You are on page 1of 7

Distinguished Author Series

Multiphase Flow in Pipes


by Peter Griffith, Massachusetts Inst. of Technology

Peter Griffith has been on the faculty of the Massachusetts Inst. of Technology (MIT)
since 1956. His primary research interests have been in two-phase flow, boiling,
condensation, supercritical heat transfer, and various applications of nuclear reactor
safety. He holds degrees in mechanical engineering from New York u., the u. of
Michigan, and MIT.

Introduction Books on Two-Phase Flow


Multiphase flow is found in many places. In the Various books on two-phase flow contain answers for
petroleum industry it occurs in oil and gas wells , many of the problems that arise. Almost all of the
gathering systems, many piping systems, and key following books describe homogeneous and separated
pieces of equipment needed in refineries and flow models for calculating void fraction and pressure
petrochemical industries, including boilers, drop, so I shall mention only those features unique to
condensers, distillation towers, separators, and each book.
associated piping. This article focuses on two-phase Wallis I contains the most complete mechanistic
flow in pipes. Though a lot has been learned about descriptions of void and pressure drop for the different
two-phase flow in the past 25 years , much of that flow regimes.
knowledge has not been collected in a convenient Hestroni 2 has a unique section on flow instability
place . In particular, much work done for the nuclear and also the best section on flow regimes.
industry remains unknown to the petroleum industry. Collier 3 is primarily a mUltiphase heat-transfer book
The primary goal of this article is to describe the kinds but has a unique section on two-phase pressure drop in
of problems we are now able to solve and to point out fittings .
where answers to these problems can be obtained. Hewitt and Hall-Taylor4 collect and rep0l1 more
When piping in which two phases are flowing is experimental observations on annular flow than any
designed , a number of questions can arise, depending other source.
on the application: Lahey and Moody 5 have a unique section on choked
I. What is the void fraction? flow. Their description of the drift-flux model is
2. What is the pressure drop? excelle nt.
3 . What is the liquid level? Govier and Aziz 6 consider both slurries and non-
4. What is the flow at a break? Newtonian fluid plus a wide variety of solid, liquid ,
5. How can one separate the phases? and gas systems.
6. Where will corrosion occur? Moore and Sieverding 7 have design data on screen
7 . What is the wear rate caused by droplet and chevron separators that are not reported elsewhere.
impingement? Hsu and Graham 8 consider cryogens.
8. What is the vibration of the pipes as a result of Szilas 9 has a design section on both pool and
two-phase flow? cyclone separators.
I shall begin by listing available books, then
recommend flow-regime maps and correlations for Flow Regimes
void, pressure drop, and critical flow, and finally The unique feature of two-phase flow is the presence
touch on the problems of separation, corrosion, wear, of flow regimes- descriptions of how the two phases
and vibration. are distributed in the pipe. Flow regimes and flow-
0149·2136/84/0031·2895$00 .25 regime maps for horizontal, vertical, and inclined

MARCH 1984 361


50

DISPERSED
BUBBLE (DB)
10

:(
ELONGATED;3 SLUG
E~_--~~~-:~~~ Stratified smooth flow BUBBLE

'-' INTERMITTENT (I)


~-~--~~
Q) DISPERSED
Stratified wavy flow If)
..... ///,
LIQUID (AD)
-.5....
~~~~o~:o9 Plug flow ::J
.1
STRATIFIED
SMOOTH (SS)
~o~~~d Slug flow

c:~. -__-=:;;-;J Annular flow


.01

.1 10 100 500

l~~"~ ?~;~~~~~
0 Dispersed bubble flow U6 (m/sec)
Fig.1-Flow regimes for a horizontal pipe (adapted from Fig. 2-Flow-regime map plotted in terms of superficial
Refs. 2 and 10). velocities of each phase for air and water in a 1-in.
[2.54-cm]-ID pipe at room temperature and pressure.
Crosshatched bands represent the data of
Mandhane. 2 ,5,10

pipes are illustrated in Figs. I through 4.2.10.11 wavy stratified annular dispersed boundary farther to
For quantities like void fraction or pressure drop. it the right. Our experiments indicate that these
has not proved convenient to use these maps as part of recommendations improve the map,
the calculation scheme. The quantities of interest are
continuous, while steps would occur at flow-regime Void Correlations
boundaries if separate correlations were used for void One of the most basic quantities in two-phase flow is
or pressure drop for each regime. Rather, the utility of void fraction or its complement, liquid fraction,
the flow-regime maps lies in their ability to help solve Various methods exists for calculating this quantity;
unconventional problems. such as: each has its advantages and faults.
Is there a liquid level? The simplest is the homogeneous model, which
Is there carryover? assumes that both phases move at the same velocity.
Is there entrainment? However, they rarely do. Void is usually
Is the flow steady? overestimated in horizontal and upflow and
Will the top of the pipe be wet? underestimated in downflow when this model is used.
They are also of great interest when one runs If pressure drop, rather than void, is the primary
"thought experiments." It is hard to imagine how a concern and the gravity contribution to the pressure
two-phase flow will behave in an untested system drop is small (say 20% of the total), this model is
without also imagining how the phases are arranged. often satisfactory,
The arrangement of the phases is the flow regime and The next most complicated expression for void
can be predicted with the maps. Often one can guess assumes that the liquid moves more slowly than the
how a flow regime will change as it passes through a vapor. These are called "slip correlations." The well-
fitting. for example. known Martinelli, Thom, or Baroczy correlations
There are also regime maps for inclined pipes. One mentioned in all the handbooks on two-phase flow 1-3,8
of the most extensive sources of these is Ref. 12. All fit into this category, Where pressure drop as such is
angles are included, from vertical uptlow to vertical the issue. these methods can be satisfactory. Martinelli
downtlow. and Baroczy have a data base that includes a wide
The tlow-regime maps recommended have a variety of fluid properties in the correlations, At low
consistent designation for the flow regimes, a broad velocity, however, they can give poor answers because
data base. and a semitheoretical basis for determining the gravity contribution to the vapor velocity is
flow-regime boundaries. The approach taken in these practically ignored in both of these correlations,
works reflects just about the right compromise between The most precise method for calculating the void
precision and simplicity. fraction relies on the drift-flux model. The most
A recent work U extends the data base for flow convenient description of this model is provided in
regimes and recommends changing the location of the Refs. 1 through 3, A recent compilation of the drift-
362 JOURNAL OF PETROLEUM TECHNOLOGY
100

A
t t--. 10
BUBBLE
0
0
0
°'0 0 IV······
:". ~ ~
: ...
ANNULAR
00 0 0
Doo 0 0 a
Q

0
O~ .-
.
~ : " "

o
,0;0
0
0
0 0
0
,
""':: :":" - u
1

--
00 0 0
II
00
o 0

000°
0'0
0
0

~ °0 00&
°0 " 10 .- : \
'"
o DO 00 00 °0
\ \0 . ' .,'
~o
0 0
00
o gV 000

~o
o ()0 0 0 A
00 °00 000 00°0 0 "'::l... ,1
00000 f:2 00 00 0
7
0°0 00 "

o°a o~c
oGGa
0 00 ,
,',

• \ J
.. _
B
~
"

C'oD 0 ' .
000 0
.. - .01
000 0 00 . ',.
00
•....-c:::::: .1 10 100 1000
t t t
Churn
U~
Bubbl. Slug Annular
flow tlow flow rtow (ft /sec)

Fig. 3-Flow regimes distinguished by Taitel and Dukler for a Fig. 4-Flow-regime map for air and water in a vertical upflow
vertical upflow pipe. 2,11 at 75°F and 1 atm [24°C and 101.3 MPa] (adapted
from Refs. 2 and 11). The coordinates are the super-
ficial velocity of each phase.

flux model constants for various flow regimes 14 has a demonstrably important, at least at some conditions.
huge data base. Properly used, the drift-flux model For example, for a single-phase, fully developed
generally gives the best predictions of void fraction flow in a pipe, the friction factor is a function of a
because it explicitly recognizes the two most important single dimensionless group, the Reynolds number.
factors that cause slip: combined velocity-density However, for a two-phase flow, the pressure drop
distributions in the channel and the direction of the (which can be calculated with a friction factor) is a
gravity vector. This model also is unique because it function of at least six variables. For exampk, one
properly predicts a liquid level for sufficiently small such set of variables identifies the friction factor as a
velocity levels. Thus, it can be used to help size function of a Froude number, the Weber number, the
devices such as separators or to tell whether some Reynolds number, the density ratio, the viscosity ratio,
heated tubes will be wet when there is only a limited and the flow-rate ratio. If we try to correlate data and
amount of liquid present. It also is unique because it leave out some dimensionless groups, we cannot
can predict void in counterflow and gives an expect a good result.
indeterminate form during a downflow when the void The same three alternatives exist for computing the
is sometimes indeterminate. pressure drop as exist for the void fraction: the
Whenever one has several methods of calculating a homogeneous model, the slip model, and the drift-flux
given quantity, guidance is needed in choosing which model. A number of comparisons between these
method to use. In general, homogeneous void can be models have been made in the literature. For example,
used only when the contribution of gravity to the total Ref. 2 makes recommendations for calculating
pressure drop is unimportant. Slip models are most pressure drop in both horizontal and vertical pipes.
convenient for engineering calculations but give poor When one looks at a large amount of two-phase
answers when the system operates outside of annular, pressure-drop data, the important differences tum out
dispersed, or bubbly flow regimes. In any regime to be caused by the different data bases underlying the
where gravity is a dominant force, a slip model will correlations. When the application for a correlation is
fail to represent an important part of the physics, so a known, the best general advice is to use a correlation
drift-flux model should be used. with a data base similar to the application. If there are
a number of differences between the data base and the
Pressure-Drop Correlations proposed application, one has a problem deciding what
Pressure drop is probably the quantity that one deals constitutes the most similar. I would rank order the
with most often in two-phase flow. In spite of this, similarities from most to least important: (1) quality
our ability to predict it in truly new situations is not and velocity level, (2) density ratio, (3) geometry (up,
very good. Differences are primarily a result of the down, or inclined), (4) diameter, and (5) other
variety of flow regimes that one tries to bridge with a properties such as viscosity and surface tension.
single correlation scheme. Another problem is the Turning now to specific models, the homogeneous
large number of dimensionless variables that are model is the simplest to use. Only one parameter is
MARCH 1984 363
needed to predict pressure drop: the friction factor. I Critical Flow
recommend that the friction factor be chosen by use of
Two-phase critical flow is an important problem in
the well-known Moody curves, assuming that only
several areas. Overpressure relief valves for devices
liquid is flowing at the mixture mass velocity. Use the
such as boilers and cryogenic storage tanks need to be
liquid density and viscosity to calculate the Reynolds
sized so the tank is protected from bursting against all
number and the Moody curves to determine the
transients. Subsurface safety valves contain choked
friction factor. This procedure gives a smooth
flow and also must be sized. Break flows must be
transition to the two-phase pressure drop in the low-
calculated for pipelines that contain two phases.
quality region and a step at 100% quality.
Over the past decade much work has been done on
Surprisingly, the step at high quality has some
break flow since this is an important factor in how a
experimental justification. In any case, the
nuclear reactor system behaves after a break occurs.
homogeneous methods that rely on a weighted
This section explains the results of nuclear work to
viscosity have practically no experimental justification
other parts of the technical community.
and make no physical sense. At best they provide a
smooth transition from a single- to a two-phase flow In this context, what we call a critical flow is
of both ends of the quality range. defined by the following experiment. A pipe
The slip models generally have a larger data base connecting a fluid reservoir close to saturation
than the homogeneous models. Thom, Martinelli, and conditions is allowed to discharge into a reservoir at a
Baroczy all are included in this category. The lower pressure. As the pressure in the lower-pressure
empirical friction pressure-drop multipliers they reservoir is dropped, the flow continues to increase to
propose are easy to use and give sensible answers to a certain point and then holds constant even though the
overall pressure drop. Average errors with these discharge pressure is decreased. This asymptotic flow
techniques are small, but errors possible for a single is the critical flow and its velocity is called the critical
calculation sometimes are huge-as much as 60 %. In a velocity. Unlike gases, there is no simple relationship
complex system where heat addition may cause a between this velocity and the velocity of a pressure
quality change, and where there are fittings and wave in the mixture. Both the frequency of the
perhaps several sources for the flow, the overall errors pressure wave and the flow regime change the
are much less because they tend to average out. measured pressure-wave velocity.
There is no suitable friction pressure-drop The homogeneous model and separated flow models
calculation procedure, which is needed to accompany both can be used to calculate choked flow for two-
the drift-flux model (used for density). Generally if the phase mixtures. At low quality and pressure the
drift-flux model is appropriate, the friction contribution homogeneous-equilibrium model has been shown to
to the pressure drop is very small. Under the underestimate the break flow greatly. Slip models for
circumstances, I recommend that the homogeneous choked flow were developed to remedy some of these
model be used to calculate the friction pressure drop. defects, but other factors, primarily the departure from
Several pressure-drop models for vertical upflow, thermal equilibrium, also apply. Because of these
including those mentioned in this section, are complications, the most successful critical flow models
compared and evaluated in Refs. 32 and 33. have an extensive data base and rely only minimally
Recommendations for calculation are included. on theory.
Fittings often are important components in piping The most convenient source of information on
systems, though little information exists that can be choked flow of steam/water mixtures is contained in
used to calculate two-phase pressure drop in fittings. Ref. 5. The results of calculations using the
Ref. 3, in any case, has a section on pressure drop in homogeneous equilibrium model and the best slip
fittings. model are included in a form that is uncommonly
Inclined pipes are a special case. Naturally the data convenient for calculation. Both models (as presented
base for any particular angle inclination is skimpy, so in this reference) are only for water, but the analytical
more extrapolation is necessary. The important factor details included allow calculations for fluids other than
to keep in mind with inclined pipes is that there is steam and water to be performed.
often a flow-regime change as the pipe changes Recently several useful reviews have been published
orientation from upflow to downflow. One often in this area. Ref. 16 discusses what goes out the break
changes from slug flow in the upflowing portions to when there is a hole in a pipe with a stratified flow.
stratified or annular flow in the downflowing portions. This is important because proportions of the two
There is little or no pressure recovery in downflow in phases that go out the break are not necessarily the
stratified or annular flow, so the effect of replacing a same as those in the pipe or those flowing. The break
section of horizontal pipe with an inclined pipe of the quality and flow rate depend on the location of the
same overall length and net elevation changes is to break and its size, among other factors. Ref. 17 is a
increase the overall pressure drop substantially. To thoughtful review of the current theories on choked
calculate this pressure drop properly, the void fraction flow and compares data with a variety of theories.
in the upflowing portions must be calculated by use of Ref. 18 examines data from a variety of sources and
the drift-flux model. For the stratified downflowing recommends calculations for the large pipes found in
regions a theory presented in Ref. 12 is most reactor systems.
appropriate. The most extensive study of inclined-pipe Though break flow is still not entirely understood,
pressure drop is Ref. 15. we know enough to make serviceable estimates of the
364 JOURNAL OF PETROLEUM TECHNOLOGY
flow and the resulting set impact forces and critical system is prone to this kind of instability, it is
pressure ratios. necessary to calculate the pressure-droplflow-rate
Other Topics curve and see whether there is a negative-sloping
region in the operating range.
The items touched on so far might be described as In principle, all the information needed to do this is
conventional two-phase flow topics. Many areas are
in the pressure-drop correlations mentioned earlier. In
affected by what we have learned about two-phase
fact, how one should do this calculation is still
flow that are not usually regarded as two-phase flow
somewhat in doubt. The calculation should be done
concerns, even though two-phase flow is an important
where the proportion of the two phases distributed to
factor. It is worthwhile to spend some time on these
the various parallel passages connecting the headers
topics because it is unlikely that the more conventional
are allowed to vary as they will. One cannot assume,
fields will be the real problems in the future. The first
for example, constant quality or equal flow split unless
of these topics is gas/liquid separation.
the system is designed to ensure such a flow split. The
Separators. Most separators are built and tested by root of the difficulty is that we don't have a method of
manufacturers with very little information provided to calculating how two phases split when they come to a
the purchasers about their operation or design. junction. This deficiency must be regarded as one of
Scattered throughout the literature are papers and the outstanding, unsolved problems in two-phase flow.
chapters in books that allow one to design separators Making a piping network predictable may well be a
and estimate their performance. This section attempts design requirement. If so, and if one has to distribute
to draw this information together. two phases, perhaps the best way is to design the
Both gravity and centrifugal separators are described system to ensure symmetry. There are at least two
in Ref. 9. Gravity separators, in essence, are tanks in ways to do this. One can arrange any number of outlet
which the velocity level is low enough to allow phase pipes in a circle around a plenum. This practically
separation. They usually have demisters at the top to guarantees that the flow out will have the same quality
remove additional small drops that might be carried in each pipe. Another procedure is to split and resplit
over. Ref. 9 gives a design procedure for separators of the flow in tee's in the horizontal plane. For equal
this kind. Properties like gas and liquid density are pressures in both branches, the quality flowing in the
considered explicitly. Cyclone separators are also men- two branches is the same.
tioned, though less information is given about their The following are examples of specific excursive
design. instabilities that have led to difficulties in various two-
Ref. 7 describes demisters of various kinds such as phase systems.
screens, knitted wire mesh, and corrugated plate 1. Small, highly heated tube. The friction term was
separators. Information on separator efficiency is found to be destabilizing when boiling began.
presented for all these kinds of separators in a form 2. N-shaped three-pass vertical boiler tube. The
that is useful for design. Flooding limits also are gravity term has been found destabilizing. 20
presented so that one can predict at what vapor 3. Heated, inverted V-tubes (in a pendant super-
velocity level the separated liquid will have difficulty heater). The gravity term was destabilizing. 21
flowing back against the wind. 4. Yankee dryer condensate drain (a "vertical"
Additional information on separators is provided in upt10w pipe sucking condensate from the inside
Ref. 2. A wider range of separators is considered, surface of a rotating drum and is discharging it into a
though some useful design information is lacking. horizontal axle).22 The gravity term was destabilizing.
Perhaps the simplest separator is a vertical Many other examples in the literature duplicate the
downflowing pipe in which the deposited liquid is failures that already have been discovered. In general
allowed simply to run out. This kind of separator can the following systems are particularly prone to two-
be designed with the information contained in Ref. 19. phase flow pressure-droplflow-rate instabilities. All
these instabilities appear when the pipe in question is
Stability. Two-phase systems often behave in an part of a multiple-tube array connecting common
unstable manner. "Instability" in this context involves headers.
two separate manifestations: excursive instability (first I. Vpflowing two-phase pipes at low velocity (where
described by Ledinegg) and oscillating instabilities. gravity is dominant). Gravity destabilizes.
Both kinds of instabilities are found in two-phase 2. Downflowing heated tubes. Gravity destabilizes.
piping systems. Ref. 2 is practically the only 3. Vpflowing chilled tubes. Gravity and momentum
compilation of the information available for describing are both destabilizing.
two-phase flow instabilities in general. 4. Heated tubes of any orientation with vigorous
Unheated two-phase systems are prone to excursive surface boiling. Friction is destabilizing.
instabilities if, for any reason, there are (1) parallel In general one stabilizes a system by putting
passages connecting common headers or plenums, or sufficient orificing in the lines to ensure a positive
(2) a negative-sloping pressure-droplflow-rate curve in pressure-drop vs. flow-rate curve over the entire
one or more of the passages connecting the two operating region.
headers. The most common cause of a negative- These instabilities also can lead to oscillating flow
sloping pressure-droplflow-rate curve is gravity. The rates. The most likely such instability, described in
slower the flow, the more liquid is held up and the Ref. 23, is where a flow delivered to a heated pipe
greater the pressure drop. To determine whether a oscillates because of compressibility in the fluid
MARCH 1984 365
delivery system. Any soft delivery system can lead to know what the mass-transfer coefficient is around the
an oscillating flow. bend.
A more common cause of an oscillating flow is a Ref. 26 reports an ingenious experiment in which
density wave instability. Ref. 2 summarizes most of pure water and air are used to simulate the
what is known about them. They are found in systems steam/water system of interest. The "pipe" is cast in
of any orientation in which heat addition causes a two pieces of plaster of Paris. The system is run for a
density change. When the oscillations occur, their while using air and water and the erosion pattern is
period is about twice the transit time in the heated observed. This shows more clearly than any other
section. method how the peculiar wear pattern observed in
For these oscillations to occur, a large proportion of steam-extraction lines comes about.
the pressure drop must be concentrated in the exit
section of the tube. If the flow is above the stable Wear. Oil and gas pipelines and wellstrings,
limit for the existing heat flux, a reduction in flow, for particularly in the vicinity of fittings, can exhibit wear
example, will cause a reduction in the exit pressure from the impact of entrained sand. This has been
drop. This will tend to increase the flow. However, it studied in a recent work and an unpublished
takes time for the resulting increase in density to thesis. 27.28
propagate to the exit section. When it gets there, the The wear theory of Finnie 29 can be adapted to the
exit pressure drop increases (because of the increase in case of sand entrained in a liquid rather than a solid.
pV2, where p=density and V=velocity) and the inlet When this is done, reasonable wear rates are
flow decreases. This causes the flow to decrease and predicted. The secondary flows in the bends are
the pV 2 to decrease, but only after a delay. The cycle important in determining how much of the sand hits
of increase and decrease occurs at a period equal to the bend, while the effects of flow regime are much
twice the transit time in the test section. In principle, smaller than anticipated. For bubbly and slug flow, the
the methods for calculating the pressure drop in heated sand is probably in the liquid but the velocities are low
sections mentioned earlier are adequate for predicting enough so that the resulting wear is not very
this instability. In fact. however, these correlations are important. In annular flow the film is apparently thin
generally too imprecise for this purpose. since pressure enough that the sand sticks out of the film and may be
derivatives as well as pressure-drop values are largely entrained. The wear pattern indicates that this
important and the correlations are not that good. is probably the case.
Again, these oscillations usually are eliminated by The homogeneous model appears adequate for
throttling at the inlet to the heated section. These predicting the sand velocity and distribution in the
oscillations can occur in any flow regime. pipe.

Vibration Caused by Two-Phase Flow. Very little


Corrosion-Erosion. One of the more peculiar two- information is available that can be used to predict the
phase flow problems concerns corrosion-erosion in wet vibration amplitudes caused by fluctuations in a two-
steam-extraction lines. Carbon-steel pipes passing wet phase flow. The mechanism of these fluctuations (as
steam from extraction points on the turbine to the feed described in Ref. 30) follows.
water heaters have suffered from wastage rates so In two-phase flow, especially slug flow, plugs of
large that pipes have to be replaced. 24 -26 The location t1uid proceed down the pipe with the density
of the wastage is entirely a result of the peculiarities of fluctuating between that of almost pure liquid and that
the two-phase flow passing through these lines. The of almost pure gas. When these fluctuations hit a
metal loss peaks at a temperature of 300°F [149°C] bend, for example, a fluctuating force resulting from
and typically is found in pipes and fittings with the momentum change in the plug or bubble as it
flowing steam of 80 to 95 % quality. The flow regime proceeds around the bend is exerted on the bend. This
is annular-dispersed. force can cause the pipe to vibrate if the fluctuations
The most peculiar facet of this wear is that it is are near a natural frequency for the system. These
sometimes found on the outside of the pipe bend and vibrations are best described as random since there
sometimes on the inside. This is because two separate usually is not a single well-defined frequency that
mechanisms are responsible for the removal of characterizes the flow. The maximum amplitude of the
material. In any case, metal removal begins by the fluctuating force can be estimated conservatively from
steel corroding to magnetite. Fe 3 0 4 , On the outside the maximum density difference between the phases
of the bend the secondary flow and centrifugal and the mixture velocity. The frequencies can be
acceleration throw the drops out onto the magnetite, estimated from the information presented in Ref. 30
fatiguing it and causing it to erode away. This exposes or 31.
new metal to the steam and accelerates the wastage. The exciting frequencies are typically from 1 to 20
On the inside, the shear stress caused by secondary cycles/sec [1 to 20 Hz] while the natural frequencies
flow in the bend draws the annular film from the of the piping systems typically range from 5 to 40
bottom or sides of the pipe to the inside, where an cycles/sec [5 to 40 Hz]. This means that a pipe
inward-flowing stagnation point occurs. This excited by a two-phase flow will vibrate at its natural
stagnation point has a very high mass-transfer frequency with a variable amplitude. The same kind of
coefficient and the oxide is dissolved away as a result. vibration would occur if a pipe were struck
To calculate the metal-removal rate, one needs to occasionally and allowed to vibrate between blows.
366 JOURNAL OF PETROLEUM TECHNOLOGY
Flow regime is of governing importance for this 19. Liu, Y.H. and Agarwal. 1.K.: "Experimental Observation of
problem. The maximum amplitude of the exciting Aerosol Deposition in Turbulent Flow," Aerosol Science 5,
(1974).
force occurs at the slug-annular boundary. Bubbly flow 20. Donner, T.l. and Bergels, A.E.: "Pressure Drop with Surface
is very smooth, whereas annular flow becomes Boiling in Small-Diameter Tubes," Report No. 8767-31, Dept. of
increasingly smooth as the velocity and quality Mechanical Engineering, Massachusetts Inst. of Technology,
increase. Slug flow, however, is very rough. (Sept. 1964).
21. Krasykova, L. Y. and Glusker, B.N.: "Hydraulic Study of Three-
Pass Panels with Bottom Inlet Headers for Once Through
Conclusion Boilers," Therlilal EnKineerinK (No.8, 1965).
Methods for calculating many of the quantities of 22. Deane, R.A.: "A Experimental Study of Some Dryer Drainage
Siphons," Technical Assn. of the Paper and Pulp Industry (March
interest in two-phase flows exist but are scattered in
1959) 42.
the literature. This article cites references where the 23. Maulbetsch, 1.S. and Griffith, P.: "A Study of Systems Induced
information can be found, stressing the handbooks, Instabilities in Forced-Convection Flows with Sub-cooled Boil-
which are the most generally available sources for this ing," AIChE 3rd IntI. H.T. Conference, Chicago (1968).
kind of information. 24. Vu, H. V.: "Erosive-Corrosive Wear in Steam Extraction Lines of
Power Plants," MS thesis, Massachusetts Inst. of Technology,
Cambridge (1982).
References 25. Coulon, A. and Thauvin, G.: "Erosion and Erosion-Corrosion of
1. Wallis. G.B.: One-Dimensional Two-Phase Flow. McGraw-Hill Mctab," Pmc., 5th IntI. Conference on Erosion by Sol id and Liq-
Book Co. Inc., New York City (1969). uid Impact (1979) 25, 1-11.
2. Hestroni, G.: Handbook of Multiphase Systems. Hemisphere 26. Sprague, P.J., Wilkin, S.K., and Coney, M.W.E: "Effects of
Publishing Corp .. Washington, DC (1982). Two-Phase Flow on Wall-to-Fluid Mass Transfer in Bends and
3. Collier, 1.G.: CO/J\'ectil'(! BoilinK and Condensation. McGraw- Straight Pipes," Pmc., European Two-Phase Flow Group
Hill Book Co. Inc., New York City (1981). Meeting, Zurich (1963).
4. Hewitt, G.F. and Hall-Taylor, N.S.: Annular Two-Phase Flow, 27. Bcnchaita, M.T., Griffith, P .. and Rabinowicz, E.: "Erosion of
Pergamon Press, New York City (1970). Metallic Plate by Solid Particles Entrained in a Liquid Set,"
5. Lahey, R.T. and Moody, F.J.: The Thermal-Hvdraulics ofa Boil- Trans" ASME (1983) 105, 215-23.
inK Water Nuclear Reactor. American Nuclear Soc., La Grange 28. Blanchard, D.: "Erosion of Metal Pipe by Solid Particles En-
Park, IL (1977). trained in Water." MS thesis, Massachusetts Inst. of Technology,
6. Govier, G.W. and Aziz, K.: The Flow of Complex Mixtures in Cambridge (1981).
Pipes, Van Nostrand Reinhold, New York City (1972). 29. Finnie, I.: "The Mechanism of Erosion of Ductile Metals,"
7. Moore, M.J. and Sieverding, C.H.: Two-Phase Steam Flow in Pmc" 3rd U.S. Natl. Congress of Applied Mechanics (1958),
Turbines and Separators, Hemisphere Publishing Corp., 527-32.
Washington, DC (1976). 30. Yih, T.S. and Griffith, P.: "Unsteady Momentum Fluxes in Two-
8. Hsu, Y.Y. and Graham, R.W.: Transport Process in BoilinK and Phase Flows and the Vibration of Nuclear System Components,"
Two-Phase Svstellls, Hemisphere Publishing Corp., Washington, ANL-7685 (May 1970).
DC (1976). 31. Hubbard, M.G. and Duklcr, A.E.: "A Model for Slug Frequency
9. Szilas, A.P.: Production and Transport (~r Oil and Gas. Elsevier During Gas-Liquid Flow in Horizontal and Near Horizontal
Scientific Publishing Co., New York City (1975). Pipes," 11111. J. Multiphase Flo\\' (1977) 3, 585-96.
10. Taitel, Y. and Dukler, A.E.: "A Model for Predicting Flow 32. Idsinga, W., Todreas, N., and Bowering, R.: "An Assessment of
Regime Transitions in Horizontal and Near Horizontal Gas Liquid Two-Phase Pressure Drop Correlations for Steam-Water
Flow," AIChE 1. 22 (Jan. 1976) 47-54. Systems," 11111. J. Multiphasl' Flo\\' (1977) 3, 401-13.
11. TaiteL Y. and Dukler, A.E.: "Modeling Flow Pattern Transitions 33. Hernandez, F.: "Comparison of Friction Factor Correlations for
for Steady Upward Gas-Liquid Flow in Vertical Tubes," AIChE Gas-Liquid Flow in Horizontal Pipes," MS thesis, U. of Tulsa
J. 26 (1980) 345-52. (1973).
12. Shohann, 0.: "Flow Pattern Transition and Characterization in
Gas-Liquid Two Phase Flow in Inclined Pipes," PhD dissertation, SI Metric Conversion Factors
Tel-Aviv U. (1982). atm x 1.013 250* E+05 Pa
13. Weisman, 1. et al.: "Effects of Fluid Properties and Pipe OF (OF-32)/l.8
Diameter in Two-Phase Flow Patterns in Horizontal Lines," 1111.
°C
1. Multiphase Flo II' (1979) 5, 437-62. ft x 3.048* E-Ol m
14. Ishii, M.· "One-Dimensional Drift-Flux Model and Constitutive m. X 2.540* E+OO cm
Equations for Relative Motion Between Phases in Various Two-
Phase Flow Regimes," ANL-77-47 (1977). * Conversion factor is exact JPT
15. Beggs, H.D. and Brill, 1.P.: "A Study of Two-Phase Flow in In-
clined Pipes," J. Pet. Tech. (May 1973) 607-17.
16. Zuber, N.: "Problems in Modeling of Small Break LOCA," Distinguished Author Series articles are general, descriptive presentations that sum-
NUREG-0724 (Oct. 1980). marize the state of the art in an area of technology by describing recent developments
17. Wallis, G.B.: "Critical Two-Phase Flow," Inti. J. Multiphase for readers who are not specialists in the topics discussed. Written by individuals
Flo\\' (Feb.! April 1980) 6, 97. recognized as experts In the areas, these articles provide key references to more
definitive work and present specific details only to illustrate the technology. Purpose:
18. Abdollahiar. P. et al.: "Critical Flow Data Review and To Inform the general readership of recent advances in various areas of petroleum
Analysis," EPRI NP-2192 (Jan. 1982). englneenng

MARCH 1984 367

You might also like