Turbo 137 04 041009
Turbo 137 04 041009
Turbo 137 04 041009
net/publication/279290776
CITATIONS READS
41 5,490
4 authors, including:
Amir Najafi
University of Tehran
31 PUBLICATIONS 458 CITATIONS
SEE PROFILE
Some of the authors of this publication are also working on these related projects:
All content following this page was uploaded by Mehrdad Raisee Dehkordi on 11 August 2015.
Fig. 2 (a) 3D models of total domain including (i) inlet duct, (ii) impeller, (iii) volute, and (iv)
outlet duct; (b) computational mesh for the investigated pump
H) of the volute (see Fig. 1). However, some discrepancies As mentioned in the Introduction, Kelder et al. [2] employed
between turbulence models at /=/n ¼ 0:825 are seen here, again Euler equation in their numerical predictions and found reasona-
all turbulence models returned similar results consistent with the ble agreement between computational and experimental data.
experimental data of Kelder et al. [2]. It should be noted that dis- However, level of agreement was not as good as those shown in
crepancies between computations and experimental data could be Fig. 5. For instance, at /=/n ¼ 0:825 on traverse A, RMS error of
partly due to the uncertainties in measurements. An uncertainty present CFD results is about 11% whereas it is about 25% for
analysis gives a relative uncertainty of about 3% in pressure and Kelder results [2].
head.
Fig. 5 Comparison of predicted and measured nondimensional radial and circumferential velocities in the volute along differ-
ent traverses in /=/n 5 0:825; 1; and 1:12
Fig. 8 2D models of volutes using (a) Pfleiderer design and (b) Stepanoff design
Fig. 9 Numerical results of volute pump using Pfleiderer, and Stepanoff methods for volute design
tongue, head magnitude, efficiency, and radial forces will alter purpose, it is essential to select a circumferential velocity (Cu ) dis-
less than 2.5%, 1.5%, and 8%, respectively. However, for quasi- tribution around impeller periphery in volute. Cu represents the
steady simulation of present work, in all cases, the tongue was tangential component of absolute velocity in the impeller outlet
similarly located in the midposition of two blades (h ¼ 25:7 deg). and circumferential velocity in volute. The difference of these two
methods would be made obvious far from impeller outlet in volute
where on every section Cu or rCu should be remained constant.
4.2 Comparison of Velocity Constant and Angular Mo- According to traditional design rules, two alternatives are avail-
mentum Constant Methods Volute Design. The preliminary able. First alternative, proposed by Stepanoff [11], assumes a con-
step for volute design is cross section area calculation. For this stant tangential velocity. This design results in uniform theoretical
Fig. 11 3D models and cross section dimensional parameters (in millimeters) of (a) rectangular cross section,
(b) circular cross section, and (c) trapezoidal cross section
static pressure distribution around impeller periphery at design numerical results of two new designed volutes. As expected, the
point. Second alternative proposed by Pfleiderer [12] takes con- radial force of Stepanoff design at normal point is somewhat
stant angular moment, rCu , in volute. lower than the radial force of Pfleiderer design due to uniform
Due to circumferential velocity variation, it is expected that the pressure distribution round impeller periphery. Pfleiderer design
uniformity of static pressure is deteriorated around impeller produces a slightly higher head and efficiency in low capacity that
periphery at design point using this method. In this paper, two vol- is consistent with Yang et al. [8] reports. At high capacity, the Ste-
utes are designed using the above mentioned methods. In Figs. panoff design yields far better performance characteristics than
8(a) and 8(b), 2D models of two volutes are demonstrated. In both the Pfleiderer design, and the radial force is much lower than that
volutes outlet cross sections were kept the same. Figure 9 shows of the Pfleiderer design. Figure 10 clearly displays that pressure
5 Conclusion
In this paper, the hydraulic characteristics of a low-Ns centrifu-
gal pump with various volutes were numerically investigated by
Fig. 14 2D views of (a) radial diffuser and (b) tangential means of commercial software. In the first part, CFD results of all
diffuser turbulence models were found to be in reasonable agreement with
the measurements. Based on the predictions, the k-x turbulence
model with automatic near wall treatments was chosen for further
contour around impeller in Stepanoff design is much more uni- analysis. In total, five various volutes were designed to study the
form in high capacity and this results in lower radial force genera- influence of various geometrical parameters on the head,
tion shown in Fig. 9(c). Another point of interest in Fig. 10 is the efficiency, and radial force.
low pressure region near leading edge of blades located toward In the second part, it is found that Stepanoff design generates
tongue due to dropping the local static pressure at /=/n ¼ 140%. lower radial force than Pfleiderer design at normal flowrate. Also,
This region is extended more in Pfleiderer design that leads to Stepanoff design encompasses more appropriate condition in
higher possibility of cavitation. This means that the design criteria head, efficiency, and radial load in high flowrate, though the Pflei-
of the volute do not take into account the fact that the impeller has derer design provides superior condition in low flowrate.
a not negligible interaction with the volute. In the last part, results showed that the circular cross section
volute provides higher head and efficiency than the other cross
sections in almost entire flowrate. Trapezoidal and circular cross
4.3 Design of Volutes With New Geometry
sections give lower radial force at high capacity rate while rectan-
4.3.1 Design of Volutes With Different Cross Section Shapes. gular cross section yields lower radial force at low capacity rate.
In this section, three new volutes with different cross section Another point of interest was shifting the minimum radial force
shapes, including rectangular, circular, and trapezoidal cross point of circular and trapezoidal cross section volutes to higher
sections, were probed into (see Figs. 11(a)–11(c)). All three vol- flowrate about 115% Qn of rectangular cross section volute. This
utes were designed using velocity constant method; and cross- means that the minimum radial force does not essentially occur at
sectional areas of three designs are the same. Figure 12 the best efficiency point depending on volute shape geometry. The
demonstrates pump performances obtained from CFD simulation volute with radial diffuser gave slightly more efficiency at normal
for corresponding cross sections. It is seen that circular cross sec- and high flowrate comparing to the volute with tangential diffuser,
tion volute is the best considering head and efficiency. Yang et al. also produced lower radial force at normal and low flowrate.
[8] reports similar results. However, circular and trapezoidal cross Finally, it can be concluded from the numerical results that gen-
sections have higher radial force, at design point and low flowrate. erally, Stepanoff volute design with circular cross section and ra-
At high flowrate, trapezoidal and circular volutes produce better dial diffuser shape provides better hydraulic performance and
condition. Hydraulic radii of circular and trapezoidal geometries lower radial force especially at high flowrate. In low-Ns pump, the
are higher than rectangular one which causes reduction in pressure generated radial force at high flowrate is significantly more than
loss. An interesting matter to be noticed here is the shift of mini- radial force at low capacity. This alarms that bearing selection
mum radial force point of circular and trapezoidal cross section based on zero-flow condition may be risky in low-Ns pumps.
volutes to higher flowrate equal to about 115% Qn of rectangular
volute. As shown in Fig. 12(b), it could be due to higher efficiency
and better flow regime of these volutes near 115% Qn. Acknowledgment
As displayed in Fig. 13, pressure distribution around impeller is The authors gratefully acknowledge the financial support of
the most uniform at /=/n ¼ 120% that is consistent with the low- Hydraulic Machinery Research Institute, School of Mechanical
est radial force shown in Fig. 11(c). Engineering, College of Engineering, University of Tehran.
Fig. 15 Numerical simulation result for volutes with radial diffuser and tangential diffuser
b0 a b rk rx
pffiffiffi
k 500t 4qk
arg1 ¼ min max 0 ; 2 ; (A12)
0.09 5=9 0.075 2 2.0 b xy y x CDkx rx2 y2
where constants b0 ; a; b; rk ; and rx are given in Table 3. [1] Gulich, J. E., and Favre, J. N., 1997, “An Assessment of Pump Impeller Per-
One of the advantages of k-x turbulence model is the near wall formance Predictions by 3D Navier–Stokes Calculations,” 3rd ASME Pumping
treatment for low-Re computations. The k-x model does not Machinery Symposium, Vancouver, BC, Canada, June 22–26.
[2] Kelder, J. D. H., Dijkers, R. J. H., Van Esch, B. P. M., and Kruyt, N. P., 2001,
involve the complex nonlinear damping functions required in the “Experimental and Theoretical Study of the Flow in the Volute of a Low
low k-e model and is therefore more robust. A low-Re, k-e model Specific-Speed Pump,” J. Fluid Dyn. Res., 28(4), pp. 267–280.
would typically require a near wall resolution of yþ < 0:2, while a [3] Feng, J., Benra, F. K., and Dohmen, H. J., 2010, “Application of Different Tur-
low-Re number k-x model would require the first grid node to be bulence Models in Unsteady Flow Simulations of Radial Diffuser Pump,”
Forsch. Ingenieurwes., 74(3), pp. 123–133.
at least yþ < 2. In turbomachinery flows, even yþ < 2 cannot be [4] Cheah, K. W., Lee, T. S., Winoto, S. H., and Zhao, Z. H., 2008, “Numerical
guaranteed, and for this reason, a new near wall treatment was Analysis of Impeller–Volute Tongue Interaction and Unsteady Fluid Flow in a
developed for the k-x models. It allows for smooth shift from k-x Centrifugal Pump,” 4th International Symposium of Fluids Machinery and
model to a wall function formulation regarding local yþ value. Fluid Engineering, Beijing, China, Nov. 24–27, pp. 24–27.
[5] Gonzalez, J. J., Fernandez, J., Blanco, E., and Santolaria, C., 2002, “Numerical
Simulation of the Dynamic Effects Due to Impeller Volute Interaction in a Cen-
trifugal Pump,” ASME J. Fluid Eng., 124(2), pp. 348–355.
A.2.3 SST Turbulence Model. The SST turbulence model [6] Spence, R., and Amaral-Texeira, J., 2009, “A CFD Parametric Study of Geo-
combines the advantages of the k-x and the k-e models. The main metrical Variations on the Pressure Pulsations and Performance Characteristics
problem with the k-x model is its well known strong sensitivity to of Centrifugal Pump,” Comput. Fluids, 38(6), pp. 1243–1257.
free stream conditions [14]. Depending on the value specified for [7] Jafarzadeh, B., Hajari, A., Alishahi, M. M., and Akbari, M. H., 2011, “The
Flow Simulation of Low-Specific-Speed High-Speed Centrifugal Pump,” J.
x at the inlet, a significant variation in the results of the model Appl. Math. Modell., 35(1), pp. 242–249.
can be obtained. This is undesirable and in order to solve the prob- [8] Yang, S., Kong, F., and Chen, B., 2011, “Research on Pump Volute Design
lem, a blending between the k-x model near the surface and the Method Using CFD,” Int. J. Rotating Machinery, 2011, p. 137860.
k-e model in the outer region was developed by Menter [15]. It [9] Torabi, R., and Nourbakhsh, S. A., 2011, “Hydrodynamic Design of the Volute
of a Centrifugal Pump Using CFD,” ASME-JSME-KSME Joint Fluids Engi-
consists of a transformation of the k-e model to a k-x formulation neering Conference, Hamamatsu, Shizuoka, Japan, July 24–29, pp. 423–427.
and a subsequent addition of the corresponding equations. The [10] Spence, R., and Amaral-Texeira, J., 2008, “Investigation Into Pressure Pulsa-
k-x model is thereby multiplied by a blending function F1 and the tions in a Centrifugal Pump Using Numerical Methods Supported by Industrial
transformed k-e model by a function 1 F1 F1 is equal to one Tests,” Comput. Fluids, 37(6), pp. 690–704.
[11] Stepanoff, A. J., 1957, Centrifugal and Axial Flow Pumps, Wiley, New York.
near the surface and decreases to a value of zero outside the [12] Pfleiderer, C., 1949, Centrifugal Pump for Liquids and Gases, Springer, Berlin,
boundary layer, that is, a function of the wall distance. At the Germany.
boundary layer edge and outside the boundary layer, the standard [13] Wilcox, D. C., 1986, “Multiscale Model for Turbulent Flows,” AIAA 24th
k-e model is therefore recovered Aerospace Sciences Meeting, Reno, NV, Jan. 6–9, AIAA Paper No. 86-0029.
[14] Menter, F. R., 1993, “Multiscale Model for Turbulent Flows,” 24th Fluid
Dynamics Conference, Orlando, FL, July 6–9, pp. 1311–1320.
[15] Menter, F. R., 1994, “Two-Equation Eddy-Viscosity Turbulence Models for
F1 ¼ tanh arg41 (A11) Engineering Applications,” AIAA J., 32(8), pp. 1598–1606.
DownloadedViewFrom:
publicationhttp://turbomachinery.asmedigitalcollection.asme.org/
stats on 12/02/2014 Terms of Use: http://asme.org/terms