← Back to AntraSIM

Pump Flow Discontinuity Simulator

Visualizing low-flow stall, recirculation, and abrupt transition to attached flow

Pump Cross-Section LOW-FLOW STALLED
Q – N Characteristic ● operating point  ·  faint red trace = path since Reset
Time Trends — 30 s rolling window Each variable on its own axis (0 → 1.0 normalised)
Speed N 0.000
Flow Q 0.000
Recirc. R 0.000
Stall S 1.000
Power P 0.0 kW
Current I 0.0 A
Speed N
0.000
Flow Q
0.000
Head H
0.000
Efficiency η
0.000
Recirc. R
0.000
Loss L
0.000
Shaft Power
0.0 kW
Motor Current
0.0 A
State Low-flow stalled
Pump Type
Speed Control
Speed Setpoint Nset0.00
manual step ±0.05
Ramp Rate0.08
Physical Parameters
Stall Severity Ls0.45
Recirculation Rmax0.75
System Resistance Ks0.75
Hysteresis Width0.10
Noise / Disturbance0.010
Motor Parameters (EU 3-phase 400 V / 50 Hz)
Rated Power Pn15.0 kW
Power Factor cos φ0.86
Motor Efficiency ηm0.92
UL-L = 400 V  |  f = 50 Hz  |  3-phase
I = Pshaft / (√3 · 400 · cosφ · ηm)
Simulation Actions

Phenomenon explained

At low flow the pump can operate in a stalled, recirculating state. A part of the fluid leaving the rotor does not become useful discharge flow; it circulates internally from the high-pressure region back toward the suction side.

The motor can therefore consume shaft power while the measured through-flow remains low. Energy is not lost physically, but it is converted into recirculation, swirl, turbulence, vibration, and heat instead of useful flow.

When the internal stall pattern collapses, the passages reattach and losses drop. The same smooth speed increase can then produce an abrupt flow jump. With hysteresis enabled, the return to stall during ramp-down occurs at a lower speed.

Theoretical Background

Physical interpretation, simplified simulation model, practical implications, and supporting references

1What the simulator demonstrates

Pump characteristics are often drawn as smooth and single-valued curves. Real pumps can behave less cleanly at low flow. During start-up or speed ramp-up, axial and centrifugal pumps may enter a low-flow stalled state, where increasing speed does not immediately produce a proportional increase in measured through-flow.

The reason is internal hydraulics. At low through-flow, blade incidence becomes unfavourable, local separation appears, and recirculation develops near the rotor, impeller eye, casing wall, or discharge region. A part of the pump work is then spent on internal circulation rather than useful discharge flow.

When a critical condition is reached, the stalled structure can collapse. Flow reattaches, hydraulic losses decrease, and the measured flow can jump abruptly to a higher-flow branch even though pump speed changes smoothly.

2Energy interpretation

The flow jump does not violate energy conservation. Before the jump, a larger part of the shaft power is dissipated in non-useful hydraulic motion:

  • internal recirculation and reverse flow pockets,
  • separated blade-passage flow,
  • swirl and pre-swirl losses,
  • turbulence, vibration, and heat generation.

After reattachment, the internal loss fraction drops and more of the same shaft power becomes useful hydraulic output. The observed discontinuity is therefore a redistribution of energy from losses to through-flow.

Core mechanism: low flow increases incidence; high incidence promotes separation; separation creates recirculation; recirculation blocks effective flow area; the reduced effective flow reinforces the low-flow condition.

3Hysteresis

The transition point can depend on operating history. During ramp-up, the pump may remain stalled until the speed and flow exceed an attachment threshold. During ramp-down, an already attached flow field can remain attached down to a lower threshold.

Direction-dependent thresholds $$N_{attach} > N_{stall}$$ $$Q_{attach} > Q_{stall}$$ The upward transition occurs at a higher speed and flow than the downward return transition.

In the simulator, the hysteresis slider controls the separation between these thresholds. With hysteresis set to zero, the two threshold pairs collapse toward a single transition region, but the dynamic stall variable can still create a short transient delay.

4Axial pump interpretation

Axial pumps are sensitive to low axial velocity because the blade inlet angle depends strongly on the ratio between axial flow velocity and blade speed. When axial velocity decreases, the relative inlet velocity no longer matches the blade angle and separation can begin.

Relative inlet velocity $$\vec{W}_1 = \vec{V}_1 - \vec{U}_1$$ Here \(\vec{V}_1\) is the absolute inlet velocity, \(\vec{U}_1\) is blade speed, and \(\vec{W}_1\) is the relative velocity seen by the blade.

At low flow, inlet backflow and rotating stall can appear. In the axial cross-section, blue arrows represent net forward flow, while red loops represent recirculation from the downstream high-pressure side back toward the suction side.

5Centrifugal pump interpretation

In a centrifugal pump, fluid enters axially through the impeller eye and is accelerated radially outward into the volute. At low flow, the impeller still adds angular momentum, but the system does not accept the design discharge flow. This can produce inlet recirculation, discharge recirculation, blade-passage separation, and pressure pulsations.

The simulator combines these local mechanisms into one scalar recirculation variable, R. This is not a CFD model; it is a compact teaching model that shows how low-flow recirculation can produce a non-smooth pump response.

6Dynamic simulation model

The simulator uses dimensionless variables. Speed is denoted by N, flow by Q, head by H, stall level by S, and recirculation fraction by R. The model is intentionally simple enough to be transparent, but it retains the main feedback loop between speed, flow, stall, recirculation, losses, and system resistance.

Ideal pump head $$H_{ideal}=aN^2-bQ^2$$ Default values: \(a=1.20\), \(b=0.85\).
Hydraulic loss factor $$L=L_0+L_sS+L_rR^2$$ The loss factor increases with stall level and recirculation intensity.
Actual pump head $$H_{pump}=H_{ideal}(1-L)$$ Negative head is clipped to zero in the numerical implementation.
System head demand $$H_{sys}=H_{static}+K_sQ^2$$ The static head and quadratic resistance define the system curve.
Flow dynamics $$\frac{dQ}{dt}=\frac{H_{pump}-H_{sys}}{\tau_Q}$$ This is a first-order pressure-head imbalance model.
Speed dynamics $$\frac{dN}{dt}=\frac{N_{set}-N}{\tau_N}$$ The actual speed follows the selected speed setpoint with a first-order lag.
Recirculation fraction $$R=S R_{max}\max\left(0,1-\frac{Q}{Q_{design}}\right)f_N$$ The factor \(f_N=\max(0,\min(1,N/0.20))\) suppresses recirculation at near-zero speed.
Stall-state dynamics $$\frac{dS}{dt}=\frac{S_{target}-S}{\tau_S}$$ $$S_{target}=0 \quad \text{if } N>N_a \text{ and } Q>Q_a$$ $$S_{target}=1 \quad \text{if } N\(S=1\) represents a fully stalled state. \(S=0\) represents attached flow.
Qualitative efficiency indicator $$\eta=\eta_{max}(1-S)(1-0.5R)$$ This is a display indicator, not a calibrated pump-efficiency correlation.

The numerical solution uses explicit Euler integration with sub-steps of at most 0.02 s. Reset starts from zero speed and zero through-flow, so the acceleration from standstill remains visible.

7Q–N characteristic used for the plotted curve

The Q–N plot is a pedagogical visualisation of the quasi-steady response. It uses a low-flow branch, a high-flow branch, and a steep logistic transition between them.

Low-flow branch $$Q_{low}(N)=q_0+c_1N^2$$
Attached high-flow branch $$Q_{high}(N)=q_2+c_2N$$
Logistic blend $$w(N)=\frac{1}{1+\exp[-k(N-N_c)]}$$ $$Q_{curve}=(1-w)Q_{low}+wQ_{high}$$

A small oscillatory perturbation is added near the transition region to make the unstable zone visually recognisable. It should be interpreted as an illustration of hydraulic instability, not as a calibrated pump curve.

8Default parameter values

ParameterDefaultDescription
\(a\)1.20Head coefficient for speed-squared term
\(b\)0.85Head reduction coefficient for flow-squared term
\(L_0\)0.05Base hydraulic loss fraction
\(L_s\)0.45Stall-related loss coefficient
\(L_r\)0.35Recirculation-related loss coefficient
\(H_{static}\)0.05Static system head
\(K_s\)0.75Quadratic system resistance coefficient
\(\tau_Q\)0.8 sFlow response time constant
\(\tau_N\)1.2 sSpeed response time constant
\(\tau_S\)0.25 sStall-state response time constant
\(N_{attach}\)0.72Nominal attachment speed threshold
\(N_{stall}\)0.62Nominal stall-return speed threshold
\(Q_{attach}\)0.42Nominal attachment flow threshold
\(Q_{stall}\)0.28Nominal stall-return flow threshold
\(R_{max}\)0.75Maximum recirculation fraction
\(\eta_{max}\)0.85Maximum qualitative efficiency indicator

9Practical plant symptoms

In an installation, this type of behaviour can appear as a fault that is difficult to interpret if only a smooth pump curve is assumed. Typical symptoms include:

  • sudden flow increase during smooth speed ramp-up,
  • unstable differential pressure near low-flow operation,
  • vibration, noise, and pressure pulsations during start-up,
  • different transition points during ramp-up and ramp-down,
  • control-loop hunting when the setpoint is close to the transition zone,
  • apparent mismatch between measured behaviour and simplified documentation.

A practical diagnosis should compare speed, flow, differential pressure, vibration, and motor current. The key indicator is a smooth actuator or speed signal combined with a non-smooth hydraulic response.

10Model limitations

This simulator is a teaching tool. It is not a vendor pump model, a CFD calculation, or a plant-specific performance guarantee.

  • All variables are dimensionless and normalised.
  • No specific impeller geometry, blade angle, tip clearance, or inlet piping is modelled.
  • Rotating stall cells are not spatially resolved; the stall state is represented by one scalar variable.
  • Inlet recirculation, discharge recirculation, and rotating stall are combined into one recirculation fraction.
  • Cavitation is not included, although it can coexist with low-flow instability in real systems.

For plant decisions, this type of simulator should be combined with vendor curves, measured data, minimum-flow protection review, and, where needed, CFD or physical testing.

11Motor power and current

The simulator also estimates shaft power and motor current for a simplified European industrial motor supply. The default electrical basis is three-phase 400 V / 50 Hz line-to-line supply.

Hydraulic power $$P_{fluid}=\rho g H Q$$ For dimensional units: \(\rho\) in kg/m³, \(g=9.81\) m/s², \(H\) in m, and \(Q\) in m³/s.
Shaft power $$P_{shaft}=\frac{P_{fluid}}{\eta_{hyd}}$$
Electrical input power $$P_{elec}=\frac{P_{shaft}}{\eta_{motor}}$$
Three-phase current $$I=\frac{P_{elec}}{\sqrt{3}U_{LL}\cos\varphi}$$ $$I=\frac{P_{shaft}}{\sqrt{3}\,400\,\cos\varphi\,\eta_{motor}}$$ The simulator uses \(U_{LL}=400\) V, selected motor efficiency, and selected power factor.
Rated-current rule of thumb $$I_n \approx 2.0\,P_n \quad \text{A, when }P_n\text{ is in kW}$$ Example: a 15 kW motor is roughly 30 A at 400 V with typical efficiency and power factor.

During stalled low-flow operation, the motor current can be high compared with the useful hydraulic output because shaft power is dissipated internally. After the flow jump, current may settle at a higher absolute level but with more useful flow per ampere.

QuantityTypical valueNote
\(U_{LL}\)400 VLine-to-line voltage for common European low-voltage three-phase supply
\(U_{LN}\)230 VLine-to-neutral voltage, approximately \(400/\sqrt{3}\)
Frequency50 HzNominal European grid frequency
\(\cos\varphi\)0.80–0.92Typical induction-motor full-load power factor
\(\eta_{motor}\)0.88–0.96Typical industrial motor efficiency range
Starting current5–8 × \(I_n\)Typical direct-on-line starting range; VFD start is lower
VFD note: The displayed current should be interpreted as motor-side current associated with the shaft-power demand. It is not a detailed model of VFD grid-side current, harmonic distortion, or drive losses.

12Research references

  • [1]Abramian & Howard (1988)An Investigation of Axial Pump Backflow and a Method for its Control. ASME Turbo Expo. Laser-Doppler velocimetry measurements confirming reverse flow and pre-swirl upstream of an axial pump rotor at low flow. PDF ↗
  • [2]Shi et al. (2022)Influence of Inlet Groove on Flow Characteristics in Stall Condition of Full-Tubular Pump. Frontiers in Energy Research. Numerical and experimental study of saddle-zone behaviour, backflow, and vortex structures in small-flow operation. Link ↗
  • [3]Cao & Li (2020)Study on the Performance Improvement of Axial Flow Pump's Saddle Zone by Using a Double Inlet Nozzle. Water, 12(5), 1493. Rotating stall and saddle-zone instability in axial-flow pumps. DOI ↗
  • [4]Kaupert, Holbein & Staubli (1996)A First Analysis of Flow Field Hysteresis in a Pump Impeller. Journal of Fluids Engineering, 118(4), 685–691. Measured pressure-discharge discontinuities and hysteresis loops in a high specific-speed radial pump. DOI ↗
  • [5]Schiavello (2004)Abnormal Vertical Pump Suction Recirculation Problems Due to Pump-System Interaction. Texas A&M Turbomachinery Symposium. System-level discussion of recirculation and pump-system interaction. PDF ↗
  • [6]Pump Industry Magazine (2013)Understanding Pump Curves #2: Stable & Unstable Curves. Practical engineering explanation of hump and drooping pump curves. Link ↗
  • [7]IEC 60038:2009IEC Standard Voltages. Defines standard voltage levels such as 400 V / 230 V for low-voltage installations.
  • [8]IEC 60034-30-1:2014Rotating Electrical Machines — Efficiency Classes. Defines motor efficiency classes IE1–IE4.
  • [9]EU Regulation 2019/1781Ecodesign requirements for electric motors and variable speed drives.