In industries such as chemical processing, water supply and drainage, and power generation, centrifugal pumps are rightly called the "heart of fluid transport" – from supplying water to residential buildings to transferring materials in chemical plants, their efficient operation is indispensable. However, details hidden inside the metal casing – how fluid flows within a centrifugal pump, where it might "get stuck", or how design can improve energy efficiency – cannot be understood simply by "disassembling the machine".
Today, we will use a complete CFD (Computational Fluid Dynamics) simulation case to thoroughly dissect the "inner workings" of a specific centrifugal pump model: from geometric modeling and meshing to convergence verification and flow field analysis. By the end, you will understand exactly how simulation helps engineers "remotely diagnose" equipment performance.
The subject of this simulation is a single-stage centrifugal pump (a common type with only one set of impellers), typically used for transporting clean water-like media.
Why perform the simulation? Engineers found that the actual energy consumption of this pump was about 4% higher than its design value, and "abnormal noise" occurred under low-flow conditions – suspecting that uneven internal flow fields, the presence of vortices, or localized high-speed zones were causing energy loss or cavitation risks.
Our simulation objectives are clear:
1. Recreate the real internal flow field of the centrifugal pump, visualizing fluid velocity distribution and motion trajectories.
2. Locate "energy loss points" within the flow channels (e.g., vortices, sudden velocity change zones).
3. Provide data support for subsequent flow channel optimization.
To make fluid "flow inside the computer", we must first build the "digital container" – the geometric model and mesh, which form the "foundation" of CFD.
1. Geometric Modeling: A 1:1 Replica of the Pump's "Core Flow Channels"
First, CAD software was used to construct the geometry of the pump's core flow channels, focusing on accurately recreating two key components:
● Impeller: The core where work is done on the fluid (blade rotation imparts kinetic energy to the fluid).
● Volute Casing: The channel that collects the fluid and converts kinetic energy into pressure energy (the channel cross-section gradually expands circumferentially).
*(Corresponding Image 1: Mesh-display model of the 3D geometry of the centrifugal pump flow channels)* Note: The geometric dimensions are identical to the physical part, especially parameters like blade angles and volute cross-section expansion rate – a deviation of even 1 mm in these could skew the flow field results.
2. Meshing: "Weaving the Net" for the Flow Channels – A Balance of Detail and Efficiency
After geometry creation, the flow channels must be divided into millions of "small cells" (the mesh), within which the software calculates the fluid state. Our meshing strategy was "refinement on demand":
● Core Zone Refinement: The impeller is a region of complex, high-speed rotational flow, so the impeller area was "mesh-refined" to accurately capture flow details.
● Non-Core Zone Simplification: The volute flow is relatively gentle, allowing for slightly larger mesh cells to balance computational accuracy and efficiency (avoiding excessively long computation times).
● Quality Assurance: Metrics like "orthogonality" and "skewness" were checked. The mesh quality pass rate for this case was ≥95%, preventing result distortion due to poor mesh quality.
(Corresponding Image 2: Mesh model of the centrifugal pump showing refinement in the impeller region)
Setting the "rules" for the fluid: telling the software "how the fluid moves, where it enters and exits".
1. Governing Equations & Turbulence Model
Fluid flow obeys the principles of "mass conservation + momentum conservation", therefore the Reynolds-Averaged Navier-Stokes (RANS) equations were used. Flow inside centrifugal pumps is "turbulent" (chaotic fluid mixing), so the k-omega turbulence model was selected – it is well-suited for simulating turbulence in rotating machinery, accurately calculating "turbulent kinetic energy (k)" and "specific dissipation rate (omega)".
2. Boundary Conditions: Defining the Fluid's "Start and End Points"
● Inlet: Velocity Inlet (set to a fixed value based on the pump's rated flow rate).
● Outlet: Pressure Outlet (corresponding to the pump's actual operating pressure).
● Impeller: Rotating Region (using the MRF – Multiple Reference Frame method to simulate the flow field during impeller rotation).
● Walls: No-Slip Condition (fluid velocity at the wall is zero).
Once the simulation runs, it's not enough to just wait for it to finish – residual curves must be monitored to judge result reliability.
(Corresponding Image 3: Residual curves from the centrifugal pump simulation)
Residuals represent the "calculation error of the equations":
● continuity: Residual for the continuity equation (mass conservation).
● x/y/z-velocity: Residuals for the momentum equations.
● k/omega: Residuals for the turbulence model equations.
In this simulation, all curves dropped below 1e-4 and stabilized – indicating the calculation converged and the results are reliable (not a product of "random computation").
Finally, the crucial step of "looking inside"! Using velocity contours and vector plots, we dissect the secrets within the flow field.
1. Velocity Distribution: From "High-Speed Energy Addition" to "Deceleration for Pressure Rise"
The velocity contour reveals:
● Impeller Zone: Significantly higher velocities (up to 15 m/s) – impeller rotation "pushes" the fluid, imparting kinetic energy.
● Volute Zone: After exiting the impeller, as the volute passage widens, velocity gradually decreases (down to 1.5 m/s) – this is the "diffuser effect" of the volute: converting fluid kinetic energy into pressure energy, forcing the fluid "out under pressure".
(Corresponding Image 4: Internal velocity contour of the centrifugal pump)
2. Flow Trajectories: From "Spiral Ejection" to "Smooth Forward Flow"
Velocity vector plots show: fluid "flung" out by the impeller advances spirally along the volute passage. The overall trajectory is smooth, but a slight velocity gradient exists at the junction between the impeller and volute (a potential point of energy loss).
(Corresponding Image 5: Internal velocity vectors of the centrifugal pump)
3. Operating Condition Comparison: The Flow Field's "Behavior" Under Different Flow Rates
We also simulated three conditions: "rated flow, high flow, low flow".
● High Flow: Velocity distribution in the impeller is more uniform, but velocity fluctuations at the volute outlet are slightly larger.
● Low Flow: Local vortices appear at the impeller edges – this is the root cause of the "low-flow abnormal noise": vortices cause fluid to "impact the walls", wasting energy and generating vibration/noise.
(Corresponding Image 6: Velocity contours of the centrifugal pump under different operating conditions)
All this analysis ultimately aims to "make the pump perform better". Through this simulation, we identified 2 key optimization points:
1. Junction between Impeller and Volute: Smooth the transition of the flow channel to reduce the velocity gradient.
2. Volute Cross-Section Expansion Rate: Adjust the rate of cross-sectional expansion locally to achieve a more uniform velocity reduction.
After optimization, this pump's efficiency increased by 6%, energy consumption decreased by 8%, and the "low-flow noise" issue was largely eliminated – equivalent to achieving "cost reduction and efficiency improvement" by merely modifying the flow channel shape, without replacing any parts.
This is the power of CFD simulation: using digital means to "test and iterate in advance", saving up to 90% of the time and cost compared to physical experiments.
