Centrifugal Pump Sizing Calculator | TDH, NPSH & Power

What does centrifugal pump sizing actually determine?

A centrifugal pump sizing calculation starts with a required flow rate and estimates the head needed to move that flow through a particular system. It then estimates power and checks suction conditions. A useful preliminary result is a duty statement such as “35 L/s at 63.6 m of head,” accompanied by the fluid properties, operating range and NPSH available.

Head is energy per unit weight of liquid, expressed in metres. It includes elevation, pressure and motion. A pump developing 40 m of head does not necessarily lift liquid vertically by 40 m: some of that energy may maintain vessel pressure or overcome pipe resistance.

The desired duty point and the actual operating point are different ideas. The desired point comes from the process requirement. The actual point occurs where the pump's head–flow curve crosses the system's head–flow curve. A power estimate alone cannot establish whether a particular pump will deliver the required flow.

This lesson connects the seven calculator modes above. It uses steady, incompressible, single-phase flow and circular pipes as its basic model. The examples are teaching calculations, with stated assumptions, rather than equipment selections.

How to use the centrifugal pump sizing calculator

1. Define the endpoints. Mark the source surface or inlet section, the destination surface or outlet section, and one elevation datum. Record what the pressure at each endpoint represents.
2. Set the flow and pressure units before entering numbers. Changing a unit converts the existing numerical inputs and clears the old result; calculate again in the current mode. Read each field label again after changing units. Use actual pipe inside diameter, not nominal pipe size.
3. Enter properties for the liquid's operating condition. Density, viscosity and vapor pressure can change with temperature and composition. The default water-like values are examples.
4. Calculate the separate checks. Flow and TDH, suction conditions, shaft demand and the operating point answer different questions. Completing one does not complete the others.
5. Test the range. Repeat for low source level, high destination pressure, maximum temperature, fouled piping and other credible operating cases. Keep each case's inputs with its results.

Quick conversions: 1 m³/s = 1,000 L/s = 3,600 m³/h; 1 bar = 100 kPa = 100,000 Pa. Enter 72% as 72 in a percentage field, but use 0.72 in a hand calculation. Dynamic viscosity μ is in Pa·s; 1 mPa·s = 0.001 Pa·s. Kinematic viscosity ν is a different quantity: ν = μ/ρ, in m²/s.

Pump sizing formulas and symbols

Use Q in m³/s, diameter D and length L in m, density ρ in kg/m³, dynamic viscosity μ in Pa·s, pressure p in Pa and gravity g in m/s². These examples use g = 9.81 m/s². Efficiencies are fractions in the equations.

Build total dynamic head without losing the pressure reference

Choose endpoints before adding head terms. Between two large reservoirs, the surface velocities are usually negligible. Between two equal-diameter pipe sections carrying the same flow, their velocity heads cancel. A discharge jet or unequal endpoint diameters can leave a significant velocity-head difference.

Enter v₁ and v₂ for the endpoints you chose. Use zero for each large-vessel surface when its velocity is negligible. If the destination is a jet, enter its outlet velocity to account for the remaining kinetic energy. If the destination is a large receiving tank, account for the relevant exit loss instead. Do not charge the same dissipation twice by mixing those two endpoint choices.

Gauge pressure versus absolute pressure

Gauge pressure is relative to local atmospheric pressure; absolute pressure is relative to vacuum. For a common atmospheric reference, p_abs = p_gauge + p_atm. The TDH pressure difference may use two absolute pressures or two gauge pressures referenced consistently. Never subtract one absolute pressure from one gauge pressure.

For example, 250,000 Pa absolute minus 101,325 Pa absolute is 148,675 Pa. The equivalent gauge-pressure pair is 148,675 Pa gauge and 0 Pa gauge when atmospheric pressure is 101,325 Pa. Entering 250,000 Pa gauge instead describes a different destination pressure.

Elevation is the vertical difference z₂ − z₁, not the pipe's total length. A horizontal 100 m pipe can have zero elevation rise and substantial friction. In a closed circulation loop returning to the same pressure and elevation, net static terms cancel around the loop, although friction remains.

The quantity ρgH is a pressure equivalent of total pump head. It equals a measured static pressure rise across the pump only when differences in measurement-section elevation and velocity head are accounted for appropriately.

Pipe friction, fittings and Reynolds-number limits

Calculate velocity first because it appears in both Reynolds number and loss equations. For a given Q, doubling D gives four times the area and one quarter of the velocity. With f held constant, substituting v = 4Q/(πD²) into Darcy–Weisbach gives h_f proportional to LQ²/D⁵. This strong diameter dependence explains why a small pipe can dominate the head requirement.

Which friction factor belongs in the equation?

• Fully developed laminar flow: f = 64/Re for a circular pipe.
• Turbulent flow: the automatic calculation uses the Swamee–Jain approximation shown below, with roughness ε and diameter D in the same units.
• Transition: roughly Re = 2,000–4,000 is a region in which the flow state and calculated loss are less certain. A smooth numerical answer does not remove that uncertainty.

The calculator rejects automatic friction results for 2,300 ≤ Re < 4,000. Pipe Loss mode allows a manually supplied Darcy factor for a justified sensitivity calculation. Even near the lower transition boundary, the real flow conditions deserve review: an automatic formula choice is not a physical validation of the flow state.

Darcy and Fanning factors differ by four: f_Darcy = 4f_Fanning. A Fanning value entered into this calculator's Darcy field underestimates straight-pipe loss by a factor of four.

Count each loss once

Use either a fitting's K value or its equivalent pipe length for that same fitting. Adding both duplicates the loss. The word “minor” refers to localized losses, not necessarily to a small contribution: a throttled valve can exceed the loss of a long pipe.

For several diameters, calculate each section using its own velocity, length and friction factor, then sum the head losses. Use each fitting coefficient's stated reference velocity. One average diameter or an unqualified combined K is generally insufficient for a branched or changing-diameter network.

Hydraulic power, shaft demand and motor input are different

The liquid receives hydraulic power. The pump shaft supplies more power because η_p is below 1. The motor draws still more electrical power because η_m is below 1. In this lesson, “brake power” and “pump shaft demand” mean the same mechanical input to the pump.

A motor's nameplate kW or horsepower rating ordinarily describes its mechanical output. Compare it with the shaft load, including relevant transmission losses and duty requirements. Electrical input is useful for estimating energy use; it is not the motor's mechanical rating. For constant input power, energy in kWh equals electrical kW multiplied by operating hours.

What does service factor mean?

Motor service factor describes a manufacturer's permitted loading capability under specified conditions. It is not another efficiency and should not automatically multiply normal running energy demand. Temperature, cooling, voltage, drive operation and the manufacturer's instructions affect whether that capability is usable.

The calculator reports a separate planning shaft-power allowance: P_shaft × M, where M is the chosen allowance multiplier. This allowance is separate from the motor's nameplate service factor and from normal electrical demand. It remains a planning value; check maximum shaft demand over the complete intended operating range before selecting a motor.

Changing pump efficiency from 72% to 60% at unchanged hydraulic duty increases shaft demand by 0.72/0.60 = 1.20, or 20%. Using a convenient efficiency everywhere on a curve can therefore hide a meaningful motor-load change.

NPSH available and the limits of a margin check

NPSH available describes suction total head above the liquid's vapor-pressure head. It is calculated from the installation. NPSH required is supplied for the pump at a stated flow, speed and test basis. Always use absolute pressure for the source pressure and vapor pressure in the NPSH calculation.

At a suction measurement section on the pump datum, the same quantity can be expressed as NPSH_a = p_s,abs/(ρg) + v_s²/(2g) − p_v,abs/(ρg). This is a different starting point from the source-vessel formula. Do not enter measured suction-flange pressure as source-surface pressure and then subtract upstream suction losses again.

Published NPSH3 identifies a test condition with a 3% head reduction caused by cavitation. It is not the onset of cavitation and does not guarantee cavitation-free operation. Confirm whether a manufacturer's “NPSHr” curve is NPSH3 or includes a different allowance.

Two useful arithmetic checks are the absolute margin NPSH_a − NPSH_r, in m, and the ratio NPSH_a/NPSH_r. No single ratio, including 1.10, is a universal acceptance criterion. The required margin depends on the pump, liquid, service and operating region.

Evaluate the minimum source level, lowest source absolute pressure, highest relevant vapor pressure and realistic suction losses. A clogged strainer or lowered tank level can consume margin even when the pump and discharge piping are unchanged. These checks supplement manufacturer and application-specific review.

Affinity laws predict corresponding points; curves predict operation

For the same pump and impeller, a speed ratio r maps a corresponding flow to rQ and its head to r²H. With approximately unchanged efficiency and density, shaft power scales as r³. These relationships help estimate a changed pump curve.

The system still imposes its own head requirement. Re-intersect the changed pump curve with the system curve to find the new operating point. Simple proportional flow reduction is especially misleading when a substantial static head remains as speed falls. Example 5 demonstrates the difference numerically.

Impeller trimming is a separate approximation. The familiar Q ∝ D, H ∝ D² and P ∝ D³ rules can estimate small trims in a given casing, subject to manufacturer limits. Scaling an entire geometrically similar pump is different: the full similarity relations include Q ∝ ND³ and P ∝ N³D⁵ at fixed density. Do not apply one diameter rule to the other problem.

Read the simplified system-curve inputs correctly

In the quadratic model, H₀ is pump shutoff head. A second pump point supplies a = (H₀ − H_rated)/Q_rated². A system point supplies K_sys = (H_system,rated − H_static)/Q_rated². Use the same reference flow for both points.

Here H_static includes any fixed elevation and fixed pressure-head requirement. A purely quadratic friction term assumes the loss coefficient is approximately constant over the range. Laminar flow, changing valve positions, pressure-dependent demands and network branches can require a different model.

If pump shutoff head is below the system's static requirement, this model has no positive forward-flow intersection. A displayed zero must not be interpreted as a usable duty point. Real operation also requires checking the manufacturer's allowable region, minimum flow and power curve.

Six fully worked centrifugal pump sizing examples

The following values are chosen for calculation practice. Carry unrounded values between steps; the displayed results are rounded. Examples 1 and 2 share a pipeline so the loss calculation can be checked independently.

Example 1: TDH, hydraulic power, shaft power and electrical demand

Given: Q = 0.035 m³/s, ρ = 1,000 kg/m³, z₂ − z₁ = 22 m, p₁ = 101,325 Pa absolute and p₂ = 250,000 Pa absolute. The source and destination are large vessels, so endpoint velocity heads are negligible. Total pipe and local losses from Example 2 are 26.399245 m. Pump efficiency is 72%; motor efficiency is 92%.

1. Pressure difference: Δp = 250,000 − 101,325 = 148,675 Pa.
2. Pressure head: Δp/(ρg) = 148,675/(1,000 × 9.81) = 15.155454 m.
3. Total head: H = 22 + 15.155454 + 26.399245 = 63.554698 m.
4. Hydraulic power: P_hyd = 1,000 × 9.81 × 0.035 × 63.554698 = 21,821.506 W = 21.822 kW.
5. Shaft demand: P_shaft = 21.821506/0.72 = 30.308 kW.
6. Motor electrical input: P_electric = 30.307647/0.92 = 32.943 kW.

Interpretation: the estimated duty is 35 L/s, or 126 m³/h, at 63.555 m head. A planning shaft allowance of M = 1.15 gives 30.307647 × 1.15 = 34.854 kW. This is an allowance on mechanical demand, while normal estimated electrical input remains 32.943 kW. It does not verify a motor rating or its allowable service-factor loading; review the full load envelope.

Calculator check: use the Pump Sizing inputs above together with L = 120 m, D = 0.10 m, μ = 0.001 Pa·s, ε = 0.000045 m and ΣK = 5. Set the flow unit to m³/s and pressure unit to Pa, and enter v₁ = v₂ = 0 for the vessel surfaces.

Example 2: Pipe velocity, friction factor, head loss and pressure loss

Given: Q = 0.035 m³/s, D = 0.10 m, L = 120 m, ρ = 1,000 kg/m³, μ = 0.001 Pa·s, ε = 0.000045 m and ΣK = 5. The K total includes the local losses required by the chosen endpoints; no fitting is also included as equivalent length.

1. Area: A = π(0.10)²/4 = 0.007853982 m².
2. Velocity: v = 0.035/0.007853982 = 4.456338 m/s.
3. Reynolds number: Re = 1,000 × 4.456338 × 0.10/0.001 = 445,634, well above the transitional range.
4. Relative roughness: ε/D = 0.000045/0.10 = 0.00045. Swamee–Jain gives f = 0.25/{log₁₀[0.00045/3.7 + 5.74/(445,633.841)^0.9]}² = 0.017568.
5. Velocity head: v²/(2g) = 4.456338²/19.62 = 1.012179 m.
6. Major loss: h_f = 0.017567997 × (120/0.10) × 1.012179 = 21.338350 m.
7. Local loss: h_m = 5 × 1.012179 = 5.060895 m.
8. Total: h_L = 26.399245 m; Δp_loss = 1,000 × 9.81 × 26.399245 = 258,977 Pa ≈ 258.977 kPa.

Interpretation: the straight-pipe loss is about four times the local loss here. Changing pipe diameter deserves investigation because velocity is high and the friction contribution exceeds the 22 m elevation rise. This numerical comparison does not establish an acceptable velocity for a particular service.

Example 3: NPSH, suction lift and a changing source level

Given: source pressure = 101,325 Pa absolute, vapor pressure = 3,169 Pa absolute, ρ = 1,000 kg/m³, source level = 2 m above the pump datum, suction losses = 1.2 m and manufacturer NPSH_r = 3.0 m at the evaluated duty.

1. Pressure head above vapor pressure: (101,325 − 3,169)/9,810 = 10.005708 m.
2. Available NPSH: 10.005708 + 2 − 1.2 = 10.805708 m.
3. Absolute margin: 10.805708 − 3.0 = 7.805708 m.
4. Margin ratio: 10.805708/3.0 = 3.601903.

Now lower the liquid surface to 3 m below the pump datum and increase suction losses to 3 m, keeping the other teaching inputs unchanged. NPSH_a = 10.005708 − 3 − 3 = 4.005708 m; the margin becomes 1.005708 m and the ratio 1.335236.

Interpretation: a 5 m level reduction plus 1.8 m extra loss removed 6.8 m of available NPSH. Both cases have positive arithmetic margins, but neither is approved by that fact alone. Confirm the required margin and whether the quoted 3 m curve value is NPSH3.

Example 4: Speed affinity laws at corresponding points

Given: at N₁ = 1,750 rpm, a pump-curve point has Q₁ = 0.035 m³/s, H₁ = 45 m and shaft power P₁ = 21.5 kW. Estimate its corresponding point at N₂ = 1,450 rpm, keeping the same impeller and assuming unchanged density and efficiency.

1. Speed ratio: r = 1,450/1,750 = 0.828571429.
2. Flow: Q₂ = 0.035r = 0.029 m³/s = 29 L/s.
3. Head: H₂ = 45r² = 30.893878 m.
4. Shaft power: P₂ = 21.5r³ = 12.230052 kW, a 43.116% reduction at these corresponding points.

Interpretation: these are mapped pump-curve points. The 29 L/s value is not a confirmed delivery rate in an unchanged installation, and the calculated percentage is not a guaranteed site energy saving. The next example performs the missing system-curve check.

Example 5: Find the operating point, then reduce speed

Given: a pump has shutoff head H₀ = 65 m and head 42 m at Q_ref = 0.040 m³/s. The system requires 22 m at zero flow and 45 m at the same reference flow. Fit the calculator's quadratic models.

1. Pump coefficient: a = (65 − 42)/0.040² = 14,375 s²/m⁵.
2. System coefficient: K_sys = (45 − 22)/0.040² = 14,375 s²/m⁵.
3. Equate the heads: 65 − 14,375Q² = 22 + 14,375Q², so 28,750Q² = 43.
4. Operating flow: Q = √(43/28,750) = 0.038673663 m³/s = 38.674 L/s.
5. Operating head: H = 22 + 14,375(0.038673663)² = 43.5 m. The pump equation also gives 43.5 m.
6. Estimated shaft demand: with ρ = 1,000 kg/m³ and η_p = 0.72, P_shaft = 1,000 × 9.81 × 0.038673663 × 43.5/0.72 = 22.921 kW.

Reduce speed to 80% of its original value. Scale the entire pump curve: H_new(Q) = 0.8²H_old(Q/0.8) = 41.6 − 14,375Q². Keep the system curve unchanged.

The new intersection satisfies 41.6 − 14,375Q² = 22 + 14,375Q². Therefore Q = √(19.6/28,750) = 0.026110135 m³/s = 26.110 L/s, and H = 31.8 m. With the same illustrative 72% efficiency, shaft demand is 11.313 kW.

Check the tempting shortcut: multiplying the original operating flow by 0.8 gives 30.939 L/s and its mapped pump head is 27.84 m. But the system needs 35.76 m at that flow. Those heads differ by 7.92 m, so that mapped point cannot be the new intersection.

Example 6: Velocity-based pipe diameter and specific speed

Part A, pipe diameter: for Q = 0.035 m³/s and an illustrative target velocity of 2.0 m/s, D = √[4 × 0.035/(π × 2.0)] = 0.149270533 m = 149.271 mm. If an available pipe has an actual inside diameter of 0.150 m, its velocity is 4 × 0.035/[π(0.150)²] = 1.980595 m/s. With ρ = 1,000 kg/m³ and μ = 0.001 Pa·s, Re = 297,089.

This establishes the velocity for a known inside diameter. It does not choose a pipe schedule, pressure class or material, and 2.0 m/s is a practice target rather than a universal recommendation. Recalculate losses and suction conditions using the proposed pipe.

Part B, specific speed: take a single-stage, single-entry pump's best-efficiency-point data as N = 1,750 rpm, Q = 0.035 m³/s and H = 45 m. Then n_q = 1,750√0.035/45^0.75 = 18.843548 in the rpm–m³/s–m convention. With NPSH3 = 3 m and the single-eye setting, n_ss = 1,750√0.035/3^0.75 = 143.625360. For a symmetric double-suction impeller, the per-eye flow is 0.035/2 = 0.0175 m³/s; the per-eye suction specific speed would be 143.625360/√2 = 101.558466.

The same hydraulic duty gives a pump specific-speed value of approximately 973.180 when flow is in US gpm and head in ft, about 51.645 times the SI-flow numerical value. The pump has not changed; the convention has. For multistage or double-entry pumps, establish head per stage and the required per-eye flow convention before comparing indices.

Common centrifugal pump sizing mistakes

• Using vertical rise as the whole TDH: include endpoint pressure requirements, losses and any remaining velocity-head difference.
• Mixing pressure references: keep both TDH pressures on one basis; use absolute pressures for NPSH.
• Confusing nominal pipe size with inside diameter: wall thickness changes area, velocity and loss.
• Counting fittings twice: use equivalent length or K for a particular fitting, not both.
• Trusting transitional-flow precision: Re near 3,000 deserves a method check, not just more decimal places.
• Calling electrical input the motor output rating: separate shaft load, motor input and any stated allowance.
• Treating NPSH3 as cavitation-free: it is a head-drop test criterion; adequate margin requires further assessment.
• Using one efficiency or one operating condition everywhere: review the relevant curves and the full expected range.
• Reading scaled affinity flow as delivered flow: the changed pump curve still has to meet the system curve.
• Comparing bare specific-speed numbers: state units, stage basis, flow-per-eye convention and the point used.

Check the supporting arithmetic and units

Use the scientific calculator to check powers and square roots. The length converter helps convert pipe dimensions to metres, and the mass converter helps distinguish mass units when interpreting fluid properties. Density still requires mass per unit volume, rather than mass alone.

Sources & References

The examples and practice solutions were calculated for this lesson. These primary references support the underlying methods and the important limits discussed above.

1. U.S. Department of Energy, Improving Pumping System Performance: A Sourcebook for Industry, second edition. Pump-system assessment, operating curves, control methods and preliminary selection considerations.
2. U.S. Environmental Protection Agency, EPANET 2.2 User Manual. The network-model chapter discusses Darcy–Weisbach, flow regimes and transitional-flow treatment.
3. U.S. Department of Energy, Determining Electric Motor Load and Efficiency. Electrical input, mechanical loading, efficiency and service-factor cautions.
4. Hydraulic Institute, Reliability Basics: Understanding NPSH, POR, and AOR. NPSH3, NPSH margin and operating-region considerations.
5. KSB Centrifugal Pump Lexicon: Affinity laws. Geometrically similar pump scaling and dimensional relationships.
6. Hydraulic Institute: Pump Principles. Unit conventions, best-efficiency-point data and stage/impeller-eye basis.