Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
To cite this article: E. Jassim, M. Abedinzadegan Abdi & Y. Muzychka (2008): Computational Fluid Dynamics Study for Flow
of Natural Gas through High-pressure Supersonic Nozzles: Part 1. Real Gas Effects and Shockwave, Petroleum Science and
Technology, 26:15, 1757-1772
This article may be used for research, teaching, and private study purposes. Any substantial or systematic
reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to
anyone is expressly forbidden.
The publisher does not give any warranty express or implied or make any representation that the contents
will be complete or accurate or up to date. The accuracy of any instructions, formulae, and drug doses should
be independently verified with primary sources. The publisher shall not be liable for any loss, actions, claims,
proceedings, demand, or costs or damages whatsoever or howsoever caused arising directly or indirectly in
connection with or arising out of the use of this material.
Petroleum Science and Technology, 26:17571772, 2008
Copyright Taylor & Francis Group, LLC
ISSN: 1091-6466 print/1532-2459 online
DOI: 10.1080/10916460701287847
Abstract: The computational fluid dynamics technique was used to study the behavior
of high-pressure natural gas in supersonic nozzles. Although many applications of gas
flow produce insignificant errors when the gas is assumed ideal, our results indicate
significant variation of gas properties. This article illustrates natural gas behavior
when it is considered to be real and how erroneous the properties may become when
the gas is assumed to be ideal. The article also presents the influences of properties
related to the flow of natural gas through supersonic nozzles. Using a quite accurate
equation of state model, real gas effects are studied and compared with the perfect gas
case. The results show a significant variation in gas properties estimation. Location
of shockwave is also analyzed. The comparison of results for two gases (methane and
nitrogen) indicated that shockwave position can significantly change when the gas is
considered as real rather than perfect.
Keywords: computational fluid dynamics, natural gas flow, real gas flow, shockwave,
supersonic nozzles
1. INTRODUCTION
The demand for natural gas has encouraged the energy industry toward
the discovery of remote offshore reservoirs. Consequently, new technologies
have to be developed to efficiently produce and transport natural gas to
consumption centers. Common design challenges in all gas processing meth-
ods for offshore applications are the compactness and reliability of process
equipment. Supersonic nozzles have been introduced as an alternative to treat
natural gas for offshore applications and to meet the offshore requirements
1757
1758 E. Jassim et al.
(Hengwei et al., 2005; Alfyorov et al., 2005, Okimoto and Brouwer, 2002;
Karimi and Abedinzadegan Abdi, 2006; Brouwer and Epsom, 2003). In a
supersonic separator, the gas temperature is lowered based on the principle
of gas expansion where no refrigerant is needed. The compact design of
supersonic nozzles is a major advantage over traditional means of natural gas-
treating technologies, particularly for offshore applications. The gas speed in
this device is very high, preventing fouling or deposition of solids and ice.
Refrigeration is self-induced, and, therefore, no heat is transferred through
the walls and, unlike external refrigeration systems, no inhibitor injection and
inhibitor recovery system are necessary. Intensive water dew points down to
Downloaded by [McGill University Library] at 13:39 06 December 2012
upstream and downstream of the shock are equal, and the energy or heat loss
is negligible. The shock can also interact with the boundary layer, and this
can delay the transition from supersonic flow to subsonic flow even further.
The thicker the boundary layer, the longer and more complex is the transition
across a shock (Saad, 1993). The increase of pressure across a shock is an
indication of the shock strength that can lead to a sound wave considered as
a shockwave of minimum strength.
The numerical solutions treating the shockwave region have been re-
ported by several authors; however, most of these methods were developed
mainly for perfect gas, particularly for air (Drikakis and Tsangaris, 1993).
Downloaded by [McGill University Library] at 13:39 06 December 2012
A number of CFD models have been developed for real gas nozzle flows,
which are divided into two categories: models that use analytical EOS and
models based on numerical fitting tables of gas thermodynamic properties
(Wark, 1995). The use of an analytical EOS makes it possible to have a
reasonably accurate and consistent description of gas properties throughout
a large range of temperature, pressure, and density for many substances.
J. D. van der Waals proposed the first analytical trial of real gas in
1873. The formula was the first model to take into account the volume of
Downloaded by [McGill University Library] at 13:39 06 December 2012
RT a
P D : (1)
b T 1=2 . C b/
Unlike the van der Waals equation, both terms on the right-hand side
of the equation contain a temperature function. The RK equation performs
quite well at small to moderate densities and is much better when T > Tc .
Note that the critical compressibility factor based on the critical state criteria
is 1/3. This value is better than the van der Walls value of 3/8, but it is still
somewhat higher than the typical range of experimental values.
A step to raise further the accurate predictions for real gases was taken
by Soave, Redlich, and Kwong (Soave, 1972) and Peng and Robinson (1976).
The addition of a third parameter greatly improves the accuracy of compress-
ibility factor data.
The van der Walls, Soave/Redlich/Kwong, and Peng/Robinson cubic
equations of state are very similar in their analytical format. They are called
cubic EOS, because, if expanded, each would contain volume terms raised
to the first, second, and third powers. These equations and many other cubic
equations of state can be expressed by a general format (Wark, 1995):
RT
P D ; (2)
b . C c/. C d /
Cubic EOS models are extremely useful for predicting vapor and liquid
P T data, vapor-pressure data, and vapor-liquid equilibrium data. However,
their accuracy is limited by the use of only two or three adjustable parameters.
To obtain greater overall accuracy, it is necessary to use equations of greater
complexity. A few dozen EOS models are available in the literature (Wark,
1995). Some are quite complex, such as the virial equation and the Benedict-
Webb-Rubin (BWR) equation. The Benedict-Webb-Rubin (BWR) equation
and its modification (MBWR) have proven to be quite accurate over wide
ranges of pressure and temperature. Although they are not as accurate as
desired in prediction of P T behavior for polar compounds, these equations
Downloaded by [McGill University Library] at 13:39 06 December 2012
accurately predict the properties of specified fluid and are capable of repre-
senting P T data in the region where VW and RK EOS fail, particularly
when < 0:8c . MBWR has the form:
9 15
X an .T / 2 X
c an
P D C exp ; (3)
n 2n 17
nD1 nD10
3. CFD SIMULATION
CFD techniques are currently used in industry for design purposes in many
different areas (e.g., propulsion, chemistry, combustion, turbo-machinery,
etc.). The importance of such techniques for integration of the Navier-Stokes
equations is steadily growing because of the fact they improve the design
reliability and flexibility. In fact, the same numerical scheme can be applied
to solve different flow systems provided the appropriate sets of boundary
conditions are implemented. Three main methodologies in the CFD modeling
research can be highlighted (Carvero and Satta, 2000): (1) flexible and robust
schemes for steady or unsteady flows for all fluid velocities, (2) turbulence
modeling (k-", k-!, Reynolds stress), and (3) fluid properties modeling (two
phase, real gas, and cryogenics). In the present work, a contribution to the
second and particularly third class of problems is rendered.
At high Reynolds numbers, turbulent fluctuations transport a far greater
amount of momentum than viscous forces throughout most regions of the
flow. Thus, the modeling of the Reynolds stress is a vital part of the flow
prediction. A turbulence model is a means of deriving the Reynolds stresses
(and other turbulent fluxes) in order to close the mean flow equations.
Turbulent flow can be modeled by using several schemes. However, the
k-" model (Pope, 2000) is used here due to its frequent use for industrial ap-
plications, its relative accuracy, and its incorporation in the most commercial
CFD codes. Diverse ranges of problems, therefore, use this model, including
1762 E. Jassim et al.
Since most published research has been concerned with the Laval nozzle, we
validated our results by applying the numerical technique for such geometry
and compared our results with the most recent available data (Arina, 2004;
Kouremenos, 1986) before proceeding and applying it to our modified nozzle
system.
Molleson and Stasenko (2005) performed their investigation for nozzle
whose geometry is shown in Figure 1a. Their working fluid was methane
at 70 bar inlet stagnation pressure and 250 K inlet stagnation temperature,
while the value of exit Mach number according to their supersonic exit radii
was 1.2. The SRK EOS model was used in their study. We used the same
geometry and conditions in comparison of the sonic condition and in the
studying of the effect of a real gas model on the sonic position. We used the
MBWR as our thermodynamic model.
Figure 2 shows the variation of Mach number with position in real case
and the comparison with results obtained from the work of Molleson and
Stasenko (2005). One can see that choke (sonic velocity, M D 1/ occurs
at the throat regardless of the EOS used. Also, our results are in very good
agreement with their results. The second comparison is performed to validate
our simulation on capturing shockwave position. The comparison is done with
recent available data (Arina, 2004). The geometry used in the comparison,
shown in Figure 1b, is adopted from Arinas work (2004). The working fluid
was CO2 . The dimensions of the assumed Laval-nozzle are:
2
x x
A.x/ D 2:5 C 3 1:5 for x xth ; (4)
xth xth
2 !
x x x
A.x/ D 3:5 6 4:5 C for x xth ; (5)
xth xth xth
where Athroat D 1 cm2 , length D 10 cm, and the throat placed at xth D 5 cm.
Since Arinas simulation was performed near CO2 condensation condi-
tions, for which T D 1:001Tc and D c , we compared our results for a
perfect gas case, as the FLUENT real gas basis could not predict multiphase
conditions. The exit pressure was 83% of the inlet pressure. Our numerical
Supersonic Flow of Natural Gas: 1. Real Gas Effects and Shockwave 1763
Downloaded by [McGill University Library] at 13:39 06 December 2012
(a)
(b)
(c)
Figure 1. Nozzle geometry considered in the study. (a) Molleson et al. (2005). (b)
Arina (2004). (c) System configuration for long-neck (long constant area/extended
throat) Laval nozzle.
1764 E. Jassim et al.
Downloaded by [McGill University Library] at 13:39 06 December 2012
Figure 2. Comparison of Mach numbers before the choke region in the Laval nozzle.
results displayed the same behavior when similar conditions and working
fluids were applied, as seen in Figure 3. After validation, we can now present
the results of studies related to the effects of a number of parameters as
follows.
The significance of using the real gas models can be more clearly shown
when comparison of the location of the shockwave within the Laval nozzle
is made for two different gases (methane and nitrogen here).
The real gas model predicts shockwave location earlier for both gases. How-
ever, the differential distance between the ideal and real shock positions is
different. In fact, when the ideal gas model is used, shock occurs earlier
for nitrogen, see Figure 4a. On the contrary, the real gas model predicts
Supersonic Flow of Natural Gas: 1. Real Gas Effects and Shockwave 1765
Downloaded by [McGill University Library] at 13:39 06 December 2012
Figure 3. Comparison of density ratios in the Laval nozzle under perfect gas
conditions.
Figure 4. Comparison of shock position for nitrogen and methane under (a) ideal
and (b) real gas conditions.
Supersonic Flow of Natural Gas: 1. Real Gas Effects and Shockwave 1767
assumed 7 MPa and at the inlet is predicted. The results of simulation are
discussed as follows:
4.1.2.1. Density. By looking at any fluid textbook, one can see that the
conservation of momentum equation directly or indirectly contains density
terms in each component. Thus, flow structure is severely affected by any
deviation in density calculation. To realize how this deviation will affect the
predictions in the nozzle, a graph of the density ratio (real/ideal) along the
nozzle system is plotted (see Figure 5). It is evident how erroneous the results
might get if the perfect gas model is used, particularly in the vicinity of the
Downloaded by [McGill University Library] at 13:39 06 December 2012
Figure 5. Centerline density ratio for real gas (methane) simulation in an extended
throat nozzle.
1768 E. Jassim et al.
Downloaded by [McGill University Library] at 13:39 06 December 2012
Figure 6. Centerline static pressure for real gas (methane) simulation in an extended
throat nozzle.
The real gas simulation predicts pressure recoveries in excess of 10% over
those predicted by the perfect gas model. Also, the difference between real
and ideal static pressures forces the calculation of total pressure to diverge.
The errors in evaluating the total pressure and temperature can lead to
incorrect predictions for friction loss, work, and heat transfer. The differences
between the predictions of ideal and real gas models for total pressure and
total temperature are considerable. These discrepancies can result in incorrect
values for friction losses and other calculated parameters.
Figure 7. Wall static temperature distribution for real gas (methane) simulation in an
extended throat nozzle.
gas law is used as the thermodynamic model. However, the real gas model
predicts different conditions in which the stagnation temperature may vary.
Stagnation temperature in the real gas case varies across the shock because
the specific heat is now varied across the shock as well. Thus, for adiabatic
process,
Z T02
Cp.T0 /:d T0 D 0: (6)
T01
5. CONCLUSION
Shockwave has been studied in this article by using the CFD approach. The
effects of real gas properties on the flow behavior were also addressed. It
is crucial to beware of the real gas fluid behavior when analyzing high
pressure natural gas flow systems. The assumption of a perfect gas may
give rise to significant errors. Comparison between perfect and real gas
1770 E. Jassim et al.
models was implemented to present the errors involved when real properties
in calculating heat, work, and other thermal processes are ignored. The results
of CFD simulation runs conducted for different gases showed that the ideal
gas assumption would lead to serious misrepresentation of the flow field,
including the position of shockwave and severe miscalculations in the design
of supersonic devices.
ACKNOWLEDGMENTS
Downloaded by [McGill University Library] at 13:39 06 December 2012
Financial support provided by the Centre for Marine Compressed Natural Gas
(CMCNG) Inc., Unconventional Gas Supply Program of the Natural Resource
Canada (NRCAN), and Atlantic Canada Opportunity Agency (ACOA) to
finance various process engineering projects within the CMCNG as well as
the presented work in this article is appreciated.
REFERENCES
Alfyorov, V., Bagirov, L., Dmitriev, L., Feygin, V., Imayev S., and Lacey, J. R.
(2005). Supersonic nozzle efficiently separates natural gas components.
Oil Gas J. 103:5358.
Arina, R. (2004). Numerical simulation of near-critical fluids. Appl. Numl.
Math. 51:409426.
Bai Sh. I. (1961). An Introduction into Theory of Compressible Liquid Flow,
Moscow: IZD. Inostrannoi Literatury (Russ. Transl.).
Brouwer, J. M., and Epsom, H. D. (2003). Twister supersonic gas conditioning
for unmanned platforms and subsea gas processing source. Proceedings
of Offshore Europe Conference, Aberdeen, UK, Sept. 25, pp. 219225.
Carvero, C., and Satta, A. (2000). A CFD model for real gas flow. ASME
paper GT-518.
Crocko, L. (1950). One-Dimensional Consideration of the Gas Dynamics of
Time-Independent Flows in Fundamentals of Gas Dynamics. Emmons,
H. W. (Ed.). Princeton, NJ: Princeton University Press.
Drikakis, D., and Tsangaris, S. (1993). Real gas effect for compressible nozzle
flows. J. Fluid Eng. 115:115120.
Ginzburg, I. P. (1939). Uch. Zap. Leningr. Gos. University, No. 42, p. 5.
Hengwei, L., Zhonggliang, L., Yongxun, F., Keyu, G., and Tingmin,Y. (2005).
Characteristic of a supersonic swirling dehydration system of natural gas.
Chinese J. Chem. Eng. 13:912.
Karimi, A., and Abedinzadegan Abdi, M. (2006). Selective removal of water
from supercritical natural gas. SPE 100442, Proceedings of the SPE Gas
Technology Symposium, Calgary, Alberta, Canada, May 1518, pp. 259
265.
Supersonic Flow of Natural Gas: 1. Real Gas Effects and Shockwave 1771
NOMENCLATURE
A area
Cp constant pressure specific heat
M Mach number
P static pressure
R gas constant
T static temperature
isentropic exponent; heat ratio
x distance from nozzle inlet
x* distance to throat
To stagnation temperature
Po stagnation pressure
L conduit length
LN nozzle length
Ld diffuser length
Lt system length
k turbulence kinetic energy
" rate of dissipation
dt conduit diameter
specific volume
1772 E. Jassim et al.
density
function of temperature
! turbulence frequency ("= k)
Subscripts
p pressure
volume
T temperature
c critical point
I ideal
Downloaded by [McGill University Library] at 13:39 06 December 2012
r real