COMPUTATIONAL FLOW ANALYSIS THROUGH A DOUBLE-SUCTION IMPELLER OF A CENTRIFUGAL PUMP

Proceedings of the Fortieth National Conference on Fluid Mechanics and Fluid Power December 12-14, 2013, NIT Hamirpur, Himachal Pradesh, India FMFP2013_141 COMPUTATIONAL FLOW ANALYSIS THROUGH A DOUBLE-SUCTION IMPELLER OF A CENTRIFUGAL PUMP Deepa Kumar Kalyan M.Tech. (Fluids Engineering) Department of Applied Mechanics MNNIT Allahabad, U.P., India Email: deepaalyan007@gmail.com Atiq ur Rehman Research Scholar Department of Applied Mechanics MNNIT Allahabad, U.P., India Email: rehman.atique5@gmail.com Ashoy Ranjan Paul Assistant Professor Department of Applied Mechanics MNNIT Allahabad, U.P., India Email: arpaul2@yahoo.co.in Anuj Jain Professor Department of Applied Mechanics MNNIT Allahabad, U.P., India Email: anujjain@mnnit.ac.in Abstract: The paper presents the behavior of flow in a double-suction impeller with five blades of a centrifugal pump using CFD code. In the various pump configuration, double-suction impeller is the one of the most complicated geometry to design and model. The geometry is modeled in a CAD software. A commercial three-dimensional CFD code- 'CFX' with a standard ε turbulence model was used to simulate the problem under different flow rates and predicted flow analysis and impeller performance. The operating characteristic curves predicted by the numerical simulation were compared with the theoretical analysis and are found in agreement. Results show that complex internal flows in centrifugal impeller can be well captured through CFD analyses. Some non-uniformities at the inlet section of the impeller is found due to blade design, especially at higher flow rate. It is observed that three dimensional steady state numerical analysis is sufficient to understand the internal flow behavior at different flow rate conditions. For each flow rate, the flow behavior, pressure distribution in the blade passage, pressure plot and velocity plots are discussed. Static pressure is observed to be increased in streamwise direction from impeller inlet to impeller outlet. Several flow circulations are observed near pressure side of the blades at low flow rate conditions considered in the study. Because of blade design, vortices are

seen on the pressure side of blades near leading edge. Keywords: double suction impeller, flow analysis, CFD-CFX, numerical simulation, pressure contour. 1. Introduction: Centrifugal Pumps are used to transfer liquids from low-pressure zones to highpressure zones, low elevation to a higher elevation, and to move liquids from one place to another. The pump impeller which is a rotating part of pump receives the pumped liquid and imparts velocity to it. The impeller is the primary component determining the pump performance. The velocity (speed) of the impeller and the diameter of the impeller will determine the head or pressure that the pump can generate. As a general rule, the velocity and the height of the impeller blades, will determine the flow rate that the pump can generate. Double suction impellers have some different design characteristics. Double suction impellers are mostly found on split case where the shaft passage completely through the impeller. This impeller is most sensitive to its geometry and the orientation. Most conventional pump impellers receive the fluid into the impeller eye at the center or inside diameter of the impeller. Double suction impellers are mostly specified for low NPSH application because area is doubled. For reducing the axial thrust, double suction impeller is used. Many pump industries are nowadays using CFD simulation tool for investigating the flow behavior and to find out the condition and causes of pump performance. Many researchers have used numerical simulation of centrifugal pumps. Denus and Osborne (2002) did some novel wor on the industrial pump impeller. They redesigned turbo machinery geometry modeling and flow simulations achieve improved impeller performance at duty point. Spence and Teixeira (2007) presented numerical results by concentrating on 15 selected locations around the pump. The regions in the pump experiencing the largest pressure pulsations are located at the impeller outlet, with large pulsations also being present in the volute at positions in close proximity to the cutwater or splitter. Stuparu et al. (2008) presented numerical study of the 3D flow in the impeller of a storage pump with double flux. The numerical simulation predicts a minimum pressure on the leading edge and towards the hub, in very good agreement with practical observation. Ushasri and Syamsundar (2010) carried out the flow simulation in a centrifugal pump impeller at five different flow coefficients and found static pressure contours, absolute velocity vectors, blade loading plots, mass averaged total pressure and static pressure, area averaged absolute velocity, pressure variation in blade to blade passage. Rajendran and Purushothaman (2012) have carried out computational analysis of centrifugal pump impeller and found the head at design flow rate. A continuous pressure rises from leading edge to trailing edge of the impeller due to dynamic head developed by the rotating pump impeller. They found the wae region at trailing edge of blade and it is seen that the total pressure loss occurs near the trailing edge of blade. Computational fluid dynamics (CFD)

analysis is being increasingly applied in the design and simulation flow in centrifugal pumps. Numerical simulation maes it possible to visualize the flow condition inside a centrifugal pump, and provides valuable information for the improvement in hydraulic design of the double-suction impeller of the centrifugal pump. By using simulation results, a great deal of labor and facility can be saved to calculate or predict the performance of a centrifugal pump, which is otherwise done by conducting experiments. Also, very few research papers are available for the hydraulic performance improvement of the double-suction impeller due to its complex geometry, which motivated the authors to tae up the present study. 2. Computational Methodology: The following sub-sections discuss on the computational methodologies used in the present study. 2.1. Geometry: In any CFD analysis, the fluid domain creation and modeling play a vital role for the solution convergence which is dependent on mesh quality. The doublesuction impeller is modeled in the Catia-v5 CAD design software. The centrifugal pump impeller model with five blades is shown in Fig. 1 while the geometrical parameters are furnished in table 1. Table1: Geometrical parameters of centrifugal pump Geometrical Parameters Dimensions Inlet eye diameter 220 mm Impeller outlet diameter 460 mm Average leading edge blade 22.5 0 angle Outlet angle 26 0 Blade number each side 5 Impeller outlet width 60 mm Vanes arrangement inline Blade thicness 7 mm 2.2 Grid Generation: The grid generation is one of the tedious and time consuming as the solution convergence criteria solely depends on the sewness of grid generation. The grid for the impeller was created in the Ansys-14.0 Worbench. Due to complexity of geometry, patch independent method is used for meshing to handle the sewness in limit. Tetrahendral element is chosen with a size of 4.8 mm. The total mesh element of impeller is is found 205771.The meshing thus generated is shown in Fig. 2. Fig. 1 3D model of double-suction impeller

dependent variables are stored at the cell centers of a uniform Cartesian mesh and the governing equations are discretized with the finite volume approach using a hybrid difference scheme. 2.4 Turbulence Models: Fig.2 mesh of double suction impeller 2.3. Governing Equations: The steady, incompressible, Reynolds Averaged Navier-Stoes (RANS) equations are employed for the flow calculations in polar coordinates and a rotating with the impeller system of reference. Continuity equation:. w 0 Momentum equation: 2 1 1 w. w 2 w. r. p.. W Here w is the relative fluid velocity, ω is the angular rotation speed of the impeller, r is the radial location and p, ρ are the fluid pressure and density respectively. The viscous stress tensor τ includes both the viscous and the turbulence viscosity terms: 2. s. w'. w' ij ij i j. In the above equation, μ is the dynamic viscosity of fluid and s ij is the strain tensor. The second term on the right side of the above relation represents the Reynolds stresses due to turbulent motion. All the There are variety of turbulence models available but one requires to choose the most suitable model depending on its applications and constraints. In the present stud, standard - ε turbulence model is used as it allows the determination of both, a turbulent length and time scale by solving two separate transport equations. The turbulence inetic energy, and its rate of dissipation ε, is obtained from the steadystate transport equations: x x x G G Y S and t ui i j j b M x x x t ui i j j 2 C G C G C S Turbulent viscosity: 1 3 b 2 The turbulent (or eddy) viscosity is t computed by combining and as 2 follows t C, where C is a constant. The model constants C, C, C, and have following 1 2 default values. C1 1.44, C2 1.92, C 0.09, 1.0, 1.3

2.5 Solution Methodology: In the impeller, the boundary conditions must be properly specified for accurate solutions. For this study, all the simulations were performed in the steady-state mode. The steady-state runs are done for a sufficient number of iterations until the flow data has converged to a constant solution. The convergence criteria were taen as 1 10 5 for all the cases. Centrifugal pump impeller domain is considered as rotating frame of reference with a rotational speed 2900 rpm. The woring fluid through the pump is water. Due to double suction, half of the total velocity applied both inlets. was used. Boundary conditions applied on the impeller are shown Fig. 3. 3. Results and Discussion: The computational results presented in this section is for the centrifugal pump impeller without any volute casing at different flow rates including design flow rate (1500 m 3 /hr). 3.1 Validation of CFD Computation: The computational results are validated with theoretical data available and is furnished in terms of operating characteristics curves in Figs. 4 and 5. Both the values are matched within acceptable limits. It shows that the hydraulic efficiency increases with the increase of flow rate and reaches the maximum value for a particular flow rate before it goes down. Fig.3 boundary condition of impeller Turbulence intensity is considered as 5%. Mass flow rate boundary condition is applied at the outlet of impeller. Except inlet and outlet, all parts of the impeller is considered wall applied with no-slip condition. The flow was assumed to be incompressible with single phase. For discretization of momentum and turbulence equation. a second-order upwind scheme Fig. 4 Discharge vs. hydraulic efficiency of a centrifugal pump Fig. 5 Discharge vs. head of a centrifugal pump

3.2 Pressure contours: There is a difference in pressure between the front face and bac face of the blade. Since front face exerts thrust, pressure is more on the front face as compared to its bac one. The pressure contours in Fig. 6 shows a continuous pressure rise from the leading edge to trailing edge of the impeller due to dynamic head developed due to rotation. As the flow rate increases, the pressure difference between inlet and outlet increases. Q=500m 3 /hr Q=1000 m 3 /hr Q=1500 m 3 /hr Q=2000 m 3 /hr Q=2500 m 3 /hr Q=500m 3 /hr Q=1000 m 3 /hr Q=1500 m 3 /hr Q=2000 m 3 /hr Fig. 7 Pressure contours at meridian plane of a centrifugal pump 3.3 Velocity contours: Q=2500 m 3 /hr Fig.6 Pressure contours on impeller of a centrifugal pump The pressure variation on the meridian plane of the impeller for various flow rates are depicted in Fig. 7. The total pressure patterns vary along the span of the impeller. Near its hub, total pressure is low. With the increase in span, total pressure are increased. In ideal case the velocity distribution in the impeller passage of a centrifugal pump impeller is uniform. But real flow through the impeller is composed of a motion displacing liquid particles from the inlet section towards the outlet section and a Circulating motion exit. Figure 8 shows that velocity in stationary frame at inlet is low and gradually increases along the outlet of the impeller.

Q=2500 m 3 /hr Fig. 9 Velocity vector at pressure side of the blade at its leading edge Fig. 8 Velocity contours on impeller of a centrifugal pump 3.4 velocity vectors at leading edge of the pressure side of blade: As the flow rate increases beyond the designed flow rate, the fluid flows smoothly along the blade profile. Due to curvature provided in the blades, a small vortex is seen at the pressure side of the blades and in the low pressure regions. Figure 9 exhibits velocity vector near the leading edge at different flow rates. It is observed that the size of vortex increases with the increase of flow rate. 3.5 Streamwise variation of total pressure in stationary frame of reference: Streamwise variation of total pressure from inlet to outlet of the impeller is shown in Fig. 10 for different flow rates. For high flow rates, a total pressure drop at leading edge is observed due to the vortex which is already shown in Fig. 9. With the increase of flow rates, a drop in total pressure is observed in Fig. 10. Q=500m 3 /hr Q=1000 m 3 /hr Q=1500 m 3 /hr Q=2000 m 3 /hr Fig. 10 Stream wise variation of total pressure at different flow rates

3.6 Streamwise variation of velocity in stationary frame of reference: The increase in area average velocity is noticed around 20-30% streamwise location from the inlet of impeller due to the dynamic energy transfer from the impeller to the fluid and is depicted in Fig. 11. Before the impeller outlet, the area averaged velocity is decreased because of the increase in static pressure at this location. of the pump. The hydraulic efficiency increases with increases the flow rate and it reached the maximum value at the designed flow rate (at 1500 m 3 /hr), beyond which it drops again. Pressure contour shows a continuous pressure raise from leading edge to trailing edge of the impeller. Velocity in stationary frame at inlet is low and gradually increases along the outlet of the impeller. Due to curvature of the blade, vortex at the pressure side of the blade at leading edge section is observed. With the increase of flow rate, a drop in total pressure is noticed. Area averaged velocity decreases because of increase in pressure at the outlet of impeller. References: Fig. 11 stream wise variation of velocity at different flow rates CONCLUSIONS A double-suction centrifugal pump impeller was modeled and analyzed with the help of CFD without volute casing. The performance results in terms of head and hydraulic efficiency, pressure contours, velocity contours, velocity vectors on pressure side of blade at leading edge, streamwise variation of mass averaged total pressure, streamwise variation of area averaged absolute velocity are predicted for five different flow rates. The increase of flow beyond the designed flow rate causes a reduction in the total head [1]. Denus, K. and Osborne, C., 2002, Hydraulic development of a Centrifugal pump impeller using the AGILE Turbomachinery design system, TASK Quarterly, vol. 6, no.1, pp.79-100. [2]. Spence R.R.G. and Teixeira, J.A., 2007, A CFD analysis of a complete double entry Centrifugal pump, Proceedings of the 15th UK Conference of the Association of Computational Mechanics in Engineering, Glasgow, United Kingdom, April 2-3. [3]. Stuparu, A., Muntean, S., Baya, A. and Anton, L.E., 2008, 3D numerical investigation of flow through the Centrifugal pump with double flux, Proceedings of the 3rd Worshop on Vortex Dominated Flows, Timisoara, Romania, pp. 75-80. [4]. Rajendran, S. and Purushothaman, K., 2012, Analysis of a centrifugal pump impeller using ANSYS-CFX, International Journal of Engineering Research & Technology (IJERT), vol.1, issue 3.

[5]. Ushasri, P. and Syamsundar, C., 2010, Computational analysis on performance of a Centrifugal pump impeller, Proceedings of National Conference on Fluid Mechanics & Fluid Power (FMFP-2010), Paper no. FMFP10- TM-07, IIT Madras, Chennai, India.