Anda di halaman 1dari 5

Shin-ichi Nakao

Senior Research Scientist,


e-mail: nakao@nrlm.go.jp

Masaki Takamoto
Group Leader
e-mail: takamoto@nrlm.go.jp
Flow Measurement Section,
National Research Laboratory of Metrology,
1-4, Umezono-1, Tsukuba, Ibaraki, Japan

Discharge Coefficients of Critical


Venturi Nozzles for CO2 and SF6
The discharge coefficients of critical Venturi nozzles were measured for CO 2 and SF 6 on
the Reynolds number range from 3 103 to 2 105 . The results showed that the measured
discharge coefficients for both gases were about 2 percent larger than the theoretical
estimation based on the assumption of isentropic flow of a perfect gas and this large
deviation could not be reduced even by introducing real gas effects. The experimental
results also showed that the large deviation for CO 2 could be explained through the
assumption of a nonequilibrium flow at the throat. On the other hand, the reason of the
deviation observed for SF 6 has not been clear yet, but one possible explanation would be
the inadequate estimation of the boundary layer at the throat because the theory is based
on the laminar boundary layer of a perfect gas. S0098-22020002004-6

Introduction

Many kinds of gases are used as process gases in semiconductor industries or as calibration gases for instruments used in environmental measurements and so on. Presently, a thermal type
mass flow controller MFC is mainly used to control and measure
gas flow rates in these fields. However, when a gas flow rate is
measured by a MFC, which is not calibrated by that gas, the
output of the MFC must be corrected with a questionable correction factor. Thus, the reliability and accuracy of flow measurements is questionable in such cases. Therefore, the development
of a highly accurate flow measurement technique independent of
the kind of gas is required.
National Research Laboratory of Metrology NRLM has developed the flow measurement technique using the critical Venturi
nozzles to transfer the standard of small mass flow rates of gases
Nakao and Takamoto 1 to accreditation laboratories. Ishibashi
et al. 2 designed super-fine manufactured critical Venturi
nozzles, which have exactly the same shape as the nozzle used in
the theoretical analyses of Geropp 3, and found that the measured discharge coefficients of these nozzles were within 0.03
percent of the theoretical estimation which was developed by Ishibashi and Takamoto 4 based on the analysis of Hall 5 and
Geropp 3. The theoretically determined discharge coefficient is
described by two parameters determined under the assumptions
that a core flow is isentropic and a perfect gas, which are a function of the specific heat ratio and the nozzle shape. Thus, if a gas
is close to the perfect gas state, the discharge coefficient of the
critical Venturi nozzle for that gas can be theoretically determined
using these two parameters. Nakao et al. 6 measured the discharge coefficients of the critical Venturi nozzle for ten gases
including CO2 and SF6 . And the results showed that the measured
discharge coefficients for the gases tested, except CO2 and SF6 ,
were within 0.6 percent of the theoretical estimation developed
by Ishibashi 4. However, the results of CO2 and SF6 were 2
percent larger than the theoretical estimation.
This paper describes the results of the discharge coefficients for
CO2 and SF6 measured for the four critical Venturi nozzles with
different throat diameters, and the behaviors of the two parameters
describing the discharge coefficient were investigated and the flow
fields at the throat were discussed to explain the large deviations
found in these gases.
Contributed by the Fluids Engineering Division for publication in the JOURNAL
OF FLUIDS ENGINEERING. Manuscript received by the Fluids Engineering Division
January 10, 2000; revised manuscript received August 17, 2000. Associate Technical
Editor: P. Bearman.

730 Vol. 122, DECEMBER 2000

2 Theoretical Estimation of the Discharge Coefficient


of a Critical Venturi Nozzle
The discharge coefficient of a critical nozzle, C d , is defined by
C d Q m /Q mth

(1)

where Q m is the actual mass flow rate and Q mth is the theoretical
mass flow rate calculated from the throat condition and is written
as
Q mth A * c * ,

(2)

where A is the throat area, is the density of gas, and c is the


sound speed. The theoretical Reynolds number, Reth , is given by
RethQ mth D/ A 0

(3)

where D is the throat diameter and 0 is the viscosity of gas at the


stagnation condition. The superscript * refers to the throat condition. The throat condition is determined from the stagnation
condition using the isentropic relation of a perfect gas. Thus, the
throat condition can be calculated from the measurements of the
upstream pressure and temperature of a critical nozzle.
When the boundary layer at the throat is laminar, the discharge
coefficient can be generally described as follows ISO 9300 7;
C d ab/ Reth 0.5.

(4)

The two parameters in Eq. 4, a and b, can be determined


theoretically on the assumption of a perfect gas isentropic flow.
The details of the theoretical analyses are described in Hall 5
and Geropp 3. Therefore, only brief explanations of the results
are given here.
2.1 Determination of a. a reflects the defect of the
mass flow rate through the nozzle due to the distortion of a velocity profile at the throat, which is caused by the acceleration flow
from the entrance region to the throat. Hall 5 calculated analytically the velocity components near the throat using the expansions
in inverse powers of the radius of curvature of the nozzle wall at
the throat, R. In his calculations, he assumed axially symmetric
irrotational flow and used the isentropic relations of a perfect gas.
Finally, a is written using the first three terms of the series
solution as follows
a1 1 /K 2 1/96 8 21 / 4608K
754 2 1971 2007 / 552960K 2 . . .

(5)

where KR2/D, D is the throat diameter and is the specific


heat ratio of gas.

Copyright 2000 by ASME

Transactions of the ASME

Downloaded From: http://fluidsengineering.asmedigitalcollection.asme.org/ on 04/27/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

Fig. 1 Schematic diagram of the ISO type toroidal throat Venturi nozzle and their throat diameters

2.2 Determination of b. Another parameter influencing


the mass flow rate through a nozzle is the displacement thickness
of a laminar boundary layer developing at the throat, which reduces the effective area of the throat. Geropp 3 and Tang and
Fenn 8 obtained analytically the exact solution of the Prandtl
boundary layer equation for axially symmetric flow with an appropriate boundary condition, which gives a nozzle profile. They
also used the assumption of isentropic flow and a perfect gas.
Finally, b is written as follows
b4/m 1 /2 1/ 2 1 3 22 3 1 / 3 (6)
where the nozzle shape m is defined by
m 8/ K 1 1/4 1 /2 2 1 / 2 1

(7)

As found from Eqs. 5 and 6, the parameters, a and b,


are functions of the specific heat ratio and the nozzle shape. The
nozzles used in the present paper have a similar shape and K is
equal to 4 because the radius of curvature of the throat is twice D
as shown in Fig. 1. Thus, a and b become functions of only
the specific heat ratio, and consequently, the discharge coefficient
of the critical Venturi nozzle can be theoretically determined using Eq. 5 and Eq. 6 with Eq. 7.

Fig. 2 The variations of the discharge coefficients versus the


theoretical Reynolds number : from Nakao et al. 6

and the fitted curve determined by the least square method from
the data in Fig. 3 was 0.05 percent for SF6 . It should be noticed
that the results for CO2 and SF6 are quite different from other
gases.
The uncertainty of the discharge coefficient consists of the
components summarized in Table 1. The standard uncertainty of
the calibration facility, by which the discharge coefficient of the
critical nozzle is determined, depends on the flow rate and the
kind of gas, and the maximum values for N2 , CO2 , and SF6 on
the present flow rate range are listed in Table 1. The repeatability
of the measurement, which is determined from the measurements
of the upstream pressure and temperature of the nozzles, was
0.043 percent under the present experimental conditions. Another
important component is the uncertainty of the throat diameter,
which depends on the resolution of the measuring tool. The throat
diameter of the critical nozzle was measured by the measuring
tool with the resolution of 1 m. Therefore, the relative standard

Experimental Results

The discharge coefficient of the critical Venturi nozzle was determined by the gravimetric calibration facility of NRLM Nakao
and Takamoto 1. When the test gas was changed, the whole
facility was purged with the new test gas for two whole days. As
the vapor rate included in the bottled gases used here was less
than 0.01 percent, the vapor effect on the gas density was neglected. The critical nozzles used here are an ISO type toroidal
throat Venturi nozzle as shown in Fig. 1. The radius of curvature
at the throat is twice D, and the diffuser with a half angle of 3
degrees is three times D in length. The throat diameters of the
critical Venturi nozzles used here are listed in Fig. 1. The critical
nozzles were manufactured at the center of the disk by the machining technique and the throat diameter was measured along
four different diameters by the profile projector with the resolution of 1 m and the averaged value of the four measurements
was used as the throat diameter.
The Reynolds number dependency of the discharge coefficient
was investigated for each gas by changing the upstream pressure
of the critical nozzle from 30 kPa to 300 kPa. The results are
shown in Fig. 2, in which the symbol is the result of the
gases in Nakao et al. 6, that is, N2 , Ar, He, H2 , O2 , CH4 ,
C2H2 , and C2H6 . It is found from this figure that the discharge
coefficients strongly depend on the kind of gas and the smaller the
Reynolds number, the stronger the Reynolds number dependency
of the discharge coefficient. The transverse axis of this figure is
the theoretical Reynolds number defined in Eq. 3. The results in
Fig. 2 are plotted in Fig. 3 versus the square root of the inverse of
the theoretical Reynolds number. The symbol is also the
results of Nakao et al. 6. Figure 3 shows that the linear relation
of Eq. 4 can be applied to all gases tested. The maximum standard deviation of the residual between the experimental results
Journal of Fluids Engineering

Fig. 3 The linear relation between the discharge coefficient


and the inverse of the square root of the theoretical Reynolds
number
Table 1 The sources and magnitude of the uncertainties of the
A1 sonic nozzle for gases

DECEMBER 2000, Vol. 122 731

Downloaded From: http://fluidsengineering.asmedigitalcollection.asme.org/ on 04/27/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

Fig. 4 The variation of a with specific heat ratio : from


Nakao et al. 6

Fig. 5 The variation of b with specific heat ratio : from


Nakao et al. 6

uncertainties of the throat diameters were 0.04 percent for the B1


nozzle, and 0.34 percent for the A3 nozzle. The thermophysical
properties like a specific heat ratio were given by polynomial
functions of the pressure and the temperature, which were determined from the tables of JSME Data Book 9. The standard
deviations of the curve fit residuals were less than 0.03 percent
in each thermophysical property. Finally, the combined standard
uncertainties of the discharge coefficients of the nozzles were 0.1
percent for the largest nozzle and 0.68 percent for the smallest
nozzle as shown in Table 1.
The a and b values of Eq. 4 were derived from the
fitted curve for each gas in Fig. 3. The theoretical results calculated from Eq. 5 and Eq. 6 are also indicated by the solid lines
in these figures. The variations of a versus the specific heat
ratio are shown in Fig. 4, which includes the results of Nakao
et al. 6 symbol . The result for CO2 deviates from the
theoretical value by more than 2 percent, but that of SF6 is very
close to the theoretical estimation. The behavior of b versus
the specific heat ratio is shown in Fig. 5. The results show that the
displacement thickness of a laminar boundary layer becomes thinner as the specific heat ratio decreases, and the result of SF6 is
rather smaller than the theoretical value, but that of CO2 is close
to the theoretical one. This suggests that the large deviations between the theoretical estimations and the experimental results for
CO2 and SF6 are caused by different reasons; the wrong estimations of the defect of the mass flow in the core flow in CO2 and
the boundary layer thickness at the throat in SF6 .

Fig. 6 The discharge coefficients of the four nozzles for CO2


versus the inverse of the square root of the theoretical Reynolds number

charge coefficients for CO2 and SF6 were recalculated introducing


real gas effects according to Johnsons method Johnson 10. By
doing this, the differences were decreased by about 0.5 percent.
The results were slightly improved, but this correction did not
bring any essential changes to the results. Therefore, the discharge
coefficients of the four critical Venturi nozzles with different
throat diameters were measured for CO2 and SF6 to investigate
how the characteristics of their discharge coefficients were
changed by the nozzle size.
The discharge coefficients of the four critical Venturi nozzles
for CO2 are shown in Fig. 6. The transverse axis is the square root
of the inverse of the theoretical Reynolds number and the solid
line indicates the theoretical estimation. All of the experimental
results are on the same curve, which is slightly curved in the range
of large theoretical Reynolds numbers. Thus, the two parameters
of the discharge coefficient, a and b were determined for
each nozzle by the least square method. The variations of a in
CO2 versus the throat diameter of a critical nozzle are shown in
Fig. 7. The figure indicates that a decreases toward the theoretical value as the throat diameter increases. This means that the
mass flow defect in the core flow depends on the throat diameter
i.e., the size of the flow field.
When a flow property such as pressure or temperature is
changed, a gas redistributes its internal energy to a new flow
condition exchanging the energy between translational, rotational,
and vibrational modes through the collisions between molecules.
In a nozzle flow, the gas temperature of the core flow quickly
decreases toward the throat so that the gas is about to redistribute
the internal energy to a new flow condition. However, when the
flow passage is short and narrow like the present nozzle, the time
that the flow passes through the passage is shorter than the relaxation time of the energy mode, and thus the gas will be unable to

Discharge Coefficients of CO2 and SF6

As the theoretical analyses of Hall 5 and Geropp 3 are based


on the assumption of isentropic flow of a perfect gas, the dis732 Vol. 122, DECEMBER 2000

Fig. 7 The variation of a versus the throat diameter for CO2

Transactions of the ASME

Downloaded From: http://fluidsengineering.asmedigitalcollection.asme.org/ on 04/27/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

Fig. 8 The comparison of the measured discharge coefficients


and the theoretical discharge coefficients with 1.38 under a
nonequilibrium flow condition for CO2

have enough collision number to relax that energy mode Levine


and Bernstein 11. As a result, that energy mode is considered to
be unable completely to recover to the equilibrium condition.
Generally, the relaxation time for vibrational mode is quite longer
than that for translational and rotational modes, and that of CO2 is
about 105 s at the present experimental condition Simpson et al.
12. And as the time that the flow passes through the present
nozzles is from 105 s to 106 s, it is the same order as the relaxation time for vibrational mode of CO2 .
Another important parameter to determine the flow condition,
equilibrium or nonequilibrium, is the characteristic temperature of
vibrational mode, v . The vibrational temperature which determines the vibrational motion is a function of the characteristics
temperature of vibrational mode. In a nozzle flow, the gas temperature decrease largely toward the throat, but the vibrational
temperature changes slightly in the case of the gases with the high
characteristics temperature, for example in N2 ( v 3340 K)
Leipmann and Roshko 13. On the other hand, in the gases with
the low characteristics temperature, for example in CO2 v
954 K: this is the lowest value. Leipmann and Roshko 13,
the vibrational temperature has a large increase. In this case, if the
relaxation time for vibrational mode is longer compared with the
time that the gas passes through a nozzle, the vibrational temperature will be unable to recover to the value of the equilibrium
condition, that is, the flow will be in nonequilibrium. A gas under
nonequilibrium condition must have a specific heat ratio between
an equilibrium flow and a frozen flow. In nonequilibrium flow, the
specific heat ratio is related to the internal freedom, and it becomes larger in a nonequilibrium flow, in which the internal freedom reduces.

Fig. 10 The variation of a versus the throat diameter for SF6

The discharge coefficients for CO2 were recalculated from both


the actual mass flow rates experimentally determined and the new
theoretical mass flow rates calculated with the specific heat ratio
of 1.38, the value between an equilibrium flow and a frozen
flow. Next, the discharge coefficients for CO2 were theoretically
determined from Eq. 5 and Eq. 6 with 1.38, that is, the
discharge coefficients for a fictitious perfect gas which has the
same viscosity and the same molecular weight as CO2 , but the
specific heat ratio of 1.38. The specific heat ratio is actually not
the same for the four nozzles because the level of nonequilibrium
is different in the different nozzles. However, the same value was
used for the four nozzles in the above calculations because there is
no way to determine the specific heat ratio of a nonequilibrium
flow. Both results shown in Fig. 8 are in very good agreement.
This suggests that the large deviation between the theoretical estimation based on Eq. 5 and Eq. 6 and the experimental results
for CO2 is explained by assuming that the flow at the throat is
nonequilibrium. This estimation that the flow is nonequilibrium at
the throat is supported by the reasons that CO2 has the low characteristic temperature for vibrational mode and the time that the
gas passes through the nozzle is the same order as the relaxation
time of vibrational mode.
The results of the discharge coefficients of the four critical Venturi nozzles for SF6 are shown in Fig. 9. All experimental results
make the line gradually curved over the whole range and here the
solid line indicates the theoretical estimation. The deviation between the theoretical and the experimental results is larger than 2
percent in the lower Reynolds number range. As in CO2 , the two
parameters of the discharge coefficient for each nozzle were determined from the data of Fig. 9. The results of a are shown in
Fig. 10. In this figure, a seems to increase with increasing a
throat diameter, but the change of a is very small, the order of
the measurement uncertainty. Therefore, it is likely that a is
independent of the throat diameter and the mass flow defect in the
core flow for SF6 must be estimated accurately by the theoretical
analyses. From the results that the standard deviations of the residual between the fitted curve and the experimental results were
around 0.1 percent in the four nozzles, it is suggested that the
linear relation of Eq. 4 based on a laminar boundary layer cannot
be applied to SF6 , that is, Eq. 6 cannot be used to estimate the
boundary layer thickness at the throat for SF6 .

Fig. 9 The discharge coefficients of the four nozzles for SF6


versus the inverse of the square root of the theoretical Reynolds number

Journal of Fluids Engineering

Discussion and Conclusion

The measured discharge coefficients for CO2 and SF6 were


about 2 percent larger than the theoretical estimation based on the
assumption of isentropic perfect gas, although those of any other
gas tested were within 0.6 percent of the results theoretically estimated. The large deviations for CO2 and SF6 could not be explained even by introducing real gas effects.
DECEMBER 2000, Vol. 122 733

Downloaded From: http://fluidsengineering.asmedigitalcollection.asme.org/ on 04/27/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

Therefore, the discharge coefficients of the four critical nozzles


with different throat diameters were measured for CO2 and SF6 to
investigate how the discharge coefficients are changed. The results for CO2 showed that the narrower the flow field, the larger
the discharge coefficient deviates from the theoretical value. This
suggests that the flow is nonequilibrium at the throat because the
gas cannot completely redistribute the internal energy to a new
flow condition due to a short and narrow flow field. The discharge
coefficients theoretically recalculated with the specific heat ratio
between an equilibrium flow and a frozen flow were in good
agreement with the experimental results. Yokogawa and Nishioka
14 carried out the numerical simulations of nonequilibrium
nozzle flow for CO2 on the basis of N-S equation using TVD
scheme and showed that the results were within 0.5 percent of the
experimental results for the four critical Venturi nozzles. And
Johnson et al. 15 confirmed numerically the influence of nonequilibrium flow on the discharge coefficients for CO2 and they
also investigated the nonequilibrium effects in vibrational mode
on the discharge coefficient by the experiments using mixtures of
CO2 and water vapor.
On the other hand, the large deviation between the theoretical
estimation and the experimental results found in SF6 may be explained by the wrong estimation of the boundary layer thickness at
the throat, not by the mass flow defect in the core flow. Back 16
reported in his paper that the laminar boundary layer becomes
much thinner when the boundary layer develops on the strongly
cooling wall in an acceleration flow. Meanwhile, a nozzle flow is
accelerated from the entrance to the throat and the temperature
drop in the boundary layer at the throat is considered to be fairly
large as the gas temperature decreases largely toward the throat.
However, as such a temperature drop in the boundary layer is
realized for other gases, it would be difficult to explain the results
of SF6 from the analogy the Backs study. Further study about the
boundary layer of SF6 , maybe from the viewpoint of the heat
transfer, will be required. And also the possibility of nonequilibrium flow like in CO2 may have to be discussed when considering
the fact that the characteristics temperatures of some vibrational
modes are comparable with that of CO2 .

734 Vol. 122, DECEMBER 2000

References
1 Nakao, S., and Takamoto, M., 1999, Development of the Calibration Facility
for Small Mass Flow rates of Gases and the Sonic Venturi Nozzle Transfer
Standard, JSME, Series B, 42, pp. 667673.
2 Ishibashi, M., Takamoto, M., and Nakao, Y., 1994, Precise Calibration of
Critical Nozzles of Various Shapes at the Reynolds Number of 0.82.5
105 , FLOMEKO94, Glasgow, Session 6.
3 Geropp, D., 1971, Laminare Grenzschichten in ebenen und rotationssymmetrischen Lavalduesen, Deutsche Luft und Raumfahrt Forschungsbericht in
Germany.
4 Ishibashi, M. and Takamoto, M., 1997, Very Accurate Analytical Calculation
of the Discharge Coefficients of Critical Venturi Nozzles with Laminar Boundary Layer, Proceedings of FLUCOME, Japan.
5 Hall, I. M., 1962, Transonic Flow in Two Dimensional and Axially Symmetric Nozzles, Q. J. Mech. Appl. Math., 15, pp. 487508.
6 Nakao, S., Hirayama, T., Yokoi, Y., and Takamoto, M., 1997, Effects of
Thermalphysical Properties of Gases on the Discharge Coefficients of the
Sonic Venturi Nozzle, Proc. 1997 ASME FED Summer Meeting, Vancouver,
Canada.
7 International Standard ISO 9300, 1990, Measurement of gas flow by means
of critical flow Venturi nozzles, p. 8.
8 Tang, S. P., and Fenn, J. B., 1978, Experimental Determination of the Discharge Coefficients for Critical Flow Through as Axisymmetric Nozzle,
AIAA J., 16, pp. 4146.
9 JSME Data Book, 1986, Thermophysical Properties, edited by Japan Society
of Mechanical Engineers in Japanese.
10 Johnson, R. C., 1964, Calculations of Real-Gas Effects in Flow Through
Critical-Flow Nozzles, ASME J. Basic Eng., Series D, 86, Sept., pp. 519
526.
11 Levine, R. D., and Bernstein, R. B., 1974, Molecular Reaction Dynamics,
Oxford University Press 1997: translated in Japanese.
12 Simpson, C. J. S. M., Bridgman, K. B., and Chandler, T. R. D., 1968, ShockTube Study of Vibrational Relaxation in Carbon Dioxide, J. Chem. Phys., 49,
pp. 513522.
13 Liepmann, H. W., and Roshko, A, 1960, Elements of Gasdynamics, Wiley,
New York 1983: translated in Japanese.
14 Yokogawa, A., and Nishioka, M., 1999, Numerical Studies on the Sonic
Nozzle for Mass Flow Calibration, Ms. thesis, Univ. of Osaka-prefecture in
Japanese.
15 Johnson, A., Wright, J., Nakao, S., Merkle, C. L., and Moldover, M. R., 1999,
The Effect of Vibrational Relaxation on the Discharge Coefficient of Critical
Flow Ventur, Flow Meas. Instrum., 11, pp. 315327.
16 Back, L. H., 1970, Acceleration and Cooling Effects in Laminar Boundary
layers-Subsonic, Transonic, and Supersonic Speeds, AIAA J., 8, pp. 794
802.

Transactions of the ASME

Downloaded From: http://fluidsengineering.asmedigitalcollection.asme.org/ on 04/27/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

Anda mungkin juga menyukai