Escolar Documentos
Profissional Documentos
Cultura Documentos
Mem. ASME
Institute of Nuclear Technology,
64 Sata, Mihama-cho, Miata-gun,
Correlation of Interfacial Friction
Fukui 919-1205, Japan
e-mail: murase@inss.co.jp Coefficients for Predicting
Takayoshi Kusunoki
Institute of Nuclear Technology,
Countercurrent Flow Limitation
64 Sata, Mihama-cho, Mikata-gun,
Fukui 919-1205, Japan at a Sharp-Edged Lower End of
e-mail: kusunoki.takayoshi@inss.co.jp
Koji Nishida
Vertical Pipes
Institute of Nuclear Technology,
64 Sata, Mihama-cho, Mikata-gun, One-region (1-R) sensitivity computations with the annular-flow model were carried out
Fukui 919-1205, Japan for countercurrent flow limitation (CCFL) at a sharp-edged lower end in vertical pipes to
e-mail: nishida.koji@inss.co.jp generalize the prediction method for CCFL there (i.e., predicting effects of diameters and
fluid properties on CCFL characteristics). In our previous study, we selected a correla-
Raito Goda tion of interfacial friction coefficients, fi, with a function of average void fraction which
Graduate School of Engineering,
gave a good prediction of the trend for air–water CCFL data, and we modified it to get
Kobe University,
good agreement with steam–water CCFL data under atmospheric pressure conditions,
1-1 Rokkodai, Nada-ku,
but it failed to predict CCFL reasonably at high pressure conditions. We recently found a
Kobe-shi 657-8501, Hyogo, Japan
Russian report on CCFL data at high pressure conditions, by which we improved the fi
e-mail: goda@cfrg.scitec.kobe-u.ac.jp
correlation using the dimensionless diameter and the viscosity ratio or density ratio of
gas and liquid phases to get good agreement with CCFL data at high pressures. The
improved fi correlation with the viscosity ratio and the improved fi correlation with the
Akio Tomiyama density ratio gave similar computed results, but the number of adjustment functions
Graduate School of Engineering,
was one for the density ratio and two for the viscosity ratio (i.e., minimum value of two
Kobe University,
functions). [DOI: 10.1115/1.4039438]
1-1 Rokkodai, Nada-ku,
Kobe-shi 657-8501, Hyogo, Japan
e-mail: tomiyama@mech.kobe-u.ac.jp
1 Introduction For CCFL in the hot leg, some experimental studies have been
done with rather large diameter models [2–5], but effects of the
Under postulated accident conditions in a pressurized water
diameter and fluid properties on CCFL characteristics were not
reactor (PWR), steam and condensate water form countercurrent
clarified. In our research group, therefore, Minami et al. [6] meas-
flows in a steam generator (SG) heat transfer tube (which is an
ured CCFL characteristics in a 1/15-scale model of a hot leg using
inverted U-tube), a hot leg (consisting of an inclined short pipe, a
air and water, and Murase et al. [7] did 3D numerical simulations
vertically-placed 50-deg elbow, and a horizontal pipe), and a pres-
for the full-scale model using a volume of fluid method imple-
surizer surge line (consisting of a vertical pipe, a vertical elbow,
mented in the computational fluid dynamics software FLUENT6.3.26
and a slightly inclined pipe with elbows), and countercurrent flow
and derived a Wallis-type CCFL correlation by using simulated
limitation (CCFL) may occur. In the safety analyses of nuclear
results and existing CCFL data, in which the characteristic length
power plants, the Wallis correlation [1] (which gives the relation-
of the Wallis parameter [1] is the diameter, D. On the other hand,
ship between superficial velocities of gas and liquid phases) is
there are only a few studies on CCFL in the pressurizer surge line
widely used to compute the falling liquid flow rate against the
[8–10]. Therefore, Futatsugi et al. [11] in our group observed
updraft gas flow rate. In the Wallis correlation, however, the
flooding locations and measured CCFL characteristics in a 1/10-
empirical constants (m and C) generally depend on the flooding
scale model of a pressurizer surge line using air and water. How-
location, the pipe diameter, the pipe length, and fluid properties.
ever, each PWR plant has its own layout of the pressurizer surge
Therefore, the common technical issue is how we can apply the
line, and so a generalized method to predict CCFL characteristics
CCFL correlation derived from small-scale test results to PWR
in the slightly inclined pipe with elbows was needed for the pres-
conditions. In our research group at Kobe University and the Insti-
surizer surge line. To generalize a method predicting CCFL char-
tute of Nuclear Safety System, Inc. (INSS), in which the represen-
acteristics in slightly inclined pipes with elbows, Murase et al.
tatives are Prof. A. Tomiyama at Kobe University and Dr. M.
[12] developed a 1D computation method, and they [13] validated
Murase at INSS, we have been working to improve CCFL correla-
1D computations in comparison with CCFL data in nearly hori-
tions for the SG heat transfer tube, hot leg, and pressurizer surge
zontal pipes and 3D simulations for the full-scale surge line
line for over 10 years. At Kobe University, we carried out small-
model. As a result, the 1D computation method was validated for
scale air–water experiments to understand CCFL behavior and to
the broad conditions of the inclination angle of h ¼ 0–1 deg, the
obtain a basic database, and at INSS we made three-dimensional
diameter of D ¼ 0.03–0.65 m, and the length to the diameter ratio
(3D) numerical simulations and one-dimensional (1D) or one-
of L/D ¼ 4.5–63 [12,13].
region (1-R) computations with a simple stratified or annular flow
For flooding in vertical pipes, many studies have been made
model to predict CCFL under PWR conditions.
and reviews are available [1,14]; but most have dealt with the
onset of flooding and studies on CCFL are limited. In our group,
1
Corresponding author.
therefore, Kusunoki et al. [15,16] did air–water and steam–water
Manuscript received May 25, 2017; final manuscript received February 15, 2018; CCFL experiments simulating the lower part of the SG heat trans-
published online May 16, 2018. Assoc. Editor: Walter Ambrosini. fer tube with the sharp-edged lower end (CCFL-L), and derived a
Journal of Nuclear Engineering and Radiation Science JULY 2018, Vol. 4 / 031001-1
C 2018 by ASME
Copyright V
Note: Databases for CCFL-U and CCFL-P are from air-water experiments.
CCFL-L correlation [17], in which the characteristic length in the Hk* in Eq. (1) becomes the Wallis parameter, Jk*, for b ¼ 0 in
Wallis parameter is the diameter, D. Doi et al. [18] also did Eq. (3) and the Kutateladze parameter, Kk*, for b ¼ 1. By using
air–water CCFL experiments simulating the upper part of the the dimensionless diameter, D*, Jk*, and Kk* can be exchanged for
pressurizer surge line with the sharp-edged upper end (CCFL-U), each other by
and Murase et al. [19] derived a CCFL-U correlation, in which the
characteristic length in the Wallis parameter is the Laplace capil- Jk ¼ Kk =D1=2 ; D ¼ D=LL ; ðk ¼ G or LÞ (5)
lary length, LL. From these results [17,19] and currently available
CCFL data inside vertical pipes (CCFL-P), Yamamoto et al. [20]
classified CCFL characteristics according to their dependence on 2.2 Effects of Diameter on Countercurrent Flow Limitation
the limiting location into CCFL-U at the sharp-edged upper end, Characteristics. Yamamoto et al. [20] classified CCFL character-
CCFL-L at the sharp-edged lower end and CCFL-P inside the ver- istics according to their dependence on the limiting location into
tical pipe with round-edged upper and lower ends, where values CCFL-U at the sharp-edged upper end, CCFL-L at the sharp-
of the characteristic length in the Wallis parameter are, respec- edged lower end and CCFL-P inside the vertical pipe with round-
tively, LL, D, and D0.5LL0.5. CCFL databases for vertical pipes, edged upper and lower ends, where values of the characteristic
however, are limited especially for large diameter pipes and high length, L, in Eq. (3) are, respectively, LL, D and D0.5LL0.5. Table 1
pressure steam–water conditions. Therefore, Kusunoki et al. [21] summarizes m and Cj in Eq. (1), and Fig. 1 shows the relationship
did 1-R computations for CCFL-L based on the annular flow between CK and D*. The maximum diameter in the database used
model by using several correlations for interfacial friction coeffi- for Table 1 and Fig. 1 is D ¼ 140 mm with air and water (D* ¼ 51)
cients, fi, to generalize a CCFL prediction method, selected an fi for CCFL-U.
correlation, and modified it with the dimensionless diameter, D*, For CCFL-L in this study, the upper limit of the dimensionless
and the ratio of viscosities of gas and liquid, lG/lL, to get good diameter, D*, may be about 31 (where CK reaches 1.79 for CCFL-
agreement with CCFL data under atmospheric pressure condi- P), and CCFL-P may become limiting in the region of D* 31.
tions, but our fi correlation could not be applied to high pressure Databases used for CCFL-U and CCFL-P are from air–water
steam–water conditions. experiments, and steam–water experiments are available only for
In this study, we did 1-R sensitivity computations with the CCFL-L with D ¼ 20 mm.
annular-flow model for CCFL-L data reported by Ilyukhin et al. [22]
under high pressure conditions (D ¼ 20 mm and pressures of 2.3 Effects of Fluid Properties on Countercurrent Flow
P ¼ 0.6–4.1 MPa) and improved our earlier fi correlation [21] by Limitation Characteristics. We recently found the literature by
using D* and lG/lL or the ratio of densities of gas and liquid, qG/qL. Ilyukhin et al. [22] that reported data for high pressure conditions
(D ¼ 20 mm and pressures of P ¼ 0.6–4.1 MPa), and Murase et al.
2 Outline of Countercurrent Flow Limitation [24] derived a new CCFL correlation for CCFL-L and CCFL-U.
Characteristics in Vertical Pipes
2.1 General Form of Countercurrent Flow Limitation
Correlation. The CCFL correlation proposed by Wallis [1] is
widely used in safety analysis computer programs and is
expressed by
1=2 1=2
HG þ mHL ¼ Cj ; ðj ¼ K or WÞ (1)
qk 1=2
Hk ¼ Jk ; ðk ¼ G or LÞ (2)
gLðqL qG Þ
where D (m) is the diameter, LL (m) is the Laplace capillary Fig. 1 Countercurrent flow limitation constants, CK, in vertical
length, and r (N/m) is the surface tension. pipes [20]
dPG dPT
¼ (9)
dz dz
Journal of Nuclear Engineering and Radiation Science JULY 2018, Vol. 4 / 031001-3
4 Results
4.1 Adjustment Factors. The major experimental conditions
for CCFL-L at the sharp-edged lower end are listed in Table 2.
Adjustment factors Nfi for the fi correlation by Bharathan et al.
[28], Eq. (12), were obtained to get good agreement between the
computed results and CCFL data for each experimental case in
Table 2. Effects of fluid properties on CCFL in vertical pipes were
well expressed by functions of the viscosity ratio, lG/lL, as shown
in Fig. 2. Therefore, values of Nfi, for Eq. (12), are plotted for lG/
lL in Fig. 3(a). Equation (12) was developed from air–water data,
but Nfi for air–water conditions was 1.7–2.0, and this means that
Eq. (12) overestimates JL for air–water data. From values of Nfi in
Fig. 3(a), a suitable function of lG/lL for Nfi was obtained by the
least-square method, but they were expressed by two functions of
lG/lL as shown in Fig. 3(a). The values of Nfi with different diam-
eters were different for air–water [15,30,31] and steam–water [16]
at P ¼ 0.1 MPa. Therefore, the adjustment factor, Na, was devel-
oped based upon the fi value after applying the Nfi adjustment fac-
tor to get good agreement between the computed results and
CCFL data as shown in Fig. 3(b). From values of Na in Fig. 3(b),
a suitable function of D* for Na was obtained by the least-square
method. Consequently, the improved correlation for Eq. (12) was
derived
n pffiffiffi o Fig. 3 Adjustment factors, Nfi and Na, for the fi correlation,
fi ¼ 0:005 þ 15:75½ð1 aÞ=a a1:96 min½0:30 ln ðlG =lL Þ Eq. (12), with functions of the viscosity ratio, lG/lL, and the
dimensionless diameter, D * (A, air; S, steam; W, water): (a)
þ 3:0; 0:78 ln ðlG =lL Þ 0:45 0:64D0:19 (15) effects of lG/lL on Nfi and (b) effects of D * on Na
Fig. 5 Adjustment factors, Nfi and Na, for the fi correlation, Eq.
(12), with functions of the density ratio, qG/qL, and the dimen-
Fig. 4 Comparison between computed values with Eq. (15) sionless diameter, D* (A, air; S, steam; W, water): (a) effects of
and experimental values qG/qL on Nfi and (b) effects of D* on Na
Journal of Nuclear Engineering and Radiation Science JULY 2018, Vol. 4 / 031001-5
5 Discussion
Countercurrent flow limitation-L at the sharp-edged lower end
in vertical pipes is basically expressed by the Wallis parameters,
but some modifications need to be applied to a wide range of the
diameters and fluid properties. CCFL-L data are limited for large
diameter and high pressure steam–water conditions. In an earlier
study [24], we derived the CCFL-L correlation, Eq. (6), from
CCFL-L data [15,16,22] with D ¼ 20 mm. In this study, we did 1-
R computations and obtained adjustment factors for the correla-
tion of interfacial friction coefficients expressed by Eq. (16) to get
good agreement between the computed results with CCFL-L data
[15–17,22,30,31] with D ¼ 14–51 mm. Similar databases were
used to derive Eqs. (6) and (16), but the difference between Eq.
(6) and results computed with Eq. (16) became relatively large for
D ¼ 51 mm and P ¼ 7 MPa as shown in Fig. 7(b). The different
approaches of Eq. (6) and of the 1-R computations with Eq. (16)
to predict CCFL are important to evaluate uncertainty of the pre-
diction, because we have to use CCFL correlations beyond their
experimental conditions in safety analyses.
One of the important technical issues for the CCFL prediction
is the correlation for interfacial friction coefficients. Interfacial
friction coefficients are generally expressed by a function of the
void fraction [25,27,28] or the liquid film thickness [26], which
can be changed to a function of the void fraction. The void frac-
tion is strongly affected by the pressure and the density ratio of
Fig. 6 Comparison between values computed with Eq. (16) gas and liquid (cf., Fig. 5(a)). On the other hand, CCFL character-
and experimental values istics are strongly affected by the liquid viscosity (cf., Fig. 2).
Nomenclature
C¼ CCFL constant
D¼ diameter, m
D* ¼ dimensionless diameter
fi ¼ interfacial friction coefficient
fw ¼ wall friction coefficient
g¼ gravitational acceleration, m/s2
H* ¼ dimensionless velocity defined by Eq. (2)
J¼ superficial velocity, m/s
J* ¼ Wallis parameter
K* ¼ Kutateladze parameter
L¼ characteristic length expressed by Eq. (3), m
LL ¼ Laplace capillary length defined by Eq. (4), m
m¼ slope in the Wallis CCFL correlation
N¼ adjustment factor for fi
P¼ pressure, Pa
Re ¼ Reynolds number
z¼ vertical coordinate, m
a¼ void fraction
l¼ viscosity, Pa
s
q¼ density, kg/m3
r¼ surface tension, N/m
Subscripts
Exp ¼ experiment
G¼ gas phase
j¼ K or W
k¼ G or L
K¼ Kutateladze parameter
L¼ liquid phase
W¼ Wallis parameter
Fig. 7 Prediction of CCFL characteristics at P 5 7 MPa (A, air; 1R ¼ one-region computation
S, steam; W, water): (a) D 5 20 mm and (b) D 5 51 mm
References
Therefore, it may be important to find another major factor suita- [1] Wallis, G. B., 1969, One-Dimensional Two-Phase Flow, McGraw-Hill, New
ble for the correlation of interfacial friction coefficients, rather York, pp. 336–345.
than use the void fraction. [2] Richter, H. J., Wallis, G. B., Carter, K. H., and Murphy, S. L., 1978,
“Deentrainment and Countercurrent Air-Water Flow in a Model PWR Hot-
Leg,” U. S. Nuclear Regulatory Commission, Washington, DC, Report No.
NRC-0193-9.
6 Conclusion [3] Mayinger, F., Weiss, P., and Wolfert, K., 1993, “Two-Phase Flow Phenomena
in Full-Scale Reactor Geometry,” Nucl. Eng. Des., 145(1–2), pp. 47–61.
In this study, we did 1-R computations based on the annular [4] Geffraye, G., Bazin, P., Pichon, P., and Bengaouer, A., 1995, “CCFL in Hot
flow model for CCFL-L at the sharp-edged lower end in vertical Legs and Steam Generators and Its Prediction With the CATHARE Code,” Sev-
enth International Topical Meeting on Nuclear Reactor Thermal Hydraulics
pipes and compared the results with CCFL-L data measured under (NURETH-7), Saratoga Springs, NY, Sept. 10–15, pp. 815–826.
the conditions of D ¼ 14–51 mm and P ¼ 0.1–4.1 MPa to general- [5] Al Issa, S., and Macian, R., 2014, “Experimental Investigation of Countercur-
ize a prediction method of CCFL-L. We used the correlation for rent Flow Limitation (CCFL) in a Large-Diameter Hot-Leg Geometry: A
interfacial friction coefficients, fi, proposed by Bharathan et al. Derailed Description of CCFL Mechanisms, Flow Patterns and High-Quality
HSC Imaging of the Interfacial Structure in a 1/3.9 Scale of PWR Geometry,”
[28], which is a function of the void fraction, a. To obtain good Nucl. Eng. Des., 280, pp. 550–563.
agreement between the computed results and the data, we [6] Minami, N., Nishiwaki, D., Nariai, T., Tomiyama, A., and Murase, M., 2010,
improved the fi correlation by using the dimensionless diameter, “Countercurrent Gas-Liquid Flow in a PWR Hot Leg Under Reflux Cooling (I)
Journal of Nuclear Engineering and Radiation Science JULY 2018, Vol. 4 / 031001-7