Annals of Nuclear Energy 63 (2014) 215227

Contents lists available at ScienceDirect

Annals of Nuclear Energy

journal homepage: www.elsevier.com/locate/anucene

Experimental study on spray characteristics of pressure-swirl nozzles in pressurizer

Zhike Lan, Dahuan Zhu, Wenxi Tian, Guanghui Su , Suizheng Qiu
Department of Nuclear Science and Technology, Xian Jiaotong University, Xian, Shaanxi 710049, PR China

a r t i c l e

i n f o

a b s t r a c t
Spraying system in the pressurizer of Pressurized Water Reactor (PWR) power plant system is of great importance for system pressure control. An experimental study on the spray characteristics, including mass ow rate, spray ux distribution, spray cone angle and drop size spectrum, was conducted. A testing loop with nine swirling nozzles was established for the study. In order to measure the spray cone angle and drop size spectrum, two original devices including a spray droplet collector and a photographic chamber were designed and employed. The former was used to collect the spray droplet along the cross-section diameter, and the latter was made to isolate and measure the targeted spray droplet. Based on the experimental data, the curves of ow rate and spray cone angle versus nozzle pressure drop were obtained. Several typical spray ux distributions were derived and the results indicated that the ux distribution changes signicantly with even small pressure changes. Thus, it was proposed that instability of the spray ux distribution should be considered in the pressurizer. Based on the spray drop pictures recorded by the high speed camera, Probability Density Function (PDF) of the drop size was obtained and compared with four standard empirical distributions. It was found that the NukiyamaTanasawa distribution provides a better t to the experimental PDF of the spray drop size. The present work introduces the experimental methodology and results of spray behaviour of the nozzle in pressurizer. The work is expected to be helpful for the optimization design of spraying systems. Crown Copyright 2013 Published by Elsevier Ltd. All rights reserved.

Article history: Received 22 May 2013 Received in revised form 3 July 2013 Accepted 28 July 2013

Keywords: Nozzle Discharge coefcient Volume ux distribution Spray cone angle Drop size spectrum Photography method

1. Introduction Liquid sprays generated by nozzles are important for many technical processes, such as energy conversion, coating, and cooling. In cooling systems, especially, spraying has signicantly superiority due to the excellent ability of heat and mass transfer. It is well known that the latent heat exchange is a major factor for the cooling efciency when a large number of droplets are injected into the ambient gaseous environment. As a result, the spraying systems have been widely used in nuclear energy engineering, such as the pressurizer (Diao, 1988; Kim and Grifth, 1987; Chen et al., 2011), the containment spray system (Lemaitre and Porcheron, 2009; Robert, 1996) and the reactor core spray system (Anglart et al., 2010; Torres et al., 1999). For the spraying system, the purpose of the nozzle is to generate a multitude of individual drops spreading over a wide angle. There have been various studies on the formation mechanism of sprays. Most of the studies showed that in the entire process of forming drops there are two stages, which are the forming of ligaments and then droplets (Kim et al., 2003; Weber, 1931; Tratnig and

Corresponding author. Tel.: +86 29 82663401; fax: +86 29 82667802.

E-mail address: ghsu@mail.xjtu.edu.cn (G. Su).

Brenn, 2010). Based on the theoretical analysis, some researches have attempted to nd a reliable way to predict the spraying performance. However, the results were unsatisfactory. As the most important parameter, spray drop size is modelled by three different methods: the empirical method, the Maximum Entropy (ME) method and the Discrete Probability Function (DPF) method (Babinsky and Sojka, 2002). But the problem with the empirical method is that it is not accurate when extrapolating the data which is outside the experimental range (Gonzalez-Tello et al., 2008; Paloposki, 1994). The most important part of applying the ME method is the appropriate adjustment to the source terms made by referring to the experimental distribution (Li and Tankin, 1988, 1989; Sellens, 1989). The DPF method may lose the predictive power because it is difcult to obtain experimental results as the input conditions (Sovani et al., 2000). Therefore, experimental investigation plays an important role in the study of sprays produced by nozzles. For the experimental studies conducted in the past, most of them mainly focused on the measurement of the drop size. Kim et al. (2003) set up a test rig for the spray experiment. The initial droplet size was measured by phase-Doppler interferometry for sprays which was generated by a planar research nozzle and a practical gas turbine air blast nozzle. The measured drop size ranged from 10 lm to 200 lm. Porcheron et al. (2010) made a special

0306-4549/$ - see front matter Crown Copyright 2013 Published by Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.anucene.2013.07.048

216

Z. Lan et al. / Annals of Nuclear Energy 63 (2014) 215227

Nomenclature A C C Cv D D d f h n P r s W x area m2 discharge coefcient constant component of discharge coefcient variable component of discharge coefcient drop diameter lm mean size of drop diameter lm diameter mm drop size distribution function height mm drop number pressure Pa radius m standard deviation mass ow rate kg/s sample

p
h

q r
D

circumference ratio shrinking angle of swirl chamber degree density kg/m3 empirical distribution parameter difference

Greek symbols a spray cone angle degree d inclination angle of swirl groove degree c coefcient of variation k area ratio of swirl ow to direct ow

Subscripts d direct ow through central direct ow conical diffuser e nozzle exit i size class iv ideal value j measuring position LH log-hyperbolic distribution LN log-normal distribution NT NukiyamaTanasawa distribution n nozzle o orice of the atomizer RR RosinRammler distribution s swirl ow through swirl groove

facility to simulate the reactor containment. In order to perform thermal hydraulic containment studies, the interferometrics laser imaging was used to measure the droplet size. The measured drop size ranged from 60 lm to 250 lm. Tratnig and Brenn (2010) carried out the experiments to study a kind of pressure-swirl atomizers. Local drop size was measured with phase-Doppler anemometry at some points in each spray cross section. The drop size was measured to be less than 100 lm. In some other studies, spray cone angle as an additional parameter was measured by digital camera (Gong and Fu, 2007; Tratnig and Brenn, 2010). An experimental investigation was made on the discharge coefcients and the spray cone angle of a solid cone swirl nozzle (Halder et al., 2004). The spray cone angle was measured by taking the photographs of spray with the help of a wide angled lens camera and ood light illumination. However, the instruments and methods mentioned above are not suitable to the nozzles used in Pressurized Water Reactor (PWR) power plant system, such as the sprinkler in pressurizer. These nozzles have the larger spray eld and mass ow. The diameter of the spray eld can have a value of P4.0 m. Diao (1988) analysed the performance of ve types of nozzles with the range of the pressure drop from 0.127 MPa to 0.245 MPa and the ow rates from 5 103 kg/h to 5 104 kg/h, which falls in the operation regime of pressurizer of PWR system. But the detailed method for the experiment was not given. The resulting spray produced by the nozzle usually has the larger drop size and spray eld. It is unfortunate that the partial drop size in these sprays exceeds the upper limits of the capacity of the advanced instrument distinctly, such as phase Doppler and laser diffraction techniques. Zaidi et al. (1998) carried out a comparative study of these techniques to investigate drop sizes in the two-phase ow, and provided their respective defects and measuring ranges. Accordingly, it appears more important to nd a suitable method to measure the drop size for the special object. In this study, a detailed experimental method is presented to characterize the spray, which is of lower pressure drop and large ow rate. As an important property which was usually neglected for conventional nozzles due to their small ow rate, the spray ux distribution was measured and analysed in this work. In addition, two photographic methods were taken respectively to characterize

the spray cone angle and the drop size spectrum. In order to carry out the above measurements, some devices, including a spray droplet collector, a waterproof light source and a photographic chamber, were designed and employed. 2. Experimental loop The experiments were carried out by using the test rig sketched in Fig. 1. This rig consisted of a spray pool, a pressure pump, pipes, a spray nozzle, a series of control valves and devices for process monitoring such as pressure sensor and mass ow meter. The loop could be run in three modes: the major-loop mode, the bypass mode and the drainage mode. Operation mode of the loop is controlled by a series of control valves. After startup of the system, the bypass valve was opened gradually to obtain an initial ow rate through the bypass pipe. Then the major-loop control valve was put into operation to get a proper ow rate close to target value. Finally, the desired ow was obtained accurately by means of vernier regulation of the reducing valve. The liquid used for the present experiment was tap water, which was stored in an open quadrate pool (4000 mm 4000 mm 800 mm). It was obvious that the pool also served as a sprays droplet collector. In order to clear the working liquid, two lters, stainless steel brous lter and Y lter, were installed at the inlets of the pump and the reducing valve respectively. In addition, the turbine mass owmeter and the pressure sensor were used to measure the values of the mass ow rate and the pressure drop of the liquid through the nozzle. The owmeter ranges from 2.0 103 to 2.0 104 kg/h and the pressure sensor is in 01.0 MPa. During the experiments, the environmental parameters including the ambient temperature, the atmospheric pressure and the medium temperature were monitored. The nozzles used in the experiments were a series of pressureswirl nozzles for spraying system in the pressurizer of the PWR power plant system. They consist of some swirling grooves, a central direct ow conical diffuser, a swirl chamber and a nozzle exit, as sketched in Fig. 2. A part of liquid is driven into the swirl grooves and then rotated into swirl chamber, where the liquid is accelerated. Due to the centrifugal inertial force, the liquid clings to the internal wall of the swirl chamber and forms a thin liquid lm.

Z. Lan et al. / Annals of Nuclear Energy 63 (2014) 215227

217

Fig. 1. Sketch of the spray test rig.

Fig. 2. Structural sketch of the pressure-swirl nozzle.

Meanwhile, remaining portion of liquid emerges into the swirl chamber directly through the central conical diffuser with high speed. At the orice, the latter impacts the liquid pre-lm produced by the swirl chamber. Then the two parts are injected into the ambient atmosphere together, where a further collapse will be caused to form the spray drop. The geometric characteristics of the nine pressure swirl nozzles used to perform the experiments are listed in Table 1. In the table, k denotes the ratio of the swirl groove ow area As to the central conical diffuser ow area Ad while other parameters have been marked in Fig. 2. 3. Experimental data reduction 3.1. Discharge coefcient According to the experimental data from the mass ow meter and the pressure sensor installed at the inlet of the nozzle, the ow characteristics were easily obtained. The typical curves that the mass ow rate of nine pressure swirl nozzles changed with the pressure drop are shown in Fig. 3. It is observed that the curves are different and each of them corresponds to a specic nozzle. The main reason is the difference of their geometric structure, as shown in Fig. 2 and Table 1.

The discharge coefcient of nozzle can be regarded as an efcient criterion for the spraying to evaluate the losses of the spray liquid supply. Since the C is mainly dependent on the nozzle geometry and the ow conditions (Guo et al., 2011), it has been widely used as an important indicator for evaluation of the nozzle ow characteristic. According to the denition of the discharge coefcient, it represents the ratio of the actual mass ow rate W to the ideal value Wiv. Thus

W ; W iv

where C is the discharge coefcient of the liquid for certain nozzle. Compared with the total pressure drop through the nozzle, the gravity pressure drop is small and can be neglected (Gong and Fu, 2007; Zeng et al., 1990). Therefore, the ideal mass ow rate Wiv is written as:

W iv Ao 2qDP0:5 :
The actual mass ow rate through the nozzle is given by:

W CAo 2qDP0:5 ;

218 Table 1 Geometric characteristics of the pressure-swirl nozzles. Nozzle type 1# 2# 3# 4# 5# 6# 7# 8# 9# Number of swirl groove 6 6 6 6 4 4 4 4 2 Inclination angle d (degree) 30 30 30 30 0 0 0 0 45

Z. Lan et al. / Annals of Nuclear Energy 63 (2014) 215227

Flow area ratio k = As/Ad 0.34 0.34 0.34 0.34 0.24 0.24 0.24 0.24 0

Nozzle height hn (mm) 88.4 72.7 69.5 64.8 104.5 75.5 60.5 44 98

Shrinking angle h (degree) 15 22.5 30 30 15 22.5 30 45 arc

Height of nozzle exit he (mm) 1.6 1.6 1.6 1.6 1.6 1.6 1.6 1.6 15

Orice diameter do (mm) 23 23 23 23 21 21 21 21 18

2 where A0 pr 2 0 is the area of the nozzle orice, m ; q is the density of the liquid, kg/m3; and DP is the pressure drop from the inlet to the orice of the nozzle, Pa. Thus

An additional parameter for characterizing the stability of the discharge coefcient is the coefcient of variation, which is dened as

W Ao 2qDP0:5

cx

Sx ; xmean

As a result, the correlation analysis of the discharge coefcient C and the pressure drop DP are obtained and shown in Fig. 4 by substituting experimental data into Eq. (4). In theory, compared with the ideal ow Wiv in Eq. (2), the discharge coefcient C should be an integrated parameter, in which the inuence of nozzle geometric and ow pattern on ow characteristic have been contained. For the pressure swirl nozzle, there is a cavity in the middle of the orice and the cavity area is usually different when the working condition changes. To evaluate the inuence of the efcient ow area at the orice, it is assumed that the discharge coefcient consist of two components, which are constant component C* and variable component Cv respectively. The relation is given by

where S(x) is the sample standard deviation, and xmean is the sample mean value of a quantity x. Here, the discharge coefcient C and C* are substitute for the variable x in Eq. (8). c(C) and cC are calculated and compared. According to the data set in Figs. 4 and 5, the variation coefcient c(C) is 2.50%, but cC is only 1.05% which is an ideal result. The comparison shows that, for the pressure swirl nozzle which has a central direct ow conical diffuser as shown in Fig. 2 and Table 1, when the orice is not full of liquid, the factor of the efcient ow area is contained in the exponent 0.43. Therefore, the remaining component C* seems more reasonable and can be regarded as an indicator to the specic geometric structure of the reference nozzle. 3.2. Distribution of spray volume ux Surveying the literature, only a few experiments are conducted for measuring distribution of spray volume ux. The main reason is that the investigated nozzles are smaller sized, which produce spray with lower mass ow rate, smaller spray eld and smaller spray drop size. So in these cases, the distribution of spray volume ux seems less signicant. However, a reasonable distribution of the spray ux is necessary to the application of spray cooling system used in a larger space. Therefore, the method of measuring the distribution of spray volume ux is introduced in this work. In order to obtain the detailed spray ux characteristic, a device for collecting the spray droplet was fabricated, as shown in Fig. 6. This device consists of 75 collecting holes. Each hole measures 16 mm in diameter and 180 mm in depth. All holes were drilled in a rectangular organic plate with the dimensions of 2300 mm 20 mm 240 mm. The center distance of the adjacent holes is 30 mm. The solid around the hole orice in the organic plate was removed to prevent the liquid droplets between the adjacent orices from owing into the holes. As observed from Fig. 6b, a scale label was pasted carefully on the middle of each pair of holes. Then, a cover board was designed to ensure all collecting holes running at a same time. In addition, an accessorial lifting jack horse was necessary because the vertical distance between the nozzle and the device orice needs to be adjusted according to the requirement. In this experiment, the vertical distance was varied from 1100 mm to 2200 mm. During the course of the experiment, the device was placed right under the nozzle and in the spray pool. In order to obtain the accurate experimental results, some caution must be exercised. The most important points are being listed here:

C C Cv :

It is observed from Fig. 4 that the coefcient C decrease signicantly with the increasing pressure drop. It is mainly caused by the change of the Cv component, which denotes the inuence of the efcient ow area at the nozzle orice. With the pressure drop DP increasing, the rotation of the liquid is strengthened and the efcient ow area at the orice decreases gradually. An additional correlation is tted according to the experimental data to eliminate the latent inuence of pressure drop DP on the discharge coefcient C. The tted mass ow rate is given as follows:

W C Ao 2qDP 0:43 ; C W Ao 2qDP0:43 :

6 7

Thus C* is the remaining component of discharge coefcient C. The result obtained by Eq. (7) is shown in Fig. 5. Its observed from the gure that the component C* is a constant which independent of the ow condition, and it can be regard as an indicator to the specic geometric structure of the reference nozzle. Halder et al. (2004) made numerical and experimental investigations on the coefcient of discharge. Hamid and Atan (2009) carried out a cold ow test for two types of spray pattern: hollow cone and solid cone. Their investigations show that the discharge coefcient (see Eqs. (3) and (4)) will be constant when the area ratio of direct ow to swirl ow increase and reach a value which is large enough that the orice is lled by the liquid. That conclusion is consistent with the analysis of the Cv in this work.

Z. Lan et al. / Annals of Nuclear Energy 63 (2014) 215227

219

Fig. 3. Typical curves of mass ow rate versus pressure drop.

(1) At different measuring height, the orices of all collecting holes ought to be kept on the same horizontal level. The point located in the central line of the nozzle should be marked as the origin of x-coordinate for representing the relative position of the holes. (2) Before each experiment, all the collecting holes must be cleaned to avoid the inuence of the residual liquid. (3) When the spraying system was running steadily, the cover plate should be opened rapidly and the stopwatch operated at the same time. Accordingly, at the end of the experiment, the cover plate was closed and the stopwatch stopped immediately. (4) In order to reduce the measuring error, the liquid collected in the holes should be as much as possible within the permitted volume range. 2-D graph of spray ux distributions in a vertical plane was obtained by processing experimental data along the diameters of the cross-section at the 2200 mm and 1100 mm vertical distance from the nozzle. According to the data of the experiment 2# and 5#, two typical distributions are shown in Figs. 7 and 8 respectively. As shown in the gures, the x-axis represents the position of the

orice level and the x-coordinate is the horizontal distance of the hole from the origin point. The right y-coordinate is vertical distance of the device orice from the nozzle, and the two left y-coordinate represent the measured liquid volume collected in the hole. It is obvious that the distribution of spray ux shown inFig. 7 (nozzle 2#) is relatively uniform. However, it is different in Fig. 8 (nozzle 5#) that the ux in the center region of the spray eld is very small, and is concentrated towards outside. The reason for this is the difference of the nozzles geometries. As shown in Table 1, 5# has small ow area ratio (k = 0.24) and stronger rotation (d = 0) compared with the nozzle 2# (k = 0.34; d = 30), so a hollow cone forms in the 5# spray eld. The hollow cone distribution is the typical spray pattern for the pressure swirl nozzle. With the pressure change, spray ux distribution of a same nozzle can show a signicant change. Figs. 911 show the spray ux of the nozzle 7# under the different pressure drop conditions. When the pressure drop through the nozzle increased from 0.15 MPa to 0.20 MPa, spray ux in the central region of the spray eld is almost unchanged, but its outside value increased signicantly. However, when the pressure drop increased from 0.20 MPa to 0.25 MPa, the shape of the distribution shows a more dramatic change. Although the outer ring ux of the eld is almost unchanged, the spray ux of the central region increase rapidly and form a peak. As shown in Table 1, the area ratiok for 7# is 0.24. With the pressure drop increasing, the mass ow rate through the central direct channel increases more signicantly than the swirl ow. It can be found that the collection device introduced in the paper can measure the spray ux distribution of the reference nozzle well. 3.3. Spray cone angle The spray cone angle is important for the application in the pressurizer especially. It determines where the liquid droplet can arrive. In this work, the spray cone angle was measured by the photography method with a digital camera. Due to the larger spray eld and ow rate, some special actions were taken and they are summarized here: (1) A supplemental lighting illuminated the spray from down to up to obtain the clear spray prole because the light is scattered by the droplets so that the spray is much brighter than its background as shown in Fig. 12.

Fig. 4. Change of the discharge coefcient C versus the pressure drop DP.

Fig. 5. Change of the constant component C* versus the pressure drop DP.

220

Z. Lan et al. / Annals of Nuclear Energy 63 (2014) 215227

collecting hole diameter: 16mm; number: 75

scale label range: 0~170mm

cover board

jack horse

rectangular organic plate length: 2300mm; width: 240mm; thickness:20mm

30mm

180mm

(a) Overall appearance of the device

(b) partial enlarged detail image

Fig. 6. Picture of the device for collecting spray droplet.

Fig. 7. Typical distribution with uniform spray ux of nozzle 2# (DP = 0.20 Mpa; Dt1100 = 90 s, Dt2200 = 240 s).

Fig. 8. Typical distribution with hollow spray ux of nozzle 5# (DP = 0.20 Mpa; Dt1100 = 45 s, Dt2200 = 180 s).

Z. Lan et al. / Annals of Nuclear Energy 63 (2014) 215227

221

Fig. 9. Spray ux distribution of nozzle 7# with 0.15 MPa pressure drop (Dt1100 = 70 s, Dt2200 = 240 s).

Fig. 10. Spray ux distribution of nozzle 7# with 0.20 MPa pressure drop (Dt1100 = 60 s, Dt2200 = 180 s).

Fig. 11. Spray ux distribution of nozzle 7# with 0.25 MPa pressure drop (Dt1100 = 60 s, Dt2200 = 180 s).

222

Z. Lan et al. / Annals of Nuclear Energy 63 (2014) 215227

Fig. 12. Arrangement scheme of the supplemental lighting for taking the photographs of spray cone angle.

Fig. 15. Change of the spray cone angle versus pressure drop.

Fig. 13. Picture of spray cone angle of nozzle 3# with 0.15 MPa pressure drop.

Fig. 14. Picture of spray cone angle of nozzle 5# with 0.15 MPa pressure drop.

(2) The spray cone angle might have a circular asymmetry. Therefore, three different lines of camera sight were selected and the angle between any two lines was 120. (3) The camera and the exit of nozzle were located at the same horizontal level. Figs. 13 and 14 are the typical spray cone angle photographs for the nozzle 3# and 5# with 0.15 MPa pressure drop. It is clear that the spray cone angle of Fig. 13 is greater than that in Fig. 14. The arithmetic mean of the spray cone angles from the different sight

lines should be regarded as the representative parameter of the nozzle. The values versus the pressure drop for all measured nozzles are shown in Fig. 15. It is observed that the spray cone angles are mainly located in the range of 80105. With the pressure drop increasing, the angle changes slightly except for the 9# nozzle. In general, the spray cone angle will not change when the pressure drop is larger than a threshold value because the ow of the liquid lm in the swirl chamber is fully developed and the velocity direction of sprays liquid emerging from the nozzle orice is related mainly to the geometric parameters of the nozzle. From Table 1, it can be seen that the 9# nozzle has two swirl grooves and the area ratio of the direct ow to swirl ow is zero. In particular, the 9# has a larger height of nozzle exit (15 mm) compared with the other nozzle (1.6 mm). In the case, the nozzle exit is as effective as a plain-orice nozzle for 9#. When the pressure drop is small, the liquid through the nozzle exit produce a cavity on the wall. Under the condition, the spray liquid diffuses outward to the ambient air and forms a larger spray angle. With the pressure drop increasing, the liquid is pushed and close to the exit wall. In the case, the role rectifying the liquid ow produced by the exit wall is strengthened. The cavity reduces and the spray angle decreases gradually. Therefore, the angle of #9 decreases signicantly with the pressure drop increasing, as shown in Fig. 15. In addition, it is also possible to obtain the cone angle in terms of the 2-D graph of spray ux distribution introduced in Figs. 711. The angle between the contour lines on both sides of the spray eld can also be considered as the spray cone angle. Figs. 1618 show the 2-D graphs and pictures of the spray cone angle for nozzle 9# corresponding to the pressure drop 0.15 MPa, 0.20 MPa and 0.25 MPa, respectively. As shown in Figs. 1618, there are some points outside the contour (red line). It can be observed that the liquid volume in these holes is almost zero and the value is barely changed. In the actual experimental process, it can be found that the tiny water in those holes is regarded as the deposition of the water vapor adhered to the hole wall. Therefore, these holes cannot be considered as efcient holes in the process to determine the boundary. Thus, it is observed that the changes of the spray cone angle with pressure drop increasing are consistent with the results shown in Fig. 15. Fig. 19 shows the comparison of the cone angles obtained from the photography method and the 2-D distribution graph. It can be found that the cone angle calculated from the 2-D graph is slightly less than the photography method. In the eld, the velocity of

Z. Lan et al. / Annals of Nuclear Energy 63 (2014) 215227

223

Fig. 16. Spray cone angle measured from the spray ux distribution and the corresponding picture for nozzle 9# with 0.15 MPa pressure drop.

Fig. 17. Spray cone angle measured from the spray ux distribution and the corresponding picture for nozzle 9# with 0.20 MPa pressure drop.

Fig. 18. Spray cone angle measured from the spray ux distribution and the corresponding picture for nozzle 9# with 0.25 MPa pressure drop.

spray drop has decreased signicantly due to the interaction with ambient air. The action of the gravity causes the increase of downward velocity component as compared to the horizontal component. Therefore, the trajectory of the outside spray droplet deects downward. Accordingly, in practical applications, the method of 2-D graph is more useful than the photography method because the effect of the gravity is considered and the symmetry of the spray cone angle could be investigated clearly. 3.4. Drop size spectrum The performance of nozzle typically depends on the spray drop size. It is undoubted that the spectrum of the spray drop size is the

most important parameter for evaluating the nozzle. In previous studies, most researches focused on the measurement or prediction of the spray drop size. Compared with them, the photography method introduced in this paper is able to measure the spray drop size in a much wider range. In order to obtain the clear picture, it is necessary to separate the measured droplets from the sprays. In this process, the key point is to ensure that the droplet formation process is not interfered by the experimental operation. To solve this issue, a photography chamber and a light source were designed to measure the spray drop size (as shown in Fig. 20). The viewing direction of the camera is regarded as the positive direction, and Fig. 20 shows the right view of the chamber.

224

Z. Lan et al. / Annals of Nuclear Energy 63 (2014) 215227

the number of the positions was determined by the radius of spray cross section and the width of photographic eld. In each local position, a calibration gauge was used as the reference substance and it was focused by the camera in the following experimental process. The distance between the gauge and the background plate could be adjusted by changing the length of the needles which was used to x it. Furthermore, the projection of the photographic eld in background plate was cut off so that the supplemental lighting at the back of the background plate could brighten the measured drops. At the top of the background plate, there was a narrow gap between the background plate and the roof sheathing of the chamber. The drop number entering into the spray-chamber was controlled by adjusting the gap width. It is emphasized that the background plate should be xed and located just below the nozzle. At the end, a waterproof light source was chosen with appropriate power and distance from the background plate. In order to decrease statistical error caused by the articial drop breakup, the following details are implemented:
Fig. 19. Comparison of spray cone angle obtained by photography method and 2-D distribution graph.

The photography chamber is divided into the antechamber and spray-chamber by the intermediate bulkhead. There is a window in the bulkhead where the camera sight can pass. The antechamber is a semi-closed space in order to protect the camera lens against the erosion of droplet. The spray-chamber is located between the intermediate bulkhead and the background plate. The spray drops to be measured enter into the spray-chamber through the narrow gap without any interference. As shown in Fig. 20, the roof sheathing, the side wall and the intermediate bulkhead of the chamber are made of synthetic glass. For obtaining global spray properties, local drop ensembles were photographed at different positions of a spray cross section. The positions were arranged equidistantly along the radius and

(1) The steel needle used to x the calibration gauge was as thin as possible. (2) To reduce the possibility of droplet breakage caused by the impact on the cross-section of the device, the ends of the background plate and the roof sheathing were cut off as the partial enlarged detail of the gap in Fig. 20. (3) The light was adjusted to be soft and uniform. (4) A spongy material was used to cover the roof sheathing. The quality of spray drop picture is determined by many factors, such as the lighting, the width of the gap, the distance between the calibration gauge and the background plate and the techniques for photography. Hence the photograph should be repeated to get the best results. The error of the measurement is mainly caused by the randomness of photographic time and the irregular contour of the spray drop. It has been mentioned that the spray drop size is in wider range, and the distribution is

Fig. 20. Schematic diagram of the photography chamber for measuring the spray drop size.

Z. Lan et al. / Annals of Nuclear Energy 63 (2014) 215227

225

Fig. 21. Spray droplet picture of nozzle 8# with pressure drop 0.15 MPa.

Fig. 24. Global Probability Density Function of the drop size and the global mean drop sizes of nozzle 8# (DP = 0.20 MPa).

Fig. 22. Spray droplet picture of nozzle 8# with pressure drop 0.20 MPa.

droplet number increased and more droplets were disintegrating when the nozzle pressure drop increased from 0.15 MPa to 0.20 MPa. When the pressure drop reached 0.25 MPa, as shown in Fig. 23, it seems that the droplet number is more and droplet diameter is smaller. In the case, as the cost, the picture clarity decreases due to an increase of the number of ne drop particles in the air. For certain local measurement, the drop size was divided into several size classes with equal interval scale when it increases from zero, and grouped samples were the number of drops falling in each interval. As the most important global spray property, the Sauter mean drop size D32 is given by Eq. (9).

PJ PI D32 PJ PI
j 1 j 1

3 i1 Di ni; j ; 2 i1 Di ni; j

where Di is the mean drop size of size class i at the measuring position j, and n(i, j) denotes the number of drops in size class i at the measurement position j. The drop size distribution of the entire spray is given by the global Probability Density Function (PDF). This function is given as

PDF Di

PJ 1 j1 ni; j ; PJ PI DD j1 i1 ni; j

10

Fig. 23. Spray droplet picture of nozzle 8# with pressure drop 0.25 MPa.

random in an instantaneous moment. Therefore, the spray drop size in a picture seems also random. To reduce the error, the camera shutter speed is set as 167 ls and the continuous shooting is taken to obtain more than 10 images in a measure point under a working condition. Then 8 different measure points are chosen to represent the global spray eld. Because the spray drop has irregular contour in the spray eld due to the interaction with the ambient air, more spray drops are expected to be involved in order to reduce the error. Total about 10 thousand drops are measured in a specic condition. The results of the spray drop picture of nozzle 8# with pressure drop 0.15 MPa, 0.20 MPa and 0.25 MPa are shown in Figs. 2123, respectively. It can be seen in the pictures that the scale as a reference size has been photographed. It is also to be noted that the

where DD is the width of the size classes. In this experiment, sufcient statistical reliability of the photographic measurement results was ensured by acquiring three different images in each local measurement. Totally, four local positions of measurement were determined along the radius of cross section. The distance between the cross section and the nozzle orice is 1100 mm and the DD is 50 lm. The global drop size PDF of the experiment 8# as an example is depicted in Fig. 24. Next to the PDF, the global Sauter mean and the number mean drop size are also depicted in Fig. 24 by different lines. The classical method of modelling drop size distributions is the empirical method. Some distributions used generally are regarded as the basis for standard empirical distributions including lognormal distribution, RosinRammler distribution, Nukiyama Tanasawa distribution and log-hyperbolic distribution as shown in Eqs. (11)(14). Log-normal distribution (Crow and Shimizu, 1988; Mugele and Evans, 1951):

226

Z. Lan et al. / Annals of Nuclear Energy 63 (2014) 215227

8 9 " #2 < 1 lnD= D 1 = p exp f D ; : 2 ln rLN ; Dln rLN 2p

RosinRammler distribution (Vesilind, 1980):

11

degree of shape freedom by a large number of parameters. As shown in Fig. 25, Log-hyperbolic distribution is reasonable and accurate, but the t is clearly inferior to that obtained from the NukiyamaTanasawa. The RosinRammler and Log-normal distribution agreement with experimental data is rather poor. 4. Conclusions

f D rRR DrRR DrRR 1 expfD= D f D rNT 1 DrNT 2 expfrNT 3 DrNT 4 g;

rRR

g;

12

NukiyamaTanasawa distribution (Li and Tankin, 1987):

13

Log-hyperbolic distribution (Bhatia et al., 1988; Bhatia and Durst, 1989, 1990):

q 2 r2 LH1 rLH2 q f D  2 2rLH1 rLH3 K 1 rLH3 r2 LH1 rLH2 r ( exp rLH1

2 LH3

D D rLH2 D D :

) 14

In Eqs. (11)(14), D denotes the mean size of the distribution and r represents the width of the distribution. The comparison of the experimental PDF and the reference four empirical distributions is shown in Fig. 25. It can be seen that the NukiyamaTanasawa distribution provides the best t to the experiment. The distribution was rst introduced by Nukiyama and Tanasawa to describe the number distribution of drops in sprays from a pneumatic atomizer (Nukiyama and Tanasawa, 1950). Paloposki (1991, 1994) analysed the range of the parameters of the NukiyamaTanasawa distribution function, especially rNT2 and rNT4. Either if rNT2 > 1 andrNT4 > 0 or rNT2 < 1 and rNT4 < 0 was proposed to the number distribution of drop size. However, when rNT2 > 1, it is singular that PDF is not zero at D = 0 for Eq. (10). The problem is obvious either 1 < rNT2 < 0 or 0 < rNT4 < 0.1. When rNT4 > 0, the width of the distribution will decrease with rNT4 increasing. Especially for the nozzles with the lower pressure drop and the larger ow rate, the problem is not acceptable. For the present experiment, the range of the parameters are 4 < rNT2 < 1 and rNT4 < 2. It may be meaningful and instructive for the NukiyamaTanasawa distribution to predict the spray with larger mass ow rate and smaller pressure drop, such as the pressurizer of PWR. The NukiyamaTanasawa curve agrees with the experimental data very well because the distribution can provide the greater

Sprays generated by the special pressure-swirl nozzles used in pressurizer were characterized experimentally for their discharge coefcient, spray ux distribution, spray cone angle and drop size spectrum. The present study provides an original and effective method to test the spray characteristics for nozzles of large spray eld. The pressure drop and corresponding mass ow rate are measured to characterize the ow characteristic of the spray nozzle. Based on the result, the amended exponent is advised for the correlation of the discharge coefcient. For measuring distribution of volume ux of spray, a special device of collecting spray droplet is designed and fabricated. In the same vertical section, the device is used to collect the spray droplet along the diameter of different cross-section, and then the 2-D spray ux distribution is obtained. The spray cone angle is measured by the photography method for the range that the pressure drop is from 0.15 MPa to 0.25 MPa. The result shows that the 2-D graph is available because the effect of the gravity is considered and the symmetry of the spray cone angle could be investigated clearly. The drop size spectrum is the most important parameter for the spray nozzle. The photography method is proposed to measure the drop size in the present study. The photographic chamber is designed to isolate and measure the targeted spray droplet, and reduce the interference of drop size statistics because of the droplet re-break caused by the manual operation. As a result, the PDF of drop size, the global Sauter mean and the number mean drop size are obtained. At the end, the experimental PDF is compared with the standard empirical distributions. The comparison shows NukiyamaTanasawa distribution provides the best t to the experimental data for the nozzles with lower pressure drop and larger ow rate. The suitable range of the parameters is 4 < rNT2 < 1 and rNT4 < 2. The present work shows the experimental methodology and results of spraying behaviours of the nozzle in pressurizer. The work is instructive for the optimization design of spraying systems. Acknowledgements The authors are grateful for the funding and the support from Natural Science Foundation of China (Grant Nos. 11075124 and 11125522). References
Anglart, H., Alavyoon, F., Novarini, R., 2010. Study of spray cooling of a pressure vessel head of a boiling water reactor. Nucl. Eng. Des. 240, 252257. Babinsky, E., Sojka, P.E., 2002. Modeling drop size distributions. Prog. Energy Combust. 28, 303329. Bhatia, J.C., Durst, F., 1989. Comparative-study of some probability-distributions applied to liquid sprays. Part Part Syst. Charact. 6, 151162. Bhatia, J.C., Durst, F., 1990. Description of sprays using joint hyperbolic distribution in particle-size and velocity. Combust. Flame 81, 203218. Bhatia, J.C., Domnick, J., Durst, F., Tropea, C., 1988. Phase-Doppler-anemometry and the log-hyperbolic distribution applied to liquid sprays. Part Part Syst. Charact. 5, 153164. Chen, Z., Hao, J., Chen, W., 2011. The development of fast simulation program for marine reactor parameters. Ann. Nucl. Energy 40, 4552. Crow, E.L., Shimizu, K., 1988. Lognormal Distributions: Theory and Applications. Marcel Dekker. Diao, W., 1988. Experiments on the spray nozzles used in the pressurizer of power reactor. China Nucl. Sci. Technol. Rep. 00, 10501064.

Fig. 25. Comparison of the four predictive model of nozzle 8# PDF (DP = 0.20 Mpa).

Z. Lan et al. / Annals of Nuclear Energy 63 (2014) 215227 Gong, J.-S., Fu, W.-B., 2007. The experimental study on the ow characteristics for a swirling gasliquid spray atomizer. Appl. Therm. Eng. 27, 28862892. Gonzalez-Tello, P., Camacho, F., Vicaria, J.M., Gonzalez, P.A., 2008. A modied NukiyamaTanasawa distribution function and a RosinRammler model for the particle-size-distribution analysis. Powder Technol. 186, 278281. Guo, L., Yan, Y.Y., Maltson, J.D., 2011. Numerical study on discharge coefcients of a jet in crossow. Comput. Fluids 49 (1), 323332. Halder, M.R., Dash, S.K., Som, S.K., 2004. A numerical and experimental investigation on the coefcients of discharge and the spray cone angle of a solid cone swirl nozzle. Exp. Therm. Fluid Sci. 28, 297305. Hamid, A.H.A., Atan, R., 2009. Spray characteristics of jet-swirl nozzles for thrust chamber injector. Aerosp. Sci. Technol. 13 (45), 192196. Kim, S.-N., Grifth, P., 1987. PWR pressurizer modeling. Nucl. Eng. Des. 102, 199 209. Kim, W.T., Mitra, S.K., Li, X., Prociw, L.A., Hu, T.C.J., 2003. A predictive model for the initial droplet size and velocity distributions in sprays and comparison with experiments. Part Part Syst Charact. 20, 135149. Lemaitre, P., Porcheron, E., 2009. Study of heat and mass transfers in a spray for containment application: analysis of the inuence of the spray mass ow rate. Nucl. Eng. Des. 239, 541550. Li, X.G., Tankin, R.S., 1987. Droplet size distribution a derivation of a Nukiyama Tanasawa type distribution function. Combust. Sci. Technol. 56, 6576. Li, X.G., Tankin, R.S., 1988. Derivation of droplet size distribution in sprays by using information-theory. Combust. Sci. Technol. 60, 345357. Li, X.G., Tankin, R.S., 1989. Prediction of droplet size and velocity distributions in sprays using maximum-entropy principle. Combust. Sci. Technol. 68, 147155. Mugele, R.A., Evans, H.D., 1951. Droplet size distribution in sprays. Ind. Eng. Chem. 43, 13171324.

227

Nukiyama, S., Tanasawa, Y., 1950. Experiments on the Atomization of Liquids in an Air Stream: reports 16. Defence Research Board, Dept. of National Defence. Paloposki, T., 1991. The physically relevant parameter space of the Nukiyama Tanasawa distribution function. Part Part Syst. Charact. 8, 287293. Paloposki, T., 1994. Drop size distributors in liquid sprays. Acta Polytech Sc Me, 27. Porcheron, E., Lemaitre, P., Marchand, D., Plumecocq, W., Nuboer, A., Vendel, J., 2010. Experimental and numerical approaches of aerosol removal in spray conditions for containment application. Nucl. Eng. Des. 240, 336343. Robert, J.L., 1996. Understanding aging in containment cooling systems. Nucl. Eng. Des. 163, 301314. Sellens, R.W., 1989. Prediction of the drop size and velocity distribution in a spray, based on the maximum-entropy formalism. Part Part Syst Charact. 6, 1727. Sovani, S.D., Sojka, P.E., Sivathanu, Y.R., 2000. Prediction of drop size distributions from rst principles: joint PDF effects. Atom. Spray 10, 587602. Torres, W.M., Baptista, B.D., Ting, D.K.S., 1999. The design and experimental validation of an emergency core cooling system for a pool type research reactor. Ann. Nucl. Energy 26, 709728. Tratnig, A., Brenn, G., 2010. Drop size spectra in sprays from pressure-swirl atomizers. Int. J. Multiphase Flow 36, 349363. Vesilind, P.A., 1980. The RosinRammler particle-size distribution. Resour. Recov. Conserv. 5, 275277. Weber, C., 1931. Zum Zerfall eines Flssigkeitsstrahles. Z. Angew. Math. Mech. 11, 136154. Zaidi, S.H., Altunbas, A., Azzopardi, B.J., 1998. A comparative study of phase Doppler and laser diffraction techniques to investigate drop sizes in annular two-phase ow. Chem. Eng. J. 71, 135143. Zeng, H.C., Han, C.Y., Wu, X.Z., et al., 1990. Combustion Technology. The Press of Huazhong University of Sci. & Tech..