Geometrically: take the single-pump curve and stretch it horizontally by a factor of 2 about the H axis.
Step 4 — find where the parallel curve crosses the system curve. That's your two-pump operating point.
The single-pump operating point lies further down its own curve — at a higher head, because each pump now has to deliver against the friction created by the *combined* flow. You can read off:
- Q_combined (flows through the force main)
- Qeach = Qcombined / 2 (flow through each pump)
- H_combined (head against which each pump operates)
Worked example
Two parallel pumps. Each pump curve, approximately linear over the duty band:
H_pump(Q) = 130 − 0.020 · Q (Q in gpm, H in ft)System: 50 ft static + friction modeled as H_f = 7.5e−5 · Q².
Single-pump operating point (one pump on):
130 − 0.020 Q = 50 + 7.5e−5 Q²
80 − 0.020 Q = 7.5e−5 Q²Solving: Q ≈ 920 gpm, H ≈ 113.6 ft.
Two-pump combined curve:
H_parallel(Q) = 130 − 0.010 Q (slope halves)Two-pump operating point:
130 − 0.010 Q = 50 + 7.5e−5 Q²Solving: Q ≈ 1,400 gpm, H ≈ 116 ft. Per pump: 700 gpm at 116 ft.
Ratio: 1,400 / 920 = 1.52× flow, not 2×.
Each pump has moved left and up its curve — from 920/113.6 to 700/116. Two practical consequences:
1. Each pump is now further from BEP. Inspect the efficiency curve at the new operating point. Many parallel installations spend their lives at 60-70% pump efficiency because the design point was computed for single-pump operation. 2. NPSHr may have changed. NPSHr usually drops as you move left on the curve, so this is the rare case where moving the operating point left helps — but check the impeller-eye flow at the reduced per-pump flow against the manufacturer's minimum.
When parallel works well
- Friction-light systems. A booster pump that lifts mostly static head doubles cleanly to almost 2× flow in parallel because the system curve is flat.
- Variable-demand systems. Where the second (or third) pump is brought on only during peaks. Each pump runs near BEP at single-pump duty; the parallel duty is the brief exception.
- Redundancy installations. Two pumps each rated 100% of duty so either can fail and the other still meets demand. Hydraulically these never run in parallel — one always idles.
When parallel works badly
- Steep system curves. Friction-dominant systems with no flat region in the system curve. Adding a second pump barely raises flow because the friction-square term resists hard.
- Dissimilar pumps in parallel. If pump A's shutoff head is higher than pump B's max operating head, B does no work and A backflows through B. Common failure mode in lift stations where a smaller "jockey" pump is paired with a larger main.
- Series-effective behavior. When two pumps with very different curves are paralleled, the steeper-curved pump dominates and the flatter one acts as a check-valved bypass. Flow gain may be 1.1× rather than 1.5×.
Designing for parallel
A few rules that catch most field problems:
1. Match pump curves. Use identical models with identical impeller trims. If you must mix, verify the shutoff heads are within ~5% of each other. 2. Check valves on each discharge. Without them, the idle pump backflows when only one is running, and the impeller spins backward — bearings and seals don't like that. 3. Design for parallel BEP, not single-pump BEP. A common error is sizing each pump so its BEP matches single-pump duty. At parallel duty the per-pump flow drops, and you may be running at 60% of BEP — inside the AOR but not great for life. 4. Plan the staging logic. When the system needs more flow, the controller turns on pump #2. But if it stages with no hysteresis, you get cycling. Hysteresis bands typically 10-20% of the staging flow. 5. VFDs work nicely on parallel. Run one pump at variable speed, the other at fixed speed (when needed). The VFD handles the trim; the fixed pump handles the bulk. Cuts capital cost vs. two VFDs.
How the calculator handles it
Plot the system curve in the Headloss Calculator, then check "Multiple pumps in parallel" in the pump panel. The calculator constructs the combined curve by horizontal-stretching the single-pump curve about the H-axis and re-solves for the intersection with the system curve. You get:
- Combined operating point (Q_total, H)
- Per-pump operating point (Qperpump, H, ηperpump)
- BEP distance check (% of BEP each pump runs at in the parallel configuration)
That last number is the one that catches the most design errors — it's how you find out you're picking a pump that'll run at 55% of BEP whenever pump #2 stages on.
References
- Hydraulic Institute. *ANSI/HI 1.1-1.2 — Rotodynamic (Centrifugal) Pumps for Nomenclature, Definitions, Application and Operation.*
- Karassik, I. J., et al. *Pump Handbook,* 4th ed. McGraw-Hill — chapter on multiple-pump operation.
- Lobanoff, V. S., and Ross, R. R. *Centrifugal Pumps: Design and Application,* 2nd ed.