432

International Journal of Fluid Machinery and Systems DOI: http://dx.doi.org/10.5293/IJFMS.2017.10.4.432 Vol. 10, No. 4, October-December ISSN (Online): 1882-9554

Original Paper

Unsteady Flow Condition of Centrifugal Pump for Low Viscous Fluid Food

Toru Shigemitsu1, Junichiro Fukutomi1 Takumi Matsubara2 and Masahiro Sakaguchi2

1 Institute of Technology and Science, Tokushima University 2-1 Minamijosanjima-cho, Tokushima-city, 770-8506, Japan, t-shige@tokushima-u.ac.jp

2 Graduate School of Advanced Technology and Science, Tokushima University 2-1 Minamijosanjima-cho, Tokushima-city, 770-8506, Japan

Abstract

Fluid machineries for fluid food have been used in wide variety of field i.e. transportation, filling, and improvement of quality of fluid food. The flow condition of these fluid machineries is quite complicated because the fluid food is different from water. Therefore, a design method based on the internal flow condition is not conducted. In this research, a turbo-pump having small number of blades was used to decrease shear loss and keep wide flow passage. In previous studies, it was found that internal flow condition was complex in the test pump, but those flow phenomena were not clarified in detail. In order to investigate the complex internal flow condition, the unsteady numerical analysis using low viscous fluid was conducted. In this paper, the relation between the blade geometry and the performance was investigated. In addition to that, the internal flow condition of the centrifugal pump using low viscous fluid was clarified by the numerical analysis results.

Keywords: Centrifugal pump, Numerical analysis, Unsteady flow, Low viscous fluid

1. Introduction

It is said that world population will increase to 10 billion in several decades, so large quantities and stable supply of food are needed. Furthermore, a concern of food safety is also increasing nowadays. Fluid machines for fluid food have been used in wide variety of fields i.e. transportation, filling, and improvement of quality of fluid food. The flow condition of these fluid machines is very complex and unclear because subject fluid of these fluid machines is different from air and water. Ota et al. clarified the influence of a tip clearance, a blade outlet angle and a blade number on the performance of a centrifugal pump using oil [1]-[3]. On the other hand, the internal flow condition of the pump using fluid food is not known, however, there are a few report related to the flow condition in a pipe, where the fluid food flows [4],[5]. Therefore, the design based on the internal flow condition is not conducted at present. The important issues requested for the fluid food pump are transportation performance, suction ability, flow rate control, quality maintenance, sanitary condition and so on. There are many kinds of the fluid food transportation pumps depending on the characteristics of the fluid, i.e. a sanitary pump made from stainless steel, a screw pump and so on. A centrifugal pump was adopted as a fluid food pump in this research because the stable continuous transportation of fluid food was possible and the structure was simple compared to a positive displacement pump. Cheah et al. investigated the impeller-volute interaction for the centrifugal pump by the numerical analysis [6] and Bellary et al. conducted the numerical analysis of the centrifugal pump using oils with different viscosity [7]. The purposes of this research are to establish the pump design method based on the internal flow condition of each fluid food and to clarify the boundary line of the fluid food, the centrifugal pump can apply to.

In this research, the centrifugal pump having small number of blade was used to decrease the shear loss and keep wide flow passage. Moreover, a semi-open impeller was adopted in this centrifugal pump in consideration of the manufacturing accuracy and simplification of the maintenance of the pump. Furthermore, the blade inlet and outlet angles were set as 1=2=90° to facilitate the maintenance and cleaning in the pump. We investigated the effect of a tip clearance and viscosity on the performance and internal flow condition of the fluid food pump for low viscous fluid food in the previous research [8],[9]. Then, the large vortex and flow separation were observed on a suction surface of a blade and strong unsteady flow could occur in the pump. However, the flow condition according to the rotor rotation wasn’t clarified because the steady numerical analysis was conducted in the previous research. Therefore, unsteady numerical flow analysis was conducted to clarify the complex internal flow condition of the pump.

In this paper, unsteady flow condition of the centrifugal pump using low viscous fluid food was investigated by the numerical

Received March 5 2015; revised May 3 2016; accepted for publication July 18 2017: Review conducted by Le-Qin Wang. (Paper number O15005J) Corresponding author: Toru Shigemitsu, t-shige@tokushima-u.ac.jp

433

analysis. In addition to that, the relation between the performance and blade geometry was considered.

2. Experimental and Numerical Analysis Condition

2.1 Experimental Apparatus and Methods

A specification of the pump, i.e. the design head, flow rate and specific speed was set based on commercial fluid pumps in this research. Then, an original rotor was designed using the conventional design method of a centrifugal pump and a blade outlet angle was set as 2=22.5° which was recommended by the design method. After that, a radial blade(2=90°) was adopted for a base model to pump fluid, of which viscosity was higher than that of water. Furthermore, a blade inlet angle was set as 1=90° to facilitate the maintenance and cleaning in the pump. Moreover, a semi-open impeller was used and small number of blades was adopted as z=4 in this centrifugal pump in order to suppress an increase of a disc frictional loss. A relation between the performance and the rotor having the special geometry, i.e. 1=2=90°, the semi-open impeller and small blade number, wasn’t known and it was difficult to estimate the performance of the pump with this special rotor geometry. Therefore, the design values of this pump were based on the original rotor having 2=22.5° and these values weren’t the specific design values for the base model having 1=2=90°. The design flow rate, head and rotational speed were Qd=65.3 l/min, Hd=2.5 m, and Nd=1650 min-1 based on the above assumptions. The specific speed was Ns=212 min-1, m3/min, m and type number was K=0.52. Figure 1 and Table.1 show the geometry of the impeller and the primary dimensions of the test impeller respectively. A two dimensional blade was used for the test rotor to facilitate the cleaning of the impeller and a tip clearance was c=0.5 mm. The pump used for this study had a single suction and a single volute casing and an impeller of 100mm in diameter and four blades with a straight blade profile. The impeller was designed to keep the wide flow passage of solids, fibres and so on, and to reduce the friction and shear loss. A suction and discharge diameter were Din=50mm and Dout=40mm respectively and a cross-sectional shape of a volute was rectangle and a clearance at a tongue of the volute casing was 3.7mm.

Figure 2 shows a schematic diagram of an experimental apparatus. For the pressure performance evaluation, the static head differences on walls between 2Din upstream and 2Dout downstream of the rotor were measured by pressure transducers. Then, the total pump head was evaluated by adding the dynamic head difference of the sectional averaged axial velocity to the corresponding measured static head difference. The rotor was driven by a motor. The flow rate Q was obtained by an electromagnetic flow meter installed far downstream of the pump and a torque was measured by the torque meter. Then the shaft power was calculated by the torque and rotational speed measured by the rotational speed sensor. The mechanical loss was

Fig. 1 Test pump rotor

Table 1 Primary dimensions of rotor

Outlet diameter : D2 [mm] 100

Inlet diameter at hub : D1h [mm] 50

Inlet diameter at tip : D1t [mm] 50

Inlet width : B1 [mm] 11.2

Outlet width : B2 [mm] 5.5

Blade number : Z [-] 4

Blade thickness : t [mm] 3

Blade inlet angle : 1 [°] 90

Blade outlet angle : 2 [°] 90

Ancillary pump

Torque meter

Flow control valve

Electro-magnetic water flow meter

Static pressure tapTank

Pump Motor

Booster pump

Fig. 2 Schematic diagram of experimental apparatus

434

obtained by a preliminary test before the performance test using a disc without the impeller and the mechanical loss was eliminated from the shaft power. Then, the hydraulic efficiency of the pump η was calculated as the ratio of the water power to the shaft power. Glycerol solution(40% volume density) was used in the experiment. The viscosity range used in this research corresponded to comparatively low viscous fluid foods such as beer, sake, saccharose solutions, and soy sauce.

2.2 Numerical Analysis Condition

A numerical flow analysis was conducted to investigate the internal flow condition in detail. In the numerical analysis, the commercial software ANSYS-Fluent was used and the numerical analysis was conducted with a numerical model which was the same with the test section of the experiment under the three dimensional unsteady condition. The fluid was assumed to be incompressible and isothermal water and the equation of the mass flow conservation and Reynolds Averaged Navier-Stokes equations were solved by the finite volume method. The standard wall function was utilized near the wall and SST model was used as the turbulence model. The numerical analysis was conducted at four different flow rate in 0.8Qd~1.5Qd, and the kinematic viscosity and density of the fluid were changed from the water to simulate the glycerol solution used for the experiment. Figure 3 show the numerical grids used for the numerical analysis. The inlet boundary was 5Din upstream of the test section and the outlet was 5Dout downstream of it. The tip clearance was set c=0.5mm as the same with the experimental apparatus. The constant velocity and constant pressure were given as the boundary conditions at the inlet and outlet respectively. The mesh numbers of the numerical domain was 4.27×106 grid points. The coupling between the rotor and casing was accomplished by the sliding mesh.

3. Results and Discussions

3.1 Performance of Fluid Food Pump

Figure 4 shows the pump performances using the 40vol% glycerol solution obtained by the experiment. Numerical analysis results using the fluid simulating the 40vol% glycerol solution are also shown in Fig.4. The experiments were conducted in wide flow rate range from the shut-off to large flow rates. The horizontal axis shows the flow rate. First, second and third vertical axes show the pump head, shaft power and efficiency. Focused on the experimental results, the head at the design flow rate was 3.55m, which was higher than the design head Hd=2.5m. The head curve had a flat shape from 0.5Qd -1.0Qd and a negative slope of the head curve were observed in the large flow rate region. The shaft power increased with the increase of the flow rate and showed the minimum value at the shut-off flow rate. The maximum efficiency =56.5% was obtained at 1.4Qd and the efficiency was low because of the blade inlet and outlet angle(1=2=90°) and the small number of blade of this impeller. On the other hand, the head,

Inlet

Outlet

Fig. 3 Numerical grids

Hexp.

[m]

Lexp.

[kW]

exp.

[%]

Hcal.

[m]

Lcal.

[kW]

cal.

[%]

0

20

40

60

80

100

[%

]

0

0.1

0.2

L [k

W]

0

1

2

3

4

5

0 50 100 150 200

H [

m]

Q/Qd [%]

Hexp.

[m]

Lexp.

[kW]

exp.

[%]

Hcal.

[m]

Lcal.

[kW]

cal.

[%]

0

1

2

3

4

5

0 50 100 150 200

H [

m]

Q/Qd [%]

0

20

40

60

80

100

[%

]

Hexp.

[m]

Lexp.

[kW]

exp.

[%]

Hcal.

[m]

Lcal.

[kW]

cal.

[%]

Fig. 4 Performance curve

435

obtained by the numerical analysis, was the same with the experimental result at the design flow rate and accorded with the experimental results at 4 different flow rate points. The shaft power of the numerical analysis was lower than that of the experiment and the efficiency was higher than that of the experiment as a result. On the other hand, the similar tendency was confirmed in both experiment and numerical analysis; the shaft increased according to the increase of the flow rate and the maximum efficiency flow rate point existed around 1.4Qd. Therefore, it was considered that the qualitative tendency of the experiment could be captured by the numerical analysis. Then, the unsteady internal flow condition was investigated by the numerical analysis results.

3.2 Unsteady Internal Flow Condition of Fluid Food Pump

Figures 5(a)-(d) show the cross sectional relative velocity vectors over the contours of the velocity in the impeller at the middle of the blade outlet width(b/B2=0.5) at each rotation angle. The length of the velocity vectors is constant, so it shows the flow direction and the contours show the velocity magnitude. The tip clearance was c=0.5mm and the flow rate was 1.5Qd. r is a rotation angle of a leading edge of the blade from the axial horizontal line. The rotational direction of the rotor is the counter crock-wise direction. The fluid simulating 40vol% glycerol solution was used in this numerical analysis. The fluid flowed out mainly from the pressure surface region of the blades, although the separation regions were observed on the suction surface of the blade at each rotational angle. There were two vortexes in the separation region near the suction surface of the blade. The rotational direction of the vortex near the leading edge was the counter crock-wise direction and vice versa for the vortex near the trailing edge on the suction surface. A part of the vortex from the leading edge of the blade mixed with the main flow and the fluid near the trailing edge flowed out along the suction surface of the blade. Therefore, the fluid could obtain works from the impeller several times in the test impeller. It was thought that these flow conditions were one of the reasons of the large head at the design flow rate. The size of the vortexes near the suction surface of each blade was almost the same and its size didn’t change drastically according to the rotor rotation. On the other hand, the vortex structure inside of the separation region near the suction surface of the blade changed with the rotor rotation.

Figures 6 (a) and (b) show radial and tangential velocity distribution in circumferential direction of the test impeller at the rotor outlet(r=50.2mm) using the fluid simulating 40vol% glycerol solution. The flow rate was 1.5Qd and the rotation angle was r=0°. shows circumferential position and counter crock-wise direction is a positive. =0° is a position according with a

(a) r=17.8° (b)

r=35.6°

(c)r=53.5° (d)

r=71.3°

Fig. 5 Relative velocity vectors and contours of velocity in impeller (1.5Qd)

7.0 m/s

0.0

436

horizontal line. The axial position, the data obtained, was the middle of the blade outlet width (b/B2=0.5) and the data were obtained at each 6° circumferential intervals. It was confirmed from Fig.6 that the radial velocity was large near the pressure surface of the blade and the radial velocity showed the positive value near the suction surface of the blade as could be seen in Fig.5. The back flow occurred in the blade-to-blade region and circumferential velocity was relatively large in this region. The circumferential velocity showed the maximum value near the suction surface of the blade and the minimum value near the pressure surface of the blade.

The test fluid food pump had small number of blades to facilitate the maintenance and cleaning. Therefore, the slip factor became small. It is important to estimate the slip factor of the test fluid food pump to clarify the relation between the performance and blade geometry. The slip factor at 1.5Qd in the blade width direction is shown in Fig.7. The slip factor was calculated by the following equation.

2

2

2

1u

v

u

C uSL (1)

It was difficult to distinguish the main flow and back flow region, so the slip velocity was estimated by the circumferential velocity at the outlet of the rotor including the back flow region. Further, the slip factor was compared to the slip factor obtained by the Stanitz’s equation, which was suitable for backward and radial blade(2=45-90°). Stanitz’s equation is shown as follows.

z

98.11 (2)

The slip factor of the test pump showed the large value at the mid of the blade width, however, small value near the hub and

shroud. Therefore, the slip was large near the hub and shroud and this could be caused by the influence of the viscous effect near

PS PS PS PS SS SS SS SS

-2

-1

0

1

2

3

4

5

0 50 100 150 200 250 300 350

Vr

[m/s

]

[deg]

Fig. 6 Velocity distribution in circumferential direction at blade outlet (r=50.2mm)

(a) Radial velocity distribution (b) Tangential velocity distribution

0

2

4

6

8

10

0 50 100 150 200 250 300 350

Vt

[m/s

]

[deg]

PS PS PS PS SS SS SS SS

°

0.4 0.5 0.6 0.7 0.8 0.90

0.2

0.4

0.6

0.8

1Test pump

Stanitz

Slip factor

b/B

Fig. 7 Slip Factor (1.5Qd)

437

the hub and shroud wall. Furthermore, the influence of the viscosity increased near the wall because the viscosity of the fluid simulating the glycerol solution was larger than that of water about 4 times. On the other hand, the slip factor of the test pump was larger than that obtained by the Stanitz’s equation(=0.505). Therefore, it was clarified that the slip of the test pump was smaller than that estimated by the Stanitz’s equation. This result could be associated with the adoption of the semi-open impeller and the fluid could obtain works from the impeller several times like a regenerative pump. The relation between the performance and special blade geometry of the fluid food pump should be clarified by further investigation using other types of the impeller.

Figures 8(a)-(d) show the cross-sectional static pressure contour in the test pump at the middle of the blade outlet width(b/B2=0.5) at each rotation angle. The tip clearance was c=0.5mm and the flow rate was 1.5Qd. The rotational direction of the rotor is a counter crock-wise direction. It was found in Fig.8 that the pressure in the casing increased when the trailing edge of the blade was approaching the casing tongue and decreased when it left the casing tongue. It was confirmed that this periodic pressure fluctuation in the casing occurred according to the rotor rotation. One of the causes of this pressure fluctuation could be the interaction between the outlet flow from the trailing edge of the radial blade and casing tongue. The pressure fluctuation could cause a harmful influence on the transportation performance, suction ability, flow rate control and so on. Then, we need to investigate the appropriate volute casing and clearance between the casing tongue and impeller having the special geometry, in addition to the establishment of the impeller geometry.

Fig. 8 Static pressure distribution at each rotation angle(1.5Qd)

(a) r＝17.8[°] (b) r＝35.6[°]

(c) r＝53.5[°] (d) r＝71.3[°]

0.0 Pa

-2.5×104

-5.0×104

438

4. Concluding Remarks

The performance and unsteady internal flow condition were investigated using the fluid food pump having four radial blades. As a result, the following concluding remarks were obtained.

1. The performance and slip factor, which is important to estimate design value, were clarified for the fluid food pump having the special geometry, i.e. semi-open impeller, 1=2=90° and small number of blades(Z=4). The head of the test pump having four radial blades was large and the slip factor was larger than that obtained by Stanitz’s equation(=0.505).

2. The separation regions were observed near the leading edge and trailing edge on the suction surface of the blade at each rotational angle. There were two vortexes in the separation region near the suction surface of the blade.

3. The radial velocity was large near the pressure surface of the blade and the radial velocity showed the positive value near the suction surface of the blade. On the other hand, the back flow occurred in blade-to-blade region and the circumferential velocity was relatively large in this region. The circumferential velocity showed the maximum value near the suction surface of the blade and the minimum value near the pressure surface of the blade.

4. The periodic pressure fluctuation in the casing occurred according to the rotor rotation. One of the causes of this pressure fluctuation could be the interaction between the outlet flow from the trailing edge of the radial blade and the casing tongue. Therefore, we need to investigate the appropriate volute casing and clearance between the casing tongue and impeller having the special geometry, in addition to the establishment of the impeller geometry.

Nomenclature

b B CSL c Din

Dout Hd

Nd

Ns Qd

Axial distance from hub [mm] Blade outlet width [mm] Slip velocity [m/s] Tip clearance [mm] Suction diameter [mm] Discharge diameter [mm] Design head [m] Design rotational speed [min-1] Specific speed [min-1, m3/min, m] Design flow rate [l/min]

u2

vu2

z1

2

r

Blade outlet circumferential velocity [m/s] Circumferential velocity of fluid at rotor outlet[m/s] Blade number Blade inlet angle [°] Blade outlet angle [°] Efficiency [%] Circumferential position [°] Rotation angle [°] Slip factor

References

[1] Ohta, H., Aoki, K. and Nakayama, Y., 1985, “Study on the Centrifugal Pump for High Viscosity Liquids (Effects of Clearance Ratio),” Trans. JSME (in Japanese), Vol.51, No.472, pp.4295-4300. [2] Aoki, K., Yamamoto, T., Ohta, H. and Nakayama, Y., 1985, “Study on Centrifugal Pump for High Viscosity Liquids (Effect of Impeller Output Angle on the Pump Performance),” Trans. JSME (in Japanese), Vol.51, No.468, pp.2753-2758. [3] Ohta, H. and Aoki, K., 1990, “Study on Centrifugal Pump for High-Viscosity Liquids (Effect of Impeller Blade Number on the Pump Performance),” Trans. JSME (in Japanese), Vol.56, No.526, pp.1702-1707. [4] Sandeep, K.P. and Zuritz, C.A., 1999, “Secondary Flow and Residence Time Distribution in Food Processing Holding Tubes with Bends,” Journal of Food Science, Vol.64, Issue 6, pp.941-945. [5] Dutta, B. and Sastry, S.K., 1990, “Velocity Distributions of Food Particle Suspensions in Holding Tube Flow: Experimental and Modeling Studies on Average Particle Velocities,” Journal of Food Science, Vol.55, Issue 5, pp.1448-1453. [6] Kean Wee, C., Thong See, L. and Sonny H, W., 2011, “Unsteady Analysis of Impeller-Volute Interaction in Centrifugal Pump,” International Journal of Fluid Machinery and Systems, Vol.4, No.3, pp.349-359. [7] Sayed Ahmed Imran, B. and Abdus, S., 2015, “Numerical Analysis of Centrifugal Impeller for Different Viscous Liquids,” International Journal of Fluid Machinery and Systems, Vol.8, No.1, pp.36-45. [8] Kubo, S., Fukutomi, J., Shigemitsu, T. and Okamoto, S., 2012, “Study on Performance and Internal Flow of Fluid Food Pump,” Proceedings of the 8th JSME-KSME Thermal and Fluids Engineering Conference, Incheon, Korea, FR03-007. [9] Kubo, S., Ishioka, T., Fukutomi, J. and Shigemitsu, T., 2012, “The Influence of Tip clearance on Performance and Internal Flow Condition of Fluid Food Pump Using Low Viscous Fluid,” Proceedings of the 26th IAHR Symposium on Hydraulic Machinery and Systems, Beijing, China, IAHRXXVI-191.