Engineering Article

Centrifugal Pump Sizing and Selection: A Complete Engineering Guide

A complete engineering guide to Total Dynamic Head (TDH), NPSH and cavitation — including HI 9.6.1 hydrocarbon derating — affinity laws, viscosity effects, and power sizing for centrifugal pumps in industrial fluid transfer systems.

Last Updated: July 27, 2026
20 min read
Reading Preferences

Centrifugal pumps are the workhorses of the chemical process industry, responsible for moving fluids safely and efficiently. Properly sizing a pump is not merely a matter of finding one that matches the desired flow rate; it requires a rigorous evaluation of the entire piping system's hydraulic resistance, liquid properties, and available suction pressure. An improperly sized pump leads to poor efficiency, motor overloading, or catastrophic mechanical failure due to cavitation.

Centrifugal Pump Sizing Schematic showing TDH, NPSH, and System Curves

Figure 1: A proper pump sizing evaluation aligns system hydraulic resistance (TDH) and suction conditions (NPSHa) with the pump's performance curve.

Pump Selection Flowchart

Selecting a pump involves matching process requirements to hydraulic calculations before consulting a manufacturer's curve. Follow this workflow:

START: Define Flow (Q) Are Fluid Properties Known? (SG, Visc, Pv) No Obtain Data Required for Calc Yes Calculate TDH Calculate NPSHa Calculate Brake HP Select Pump from Curve Verify NPSHa > NPSHr and Operating near BEP

Figure 2: A structured workflow for sizing a centrifugal pump. Scroll horizontally to view the full diagram on smaller screens.

Advertisement

1. Principles of Pump Sizing

Pump sizing relies heavily on Bernoulli's principle. Unlike compressors which deal with highly compressible gases, pumps handle mostly incompressible liquids. Therefore, process engineers talk about fluid energy in terms of "Head".

Head represents the height of a fluid column that a pump can support. Using Head instead of Pressure (e.g., psi or bar) is advantageous because a centrifugal pump operating at a specific speed will generate the exact same Head regardless of the fluid's density. A pump that produces 100 ft30 m of head will push water to 100 ft30 m, and it will push heavy sulfuric acid to exactly 100 ft30 m—although the pump motor will have to work much harder (consume more power) to move the denser acid.

2. Calculating Total Dynamic Head (TDH)

The Total Dynamic Head (TDH) is the total equivalent height that a fluid is to be pumped, taking into account friction losses in the pipe. It is the core metric used to read a pump curve.

$$TDH = H_{static} + H_{pressure} + H_{friction}$$ (Eq. 1)
$H_{static}$ = Elevation change · $H_{pressure}$ = Pressure difference between vessels · $H_{friction}$ = Piping losses
Head Component Description Formula / Origin
Static Head ($H_{static}$) The physical change in elevation between the liquid level in the suction tank and the discharge point. $Z_{discharge} - Z_{suction}$
Pressure Head ($H_{pressure}$) The difference in surface pressure between the destination vessel and the source vessel. E.g., pumping from an open tank to a pressurized reactor. $\frac{P_{discharge} - P_{suction}}{\rho \times g}$
Friction Head ($H_{friction}$) Energy lost due to fluid rubbing against pipe walls and passing through valves/fittings. Calculated via Darcy-Weisbach or equivalent equations.

3. Frictional Losses & Equivalent Length

Calculating friction head is often the most tedious part of pump sizing. For liquids, the standard method utilizes the Darcy-Weisbach equation:

$$H_{friction} = f \left( \frac{L_{eq}}{D} \right) \left( \frac{v^2}{2g} \right)$$ (Eq. 2)
$f$ = Darcy friction factor · $L_{eq}$ = Total equivalent length [ft] · $D$ = Inner pipe diameter [ft] · $v$ = Velocity [ft/s]

The length $L_{eq}$ must include not just the straight run of pipe, but the Equivalent Length of all fittings (elbows, tees, valves) — typically expressed either as an L/D multiple (e.g. a standard threaded 90° elbow ≈ 30 pipe diameters of straight pipe) or as a dimensionless resistance coefficient $K$, where $h_{f,fitting} = K \left( \frac{v^2}{2g} \right)$. Both methods are equivalent ($K = f \times L_{eq}/D$); the CheCalc Pump Sizing Calculator's fittings library uses the $K$-factor form directly (per Crane TP-410), since $K$ is largely independent of pipe diameter and is easier to look up for a specific valve or fitting type.

The friction factor $f$ itself depends on the flow regime: for laminar flow ($Re < 2300$), $f = 64/Re$; for turbulent flow ($Re > 4000$), it is obtained from the Swamee-Jain explicit approximation to the Colebrook-White equation. The narrow transition zone ($2300 < Re < 4000$) is neither reliably laminar nor fully turbulent, and is conservatively evaluated using the turbulent formula — the same approach used by the calculator's hydraulic engine.

Velocity Rule of Thumb: To balance pipe cost against pumping energy costs (and, on the suction side, to protect NPSHa), engineers typically target:
  • 7 to 12 ft/s2.0 to 3.5 m/s for general pump discharge lines (long transfer lines are often kept lower, around 5 to 8 ft/s1.5 to 2.5 m/s, to limit friction loss and water-hammer risk).
  • 3 to 5 ft/s0.9 to 1.5 m/s for general pump suction lines, tightening to 2 to 3 ft/s0.6 to 1.0 m/s for NPSH-sensitive services (hot or volatile liquids, high suction-specific-speed pumps).
Advertisement

4. Net Positive Suction Head (NPSH) & Cavitation

When pressure inside the pump impeller drops below the liquid's vapor pressure, the liquid boils. Vapor bubbles form and then rapidly collapse as they move to higher pressure regions, causing severe shockwaves. This is cavitation, and it sounds like pumping gravel. It will quickly destroy an impeller.

To prevent this, the energy available at the pump suction must exceed what the pump requires.

Rule for safe operation: $NPSHa > NPSHr + \text{Safety Margin (usually 3 to 5 ft / 1 to 1.5 m)}$

$$NPSHa = H_a \pm H_z - H_f - H_v$$ (Eq. 3)
$H_a$ = Absolute pressure on liquid surface · $H_z$ = Static head of liquid over pump centerline · $H_f$ = Suction line friction loss · $H_v$ = Vapor pressure of liquid at operating temp
Pumping Hot Liquids: As a liquid's temperature rises, its vapor pressure ($H_v$) increases exponentially. Pumping boiling water from a reboiler, for example, means $H_a$ and $H_v$ essentially cancel out, making NPSHa entirely dependent on physical elevation ($H_z$) above the pump.

NPSH Margin: Rule of Thumb vs. API 610 Ratio

A flat additive margin (a few feet/meters of NPSH cushion, as used above) is a useful hand-calculation rule of thumb, but it does not scale well: a 3 ft1 m margin is generous for a low-NPSHr pump and dangerously thin for a high-suction-energy pump with an NPSHr of 20+ ft6+ m. Modern practice, codified in API 610 (Section 6.1.6), instead specifies a minimum NPSH margin ratio:

$$\frac{NPSHa}{NPSHr} \geq 1.3$$ (Eq. 3a)
Some operators apply higher ratios (1.5–2.0×) for high-suction-energy or boiler feed water services, where cavitation damage is especially severe.

NPSHa should also be checked at the minimum suction liquid level (not just the normal level), since batch draindown or low-inventory upset conditions represent the worst-case NPSHa the pump will see in service. Both checks — ratio-based margin and minimum-level NPSHa — are what the CheCalc Pump Sizing Calculator's Engineering Alert Panel screens automatically.

NPSH Derating for Hydrocarbons (HI 9.6.1)

Pump vendor curves publish NPSHr from a cold-water test. For fluids with a relatively flat vapor-pressure-vs-temperature curve near the operating point — light hydrocarbons in particular — the pump can tolerate more local vaporization at the impeller eye before performance actually degrades, because the collapsing vapor re-condenses almost immediately. The Hydraulic Institute (HI 9.6.1) captures this with a Vapor Suppression Correction Factor (VSCF):

$$NPSH_{r,eff} = NPSH_r \times VSCF$$ (Eq. 3b)
Fluid Indicative VSCF (HI 9.6.1)
Propane / LPG 0.15 – 0.25
Light Naphtha 0.30 – 0.40
Benzene 0.50 – 0.60
Kerosene 0.50
Water / Amines 1.00 (no credit taken)

Applying VSCF is a credit, not a requirement — when in doubt, or for aqueous and amine systems, use VSCF = 1.0 (i.e., take no derating credit) and rely on the standard cold-water NPSHr.

5. Power Calculations (Hydraulic & Brake)

The power imparted directly to the fluid is hydraulic power. However, mechanical losses (bearings, seals) and volumetric losses mean the motor must supply more power. This is the Brake Horsepower (BHP) or Brake Power.

$$BHP = \frac{Q \times TDH \times SG}{3960 \times \eta}$$ (Eq. 4 - US)
$Q$ [gpm] · $TDH$ [ft] · $SG$ = Specific Gravity · $\eta$ = Pump Efficiency (decimal) · $BHP$ [hp]
$$Power (kW) = \frac{Q \times H \times SG}{367 \times \eta}$$ (Eq. 4 - Metric)
$Q$ [m³/h] · $H$ [m] · $SG$ = Specific Gravity · $\eta$ = Pump Efficiency (decimal) · $Power$ [kW]

6. Solved Examples: System Head & Power

Example: Sizing a Water Transfer Pump

Scenario: You need to transfer water ($SG = 1.0$) from an atmospheric tank at grade to a higher elevation tank.

  • Desired Flow Rate ($Q$): 500 gpm
  • Static Elevation Change ($H_{static}$): 50 ft
  • Calculated Friction Losses ($H_{friction}$): 20 ft
  • Pressures are both atmospheric, so $H_{pressure} = 0$.
  • Pump Efficiency ($\eta$): $75\%$ ($0.75$)
  1. Calculate TDH:
    $TDH = H_{static} + H_{pressure} + H_{friction}$
    $TDH =$ 50 + 0 + 20 = 70 ft
  2. Calculate Required Power:
    Using the power formula for the selected unit system:
    $BHP = (500 \times 70 \times 1.0) / (3960 \times 0.75)$ $Power = (113.5 \times 21.3 \times 1.0) / (367 \times 0.75)$
    Power required at the pump shaft = 11.78 hp.
    Standard practice dictates selecting the next standard motor size up to prevent overloading (e.g., a 15 hp motor).

Example: NPSHa with Negative Suction Lift

Scenario: A centrifugal pump draws water ($SG = 1.0$) from an underground sump pit. The liquid level is below the pump centerline, creating a "suction lift" which severely impacts available NPSH.

  • Atmospheric Pressure Head ($H_a$): 34.0 ft
  • Static Suction Head ($H_z$): -15.0 ft (Negative because it's below the pump)
  • Suction Line Friction Loss ($H_f$): 3.0 ft
  • Water Vapor Pressure Head ($H_v$) at 68°F (20°C): 0.8 ft
  1. Calculate NPSHa:
    $NPSHa = H_a + H_z - H_f - H_v$
    $NPSHa =$ 34.0 - 15.0 - 3.0 - 0.8 = 15.2 ft
  2. Conclusion:
    The pump manufacturer's curve must show an NPSHr of less than 12.2 ft (leaving a typical 3 ft flat safety margin). Checked against the API 610 ratio method instead ($NPSHa/NPSHr \geq 1.3$), the vendor's NPSHr would need to be at or below 11.7 ft — the two methods land close together here, but the ratio method is the more defensible basis for a formal pump specification. If the required NPSH is higher than either limit, the pump will cavitate. To fix this, you must either lower the pump elevation or increase the suction pipe diameter to reduce friction ($H_f$).

Example: Viscosity-Corrected Power for a Glycol Circulation Loop

Scenario: A closed-loop circulation pump handles a 50% MEG (mono-ethylene glycol) solution ($SG = 1.05$) at a cold-start condition, where viscosity is significantly higher than in normal running. TDH is assumed already established from the system curve, as in Section 2.

  • Design Flow Rate ($Q$): 300 gpm
  • Total Dynamic Head ($TDH$): 100 ft
  • Kinematic Viscosity ($\nu$) at a cold winter start-up: 150 cSt
  • Pump Efficiency at duty point ($\eta$, from vendor's water-test curve): $70\%$ ($0.70$)
  1. Calculate the water-equivalent shaft power (Eq. 4), ignoring viscosity for now:
    $BHP_{water} = (300 \times 100 \times 1.05) / (3960 \times 0.70)$ $Power_{water} = (68.1 \times 30.5 \times 1.05) / (367 \times 0.70)$
    = 11.36 hp
  2. Apply the HI 9.6.7-inspired viscosity correction (Eq. 7): since $\nu = 150$ cSt exceeds the 10 cSt threshold,
    $C_{eff} = \max(0.4,\ 1 - 0.005\sqrt{150}) = 0.94$
  3. Corrected shaft power = water-equivalent power ÷ $C_{eff}$:
    Power required at the pump shaft, corrected for viscosity = 12.10 hp
  4. Conclusion: Applying the standard 25% motor margin (the corrected shaft power is below 30 hp22 kW), the required minimum becomes 15.13 hp, so the next standard motor size is 20 hp — one frame size larger than the 15 hp motor an uncorrected, water-curve-only calculation of this same duty would suggest (11.36 hp × 1.25 = 14.20 hp, which the 15 hp/11 kW frame still covers). Skipping the viscosity correction on a glycol, amine, or other non-aqueous service can silently undersize the driver for cold-start conditions.

Example: NPSH Derating for a Propane Condensate Sump Pump

Scenario: A pump lifts propane condensate ($SG = 0.50$) from a sump vessel operating close to, but slightly below, its bubble point, through a short suction-lift arrangement.

  • Absolute Pressure Head at liquid surface ($H_a$): 640 ft
  • Vapor Pressure Head at pumping temperature ($H_v$): 630 ft
  • Static Suction Head ($H_z$), pump centerline above sump low level: -3.3 ft
  • Suction Line Friction Loss ($H_f$): 1.3 ft
  • Vendor Cold-Water NPSHr (from curve): 9.8 ft
  1. Calculate NPSHa (Eq. 3):
    $NPSHa = (H_a - H_v) + H_z - H_f$ = 5.4 ft
  2. Check against the cold-water NPSHr with no derating credit (Eq. 3a): the required margin is 9.8 × 1.3 = 12.7 ft, well above the 5.4 ft NPSHa available — this pump fails a naive cold-water check.
  3. Apply the HI 9.6.1 derating for propane (Eq. 3b), using a mid-range VSCF of $0.20$:
    $NPSH_{r,eff} =$ 9.8 $\times 0.20 =$ 2.0 ft
  4. Conclusion: Re-checking the ratio with the derated value, $NPSHa/NPSH_{r,eff} \approx$ 2.76, comfortably above the 1.3 minimum — the selection passes. The same pump and layout that fails a naive cold-water check clears with a healthy margin once the legitimate vapor-suppression credit for this light hydrocarbon is applied. This is exactly the case HI 9.6.1 derating exists for — it is not a way to force a marginal design through, only to remove an artificial penalty that a cold-water test never actually applies to this fluid.

7. Reading a Pump Curve, BEP & Specific Speed

Once you have your Design Flow ($Q$) and your calculated $TDH$, you plot this coordinate on manufacturer pump curves. A pump curve graphs the pump's Head generation against Flow Rate. Because centrifugal pumps have "slip", as flow rate increases, the head the pump can generate drops off.

You should aim for your design point to fall near the Best Efficiency Point (BEP) on the curve. Operating too far to the left (low flow) causes internal recirculation, overheating, and shaft deflection. Operating too far to the right (high flow/run-out) causes cavitation and extreme power draw. API 610 quantifies this as a preferred operating region of 70–120% of BEP flow, with 50–140% considered the outer acceptable limit; running much below 50% or above 140% of BEP flow measurably increases the risk of vibration, seal failure, and bearing wear.

The impeller geometry that gives a pump its curve shape (and determines whether it is best suited to a high-head/low-flow or low-head/high-flow duty) is characterized by the Specific Speed ($N_s$), evaluated at the BEP:

$$N_s = \frac{N \sqrt{Q}}{H^{0.75}}$$ (Eq. 5)
$N$ = rotational speed [rpm] · $Q$ = flow at BEP [m³/sgpm] · $H$ = head at BEP [ft]

Low $N_s$ values correspond to radial-flow impellers (high head, low flow); high $N_s$ values correspond to mixed- and axial-flow impellers (high flow, low head). Note that the numeric value of $N_s$ depends on which unit convention is used — the US customary form (rpm, gpm, ft) and the metric form (rpm, m³/s, m) are not interchangeable inputs to the same chart, so always match your calculated $N_s$ against a classification chart built for the same units. $N_s$ is mainly a classification/verification tool — it confirms a vendor's proposed impeller type is a reasonable match for the duty — rather than a quantity you solve for directly during sizing.

8. Affinity Laws & Off-Design Performance

If you already have a pump curve at one speed or impeller diameter and need to predict performance at another, the Affinity Laws relate flow, head, and power to speed ($N$) and impeller diameter ($D$):

$$\frac{Q_1}{Q_2} = \frac{N_1}{N_2}\left(\frac{D_1}{D_2}\right) \qquad \frac{H_1}{H_2} = \left(\frac{N_1}{N_2}\right)^2 \left(\frac{D_1}{D_2}\right)^2 \qquad \frac{P_1}{P_2} = \left(\frac{N_1}{N_2}\right)^3 \left(\frac{D_1}{D_2}\right)^3$$ (Eq. 6)

These are most reliable for small changes in speed or trimmed impeller diameter (roughly ±10–15%) on the same pump casing; larger changes shift the efficiency curve and should be confirmed against vendor data. Practical uses include: sizing a VFD's speed turndown range for a control application, estimating the effect of trimming an impeller to meet a lower duty point without buying a smaller pump, and predicting how NPSHr shifts (approximately with $N^2$) at a different operating speed.

9. Viscosity Effects on Pump Performance

The equations above assume a low-viscosity fluid (close to water). As viscosity rises, disc friction and internal recirculation losses inside the impeller increase, which reduces the head and flow the pump can deliver and reduces efficiency relative to its water-test curve, while increasing the shaft power required for a given duty.

The rigorous method is the Hydraulic Institute (HI 9.6.7) viscosity correction, which applies separate correction factors to flow, head, and efficiency as a function of viscosity, flow, and head. For a quick screening estimate of the power penalty alone (not a substitute for the full HI 9.6.7 chart near design limits), a simplified single-factor form is sometimes used:

$$C_{eff} = \max\left(0.4,\ 1 - 0.005\sqrt{\nu_{cSt}}\right) \quad \text{for } \nu > 10\ cSt$$ (Eq. 7)
$\nu_{cSt}$ = kinematic viscosity in centistokes · $C_{eff}$ derates the water-curve efficiency to estimate corrected shaft power

As a rule of thumb, viscosity corrections become worth applying above roughly 10–20 cSt; below that, water-curve performance is normally close enough for standard process services. For viscous duties near a pump's operating limits (e.g. cold-start conditions, high-viscosity glycols or heavy fuel oils), always confirm the final selection against the vendor's own viscosity correction data.

10. Common Mistakes to Avoid

Engineering Pitfalls in Pump Sizing
  • Ignoring Specific Gravity in Motor Sizing. A pump moving a fluid with SG=1.5 requires 50% more horsepower than one moving water, even if flow and TDH are identical. Always size motors for the heaviest fluid expected (e.g., cold start conditions).
  • Oversizing pumps "just in case". Adding excessive safety margins to TDH pushes the operating point to the left of the BEP. The pump will be forced to operate partially shut-in (throttled), wasting immense amounts of electrical energy over its lifespan and increasing wear.
  • Using nominal pipe size for velocity calculations. A 2-inch Schedule 40 pipe does not have a 2.0-inch inside diameter (it's 2.067 inches). Always use exact ID for calculating velocity and friction.
  • Applying a flat NPSH margin at every flow. A fixed 3–5 ft (1–1.5 m) cushion is fine for a quick hand check, but for a formal specification use the API 610 ratio ($NPSHa/NPSHr \geq 1.3$) and re-check it at the minimum suction level, not just the normal one.
  • Ignoring viscosity on non-aqueous services. Above roughly 10–20 cSt, a pump delivers less head and flow than its water-test curve suggests, while drawing more shaft power. Skipping the HI 9.6.7 correction can leave both the pump and its motor undersized for cold-start or winter conditions.
Advertisement
Key Takeaways
  • Evaluate fluid energy in terms of Head (length) rather than Pressure, as Head is independent of the pumped fluid's density.
  • Calculate TDH comprehensively by summing Static Head, Pressure Head, and Frictional Head.
  • Always verify that available suction head (NPSHa) clears the pump's required suction head (NPSHr) with an adequate margin — preferably a 1.3× ratio per API 610, checked at the minimum suction level, rather than only a flat additive cushion.
  • Target pump selection so the operating point falls within the API 610 preferred region of 70–120% of Best Efficiency Point (BEP) flow.
  • Use the Affinity Laws for small speed or trimmed-impeller changes, and apply an HI 9.6.7 viscosity correction above roughly 10–20 cSt — both matter once the duty moves away from a water-like fluid at fixed speed.

Conclusion

Proper pump sizing is a rigorous procedure linking fluid mechanics to mechanical equipment selection. By meticulously calculating Total Dynamic Head and evaluating Net Positive Suction Head, engineers ensure reliable process operation, minimize maintenance costs, and reduce lifetime energy consumption.

Automate Your Pump Sizing Calculations

Avoid manual friction factor iterations and unit conversion errors. The CheCalc Pump Sizing Application instantly calculates equivalent lengths, pipe friction, TDH, NPSHa, and required motor power based on standard pipe schedules.

Open Centrifugal Pump Sizing Calculator
Further Reading
  1. Crane Co. Flow of Fluids Through Valves, Fittings, and Pipe (Technical Paper No. 410). — The industry standard reference for frictional pressure drop and fitting $K$-factor data.
  2. Karassik, I.J., et al. Pump Handbook, 4th ed. McGraw-Hill Education, 2007. — Comprehensive resource on pump design, application, and theory.
  3. API Standard 610 (11th/12th Ed.). Centrifugal Pumps for Petroleum, Petrochemical and Natural Gas Industries. — Duty point margins, NPSH margin ratio requirements, and the preferred BEP operating region.
  4. API Standard 682 (4th Ed.). Pumps — Shaft Sealing Systems for Centrifugal and Rotary Pumps. — Seal arrangement categories and flush plan piping (Plans 11–76).
  5. Hydraulic Institute (HI) 9.6.1. Rotodynamic Pumps — Guideline for NPSH Margin. — Basis for the Vapor Suppression Correction Factor (VSCF) used for hydrocarbon NPSH derating.
  6. Hydraulic Institute (HI) 9.6.7. Rotodynamic Pumps — Guideline for Effects of Liquid Viscosity on Performance. — Full multi-parameter method for correcting flow, head, and efficiency for viscous fluids.
  7. Swamee, P.K. and Jain, A.K. (1976). Explicit Equations for Pipe-Flow Problems. Journal of the Hydraulics Division, ASCE. — Explicit approximation to the Colebrook-White equation used to solve for the Darcy friction factor.