2nd International Conference on Experiments/Process/System Modelling/Simulation & Optimization 2nd IC-EpsMsO Athens, 4-7 July, 2007 IC-EpsMsO

ON THE INFLUENCE OF THE OUTLET BLADE ANGLE IN THE PERFORMANCE OF A RADIAL PUMP IMPELLER
Michalis D. Mentzos1, Evangelos. C. Bacharoudis1, Andronikos E. Filios*, and Dionissios P. Margaris1 1 Fluid Mechanics Laboratory, Mechanical Engineering and Aeronautics Department University of Patras GR-26500 Patras, Greece e-mail: margaris@mech.upatras.gr
*

Fluid Mechanics & Turbomachinery Laboratory, Department of Mechanical Engineering Educators, School of Pedagogical and Technological Education, GR-141 21 Athens, Greece e-mail: fmtulab1@otenet.gr

Keywords: Centrifugal pump, impeller, CFD. Abstract. With the aid of computational fluid dynamics, the complex internal flows in water pump impellers can be well predicted, thus facilitating the design of pumps. The pump performance and energy consumption is affected by various design parameters among which impeller outlet diameter, blade angle and blade number are the most critical. The present paper describes the simulation of the flow into the impeller of a centrifugal pump in a parametric manner. In this study, the performance of impellers with the same outlet diameter having different outlet blade angles (2= 20, 30 and 50 deg) is thoroughly evaluated. The one-dimensional approach along with empirical equations is adopted for the design of each impeller. The predicted performance curves result through the calculation of the internal flow field and a successful correlation of local and global parameters. The numerical solution of the discretized three-dimensional, incompressible Navier-Stokes equations over an unstructured grid is accomplished with a commercial CFD finite-volume code. For each impeller, the flow pattern and the pressure distribution in the blade passages are calculated and finally the head-capacity curves are compared and discussed.

1 INTRODUCTION Pump designers are continually challenged to provide machines that operate more efficiently, quietly, and reliably at lower cost. The design of a hydraulic turbomachine premises a detailed understanding of the internal flow within its stationary and rotating passages prior to its performance calculation during design and off-design conditions. With the aid of the Computational Fluid Dynamics (CFD) approach, the complex internal flows in water pump centrifugal impellers, which are not fully understood yet, can be well predicted and therefore establishing the CFD as a key tool for pump designers. The use of CFD in turbomachinery industry is quite common today since many tasks can numerically be solved much faster and cheaper than by means of experiments. Especially, commercial CFD software, is increasingly using to study pump design and off-design performance. Various researchers have considerably contributed in revealing the flow mechanisms inside centrifugal impellers aiming to the design of high performance centrifugal turbomachines. The reported works by Eckardt[1], Johnson and Moore[2], Kjork and Lofdahl[3], Denton[4], Dawes[5], Casey et al[6], Bansod and Rhie[7], Krain and Hoffman[8], Farge and Johnson[9] and Zhang et al[10] are an indicative collection of research efforts on the computation and the experimental verification of the flowfields within centrifugal impellers. On the numerical simulation of the flowfield of a centrifugal impeller, several algorithms have been proposed and developed, but there isnt still a robust approach distinguished for its modeling accuracy along with the efficient computation. Pressure-based methods initially developed for the incompressible flow regime, obtain the pressure field through a pressure or a pressure correction equation which is formulated by manipulating the continuity and momentum equations. The solution procedure is conventionally sequential in nature, and hence can more easily accommodate a varying number of equations depending on the physics of the problem involved, without the necessity of reformulating the entire algorithm. Lakshminarayana[11], Rodi et al[12], Thakur et al[13], provides a review of the techniques that are useful as an assessment of the state of the art.

Michalis D. Mentzos, Evangelos. C. Bacharoudis, Andronikos E. Filios, and Dionissios P. Margaris.

Denomination Suction pipe diameter Impeller diameters Impeller widths Impeller angles Number of blades Flow rate in best efficiency point, rpm Specific speed (nq = n Q1/2/H3/4)

Value Ds = 100mm D1 = 150mm, D2 = 280mm b1= 20mm, b2= 20mm 1 = 14, 2 =20, 30, 50 z=6 Q = 45m3/h, n = 925rpm nq = 18.4

Table 1. Main design data of the experimental radial pump impeller The present paper summarizes the under progress research work regarding the flowfield prediction an experimental centrifugal pump as well as its performance curves. Towards to study the design and off-design performance of the pump, flow and pressure fields are analyzed numerically prior to the comparison of the computational and experimental data. The experimental radial pump which is designed according to Pfleiderer method consists of a shrouded impeller of constant width rotating in a volute of rectangular section with a diffuser in the radial direction. Initially, three impellers with six untwisted blades backward facing with a circular arc camber line are studied. All impellers have the same diameters in suction and pressure side and they vary on the outlet blade angle which is 2=20, 30 and 50 deg, respectively. The main design data of the experimental pump impeller are shown in Table 1. All gaps between the impeller shroud and the pump casing are neglected in the present initial numerical simulation. In the lack of completing the experimental study, only the computational study of the flow through the impeller with the use of Fluent is discussed. The commercial software package Fluent has been widely used in the field of turbomachinery, and the simulation results have been proven by Sun and Tsukamoto[14] and Gonzalez et al.[15] to be reliable. Fluent overcomes the meshing difficulties that arise in complex geometry by using a powerful CAD-based preprocessor, Gambit. The authors have already used Fluent to study three-dimensional turbulent flow through a commercial water-pump during design and off-design conditions [16, 17]. 2 COMPUTATIONAL MODELING 2.1 Geometry and grid A geometrical discretization of the centrifugal pump is made for the numeric treatment. The whole domain consists of three sub-domains. The first and third zones are stationary while the second zone that incorporates the blade is moving with the applied rotational speed of n=925 rpm. The first zone represents the suction or inlet pipe and the third zone is the discharge or outlet portion where the flow is fully developed with a less possible reacting outlet boundary condition. Generally all the domain is separated in sub-domains with some additional inner faces were introduced, in a way that the density and the quality of the cells in local flow field regions can be suitably controlled and handled depending on pressure gradients and velocities.

(a)

(b)

Figure 1. Sketch of the pump unstructured mesh (a) and details of the pump mesh in the region of the tongue (b)

Michalis D. Mentzos, Evangelos. C. Bacharoudis, Andronikos E. Filios, and Dionissios P. Margaris.

Unstructured wedge are generated to define the inlet and outlet zones. An unstructured mesh but with tetrahedral cells are also used for the zones of impeller and volute (Figure 1a). In the near tongue region of the volute, a mesh refinement applied just as and for some other regions such as the inlet and outlet of the blade (Figure 1b). Structured hexahedral cells are generated all around the blade. The size of the resulting cells is not adequate for a full boundary layer simulation, but however it provides correct values for the pump performance and allows the analysis in details of the main phenomena involved. 2.2 Description of numerical simulation procedure All the calculations have been performed with Fluent. The code uses the finite volume method and solves the fully 3D incompressible Navier-Stokes equations, including the centrifugal force source in the impeller and the unsteady terms. Turbulence is modeled selecting the standard k- model. The involved parameters regarding the turbulence intensity and the hydraulic diameter in the lack of realistic turbulent inflow conditions in industrial applications are estimated with values of 5% and D/2 respectively. The pressure-velocity coupling is performed through the SIMPLE algorithm. Second order, upwind discretization have been used for convections terms and central difference schemes for diffusion terms. The modeled boundary conditions are those considered most physically meaningful for turbomachinery flow simulations and those which give a flow neither limited nor restricted by them. At the inlet and exit pipes, there is an unavoidable effect on the final flow solution as a result of the boundary conditions. A reasonable length must be added to real machine geometry to avoid this effect as much as possible and to better simulate the pumping circuit influence. At the inlet zone, the axial velocity is a constant based on the through flow for the pump. The absolute tangential velocity at the inlet is zero, which implies in the rotating frame the relative velocity is r, and the radial velocity is zero. At the exit of the discharge nozzle, assuming a fully developed turbulent flow, a practical zero velocity gradients was set. The walls of the model are stationary with respect to their respective frame of reference, and the no slip condition is applied. Although grid size is not adequate to investigate local boundary layer variables, global ones are well captured. For such calculations wall functions, based on the logarithmic law, have been used. 2.3 Numerical solution control The code was run in a 3GHz Pentium IV PC. The number of iterations has been adjusted to reduce the scaled residual below the value of 10-5 which is the criteria. For each run, the observation of the integrated quantities of total pressure, at suction as well as at discharge surface was appointed for the convergence of the solution. In many cases this drives the residuals in lower values than the initially set value. Depending on the case, the convergence was achieved at difference iterations, as the result at a specific mass-flow was used to initialize the computations at another mass-flow. Aiming to smooth convergence, various runs were attempted by varying the under-relaxations factors. In that way a direct control regarding the update of computed variables through iterations was achieved. Initializing with low values for the first iterations steps and observing the progress of the residuals, their values were modified for accelerating the convergence. Since the problem involves both stationary and moving zones, the multiple reference frame (MRF) model was selected. It is a quasi-steady state approximation in which individual cell zones move at different rotational speeds. As the rotation of the reference frame and the rotation defined via boundary conditions can lead to large complex forces in the flow, calculations may be less stable as the speed of rotation and hence the magnitude of these forces increases. To control this undesirable effect, each run starts with a low rotational speed and then slowly is increased the rotation up to desired level. 3 RESULTS

The outlet angle of three different impellers (2=20, 2=30 and 2=50) is correlated with the slope of the performance characteristic curve of each impeller. Increasing the outlet blade angle, the shape of the characteristic curve becomes smoother and flatter. This fact is expected and is consistent with theory, i.e.
H = nH
2 c u2 (1 m 3 cot 2 ) g u2

(1)

The above equation enables the ascertainment of the effect of the two design parameters cm3/u2 and 2 on the total head. If cm3/u2 is reduced, the head increases. The effect of the vane discharge angle is not so straightforward. If 2 is increased cot2 is decreased and the head become larger. But the slip factor formulas indicate that larger angles 2 decrease the slip factor . Therefore the effects of changes in 2 are partly canceled. The change of the shape of the curve is noticeable for blade angles greater than 25o. The later is in agreement with the results of the numerical prediction of the H-Q curves for the examined impellers which are shown in Figure 2.

Michalis D. Mentzos, Evangelos. C. Bacharoudis, Andronikos E. Filios, and Dionissios P. Margaris.

10,0

9,5 H (m)

9,0 2=20 deg 2=30 deg 2=50 deg

8,5

8,0 10 20 30 40 Q (m3/h) 50 60 70

Figure 2. H-Q curves for the examined pump impellers. The pump is designed to operate at a flow rate Q=45 m3/h with H=10 m head, when the rotational speed is n=925 rpm. The numerical simulations of the pump predict with very well accuracy the total head which corresponds to the prescribed operating conditions. The static pressure field for the plane near the back shroud is presented in the Figure 3. It is obvious that the static pressure drops from the pressure side to the suction side of the impeller blade. Moreover the value of the static pressure drop between the pressure and suction side is reduced at the exit of the blade passage. The view of static pressure field appears different for the hub and shroud plane of the impeller. That alteration is more evident between the shroud and middle span plane while becomes negligible between middle span and hub plane. Such pressure distribution in the actual pump does not entail any additional losses, it simply means that each vane can only transmit, and the liquid can only absorb a fixed amount of energy. This is lower than that given by Eulers equation. We can observe the same behavior either when the exit blade angle 2=20o or 2=50. The only difference between them is the values of static pressure that developed at the flow domain at each case. From the observation of the Figure 3 it is evidence that the bigger values correspond to the impeller with 2=50. The minimum value of the static pressure inside the impeller is located at the leading edge of the blades at the suction side except for the blade with trailing edge which end up at the region of the tongue. For that case (Figure 4) the minimum pressure is located at the leading edge of the blade at the pressure side. This is explained by the interaction between the impellers blade and the volutes tongue. A close examination of the relative velocity field at the blade leading edge reveals that relative velocity is augmented which imply reduced static pressure. Hence, the minimum pressure is observed at these regions. The same situation stands for flow rates greater or lower than the nominal one, as also for the difference exit angle. As expected from other published data [18], the position of stagnation represented to be placed at the middle of the tongues radius (Figure 5). Furthermore the outlet blade angle affects and changes the relative velocity field. Specifically as the outlet angle increases a recirculation zone of fluid appears at the trailing edge of the blade (Figure 6).

Figure 3. The static pressure contours (atm) at nominal flow (a) 2=20 (b) 2=50

Michalis D. Mentzos, Evangelos. C. Bacharoudis, Andronikos E. Filios, and Dionissios P. Margaris.

Figure 4. The static pressure (atm) overlay with the relatives velocities (m/s) at the leading edge of the blade which end up at the tongue

Figure 5. The stagnation point at the nominal flow at the tongue

(b) (a) o o Figure 6. The relatives velocities (m/s) at the trailing edge of the blade with: a) 2=20 and b) 2=50

(a)

(b)

(c)

Figure 7. The absolute velocities at the region of leading edge for a) Q=QN b) Q=0.6xQN c) Q=1.45xQN in the front and back shroud of the impeller with 2=20

Michalis D. Mentzos, Evangelos. C. Bacharoudis, Andronikos E. Filios, and Dionissios P. Margaris.

(a)

(b)

Figure 8. Secondary flow at nominal flow in different planes along the volute. a) 2=20 and b) 2=50
2=20

(a) 2=50

(b)

(c)

(a)

(b)

(c)

Figure 9 The static pressure (atm) for a) Q=0.6xQN b) Q=QN c) Q=1.45xQN The developed velocity field at the impeller middle-span plane displays significant variations between the nominal flow rate Q=QN, the over flow rate Q=1.45QN and the part flow rate Q=0.6QN. For the nominal flow rate and higher the fluid flows smoothly through the impeller passage. However, when the flow rate is reduced from the nominal one, recirculation zone is established. At the leading edge of each blade in the impeller, eddies appear as is shown in Figure 7. The flow field is not the same at all sections from hub to shroud. Especially, it changes with the width of the impeller mainly in the region of the leading edge. At the exit of passage the flow field becomes symmetry and uniform. This phenomenon is explained partly by the fact that varying boundary layers are created adjacent the hub and shroud due to the different shape of these walls. Another reason is that the entrance of our fluid is impossible to achieve with the same way in all the cross section of the suction. It is obvious that the elements of fluids at the inside of turn have smaller velocities than that at the outside of curvature. The above variations are more significant as the flow rate of the pump is reduced. This is anticipated as at reduced flow rate we have the presence of inlet recirculation and is located at the side of the suction shroud, a phenomenon that makes the situation more complicated. The presence of secondary flows[19] along the duct of the volute is also detected and it is shown in Figure 8. As the fluid approaches the discharge, the eddies maintain their existence with small changes of their size and the position of their core. The static pressure variation within the pump for three flow rates and for two different outlet blade angles, i.e. 20deg and 50 deg, is shown in Figure 9. The pressure forces seem to be the main driven mechanism to establish the flow features both in the impeller and volute.

Michalis D. Mentzos, Evangelos. C. Bacharoudis, Andronikos E. Filios, and Dionissios P. Margaris.

CONCLUSIONS

The flow and pressure field through a centrifugal pump impeller is predicted solving the fully 3d incompressible Navier-Stokes equations, including the centrifugal force source. Turbulence is simulated with the standard k- model. Although grid size is not adequate to investigate local boundary layer variables, global ones are well captured. The present approach is useful for basically understanding the flow states in various operating points. Further work will focus on validating the CFD results with the planned experiments in the near future. REFERENCES [1] Eckardt, D. (1976), Detailed flow investigations within a high-speed centrifugal compressor impeller, ASME J. Fluids Eng. 98, pp.390-402. [2] Johnson, M. W. and Moore, J., (1980), The development of wake flow in a centrifugal impeller, ASME J. Eng. for Power 102, pp.382-390. [3] Kjork, A. and Lofdahl, L., (1989), Hot-wire measurement inside a centrifugal impeller, ASME J. Fluids Eng. 111, pp.363-368. [4] Denton, J.D. (1992), The calculation of three-dimensional viscous flow through multistage turbomachinery. ASME J. Turbomachinery 114, pp.18-26. [5] Dawes, M.W. (1992), Toward improved through flow capability: the use of three-dimensional viscous flow solvers in a multistage environment, ASME J. Turbomachinery 114, pp.8-17. [6] Casey, M.V., Dalbert, P. and Roth, P. (1992), The use of 3D viscous flow calculations in the design and analysis of centrifugal compressors, ASME J. Turbomachinery 114, pp.27-37. [7] Bansod, P. and Rhie, C.M. (1990), Computation of flow through a centrifugal impeller with tip leakage, AIAA Paper No 90-2021. [8] Krain, H. and Hoffman, W. (1989), Verification of an impeller by laser measurement and 3D viscous flow calculations, ASME Paper 89-GT-150 [9] Farge, T.Z. and Johnson, M.W. (1990), The effect of backswept blading on the flow in a centrifugal compressor impeller, ASME Paper 90-GT-231. [10] Zhang, M.J., Gu, C.G. and Miao, Y.M. (1994), Numerical study of the internal flow field of a centrifugal impeller, ASME Paper 94-GT-357. [11] Lakshminarayana, B. (1991), An assessment of computational fluid dynamic techniques in the analysis and design of turbomachinery, ASME J. Fluids Eng. 113, pp.315-352. [12] Rodi, W., Majumdar, S. and Schonung, B. (1989), Finite volume methods for two-dimensional incompressible flows with complex boundaries, Computer Meth. in Applied Mechanics and Engineering 75, pp.369-392. [13] Thakur, S., Wright, J., Shyy, W., Liu, J., Ouyang, H. and Vu, T. (1996), Development of pressure-based composite multigrid methods for complex fluid flows. Prog. Aerospace Science, Vol. 32, pp.313-375. [14] Sun, J., and Tsukamoto, H. (2001), Off-design performance prediction for diffuser pumps. Journal of Power and Energy, Proceedings of I. Mech. E., Part A, Vol. 215, pp. 191201. [15] Gonzalez, J., Fernandez, J., Blanco, E., and Santolaria C. (2002), Numerical simulation of the dynamic effects due to impeller-volute interaction in a centrifugal pump, ASME Journal of Fluids Engineering, Vol. 124, pp. 348-355. [16] Mentzos, M.D., Filios, A.E., Margaris, D.P. and Papanikas, D.G. (2004), A numerical simulation of the impeller-volute interaction in a centrifugal pump, 1st International Conference From Scientific Computing to Computational Engineering, Athens, 8-10 September, pp.1-8. [17] Mentzos, M.D., Filios A.E., Margaris, D.P. and Papanikas, D.G. (2005), CFD predictions of flow through a centrifugal pump impeller, 1st International Conference on Experiments/Process/System Modelling/Simulation/Optimization, Athens, 6-9 July, pp.1-8. [18] Gonzalez, J., Parrondo, J., Santolaria, C. and Blanco, E. (2006), Steady and unsteady forces for a centrifugal pump with impeller to tongue pump variation, ASME Journal of Fluids Engineering, Vol. 128, pp. 454-462. [19] Gonzalez, J., and Santolaria, C. (2006), Unsteady flow structure and global variables in a centrifugal pump, ASME Journal of Fluids Engineering, Vol. 128, pp. 937-946.