Centrifugal pump calculator (head, NPSH, specific speed)

Work out a centrifugal pump's net head from gauge readings (with the velocity heads), hydraulic and shaft power, efficiency, the H-Q curve, best efficiency point and specific speed. Size an impeller with velocity triangles, Euler head and slip, find the operating point on a pipe system, compare throttling with a speed change, run two pumps in series or parallel, and check NPSH available. A fourth tab checks SolidWorks Flow Simulation pump results. Free, no sign-up, runs in your browser.

IAPWS-IF97 vapour pressure · Colebrook friction · Wiesner and Stodola slip · KSB specific speed bands · Purdue, IIT (ISM) and NPTEL examples reproduced · checked against an independent solution (3,157 checks at build) · Updated 22 Nov 2026

Water at 20 °C is 998.21 kg/m³ (IAPWS-IF97). g = 9.81 m/s². Leave the diameters blank to ignore velocity heads; with total pressures they are already included.

Shutoff head36.00 mfree delivery 169.1 m³/h
Best efficiency point72.1 m³/h, 29.0 mη = 82.2%
Specific speed at BEPNs = 16.4US 847, Ω 0.310: radial impeller
Head against flow: test points and the quadratic fit0306090120150010203040Q (m³/h)H (m)BEP
Efficiency against flow03060901201500306090120Q (m³/h)η (%)82.2%
Shaft and hydraulic power against flow030609012015002.557.510Q (m³/h)P (kW)shaft powerhydraulic ρgQH
Q (m³/h)H (m)ΔV²/2gPh (kW)Ps (kW)ηp₀ ratio

    What does this centrifugal pump calculator do?

    It turns pump readings into the numbers a pump is judged by: net head H in metres, hydraulic power ρgQH, shaft power, efficiency and the best efficiency point (BEP). From the BEP it gives the specific speed and the impeller type it implies. Tab 2 goes the other way, from impeller dimensions to the head the blades can give. Tab 3 puts the pump on a pipe system, and tab 4 checks a SolidWorks Flow Simulation pump study.

    Every number on this page comes from one TypeScript library that an independent Python script re-computes; the page does not build if the two disagree. The same script reproduces worked examples from Purdue University, IIT (ISM) Dhanbad and NPTEL (table below).

    Why is head, not pressure ratio, the pump number?

    A pump adds energy per unit weight: head. The pressure ratio of a liquid pump is

    so it depends on the inlet absolute pressure, not just on the pump. The example pump at BEP adds 29.0 m; with the suction at atmosphere that is a ratio of 3.81, and the same pump at the same flow fed from a tank at 3 bar would show 1.95. Pressure ratio is how compressors and fans on gas are rated, because gas density changes with pressure. Pump curves, catalogues and acceptance tests use head.

    How do you find the best efficiency point and specific speed?

    Tab 1 fits H = a + bQ + cQ² through the points by least squares. Shutoff head is a; free delivery is the flow where the fitted head reaches zero (an extrapolation if no point is near it). The BEP is the top of the parabola through the best measured efficiency and its two neighbours; in the example it is 72.1 m³/h at 29.0 m and 82.2%. Specific speed is taken there:

    The factor 51.65 is √(15,850 gpm per m³/s)/3.28080.75; KSB and the Hydraulic Institute round it to 51.6. KSB's pump lexicon gives radial impellers ns ≈ 12 to 80, mixed-flow 80 to 160 and axial 160 to 400 and higher (rpm, m³/s, m). The example's Ns = 16.4 is a radial impeller. Specific speed is also how the turbomachinery book compares pumps.

    What is the Euler head of an impeller?

    The blades change the fluid's angular momentum. Per unit weight that is the Euler head, and it splits into three energies:

    Blade speed, meridional velocity and the ideal whirl at the outlet come from the dimensions, with τ the blade blockage:

    For the example impeller (D₂ 250 mm, β₂ 25°, 6 blades, 60.0 m³/h at 1450 rpm) u₂ = 18.98 m/s, Vu2,∞ = 15.48 m/s and Hth,∞ = 29.94 m. The ideal head falls linearly with flow:

    It falls for backward-curved blades (β₂ < 90°), is flat for radial blades and rises for forward-curved blades. Pumps use backward-curved blades because a rising curve can be unstable on a system.

    What is slip, and how do Wiesner and Stodola estimate it?

    With a finite number of blades the flow leaves the impeller turned less than the blade: the whirl Vu2 is lower than the ideal. The slip factor σ measures that. With β₂ from the tangent:

    The Wiesner form and its limiting radius ratio are as quoted by Zulkifli (Hochschule Coburg, 2017); Stodola's ΔVu = πu₂cosβ/Z with β from the radial direction is in von Backström (2007), which also notes that Stodola and Stanitz agree at 51° from radial. For the example σWiesner = 0.8145, so the head with slip is 23.13 m, and with a hydraulic efficiency of 0.85 about 19.7 m. Slip is not a loss: the impeller simply does less work. At zero flow the ideal head is u₂²/g = 36.72 m.

    Tab 2 also checks inlet incidence: the relative flow angle at the eye, arctan(cm1/(u₁ − Vu1)), against the blade angle β₁. The example blade is 22.0° and the flow 19.3°, an incidence of 2.7°.

    How do you find a pump's operating point?

    The pump runs where its curve meets the system curve: the head the pipework needs at each flow.

    The example pump on a 100 mm pipe, 250 m long, ΣK = 6 and 15 m static lift runs at 67.3 m³/h and 29.9 m. To get 53.8 m³/h instead, a throttle valve has to burn 7.3 m; running the pump at 89.2% speed (1,294 rpm) gives the same flow without the valve, and the shaft power drops from 6.03 kW to 4.46 kW. The affinity laws for that:

    Speed changes follow these laws closely. Trimming the impeller is only approximately like a speed change, and the efficiency drops with large trims, so trim in steps and test. The affinity laws move a point along a parabola H ∝ Q²; with a static lift the system curve is not that parabola, which is why tab 3 finds the new operating point on the system instead of scaling the old one.

    Two pumps in series or in parallel: which gives more flow?

    On the example system one pump gives 67.3 m³/h, two in parallel 76.1 m³/h (13% more, not double) and two in series 99.1 m³/h. Parallel pumps help most on a flat system curve (mostly static lift); series pumps help on a steep one (mostly friction) or where one pump cannot reach the static lift. See pumps in series and parallel in the turbomachinery book.

    How do you calculate NPSH available?

    Vapour pressure pv is the IAPWS-IF97 saturation equation (Region 4, Eq. 30) and density its Region 1, so the page covers 0 to 100 °C: 2,339 Pa at 20 °C, 47,415 Pa at 80 °C. The example (water surface 3 m below the pump, 8 m of 100 mm suction pipe with ΣK = 1.5, 60.0 m³/h) has NPSHa = 6.43 m with 20 °C water and 1.97 m with 80 °C water. Keep NPSHa above the pump's NPSHr with a margin; Wassgren's Purdue notes say regulations typically ask for at least 10%, and critical services often use 100%. Cavitation and suction lift are covered in the turbomachinery book's NPSH chapter.

    The SolidWorks centrifugal pump challenge: worked check of one design point

    The challenge asks for a pump in SolidWorks Flow Simulation with a rotating region, environment pressure at the inlet, an outlet velocity of 10 to 30 m/s and goals for the average total pressure on both lids, then "the relationship between pressure ratio and mass flow rate". The upper bound for the impeller's work is the Euler head with no pre-swirl:

    Design point 2 of a typical report's table (15 m/s), with D₂ = 100 mm assumed:

    So this point has the total pressure falling by 45.3 m across the pump, a negative absolute pressure, a pressure ratio printed without its minus sign, and an outlet dynamic pressure four times the most a 1000 rpm, 100 mm impeller could add (27.4 kPa). The 8 mm outlet is forced to 15 m/s: the outlet sucks water through the pump faster than it can deliver it. In tab 4, sweep the outlet static pressure (or the volume flow) from shutoff to free delivery instead, turn on cavitation, and plot head against flow.

    The other typical table runs at 1000 rad/s: there u₂ = 50 m/s and ρu₂² = 2.50 MPa, so its 1.28 to 3.73 MPa rises fit a 9,549 rpm impeller (one point is still above the ceiling for D₂ = 100 mm). Read the same table as 1000 rpm and every positive rise is 47 to 136 times the ceiling.

    What the challenge reports do

    We read the 195 public reports of the challenge on 22 Nov 2026; 147 of them use SolidWorks Flow Simulation (21 are the CONVERGE version and 27 do not name the solver). The counts below are out of the 147.

    • Pressure ratio, not head. 141 report a pressure ratio; 6 mention head together with ρg. 128 force an outlet velocity, most from 10 to 30 m/s.
    • rpm or rad/s. 55 state 1000 rad/s (9,549 rpm) and 28 state 1000 rpm. Flow Simulation's angular velocity field is in rad/s by default. All 5 of the "1000 rpm" reports that print a pressure table above atmosphere show more than 1 MPa. For the 80 to 150 mm impellers the reports state, ρu₂² is 17.5 to 61.6 kPa at 1000 rpm and 1.60 to 5.61 MPa at 1000 rad/s. MPa results fit 1000 rad/s, so a "1000 rpm" report with MPa results most likely typed 1000 into a rad/s field.
    • Negative absolute pressure. 3 print a negative absolute total pressure in their tables (down to -2.42 MPa); none of them comment on it. 12 mention cavitation, 5 vapour pressure and 2 dynamic pressure or velocity head.
    • The mass flow sign is fine. 41 explain the negative outlet mass flow as Flow Simulation's sign convention, which is right.

    Tab 4 flags each of these in your own table.

    Published pump examples this calculator reproduces

    ExamplePrinted answerThis calculator
    Purdue ME308 pump_05: head from two gauges, torque, efficiencyΔH 34.1 m (from 55.7 − 21.6), 1.07 kW, 1.35 kW, η 79.4%34.18 m, 1.07 kW, 1.35 kW, 79.4%
    Purdue pump_19: brine, mercury manometer, gauge 1.2 m upH 15.3 m, 22.5 kW into the fluid15.27 m, 22.47 kW
    Purdue pump_06: specific speed at 2875 rpm, 252 gpm, 138 ftNsd 1130, Ns 0.4141,134, 0.4146
    Purdue pump_18: NPSH available, 4 m suction liftV 1.28 m/s, Re 131,000, f 0.0195 (chart), HL 1.61 m, NPSHA 4.5 m1.28, 131,069, 0.0194 (Colebrook), 1.60, 4.51
    Purdue pump_02: 12 in impeller cut to 10 in1850 gpm, 41.7 ft, 24.1 hp (similarity)1,852, 41.7, 24.1
    IIT (ISM) Dhanbad, problem 1: 1450 rpm, D 150/350 mmu1 11.39, inlet angle 4.02°, Hm 51.90 m, η 72.5%, ηman 0.74411.39, 4.02°, 51.89, 72.5%, 0.7433
    IIT (ISM) Dhanbad, problem 3: impeller for 20 m at 1000 rpmu2 16.96 m/s, D2 0.324 m, B2 78.61 mm16.96, 0.3239, 78.62
    NPTEL 112104117 ch 8.11, problem 4: Thoma numberσ 0.0840.0843

    Printed answers are matched to their last printed digit, or within one unit of it where the source rounds intermediate values (IIT (ISM) problem 1 rounds its velocities first; with those rounded values the verifier gets its 51.90 m and 0.744 exactly). Purdue pump_18 reads f = 0.0195 off a Moody chart; Colebrook gives 0.0194 and the NPSHa still rounds to 4.5 m.

    How we checked the calculator

    An independent Python script, scripts/verify/pump.py, rebuilds every calculation, and this page refuses to build if the browser code differs from it in any of 3,157 numbers (251 further assertions run inside the script). Water properties come from the iapws package (IF97 Region 4 and Region 1, IAPWS 2008 viscosity) and are checked against Table 35 of the IF97 release, the NIST WebBook and CoolProp's IAPWS-95. Fits are checked against numpy.polyfit, operating points against scipy's brentq with its own Colebrook solve, and the Euler head against its three-term form for 26 impellers. Two deliberate bugs (head scaled with N instead of N², and vapour pressure in kPa read as Pa) each make the build fail.

    What this calculator does not cover

    • Losses inside the pump. Tab 2 gives the Euler head and slip; hydraulic efficiency is your input, not predicted. Disc friction, leakage and recirculation are not modelled.
    • NPSH required. NPSHr comes from the pump maker's test (NPSH3). The page does not predict it, and it does not reproduce the ANSI/HI 9.6.1 margin tables.
    • Viscous liquids. Head and efficiency corrections for oils (ANSI/HI 9.6.7) are not applied; the fluid is water, or a density you type in tab 1.
    • Trim laws. Q~D, H~D² for a trimmed impeller is an approximation; the page shows two common forms, not a test-based correction.
    • The CFD itself. Tab 4 checks numbers for consistency and physics limits; it cannot tell a converged, mesh-independent run from a lucky one.
    Learn pump CFD

    The centrifugal pump challenge is part of Skill-Lync's CFD courses with SolidWorks: moving reference frames, rotating regions and pump performance curves, next to the rest of the CFD toolkit.

    Frequently asked questions

    How do you calculate pump head from pressure gauge readings?

    Subtract the suction reading from the delivery reading and divide by ρg, then add the difference in velocity heads V²/2g (from the flow and the two pipe diameters) and the height of the delivery gauge above the suction gauge. Use both readings as gauge or both as absolute. Tab 1 does this for every test point.

    Why is pump performance given as head and not pressure ratio?

    Head is the energy the pump adds per unit weight of liquid, so it does not change with the inlet pressure or the liquid. A pressure ratio is 1 + ρgH/p₀,in: the same pump gives a different ratio on a different day. Pressure ratio is a compressor number. Pump curves, catalogues and IS/ISO tests all use head.

    What is the specific speed of a centrifugal pump?

    Specific speed Ns = N√Q/H^0.75 at the best efficiency point describes the impeller's shape, not its size. With rpm, m³/s and m, KSB gives radial impellers about 12 to 80, mixed flow 80 to 160 and axial 160 and above. Multiply by 51.65 for the US (rpm, gpm, ft) number.

    How do you calculate NPSH available?

    NPSHa = (p_surface − p_v)/(ρg) + z_s − h_f: the absolute pressure on the liquid surface less the vapour pressure, as head, plus the height of the surface above the pump (minus for a suction lift), minus the suction pipe loss. Vapour pressure rises fast with temperature: 2.34 kPa at 20 °C, 47.4 kPa at 80 °C.

    What are the pump affinity laws?

    For one pump at a new speed, flow scales with speed, head with speed squared and power with speed cubed. For a trimmed impeller the same exponents are used with diameter, but only as an approximation for small trims. The laws move a point on the pump curve; the new operating point still comes from the system curve.

    What happens when two identical pumps run in series or in parallel?

    In series the heads add at the same flow; in parallel the flows add at the same head. Against a real system neither doubles the result: on this page's example system one pump gives 67.3 m³/h, two in parallel 76.1 m³/h and two in series 99.1 m³/h. Which helps more depends on how steep the system curve is.

    Why does my SolidWorks Flow Simulation pump show a negative pressure?

    Flow Simulation prints absolute pressure, so a negative value is impossible. Forcing a high outlet velocity on a small pump pushes it past free delivery; with no cavitation model the solver lets the pressure fall below zero instead of letting the water boil. Sweep the outlet pressure or flow, and report head, not pressure ratio.

    Is this centrifugal pump calculator free?

    Yes. It runs in your browser with no sign-up, and nothing you type is sent anywhere. It reproduces published examples from Purdue, IIT (ISM) Dhanbad and NPTEL, and the page will not build if the browser code and an independent Python check disagree on any of 3,157 numbers.

    Related on Skill-Lync

    Sources

    • IAPWS R7-97(2012), Revised Release on the IAPWS Industrial Formulation 1997 for the Thermodynamic Properties of Water and Steam, Eq. 30 and Table 35; IAPWS R12-08, viscosity of water (through the iapws Python package).
    • KSB, Centrifugal Pump Lexicon, "Impeller" and "Specific speed" (ns bands, Ns = 51.6 ns, K = ns/52.9); Hydraulic Institute, Pump Fundamentals: Pump Principles (51.6 conversion).
    • M. A. Zulkifli, Evaluation of known slip factor models using CFD, B.Eng. thesis, Hochschule Coburg, 2017, §7.7 (Wiesner and Gülich forms); T. W. von Backström, "Relative-eddy induced slip in centrifugal impellers for engineering students", R&D Journal of SAIMechE 23(1), 2007 (Stodola, Stanitz, Wiesner).
    • C. Wassgren, Purdue ME308 pump examples 02, 05, 06, 07, 18, 19, and notes on specific speed and NPSH.
    • S. Sarkar, IIT (ISM) Dhanbad, MMC16101 Fluid Machines, "Centrifugal pump problems"; NPTEL course 112104117 Fluid Machinery, chapter 8.11.
    • Skill-Lync, the 195 public reports of the "Centrifugal pump design and analysis" challenge, read 22 Nov 2026.
    Free counselling

    From a pump simulation challenge to a CFD job.

    Rotating machinery CFD, pump curves and reading results critically are what CFD teams test for. A Skill-Lync counsellor can map the course that fits the role you want.

    Talk to a CFD course counsellor

    Free. We call you back, usually the same day.

    By submitting you agree to our Terms and Privacy Policy.

    Talk to a career counsellor

    Talk to a career counsellor

    Free 20-minute call. We'll map the right course to where you want to work.

    By submitting you agree to our Terms and Privacy Policy.