This article was originally published in a journal published by Elsevier, and the attached copy is provided by Elsevier for

the authors benet and for the benet of the authors institution, for non-commercial research and educational use including without limitation use in instruction at your institution, sending it to specic colleagues that you know, and providing a copy to your institutions administrator. All other uses, reproduction and distribution, including without limitation commercial reprints, selling or licensing copies or access, or posting on open internet sites, your personal or institutions website or repository, are prohibited. For exceptions, permission may be sought for such use through Elseviers permissions site at: http://www.elsevier.com/locate/permissionusematerial

Experimental Thermal and Fluid Science 31 (2007) 505512 www.elsevier.com/locate/etfs

Ming-Jyh Chern *, Chin-Cheng Wang, Chen-Hsuan Ma

Received 20 January 2005; accepted 27 April 2006

Abstract The performance, ow patterns and cavitation phenomena of a ball valve are studied experimentally. Various patterns of ows in and downstream the ball valves with respect to dierent valve openings and inlet velocities are visualized using a particle tracking ow visualization method (PTFV). Meanwhile, cavitation phenomena are observed under certain conditions. Coecients regarding to the performance of valve are determined by pressure and ow rate measurements. The correlations between the valve performance and the ow patterns are presented and discussed. The proposed method provides an eective way to determine the performance coecients of a valve and to understand the condition for the inception of cavitation. 2006 Elsevier Inc. All rights reserved.
Keywords: Ball valve; Flow visualization; Cavitation

pe

1. Introduction

Valves have been used widely in various industries. There are many purposes for valve usage in a piping system. One of them is to control the ow rate. For a manufacturer of ball valves, it is very important to know the characteristics of ows inside a ball valve. In addition to the ow coecient, there are other points which a manufacturer must consider. For example, the cavitation due to local low pressure has to be avoided. Cavitation causes noise and vibration during operation of a ball valve. For a long term operation, it may cause serious damage to a piping system. In the past, it was dicult to investigate details of the ow inside a ball valve, because it is not transparent. Information regarding the ball valve performance was obtained from pressure and ow rate measurements. Hutchison [1], Kirik and Driskell [2] and Pearson [3] provided useful information on the design of a ball valve. Ota and Itasaka [4] measured the surface pressure distribution behind a blunt body to understand the structure of the recirculation
*

Corresponding author. Tel.: +886 2 27376496; fax: +886 2 27376460. E-mail address: mjchern@mail.ntust.edu.tw (M.-J. Chern).

0894-1777/$ - see front matter 2006 Elsevier Inc. All rights reserved. doi:10.1016/j.expthermusci.2006.04.019

Au

th o

r's

rs
behind a blunt body. In addition, Kelso et al. [5] also measured the surface pressure distribution and investigated the balance between the pressure distribution and Reynolds shear stress along the separation streamline and the surface behind a surface-mounted blunt plate. If the ow structure in a ball valve is available, then it would help engineers improve the performance of ball valves. Due to progress in the areas of ow visualization and computational analysis, it is now possible to observe the ow inside a ball valve. The ow visualization can provide ow patterns of a ball valve, especially the cavitation phenomenon. Structures of vortices in ows inside a ball valve can be obtained from the results of ow visualization. These vortices mainly determine the energy loss (or the pressure drop). According to the information obtained, cavitation can be predicted using the cavitation coecient. Many previous researches have considered the ow inside a valve. Computational approaches are becoming popular in this area. For example, Kerh et al. [6] utilized the nite element method to simulate transient interaction of uid and structure in a control valve. Mertai et al. [7] adopted a commercial package, FLUENTTM, to investigate the ow around a V-sector ball valve. Van Lookeren Campagne et al. [8] also used a commercial package, AVL-FireTM, to simulate ows

on

al

co

Department of Mechanical Engineering, National Taiwan University of Science and Technology, 43 Sec. 4 Keelung Road, Taipei 106, Taiwan

py

Performance test and ow visualization of ball valve

506

M.-J. Chern et al. / Experimental Thermal and Fluid Science 31 (2007) 505512

Nomenclature Ccs Cv D g K P Pv cavitation index, Eq. (3) ow coecient, Eq. (2) diameter of straight pipe, m gravitational acceleration, m s2 loss coecient, Eq. (1) pressure, bar saturated vapor pressure, Pa q ReD Ui a q m / volumetric ow rate, m3 h1 iD Reynolds number, Um averaged velocity at the inlet boundary, m s1 angle of ball valve density of water, kg m3 kinetic viscosity of water, m2 s1 opening of ball valve

2. Experimental conguration

Fig. 1. Schematic diagram of experimental apparatus.

Au

th o

Fig. 1 shows the experimental facility for both the valve performance measurement and the ow visualization. The clean water contained in a reservoir tank of 600-l is drawn by a high pressure pump and delivered to a pressurized buffer tank. The water ows sequentially through a pressure regulator, upstream ow conditioning section, transparent valve section, downstream ow conditioning section, ow meters, and lters, then recirculates back to the reservoir

r's

pe

containing bubbles in ball valves. Davis and Stewart [9] adopted FLUENTTM to study ows in global control valves. For three-dimensional analysis, Huang and Kim [10] utilized FLUENTTM to simulate turbulent ows in a buttery valve. The ke model was employed in their numerical simulation. Experiments were also conducted by several researchers. Mertai et al. [7] established a water tunnel system to conduct the performance test of a V-sector ball valve. They used an LDV measuring system to investigate the ow in ball valves. Davis and Stewart [11] employed a closed piping system to test and observe a global control valve. The main purpose of this study is to provide ow characteristics and ow patterns inside ball valves using an experimental approach. The correlation between the ow patterns and the valve performance are discussed. In general, the openings of a ball valve and the inlet velocity play vital roles in the ow characteristics of ball valves. The following sections describe the details of the proposed experimental procedure.

tank. The upstream pressure is provided by high pressure air which is supplied to the buer tank from an air-compressor system. The water in the buer tank gains the pressure from the pressurized air and goes through the pressure regulator, which is installed at the outlet of the tank, to maintain a stable, desired pressure for valve operation. Fig. 2 demonstrates the pictures of the established experimental facility. Water ows from the left hand side to the right hand side through the white pipe in Fig. 2(a). The transparent test section can be found at the black frame in Fig. 2(a). The white buer tank, which provides the stable stream to the test section, can be seen at the left hand side of Fig. 2(b). The length of the pipe from the buer tank to the test section is 120D to get the fully developed turbulent ow at the entrance of the valve. The control panel and the readings of the owmeters are shown at the right hand side of Fig. 2(b). When the valve performance is tested, the static pressures, Pin and Pout are measured at 2D upstream of the valve and 6D downstream of the valve, respectively, by using the mercury manometers, according to the standard test method of ANSI/ISA-75.02-1996. The diameter of the pipe is 38 mm. The pressure drop, DP, across the valve is also detected by an inclined mercury U-tube manometer. The accuracy of the manometer is within 3% of reading. The volumetric ow rate, q, is measured by several calibrated turbine ow meters which are installed in the downstream area of the valve. The accuracy of the owmeter is within 2%. The range of operation for pressure (gauge) and inlet velocity are 0.51.8 bars and 210 m s1, respectively. In terms of the Reynolds number, it is 0.64 105 3.18 105. The test section, as shown in Fig. 3, includes a plexiglass tube and acrylic ball valve of nominal diameter 50.8 mm (2 in.) and is CNC machined. The transparent test section is connected to the upstream and downstream pipes via the specially-designed fast-coupling mechanisms. The laser-light beam emitted from an 5 W argon-ion laser is transmitted through an optical ber and then connected to a 20 laser-light sheet expander. The laser-light sheet is adjusted to a thickness of about 0.5 mm. In order to compensate for the dierence of the refraction indices between the air and the curved plexiglass, a rectangular glass tank lled with still water is enclosed around the whole test section. By using this method, the refraction of the laser-light

rs

on

al

co

py

M.-J. Chern et al. / Experimental Thermal and Fluid Science 31 (2007) 505512

507

Fig. 2. Pictures of experimental apparatus.

1W Inova-70 Ar-ion laser optical fiber

alignment rake

pe

rs

on
support 20 lase r-light sheet generator laser-light sheet
o

al
rotatory encoder angularity regulator test section frame

r's

ball valve inlet flow stainless steel tube fast-coupling connector

water

ball valve
outlet flow

th o

stainless steel tube plexiglass

valve set clamping mechanism

Fig. 3. Schematic diagram of ow visualization.

plexiglass

Au

digital video camera

Fig. 4. Schematic of angularity regulator.

sheet is eectively reduced. The angle of the model valve or the corresponding opening is adjusted by an accurate mechanism. Fig. 4 reveals the angle adjustment mechanism. A circular cylinder is used to connect the ball valve and the mechanism. An O-ring is employed to seal the ball valve. The adjust mechanism contains a rotary encoder whose resolution is 500 pulses per revolution.

The particle tracking ow visualization method (PTFV) is employed to reveal the ow patterns. When the ow visualization is conducted, plastic particles made of polystyrene (PS) are seeded to the piping system to scatter the laser light via a particle seeding mechanism. The diameters of polystyrene particles are between 30 and 70 lm and the specic weight is 1.03 at 25 C. Ignoring the eect of

co
o-ring

py

508

M.-J. Chern et al. / Experimental Thermal and Fluid Science 31 (2007) 505512

turbulent diusion, the relaxation time constant is estimated to be less than 6.25 105 s and the Stokes number is in the order of 106 within the range of experiment so that slip between the ow and particles could be neglected (Richard and John [12]). A digital camera and a CCD camera which has 1018 1008 pixels are used to take the pictures of path-line patterns and the animations of the ows, respectively. 3. Results and discussion 3.1. Flow conditions Several inlet ows and valve openings have been considered. The values of inlet velocity for experiments range from 2 to 6 m s1. The corresponding Reynolds numbers, iD ReD qU , are 0.64 105, 1.27 105 and 1.91 105. The l valve openings, /, under tests are 100%, 88.9%, 77.8%, 66.7%, 55.6% and 44.4%. These values correspond to the angles of the ball valve, a = 0, 10, 20, 30, 40 and 50. 3.2. Characteristic coecients In general, a valve is evaluated using three coecients: the loss coecient, the ow coecient and the cavitation index. Kirik and Driskell [2] provided formulae for these coecients. The loss coecient represents the energy loss due to a valve. It can be denoted as K1 DP ; qU 2 i 2 1

where Pv refers to the saturated vapor pressure (2.339 kPa at 20 C). The curve of cavitation index with respect to the valve opening indicates the lowest value for which cavitation occurs. Severe cavitation is possible above the cavitation index curve. No cavitation is expected below the curve. 3.3. Flow patterns Fig. 5 shows the path-line ow patterns at ReD = 0.64 105. Three vortices can be observed in the ow eld when the valve is not fully open. Two vortices are inside the

Au

where q is the volumetric ow rate, G is the specic gravity relative to water at 4 C and Fp is the geometric factor. G and Fp are set to unity in this study. To predict the conditions where cavitation happens, the cavitation index is used and is denoted as DP C cs ; 3 P in P v

th o

r's

where Ui is the mean inlet velocity and DP is the pressure drop measured between 2D in front of the valve and 6D behind the valve. The ow coecient refers to the capability of ow in a valve, which is dened as r q G Cv ; 2 0:865F p DP

pe
Fig. 5. Results of ow visualization at ReD = 0.64 105. Valve angle varies from 0 to 50.

rs
ball valve. The other one is behind the exit of the ball valve. It is larger than the two inside the valve. The size of each vortex can be estimated. Fig. 5(f) reveals a cavitation phenomenon at / = 44.4%. A strong light scattering is found when cavitation incepts. The rst inception of cavitation appears at the edge of the vortex behind the ball valve. When cavitation happens, noise is heard at the same time. As the valve angle increases, cavitation is also found inside the valve. Fig. 6 shows the schematic diagram of the ow patterns. As the opening of the valve decreases, these vortices grow and may cause more pressure drop. In other

on

al

co

py

M.-J. Chern et al. / Experimental Thermal and Fluid Science 31 (2007) 505512

509

Fig. 6. Schematic diagram of ow in a valve.

3.4. Valve performance

Fig. 7. Results of ow visualization at ReD = 1.0 105.

Au

th o

r's

pe
Fig. 8. Results of ow visualization at ReD = 1.2 105.

rs

on

The results of the ow visualization provide the ow patterns in a ball valve and in the vicinity region of the valve, while the measurements of pressure drop and volumetric

al

words, more energy is lost due to these growing vortices. It is interesting that many small eddies are observed at these main vortices and constitute large eddies. Figs. 7 and 8 demonstrate the ow patterns at ReD = 1.0 105 and 1.2 105, respectively. It is clear that these vortices grow when the opening of the valve decreases as that has been shown in Fig. 8. In addition, when the inlet velocity increases, the critical valve opening at which the

cavitation happens increases in terms of Figs. 5, 7, and 8. The critical valve openings are 55.6% and 66.7% for ReD = 1.0 105 and 1.2 105, respectively. To the authors knowledge, a picture of cavitation in a valve has not been mentioned in other papers. In the past, the inception of cavitation is judged by noise. The proposed experimental rig successfully provides the cavitation pictures. It is very useful for valve designers and users.

co

py

510

M.-J. Chern et al. / Experimental Thermal and Fluid Science 31 (2007) 505512
7

ow rate are used to estimate the performance of a ball valve. In general, the inlet velocity and the valve opening play very important roles to the performance of a ball valve. 3.4.1. The recirculation length behind the ball valve Fig. 9 shows the locations of reattached points of the recirculation behind the valve at ReD = 0.64 105. The arrows indicate the reattached points at those pictures, which have to be zoomed in to nd the attached points. The reattached point of the recirculation is not steady, so its time-averaged length determined by two hundred pictures is provided. Fig. 10 reveals the estimation of the time-averaged recirculation length changing with various Reynolds number and valve openings. The recirculation zone behind the ball valve dominates the pressure drop. Energy is dissipated in this region. Hence, the larger the recirculation length, the more the pressure drop. It is found that the experimental results at various Reynolds numbers are the same at dierent openings. It turns out that the recirculation length is not aected by the

Uin (m/s)
6 5 4

2 3 4 5

ReD 5 0.64x10
0.95x10 1.27x10 1.59x10
5 5 5

Lbubble/D
3 2 1 0 30

40

co
50 60 50 40

py
(%)
30 70 80 90 100 60

closed

(deg.)

20

10

fully open

Fig. 10. Variation of circulation length. The error bars show the 95% condence intervals of the data points plotted according to Smith and Wenhofer [14].

r's Au th o
Fig. 9. Locations of reattached points of the recirculation with various openings. An arrow refer to a reattached point.

pe

rs

on

Reynolds number. As a result, the molecular viscosity does not play a vital role in the ow at high Reynolds numbers. It is the Reynolds stress that dominates the recirculation length. For ows at high Reynolds numbers, the averaged ow behaviors are not signicantly aected by the variation of the Reynolds number but by the uctuation quantities [13]. Therefore, the Reynolds number does not apparently aect the ow pattern until the cavitation appears. Hence, the recirculation length or the energy dissipation is only aected by the valve position. It is clear that the recirculation length gets increased as the valve opening get decreased. In other words, the system has to pay more energy to maintain the same volumetric ow rate or the same inlet velocity at small valve opening. 3.4.2. The loss coecient The pressure drop is nondimensionalized as the loss coecient, K, in Eq. (1). As the valve opening decreases, the pressure drop increases due to the growth of the vortices in terms of Figs. 10 and 11. Fig. 11 shows that the loss coecient is inversely proportional to the valve opening, and it does not vary with the Reynolds number in the present range of experiments. Experimental data are tted by an equation, K = 385.3 12.67/ + 0.141/2 0.000522/3. Designers and users of ball valves can obtain the variation of K according to this equation. Although this is not a general equation for all of ball valves, subsequent experiments regarding valves can also provide such an empirical equation for designers and users. 3.4.3. The ow coecient The ow coecient, Cv, refers to the capability of ow of a ball valve. The denition of Cv is shown in Eq. (2).

al

M.-J. Chern et al. / Experimental Thermal and Fluid Science 31 (2007) 505512
60

511

Uin (m/s)
50

ReD
5 5 5

40

2 3 4 5

0.64x10 0.95x10 1.27x10 1.59x10

valve opening changes the ow coecient. For valve openings larger than 80%, the Reynolds number plays a role which inuences Cv as well. The increase of the ow coecient with the valve opening is nonlinear. Hence, in order to conduct a linear control of the ow coefcient, it requires another accessory. The tted equation, Cv = 2.23 + exp(0.0034/), is provided in Fig. 12. It is useful for designers and users. 3.4.4. Cavitation in ball valve Another important eect in a valve operation is cavitation. The cavitation is caused by a local low pressure induced by the ow conditions as shown in Fig. 5(f), Fig. 7(e), and Fig. 8(d). The inception and burst of bubbles moving with liquids are observed in the experiments. Theoretically, the cavitation should appear at the critical condition that the local pressure is lower than the saturated vapor pressure. Cavitation causes not only noise but also vibration. Using the ow visualization technique, recording the inlet pressure and pressure dierence across the valve at cavitation, and employing the vapor pressure at valve outlet, the cavitation index, Ccs, which indicates the critical condition for occurrence of cavitation, can be calculated. Fig. 13 shows the results. It is clear that the critical cavitation indexes vary with the inlet velocity and the valve position. Above the curves, the cavitation is observed. In other words, the lower the value of the cavitation index, the more dicult the cavitation appears. For a constant inlet velocity, Ccs decreases with the increase of valve openings. Large pressure drop and high shear eect present in the small opening region may cause

K 30

20

10

0 30

40

50

60

(%)
30

70

80

90

100

closed

(deg.)

fully open

200 180 160 140 120 Cv 100 Uin (m/s) ReD

0.64x105 0.95x105

4
5

1.27x105
1.59x105

pe

2 3

rs

80

on
1.8

Fig. 11. Variation of loss coecient. The error bars show the 95% condence intervals of the data points plotted according to Smith and Wenhofer [14]. The solid line refers to the tted equation, K = 385.3 12.67/ + 0.141/2 0.000522/3.

al

60

50

40

20

10

co
50 60 50 40

py
(%)
30 70

r's

Uin (m/s)
1.6 1.4 1.2 1.0

ReD

60 40 20 0
30 40

50

th o

2 3 4 5

0.64x10 0.95x10 1.27x10 1.59x10

5 5 5 5

60

(%)
30

70

80

90

100

Ccs
0.8

60

50

40

20

10

Au

closed

(deg.)

0.6 0.4 0.2 0.0 30 40 80 90 100

fully open

Fig. 12. Variation of ow coecient. The error bars show the 95% condence intervals of the data points plotted according to Smith and Wenhofer [14]. The solid line refers to the tted equation, Cv = 2.23 + exp(0.0034/).

Fig. 12 demonstrates the ow coecient determined by the measured volumetric ow rate and pressure drop. For ows at the same Reynolds number, Cv increases as the valve opening becomes large. The tendency can be found in Fig. 12. For valve opening smaller than about 80%, Cv is not aected by the Reynolds number as well. Only the

60

20

10

closed

(deg.)

fully open

Fig. 13. Variation of cavitation index. The error bars show the 95% condence intervals of the data points plotted according to Smith and Wenhofer [14].

512

M.-J. Chern et al. / Experimental Thermal and Fluid Science 31 (2007) 505512

early occurrence of cavitation. Besides, since the cavitation indexes at large inlet Reynolds numbers are higher than those at lower Reynolds numbers, it is easier to observe the cavitation occurring in the high velocity regime. 4. Conclusions By employing the particle tracking ow visualization method and the ow rate/pressure measurement technique, the performance characteristics and the inside-valve ow patterns are studied in this paper. The following conclusions are drawn from the results. First of all, features of ows in a ball valve, including three vortices and cavitation, can be observed by experimental results at varying inlet ows (210 m s1, ReD = 0.64 1051.59 105) and valve openings (10044.4%). Subsequently, the valve performance can be determined by the loss coecient K and the ow coecient Cv using the pressure and ow rate information in experimental data. Two gures are shown in this study to reveal variations of K and Cv with the valve opening. Correlation equations for K and Cv are provided for practical use. A designer of valve can nd the corresponding K and Cv for a given valve opening in those gures. The inception of cavitation can be predicted using the cavitation index Ccs which is also determined by the pressure data. A gure is provided to present the variation of Ccs with ReD and the valve opening. The critical conditions of inception of cavitation can be found in the gure. A designer of valve can nd the critical values of inception of cavitation at various ReD and valve openings in the gure. The region above curves in the gure refers to the appearance of cavitation and should be avoided. The same approach can be applied to other control valves to determine their performance and to observe cavitation phenomena.

Acknowledgements We are truly grateful to MIRDC Taiwan for their support (Project No.: 90EC2A179179). References
[1] J.W. Hutchison, ISA Handbook of Control Valves, second ed., Instrument Society of America, Pittsburgh, 1976. [2] M.J. Kirik, L.R. Driskell, Flow Manual for Quarter-Turn Valves, Rockwell International Co., 1986. [3] G.H. Pearson, Valve Design, Mechanical Engineering Publication, London, 1978. [4] T. Ota, M. Itasaka, A separated and reattached ow on a blunt at plate, Journal of Fluids Engineering 98 (1) (1976) 7986. [5] R.M. Kelso, T.T. Lim, A.E. Perry, The eect of forcing on the timeaveraging structure of the ow past a surface-mounted blu plate, Journal of Wind Engineering and Industrial Aerodynamics 49 (13, Part 1) (1993) 217226. [6] T. Kerh, J.J. Lee, L.C. Wellford, Transient uid-structure interaction in a control valve, Journal of Fluids Engineering 119 (2) (1997) 354359. [7] P. Merati, M.J. Macelt, R.B. Erickson, Flow investigation around a Vsector ball valve, Journal of Fluids Engineering 123 (3) (2001) 662671. [8] C. van Lookeren Campagne, R. Nicodemus, G.J. de Bruin, D. Lohse, A method for pressure calculation in ball valves containing bubbles, Journal of Fluids Engineering 124 (3) (2002) 765771. [9] J.A. Davis, M. Stewart, Predicting globe control valve performance Part I: CFD modeling, Journal of Fluids Engineering 124 (3) (2002) 772777. [10] C. Huang, R.H. Kim, Three-dimensional analysis of partially open buttery valve ows, Journal of Fluids Engineering 118 (3) (1996) 562568. [11] J.A. Davis, M. Stewart, Predicting globe control valve performance Part II: Experimental validation, Journal of Fluids Engineering 124 (3) (2002) 778783. [12] C.F. Richard, H.S. John, Fundamentals of Air Pollution Engineering, Prentice Hall, New-Jersey, 1998, pp. 290357. [13] H. Tennekes, J.L. Lumely, A First Course in Turbulence, MIT Press, Cambridge, 1972, pp. 5965. [14] R.E. Smith, S. Wenhofer, From measurement uncertainty to measurement communications, credibility and cost control in propulsion ground test facilities, Journal of Fluids Engineering 107 (2) (1985) 165172.

Au

th o

r's

pe

rs

on

al

co

py