Proceedings of ITP2007, Paper Number ITP-07-20,
Interdisciplinary Transport Phenomena V:
Fluid, Thermal, Biological, Materials and Space Sciences
October 14-19, 2007, Bansko, Bulgaria
COMPUTATIONAL AND GROUND BASED EXPERIMENTAL INVESTIGATIONS OF THE
EFFECTS OF SPECIFIED AND UNSPECIFIED (FREE) PRESSURE CONDITIONS AT
CONDENSER EXIT FOR CONDENSING FLOWS IN TERRESTRIAL AND MICRO-GRAVITY
ENVIRONMENTS
1
Amitabh Narain , Shantanu Kulkarni, Soumya Mitra, Jorge H. Kurita
and Michael T. Kivisalu
Department of Mechanical Engineering-Engineering Mechanics
Michigan Technological University
Houghton, MI-49931
ABSTRACT
Reported experimental and computational results confirm that both the flow features and heat transfer rates
inside a condenser depend on the specification of inlet, wall, and exit conditions. When exit pressure is changed
from one steady value to another, the changes required of the interior vapor flow towards achieving a new steady
duct flow are such that they do not demand removal of the new exit pressure imposition back to the original
steady value. This behaviour of condensing flow is unlike that of any incompressible single phase duct flow
which cannot allow imposition of an exit pressure other than the “required” value. Instead, new steady flows may
be achieved for different steady exit pressure prescriptions (termed category I flows) through appropriate changes
in the vapor/liquid interfacial configurations and associated changes in interfacial mass, heat transfer rates (both
local and overall), and other flow variables. This special feature of these flows is for the commonly occurring
large heat sink situations for which the condensing surface temperature (not heat flux) remains approximately the
same for any given set of inlet conditions while exit condition changes. In this paper’s experimental context of
flows of a pure vapor that experiences film condensation on the inside walls of a vertical tube, the reported results
provide important quantitative and qualitative understanding. The theorotical and experimental results presented
in this paper allow us to propose important exit-condition based categorization of these flows. Of these, category
II flows, which are defined to be the cases for which exit pressures are left unspecified, have typically been
assumed to hold for most actual steady realizations of gravity driven flows. However it is shown here that steady
flows under self selected exit pressure conditions are achieved under these conditions only if special hardware
arrangements are made that allow vapor to freely choose an exit pressure. If this is not so, one often has an
inadvertent category I flow without the explicit knowledge of the imposed exit pressure condition. It is
theoretically shown here that though this category II flow situation in the absence of specification of exit
condition constitutes an “ill posed” steady boundary value problem, steady flows may still be experimentally
realized under special arrangements. This fact is also validated by unsteady theory which shows that solutions
from many different initial conditions may tend to a unique steady attracting solution (as t
) for some of the
internal and/or external flows. These “attractors” exist and are strong for gravity driven flows and are often weak
to non-existent for zero gravity or horizontal shear/pressure driven flows (both for internal and external flows). In
microgravity, the “attractors” for category II internal flows are weak in the sense that they significantly lose their
strength of “attraction” after a certain downstream distance – after which the flows are indeterminate if exit
pressure remains unspecified. For these microgravity situations, the remedy is to run condensers under suitably
specified inlet and exit pressures (category I conditions) as well as a proper cooling strategy (i.e. proper wall
temperature variations).
Key-words: film condensation, phase-change heat transfer, two-phase flows, interfacial waves.
1
To whom correspondence should be addressed. E-mail: narain@mtu.edu, Phone: 906-487-2555 Fax: 906-487-2822
1
NOMENCLATURE
Cp1
Cp2
D
G
hfg(pin)
h
h
Ja
Ja|V
k1
L
LC
M in
ML
MV
pB
pD2
pT"
pin
pexit
Pr1
pxi
q′′
Qb
Q out
Re
TB
TR
Tsat(p)
Ts-xi
Tw
Tw( )
TV-in
TC-in
u
U
v
x
xe
Specific heat of the liquid condensate, J/(kg-K)
Specific heat of the vapor, J/(kg-K)
Inner diameter of the test-section, m
Inlet mass flux, 4· M / (π·D2), kg/(m2/s)
Heat of vaporization at pressure pin, J/kg
Average heat transfer coefficient, Qout / (π·D·L), W/(m2-K)
Channel gap, m
Condensate Jakob number, Cp1· ∆ T / hfg(pin)
Vapor Jakob number, Cp2 Tsup/hfg(pin)
Conductivity of condensate liquid, W/(m-K)
Length of the test-section, m
Characteristic length, LC = D for tubes and LC = h for channels, (m)
Vapor flow rate at test-section inlet, g/s
Liquid flow rate at test-section exit, g/s
Vapor flow rate at test-section exit, g/s
Evaporator (boiler) pressure, kPa
Pressure at the location near the rotameter, kPa
Pressure at a location downstream of the test section, kPa
Pressure at the test-section inlet, kPa
Pressure at the test-section exit, kPa
Condensate liquid Prandtl number, µ1·Cp1 / k1
Test-section pressures at different locations = i (i = 1, 2, …), kPa
Average convective heat flux, W/m2
Net heat rate into the evaporator, W
Net heat rate out of the test-section, W
Inlet vapor Reynolds number, 4· M /(π·D·µ2)
Evaporator fluid temperature, oC
Rotameter fluid temperature, oC
Saturation temperature at pressure p, oC
Condensing surface temperatures at different locations = i (i = 1, 2, …), oC
Mean condensing surface temperature, oC
Non-uniform steady condensing surface temperature, oC
Vapor temperature at test-section inlet, oC
Temperature of the counter-current coolant water flow at the approach to the testsection, oC
Non-dimensional velocity in the x-direction
Velocity in the x-direction, m/s
Dimensional velocity in x-direction, m/s
Non-dimensional velocity in y-direction
Dimensional velocity in y-direction, m/s
Non-dimensional distance in x direction ( /LC)
Ratio of test section length to characteristic length (L/LC).
in
in
2
fc
y
∆T
Tsup
p
Ze
Z( )
δ
2
1
2
1
D
Distance in x direction, m
Approximate length needed for full condensation (estimted by computations), m
Non-dimensional distance in y direction ( /LC)
Distance in y direction, m
Tsat(p) - T w , oC
Vapor superheat, TV-in - Tsat(p), oC
pin – pexit, kPa
Ratio of exit vapor mass flow rate to total inlet mass flow rate
Ratio of vapor mass flow rate to total mass flow rate at any location x along the test
section
Non-dimensional value of condensate thickness
Density of vapor, kg/m3
Density of liquid, kg/m3
Viscosity of vapor, kg/(m-s)
Viscosity of liquid, kg/(m-s)
Transient decay time for disturbances, s
Subscripts
Exit
In
Na
Exp
comp
Test-section exit
Test-section inlet
Natural steady case
Obtained from experiments
Obtained from computations
1. INTRODUCTION AND BACKGROUND
This paper presents a synthesis of results obtained from a fundamental and novel experimental
investigation (see Narain et al. [1]) of effects of exit conditions on internal condensing flows and
computational investigations ([2]-[6]) of internal and external condensing flows. In section-6, the paper
also presents a new set of generalized results for this and other internal two-phase flows. Reported
experimental results confirm the existing computational results for internal condensing flows ([4], [5],
[6]) that both the flow features and heat transfer rates inside a condenser depend on exit pressure
conditions over and above inlet and wall conditions (i.e., the method of cooling for the condensing
surface). The steady condensing surface temperature (not heat flux) variation is assumed fixed and
known or knowable (through consideration of the appropriate conjugate problem). This paper identifies
and establishes a multiplicity of steady/quasi-steady solutions – and/or oscillatory flows – under
different steady conditions at the exit. Experiments support the existing simulation results ([4], [5], [6])
that have already shown, among other results, the presence of multiple steady/quasi-steady solutions
under multiple steady specifications of the exit condition. The condenser exit condition (for partial as
well as full condensation) is specified by exit pressure. For partial (incomplete) condensation flows,
specification of exit pressure is equivalent (see Fig. 6 in [4]) to specification of exit vapor quality i.e. the
ratio of vapor mass flow rate at the exit to the inlet mass flow rate (see simulation results from [4]-[6]).
The computational simulations ([4], [5], [6]) for the gravity driven partial condensation cases (such as
flows inside a vertical tube) also predict that, for a certain set of inlet and wall conditions, even if the
exit condition is not specified and a suitable range of exit conditions is available for the flow to choose
3
from, the condensate flow under gravity makes the overall flow seek and attain a specific “natural” flow
and associated exit pressure condition (in Narain et al. [4] this situation is also termed as a natural
unspecified steady exit condition due to the presence of an “attractor,” i.e., an “attracting” solution).
Unfortunately many planned system designs incorporate a condenser and assume that the condenser will
always find a “natural” steady flow since no exit conditions are specified. This is not generally true. The
attainment of natural steady flows under unspecified exit conditions require special hardware
arrangements and, even then, they occur more readily (over a larger parameter zone) for gravity driven
condensate flows than they do for shear driven flows (see [2] – [3] and [6]). The reported experimental
results for gravity driven partial condensation cases under unspecified exit conditions support the
computational results – both qualitatively and quantitatively. Therefore, for an unspecified exit condition
flow, the steady “natural” exit condition may or may not exist depending on whether or not a steady
“attractor” exists. For example, as shown in [6], an attractor does not exist for many slow to moderate
inlet flow rates in horizontal or zero gravity situations. Furthermore even if an “attractor” does exist, its
realization depends on whether the attracting “natural” exit pressure value falls within the range of exit
pressures made available to the condensing flow test-section by the rest of the system/flow-loop. In the
absence of active specification of a steady exit condition, the available range of steady exit conditions is
determined by the components downstream of the condenser as well as the specific nature of the
hardware used in the design of the rest of the system (or flow loop). Furthermore, because of the small
pressure drops (see [6] or experimental runs reported here), when a steady “natural” exit condition is
achieved under unspecified conditions, or when the specified exit condition is not too far from this
“natural” steady exit value, typically, vapor flows are typically close to incompressible.
The vertical in-tube internal partial condensing flows are investigated here for a downflow
configuration. Though numerous in-tube condensation experiments have been done, most of the well
known in-tube vertical downflow experiments done by Goodykoontz and Dorsch [7]-[8], Carpenter [9],
etc. either limit themselves to sufficiently fast flow portions of the duct that do not significantly depend
on exit conditions or operate under a particular (natural or otherwise) set of exit conditions (that gets
specified or remains unspecified depending on the employed experimental set up) and, therefore, results
may vary from one experimental system to another. In addition to our group’s very early (see Yu [10])
and subsequent ([4], [5]) computational findings on the importance of exit conditions for internal
condensing flows, experimental findings of Rabas and Arman ([11]) have also indicated the significance
of exit conditions through their observation that the presence or absence of valves at the exit affected
some of their in-tube vertical downflow results.
Furthermore, several experimental results and analyses ([12] - [19]) indicate that, for certain physical
arrangements leading to a specific class of inlet and outlet conditions, transients and instabilities are
expected in complete condensation horizontal (or slightly tilted) in-tube internal condensing flows.
Since these experiments and the corresponding modeling techniques in the literature limit themselves to
a particular type of inlet and exit conditions, they do not directly apply to the presence or absence of
observed transients and instabilities in other feasible categories of exit-condition specifications.
In the context of boundary value problems for internal condensing flows, what is new is that our
computational and experimental results unequivocally show that the commonly occuring condensing
flows are very sensitive to the nature of exit conditions as well as to the changes in exit conditions (due
to changes in exit pressure). This sensitivity arises from the ease with which these changes alter the
vapor flow field in the interior. Therefore when only exit pressure condition is changed from one steady
value to another, the changes required for the interior vapor flow towards attaining a new steady flow
4
are such that they do not demand removal of the new exit pressure imposition - as is the case for
incompressible single phase duct flows with only one allowed exit pressure. Instead, for condensing
flows, new steady flows are achieved for new exit conditions through appropriate changes in the
vapor/liquid interfacial configurations and associated changes in interfacial mass, heat transfer rates
(both local and overall), and other flow variables. This happens because the “interface tracking
equation” (see, e.g., eqn. (20) of [4] which arises from setting the interfacial mass flux L K given in
eqn. (A.10) of the Appendix equal to another interfacial mass-flux Energy given in eqn. (A.11) of the
Appendix) which locates the interface involves a key parameter Energy . This key parameter Energy
which is obtained from the interfacial heat flux associated with the local temperature gradient easily
changes with changing interface location; particularly if the interface is near a heat transfer surface at a
prescribed wall temperature. It is the absence of an available adjustable range for interfacial mass-flux
Energy for single-phase flows (which have no interface) and for adiabatic gas-liquid flows (where this
interfacial mass-flux is constrained to be zero) that makes it difficult to externally impose different exit
pressures for these better understood flows.
Fig. 1: The schematic for the flow through test-section and exit condition issues.
Fundamentally, internal condensing flows’ exit condition sensitivity arises from governing equations
being “elliptic” or requiring “two-way” space co-ordinates in the flow direction (i.e. flow variable at a
point P is influenced both by upstream and downstream local neighbors as shown in Fig. 1c) to
determine the flow fields in the two phases and the interface location. However, this sensitivity to exit
5
conditions, or “two-way” behavior, is special and does not result from the typical “ellipticity” associated
with slow flows and flow reversals (which are, as described in Patankar [20], associated with the
changes in the sign of local Peclet numbers). In fact, it is found (for simulations in [4]-[6]) that even
when vapor and liquid flows are unidirectional (or “one-way” or “parabolic”) due to local Peclet
numbers (which appear in local discretization equations for the velocity components - see eqs. (5.61) –
(5.64) in [20]) being very large (greater than 50,000 in both x and y directions for discretizations used in
exit-condition sensitive simulation results reported here), the flows exhibit “elliptic” or “two-way”
behavior leading to sensitivity to exit conditions. This special sensitivity to exit condition is due to “twoway” behavior of the vapor pressure fields (reflected, in the context of our computational methodology
[4], by the “two-way” behavior of the pressure equations given by eqs. (6.30) – (6.31) in [20]) combined
with the degree of freedom available with regard to relocation of the interface towards accommodating
changed vapor flow fields through changed interfacial mass-fluxes. These equations for condensing
flows are such that the coefficients that multiply the pressures at the locally upstream and downstream
neighbors are comparable even for large Peclet numbers (see pressure discretization equations for
SIMPLER procedure in Patankar [20]). Due to this, effects of changes in the exit pressure are felt by the
entire vapor flow field which is then able to accommodate this change by changing the interface location
and achieving a new flow field with different interfacial mass-transfer rates (and hence heat-transfer
rates).
A much simpler support for the above result comes from simple one-dimensional modeling (leading
to ordinary differential equations) of laminar/laminar internal condensing flows in a channel. It is
appropriate to ignore the integral form of vapor momentum balance only for the gravity dominated
external flow problem of Nusselt [21] and, occasionally, for some gravity driven channel flow analyses.
For gravity driven internal (e.g. channel) flow analyses this approximation means condensate motion is
not affected by vapor momentum balance because of negligible interfacial shear and freedom of the
vapor to choose its exit pressure – which is a freedom that does not always exist. If one does not ignore
the integral form of vapor momentum balance, the correct formulation of the problem is achieved (see,
e.g., eqn. (53) of [22] for a horizontal channel). The ordinary differential equations for the correct
formulation involves first order derivatives for other flow variables (interface speed uf(x) and film
thickness δ(x) in [22]) but second order derivatives (d2π/dx2) of the variable π(x) that represents the nondimensional value of cross-sectional pressure. To solve this set of equations correctly, which was not
done in [22], one needs exit pressure specification over and above specifications of inlet (pressure,
velocity profile, and film thickness values) and wall temperature conditions. At the time of the analysis
reported in [22], though eqn. (53) in the formulation given in [22] was correct, the solution was not.
Instead of solving the equations for different exit pressures, the reported solution effectively chose only
one exit pressure by an ad hoc mathematically specified value for the pressure gradient (dπ/dx) at the
inlet. In hindsight this was done to conform to the popular notion that the flow should be parabolic, even
though the analytical formulation in [22] suggested that the flow is “elliptic” and requires exit pressure
prescription. A serious problem with this type (as in [22]) of correct one dimensional steady formulation
is that it always requires exit pressure specification as a boundary condition. Such an analysis is
incapable of identifying the cases where exit pressure prescription is not required, such as fast partially
condensing flows in ducts of small lengths or gravity driven steady flows in a vertical channel under
unspecified exit conditions. In these cases, if the flow is allowed to choose its own exit pressure, it
should self select its own “natural” exit pressure. The results based on unsteady solutions for these
situations, which are presented here, exhibit presence of an “attracting” solution for the mathematically
“ill posed” steady formulation. These results assure us that there exists a steady solution even in the
6
absence of exit pressure specification and the associated “ill-posed” steady formulation. There are other
types of one-dimensional parabolic flow formulations/analyses that are frequently employed to allow
turbulent vapor or wavy interface through introduction of an independently assumed model for
“interfacial friction” Si. This interfacial shear Si relates to laminar liquid velocity profile through first
equality in eqn. (A.7) of the Appendix but does not require satisfaction of the second equality in eqn.
(A.7) of the Appendix – this is because vapor flow is allowed to be turbulent near the interface. The
integral momentum balance for the vapor directly uses the interfacial friction Si. Such steady
formulations involve assumed “models” for interfacial shear Si (see, e.g., Narain et al. [23]) and they
work because the model functions are such that they can yield known values of Si during forward
marching of the solution through a first order system of non linear ordinary differential equations under
prescribed initial conditions at the inlet. As shown later, with the help of the complete solution of the full
two-dimensional steady/unsteady problems, results obtained from one dimensional theory employing
such models can be good only over finite length problems involving partial condensation and sufficient
vapor velocities (with either laminar or turbulent vapor flows). Therefore, these models are
fundamentally flawed and need to be changed to allow second order derivatives in pressure (as is the
case in [22]) if one is looking at complete condensation problems or sufficient downstream distances
where vapor flows are slow and are sensitive to exit pressure.
Fig. 2: For tube flow situations specified as in Phan et al. [6] (see their Table 1 and Fig. 2), the
figure depicts three steady film thickness profiles (curves C1 to C3) for three different exit
conditions for category I steady flows. The same figure is also used to show results for category II
gravity driven flows once the exit condition specification is removed and the flow is allowed to seek
its own exit condition (curve C4). For category II flows, the figure indicates time trends for two
sets of δ(x,ττ) predictions for τ > 0. The curve C1 starts at Ze = 0.3 at τ = 0, and, as τ → ∞, tends to
the curve C4 which represents the solution for Ze = Ze|Na = 0.215. The curve C2 starts at Ze = 0.15 at
τ = 0 and tends, as τ → ∞, to the same curve C4. The curve C3 represents category I solution when
exit condition is held fixed exactly at the natural exit condition. As expected, the curves C3 and C4
overlap each other.
In the literature, condensing flows have been classified as to whether they are shear dominated or
gravity dominated, internal or external, smooth or wavy at the interface, laminar or turbulent in the two
phases, etc. It is proposed here that one can only make sense of the vast literature on forced internal
7
condensing duct flows if they are also classified in different categories based on the conditions imposed
at the inlet and the exit. The following three categories (termed categories I – III) proposed here cover
most cases of interest.
Category I (with complete or incomplete condensation under specified exit conditions)
•
Prescribed or known values of total inlet vapor mass flow rate Min (kg/s), inlet vapor quality, inlet
pressure pin, and inlet temperature at all times t. Without loss of generality, one can focus on an all
vapor flow (an inlet vapor quality of unity) at the inlet with known values of total inlet (all vapor)
mass flow rate M in (kg/s), inlet pressure pin, and inlet temperature (at saturation temperature Tsat(pin)
or at some superheat) at all times, t. This means the prescription could be steady or unsteady.
•
Prescribed method of cooling that leads to known condensing surface (wall) temperatures Tw( ,t) <
Tsat(pin) for all -locations over which film condensation occurs. Typical wall temperature
conditions of interest are steady, but unsteady conditions are relevant to start-up and shutdown.
•
Specified or known exit pressure conditions. For example, for steady exit conditions, exit pressure pe
= constant, which is equivalent, for steady partial condensation flow cases, to setting exit mass
quality Ze ≡ M V (kg/s) M in (kg/s) equal to an appropriate constant, where M V (kg/s) is the vapor
mass flow rate at the exit.
For category I partial condensation flows with steady specified exit pressure and other conditions,
the computational simulation results shown in Fig. 2 are the solutions of the formulation given here in
the Appendix. The flow conditions and formulation are the same as the ones specified for Fig. 2 and
Table 1 of Phan et al. [6]. The results show three different steady solutions (curves C1 to C3) for three
different specified vapor qualities Ze at the exit (viz. Ze1 = 0.3, Ze2 = 0.15 and Ze3 = 0.215). Some of the
Garimella et al. experimental investigations ([24]-[25]) are for flows in this category with inlet quality Zi
(< 1) and exit quality Ze specified at values incrementally smaller than Zi.
Category II (with complete or incomplete condensation under unspecified exit conditions)
•
Prescribed or known values of inlet mass flow rate M in (kg/s), inlet pressure pin, and inlet
temperature (at saturation temperature Tsat(pin) or at some superheat) at all times, t. Without loss of
generality, it is assumed that the flow is all vapor at the inlet.
•
Prescribed method of cooling leading to known wall temperatures Tw( ,t) < Tsat(pin) for all
locations over which film condensation occurs.
•
Exit condition is not specified. However, due to some system hardware limitations and constraints, if
only a range of exit conditions (i.e. pressures) is available and the condenser is freely able to choose
its “natural” exit pressure within this range, we still call it a category II flow. On the other hand, in
the case of the availability of only a limited range of exit pressures, if the flow seeks a “natural”
value that is at or outside the boundary of the limited range, the flow is not considered a category II
flow.
8
-
For category II unsteady partial condensation flows with unspecified exit pressure or exit quality, the
computational simulation results shown in Fig. 2 are the solutions of the formulation given here in the
Appendix. The steady formulation for this problem is however “ill-posed.” The flow conditions and
formulation are the same as the ones specified for Fig. 2 and Table 1 of Phan et al. [6]. For these gravity
driven condensate cases of category II flows shown in Fig. 2, it is computationally shown that if exit
vapor quality specification constraints at Ze1 = 0.3 or Ze2 = 0.15 are removed at some time (τ = 0) – and
subsequently (τ > 0) one does not specify any exit condition in the unsteady mathematical model
(which is equivalent to not specifying exit pressures and allowing them to vary over a certain range), the
solutions show that the experimental flow in this category should be able to select, in accord with the
predictions, a quasi-steady flow with a “natural” value of exit quality (curve C4, representing Ze3 = Ze-Na
= 0.215). However, as discussed later, experimental arrangements for category II flows require special
design and the available range of experimental conditions may not include the “natural” exit condition
the flow is seeking. It is also possible that natural “attractors” for category II flows are weak or do not
exist, as is shown to be the case for some horizontal and zero gravity condensing flows discussed later.
For these cases, the concave bowl analogy schematic for “attractors” given in Fig. 9 of Narain et al. [4]
needs to be replaced by a flat or weakly concave shape for this bowl. Therefore, the existence of steady
solutions is at the mercy of other factors such as whether or not an “attracting” steady solution exists
and, if it does, whether or not downstream conditions in the experimental set up are conducive to the
attainment of this “attracting” solution and associated exit condition.
As a matter of semantics, it should be noted that, for category II flows, in Phan et al. [6], existence of
an “attractor” leading to a long-term steady exit condition was termed differently – it was called a longterm “one-way” or “parabolic” behavior. Similarly, non-existence of an “attractor” (typically an
indicator of flows that lie outside the annular regime and are more complex in the sense that they exhibit
certain degrees of randomness or indeterminacy) was termed differently in [6] – it was called long-term
“two-way” or “elliptic” behavior.
Category III (Complete condensation involving special specified conditions
at the inlet and the exit)
Though, technically, this is a special subcategory of a general unsteady and/or steady category I
flows, it is listed separately because it typically involves and often used specialized experimental set-up
and hardware facilities. This class of condensing flows has been extensively investigated in the
experimental literature ([12]-[19]) for steady or oscillatory flows.
•
In this case there is a constant pressure reservoir, with a stagnation pressure pTank-in, that feeds the
⋅
vapor flow (at inlet mass flow rate Min (t) , pressure pin, temperature TV-in, and density V-in) into the
test section through an inlet valve (with valve coefficient ki). This requires the inlet pressure, pin, to
satisfy at any time t:
M (t)2
pin ≅ pTank-in − ki in
(1)
V-in
• Also, there is a constant pressure exit tank, with a stagnation pressure pTank-exit, to which the
condensate flows through an exit valve of valve coefficient ke. The exit valve handles an all liquid
flow because this case is only for complete condensation flows. At the exit of the condenser, the
9
liquid flow rate is Mexit at any time, t, and the liquid density is
section exit pressure, pexit, satisfy at any time t:
Mexit (t)2
pexit ≅ pTank −exit + k e
L-exit.
L − exit
•
Prescribed or known steady wall temperatures Tw( ) < Tsat(pin) for all
condensation occurs.
This requires that the test
(2)
-locations over which film
The experimental and/or modeling analysis papers of Wedekind et al. ([12], [15]), Bhatt et al. ([13][14], [16]), Kobus et al. ([19]), Liao et al. ([26], [27]), etc. focus on category III flows for a horizontal
condenser. Our theory and experiments also suggest what they find, namely, unless pTank-in and pTank-exit
are sufficiently compatible with system parameters (valve coefficients ki and ke) for a desired steady
⋅
value of M in , an oscillatory flow may result with unsteady and oscillatory inlet and exit pressures (viz.
pin and pexit).
The experiments and computations in this paper, however, focus only on steady category I and
category II partial condensation flows. As a result, the flow transients and system instabilities reported
in this paper, as far as flows within the test-section are concerned, are necessarily of different origin.
However, at a system level, the experimentally observed flow oscillations’ relationship to the better
known ([14], [16]) results for category III flows in the downstream auxiliary condenser is also discussed
here.
The experimental runs reported here largely involve laminar condensate and turbulent vapor
situations with possible vapor compressibility effects for some of the category I flows. Despite this, both
qualitative and quantitative comparisons with simulation results based on the laminar-vapor/laminarcondensate methodology given in [4]-[6] are possible for a feasible subset (within the boundaries for
steady annular flows) of experimental runs. This comparison is presented here and is possible because
turbulent vapor often laminarizes in the vicinity of laminar condensate as the condensate is slow and
remains laminar approximately up to Re 1800 (see film Reynolds number, Re , definied in Phan and
Narain [28]). Also, for gravity driven condensate cases considered here, the existence of turbulent vapor
zones in the core and entrance zone of the condenser has only minor second order impact on most of the
flow variables with first order effects limited to pressure variations in the condenser. The far field vapor
turbulence often tends not to be a significant player because the overall flow features (local and average
heat transfer coefficients) are dominated by interfacial mass and heat transfer rates, which are governed
by the typically laminar nature of the gravity driven condensate flow and the associated laminar nature
of vapor flow in the vicinity of the interface. Because of the above, all experimental runs reported here
for partial (or incomplete) condensation cases involving laminar condensate show a very good
qualitative agreement with the simulations as far as the existence of multiple steady solutions for
multiple steady exit conditions (category I) and a “natural” steady solution for the unspecified exit
condition cases (category II) are concerned. The agreement with simulations, with regard to exit vapor
quality and general consistency with overall heat transfer rates, are also quantitatively very good for
category II experimental runs that fit the annular flow assumption for the simulations.
The experiments reported here involve a single pure working fluid (viz. FC-72 by 3M Corp.) and
focus on inlet mass flow rates that correspond to inlet vapor Reynolds numbers in the range of 10,000 40,000 and vapor to wall temperature differences of 3 - 60oC (i.e. 0 Ja 0.4).
10
Furthermore, computational results for external condensaing flows that are obtained from this
simulation tool have proven record (see Phan and Narain [28] and Kulkarni et al. [2]–[3]) of making
qualititively and quantitiavely correct predictions for steady flows. The steady flow predictions are in
excellent aggrement with the classical solutions of Nusselt [21] and Koh [29]. Additionally, the unsteady
results obtained from this computational tool are also able to accurately predict wave phenomena and
their effects for these benchmark classical problems.
Hitherto nonexistent accounting of exit condition categories (and exit condition parameter “ϕ”
introduced later in this paper for internal condensing flows, as discussed in section 6), is perhaps one of
the reasons for the large uncertainities and deficiencies noted by Palen et al. [30] with regard to the poor
usefulness of quantitative information available from existing correlations for heat transfer coefficients.
Fig. 3: For a channel (h = 0.004 m) in 0g environment and flow of R113 at uniform inlet speed U =
0.6 m/s and temperature difference ∆ T = 5°C, the figure shows non-dimensional film thickness at
different non-dimensional times (see [6]). A zone with strong steady attractor is seen in the initial
part of the channel and the zone after x* > 23 exhibits weak unsteady attraction to a quasi-steady
solution.
Instead of the strongly attracting solutions for gravity driven category II flows in a vertical channel
(of the type shown in Fig. 2), one obtains significantly weaker attracting solution for horizontal channel
or 0g channel flow situations and this is exemplified by Fig. 3 where an attractor exists only for 0 x
x* 23 with the attractor’s strength diminishing with increasing distance from the inlet. For x* x
x** 45 in Fig. 3, the “attractor” can be seen to have morphed into a rough “attracting zone” that is
sufficiently weak and wavy and thickens with increasing x (no initial disturbance were imposed – the
simulations are showing amplified computational disturbances). For x >> x** 45 locations that are not
shown in Fig. 3, the “attracting zone” is non-existent and true “elliptic” nature of the problem manifests
11
itself for unspecified exit condition category II flows - making the flow physically and computationally
indeterminate. Furthermore, it is found that x* increases/decreases with increasing/decreasing inlet mass
flow rate (per unit depth of the channel) M in = ρ2Uh and, for x*
0, there is a positive non-zero lower
threshold of mass flow rate – termed M in − cr - for which no steady flow exists over any length of the
channel. Therefore, if the category II flow is one of partial condensation and the length of the channel is
xe, then we have the following possibilities: the flow has an “attractor” if xe < x*, it has a weak
“attractor” over x* xe x** if xe is in this range, and the flow has no solution possible under category
II conditions if xe >> x**. In other words, fully condensing steady flows of this type (stratified/annular)
in micro-gravity or horizontal channel is an impossibility for category II flows. However, as shown later,
a fully condensing steady flow under category I or III flow conditions with suitably prescribed inlet
pressure, exit pressure, and cooling method (i.e. condensing surface temperature Tw( )) is possible and
can be arranged (this allows for non-annular plug-slug flow regime near the point of full condensation)
for zero gravity or shear driven horizontal situations. If the exit pressure chosen for realization of
category I flow is not compatible with inlet and other conditions, oscillatory flow may result.
The above results also make physical sense. The attractor for gravity driven cases under category II
flows arise because of the irrelevance of vapor motion as the condensate is driven by gravity and the
vapor motion is eventually (near the point of complete condensation) determined by a liquid flow which
is allowed to flow and get whatever pressure it seeks. However since the entire flow morphology can
easily be changed by a “non-natural” externally imposed prescription of exit-pressure under category I
conditions, the category II attainment remains sensitive. For shear driven horizontal or micro-gravity
flows, existence of an annular/stratified “attractor” of the type being sought here requires sufficiently
fast vapor flows M in > M in − cr that can provide sufficient shear and determine pressure fields over a certain
length x* of the duct. However, the definition/existence of this attractor is in jeopardy if the vapor
motion is not sufficiently well defined; this which happens, for unspecified exit pressures, after a certain
distance x*.
Fig. 4a: The figure shows the non-dimensional film thickness and the u versus y velocity profile (at
x = 25) for steady solution of the typical Nusselt problem (see [28]) with ∆ T = 5°C.
12
Fig. 4b: For a forced flow with U = 0.2 m/s and all other conditions remaining the same as in Fig.
4a, this figure shows the non-dimensional film thickness and the u velocity profile (at x = 25)
obtained from a steady solution of the problem. It can be seen that the film thickness values do not
change, as compared to the those in Fig. 4a, but u velocity profiles and associated shear values are
different for the two cases.
Fig. 5: For a horizontal condensing flow over a flat plate (see [3]) with U = 1.7 m/s and ∆ T =
5°C, this figure shows non-dimensional film thickness at different non-dimensional times given by
the unsteady solution of the problem. An initial guess given at time τ = 0 is seen to get attracted to
the long term steady solution at different rates. This rate decreases along the length of the plate.
The markings demarcate the zones that converged into steady solution at different times.
13
In fact the strong versus weak attractor situations depicted, respectively, in Figs. 2 and 3 for vertical
and 0g channel flows is quite typical and are also present, respectively, for their external flow
counterparts, namely the Nusselt problem (see [28]) and the Koh problem (see [2]-[3]) as well. Note,
however, both the Nusselt [21] and the Koh [29] problems have well defined uniform pressure values at
the inlet, far field, and at the finite exit location used to model the infinite plate. Therefore, in a sense,
the solution of these problems, particularly the Koh problem [29] for which vapor motion is important,
should behave more like a category I than a category II flow problem. The steady solution shown in
Figs. 4a and 4b are for Nusselt problem and Nusselt problem in the presence of forced convection
respectively. The depicted steady interface locations shown in Figs. 4a and 4b are strong “attractors” as
they have also been obtained, though not shown here, in the limit of τ
solutions of the unsteady
problem with an initial condition different than the long term steady solution itself. As seen from the
comparison of Figs. 4a and 4b, for the case considered, the film thickness does not change in the
presence of forced convection as the vapor motion remains unimportant. This is because, for both cases,
gravity is the dominant force in driving the condensate and the steady solutions shown and generally
obtained (see [28]) satisfy the well known analytical results (see [21] and [28]) for film thickness in the
presence or absence of waves. In comparison, the representative 0g solutions (obtained by solving the
unsteady equations in [2] – [3] for the Koh problem [29]) shown in Fig. 5 are possible only if U is above
a certain lower threshold (termed U |L-cr in [3]). Furthermore the solution in Fig. 5 exhibits the presence
of an “attractor” whose attracting strength decreases with increasing x. The existence of a threshold
inlet speed is similar to what is needed for the channel flow in Fig. 3 – but the Koh problem [29], being
a category I external flow problem has an “attractor” which theoretically exists at all x locations (i.e.
there is no x* as in Fig. 3). Despite this, in Kulkarni et al. [2]-[3], it is shown that the weak attractor at
downstream locations becomes more noise-sensitive and amplifies ever-present minuscule noise
(condensing surface vibrations, etc.) leading to persistent and large amplitude waves around Rex|wavy-cr
5 x 105. Further downstream of this, the condensate is expected to become turbulent.
2. EXPERIMENTAL FACILITY
The condenser section, which is of the type shown in Figure 1a, is typically a part of a closed flow
loop. The flow loop, which maintains a steady inlet pressure (pin) and mass flow rate ( M in ) at the inlet,
while maintaining a prescribed steady (and nearly uniform) condensing surface temperature, may be
designed to provide different categories of exit conditions. Exit condition specifications for category I
and category II partial condensing flows defined earlier in section 1 are realized through flow
arrangements indicated in Fig. 6 and Fig. 7 respectively.
A 0-500 W evaporator/boiler in Figs. 6-7 is used to evaporate the working fluid (FC-72). The vapor
mass flow rate out of the evaporator, Min , is fed into the test section. This mass flow rate is measured by
a Coriolis flow meter F1 and, during transients, this value can be controlled by the pneumatically
actuated control-valve V1 (shown connected to F1 in Figs. 6-7). Under steady conditions though, the
value of Min gets approximately fixed by the net steady electrical heating rate for the evaporator. This is
due to the restriction imposed by the evaporator energy balance, viz., M in ≈ QB / h fg (pB ) . Here QB is the net
heat rate into the evaporator, pB is the steady evaporator pressure, TB Tsat(pB) is the steady evaporator
temperature (which is nearly equal to the saturation temperature of the fluid at pressure pB), and hfg is
the heat of vaporization at the liquid/vapor surface pressure pB in the evaporator. Towards reduction in
start-up time to steady state in the evaporator, the liquid flowing in the evaporator is warmed up,
14
(between points P' and B' in Figs. 6 – 7) so its temperature is nearly equal to the evaporator temperature
TB Tsat(pB).
Fig. 6: The schematic of the flow loop for achieving specified exit condition category I flows for
partial condensation cases.
The test-section is a 0.8 m long, vertical, stainless steel (316 SS) tube with 6.6 mm inner diameter,
D, and 12.7 mm outer diameter. At the entrance of the test-section, the inlet vapor temperature is
denoted as TV-in, the inlet pressure is denoted as pin, and the inlet vapor is kept slightly superheated (210oC superheat obtained by heating a relevant portion of connecting tubes by a rope heater). A suitable
thermocouple and an absolute pressure transducer respectively measure the temperature, TV-in, and
pressure, pin, of the vapor at the inlet. The dynamic view from an axial boroscope, mounted at the top of
the test-section shown in Fig. 8, is used to visualize and ascertain the nature of the flow in the first half
of the test-section. However, because of sharpness and contrast improvements that are needed for better
quality images, snapshots and video clips of the flows are not included in this paper. They are, however,
expected to be available in the near future. We are currently able to use these views to ascertain whether
or not annular film condensation begins near the indicated “start of condensation” point in Fig. 8 and,
also, to ascertain (and then to ensure) dryness of the vapor up to the test-section inlet.
The test-section (see Fig. 8) is suitably instrumented with various sensors (thermocouples, pressure
transducers, etc.). For future work, it will be possible to obtain local film thickness data through
integration of our recently invented, non-intrusive film thickness sensors that utilize the principle of
fluorescence and fiber-optic technology (see Ng [31]) and are able to measure the “local” time-varying
15
thickness of dynamic liquid films. The technique used for mounting all the sensors in Fig. 8 is described
in detail in [32] - [34].
Fig. 7: The schematic of the flow loop for achieving unspecified exit condition category II flows for
partial or full condensation cases.
The test-section in Fig. 8b (not shown to size relative to the outer tube in Fig. 8a) is centrally aligned
in the hollow space of a larger diameter stainless steel (314 SS) tube. This outer tube has an inner
diameter of 23.62 mm and an outer diameter of 25.40 mm. The test-section tube is cooled by the flow of
cold water in the annulus formed by the outer surface of the test-section tube and the inner surface of the
outer tube. As shown in Figs. 6-7, the flow of coolant water is arranged by a separate closed loop
consisting of the shell-side of the shell-in-tube heat exchanger (flow is on shell-side) and a pump. A
separate loop, not shown in Figs. 6-7, assures secondary coolant (cooler-water) flow at a steady constant
temperature and a steady flow rate through the tube-side of the heat exchanger in Figs. 6-7. This loop
(see Kurita [33]) replaces the open drain water loop used in the preliminary experiments. This loop has
two chillers in series (one for coarse and one for finer control of temperature) and this provides for a
good control of the steady value of temperature TC-in (marked in Fig. 8a) at the coolant inlet location in
Figs. 6-7. This, in turn, enables repetition of experimental runs regardless of seasonal variations in drain
water temperature.
Resistance temperature detectors (RTDs) and type-T thermocouples measure temperatures at
different locations of the test-section (see Fig. 8) and at other flow loop locations marked by points B,
B′ , T ′ , C1, C2, D1, D2, P1′ , P2′ , and P′ in Figs. 6-7. A barometer measures outside atmospheric pressure.
16
Flow meters at locations marked F1 (Coriolis meter that directly measures mass flow rate), F2 (a volume
flow rate measuring rotameter), P1 (volume flow rate meter imbedded in pump P1), and P2 (volume flow
rate meter imbedded in pump P2) yield mass flow rates through those locations. Absolute pressure
transducers measure pressures at test-section inlet (location 1 in Fig. 8), and at locations B and D2 in
Figs. 6 - 7. Differential pressure transducers measure pressure differences in the test-section (in Fig. 8a,
this is between locations 1 and 9, location 3 and outside atmosphere, and location 6 and outside
atmosphere). Two electronically controllable displacement pumps P1 and P2 (see Figs. 6-7) can pump
liquid FC-72 at a steady or unsteady specification of volume or mass flow rates. A pneumatically
controlled valve V1 is used, as needed, to control mass flow rate through F1. Most of the details of the
employed data acquisition system are explained in Narain et al. [34] and Siemionko [32].
Fig. 8: (a) The photograph of condenser test-section. (b) The test-section schematic (diameters in
(a) and (b) are not to the same scale). The condensing surface covers the zone 0
10.
17
For convenience, the system in Figs. 6-7 is broken into the following sub-systems. (i) Sub-system
is the portion of the flow loop between points P′ and T ′ (this portion contains the flow into the
evaporator, the evaporator, the flow meter F1, valve V1, and the tubing leading the flow into the testsection). (ii) Sub-system is the portion of the flow loop between points T ′ and T" (this portion consists
of the test-section). (iii) Sub-system is the portion of the flow loop between points T" and P′ (this
portion consists of the L/V separator, the two branches of the flow in the liquid line and the auxiliarycondenser line, and the pump or pumps). (iv) Sub-system is the portion that consists of the primary
coolant loop shown in Figs. 6-7 and a secondary coolant loop (not shown here but shown in Kurita [33]).
The sub-systems
defined above are not marked in Figs. 6-7 but the definitions introduced here are
necessary for later discussions of the experimental results.
Additional details of this experimental facility (rather of a similar earlier version), data-acquisition,
and LabView 7.1 based data processing and instrument control strategies are available in Siemionko
[32].
3. EXPERIMENTAL PROCEDURE
Here we describe the procedure for investigating partial condensation cases under different exit
conditions. Some procedure and results for fully condensing cases are also given in Narain et al. [1].
Note that results from different exit conditions are to be compared for approximately the same inlet mass
flow rate M in , inlet pressure pin, and temperature difference T = T sat ( p in ) − Tw , where Tw is the mean
condensing surface temperature. Pure vapor may be allowed to enter the test section with a superheat of
2-10oC. The purging process (see [32]) ensures that the vapor flowing in the test section is pure over the
duration of the experimental run and that non-condensable air in the flow loop has zero to insignificant
presence. Note that vapor Jakob numbers Ja|V ( Cp2 Tsup/hfg(pin)), where Cp2 is the specific heat of the
vapor and hfg(pin) is the heat of vaporization at pressure pin) represent the ratio of sensible cooling of
vapor to heat of vaporization. Since these numbers are very small (< 1.0 ⋅ 10-5) in comparison to liquid
Jakob numbers Ja ( Cp1 ∆ T /hfg(pin)), where Cp1 is the specific heat of the liquid condensate), vapor
temperature can be effectively modeled as a steady constant equal to the inlet saturation temperature.
The steadiness of Min primarily depends on the constancy of heat supply to the evaporator (which is
easily achieved by constant electric heating at a known wattage) and the eventual approximate
steadiness of the evaporator pressure pB. Even if the steady value of pB changes somewhat for different
start-ups, as long as the corresponding evaporator saturation temperature TB = Tsat(pB) changes
negligibly, it is found that the remaining flow loop pressures relative to a single effective boiler pressure
are the same for two independent repetitions of the same experimental run (i.e. same Min , ∆ T , Tw , and
Ze). This is true because all other pressures, p, are effectively characterized by their pressure differences,
p-pB, relative to this pressure pB. Alternatively, a previously obtained steady value of pB can be regained
by bringing the new evaporator pressure back to roughly this same value by: (i) use of a newly
developed electronically controlled accumulator discussed in section-6, or (ii) by suitably switching the
heater on and off under same steady heating rate, or (iii) by gradually venting the evaporator from a
higher pressure to the desired earlier value of pB. The last two of the above described processes have
been used to successfully assess the repeatability of a few representative experimental runs. With regard
to constancy of water temperature TC-in at the coolant inlet for the test-section (see Fig. 8) over time;
achieving constancy of water temperature and its flow rate in the secondary coolant loop (not shown
here) proved to be sufficient.
18
Specified Exit Conditions (Category I flows)
Incomplete or partial condensation flows
For investigation of specified conditions (through a steady and specified exit vapor quality) at the
exit that involve partial condensation flows through the test-section, the arrangement in Fig. 6 is used. In
this arrangement, the liquid at the exit flows out of the test-section at a mass rate of M L , goes through
the liquid/vapor (L/V) separator, and is pumped by pump P1 back into the evaporator. Both the pumps P1
and P2 (displacement pumps made by Masterflex) in Fig. 6 allow digital control of flow rates. The vapor
at the exit flows out of the test-section at a mass flow rate of MV and is measured through a volume and
mass flow rate measuring rotameter F2. This vapor then flows through an auxiliary condenser where the
vapor is completely condensed into liquid, goes through the pump P2, and then, on its way to the
evaporator, merges near point P′ (see Fig. 6) with the liquid flowing out of pump P1.
The control strategy, to achieve a specified steady flow with a prescribed exit vapor quality
( M V / Min Ze) for a given inlet and wall conditions, is to initially hold valve V1 open at a fixed level of
opening while ensuring (as described in the first paragraph of this section) desired steady values of Min ,
. Then the exit vapor mass rate M V through pump P2 (or rotameter F2) is held fixed at a
value less than the inlet mass rate Min while exit liquid mass flow rate M is varied through pump P1 at a
pB, and
∆T
L
value given by the tracking equation: M L |P1 = M in − M V |Rotameter .
As the flow through the evaporator becomes steady, Min becomes steady, and, at that time, we may
or may not need to hold this value actively fixed with the help of controllable valve V1. At this stage,
active control of valve V1 does not achieve much except that it eliminates some unwanted minuscule
drifts in the inlet mass rates. For a given set of inlet ( Min , pin, TV-in) and wall ( Tw ) conditions, different
specified steady states are achieved by the above strategy for different values of M L . Examples of
experimentally achieved partial condensation flows under specified exit conditions (category I flows)
are given and discussed in the next section.
Unspecified Exit Condition Cases (Category II flows)
“Natural” Partial Condensation
For obtaining/investigating existence of a “long term,” steady, “natural” exit condition for category
II flows (under unspecified exit conditions) with all other conditions being kept the same as in a
corresponding specified exit condition case in category I, the flow is required to go through the testsection and onwards under the arrangement shown in Fig. 7. Note that this arrangement has a single
displacement pump as opposed to the two displacement pumps used in the arrangement of Fig. 6. The
approach is to hold values of M in , pB, and ∆ T nearly the same as in one of the specified category I
cases while the pump P1 in Fig. 7 is controlled such that the mass flow rate through it tracks the
equation: M L| P1 = M . If start-up and other conditions (such as the range of pressures allowed in the
L/V separator) allow a steady state flow, with gravity driven condensate, is attained in which, at the exit
of the test-section, the inlet vapor mass flow rate is split, by a natural selection process, into a liquid
condensate flow rate ML |Na and a vapor flow rate M V | Na . Clearly these values satisfy the
equation: M in = M L | N a + M V | N a . Examples of experimentally achieved partial condensation flows
under unspecified exit conditions (Category II flows) are given and discussed in the next section.
in
19
4. EXPERIMENTAL RESULTS, DISCUSSIONS, AND COMPARISONS WITH
SIMULATIONS
The column headers in Tables 1-2 indicate accuracies of the values of key measured variables
obtained through flow loops’ instruments and sensors. Overall accuracy bounds for the reported
calculated variables (such as q′′ , h , etc.) are also shown. The non-dimensional numbers Rein, xe = L/D =
106, Ja, ρ2/ρ1, and µ2/µ1 in Tables 1-2 define the flows and they are defined in [4]-[6]. In Tables 1-2, the
heat flow rate Q out and associated average heat transfer coefficients ( h ) are obtained through the
relation: Q out ≈ M L h fg = h ( DL) ∆T . The inlet vapor mass-flux G in Tables 1-2 is defined as 4 M in /
(π⋅D2).
All (except the Coriolis meter F1) of the instruments’ accuracies for measured variables were
established after their in-house calibrations with the help of suitable and reliable reference instruments
of known resolution and appropriate reference physical conditions (temperature, flow rate, pressure,
etc.). The accuracy of the Coriolis meter was established by the vendor support staff at the time of its
installation. The error estimates for the calculated variables reported in Tables 1-2 were obtained by
well-known standard procedures (see, e.g., eqs. (3.27)-(3.28) in Parratt [35]). The accuracies of
individual calculated variables in a column were taken into account to report maximum values of the
errors in the column headers of Tables 1-2. All the individual values of errors were either less than or
equal to these reported error values. The error definitions, associated error analyses and calibration
accuracies can be found in Narain et al. [34].
Table 1: Experimentally measured data and some key calculated and computed variables for
steady states achieved for category II (unspecified exit condition) partial condensation
flows.
Run
M in
MV
Ze
Ze
EXP
COMP
p
pD2
2/ 1
2/ 1
G
Re
Ja
Pr1
q′′
Tw
Tsat
∆T
pin
px6
pT"
(K)
(K)
(K)
(kPa)
(kPa)
(kPa)
(kPa)
(kPa)
±1
±0.15
±1
±0.6
±0.6
±0.7
±0.05
±2
±0.0001
±0.0001
±1.5
±900
±0.02
±0.02
±5
±800
±80
Qout
h
No.
(kg/m2 s)
(g/s)
±0.05
±0.04
±0.04
1
1.44
0.48
0.33
0.33
320
331.49
11
107.3
109.3
106.5
-0.82
105
0.0085
0.0243
42.1
23900
0.14
9.61
81
5200
453
2
1.76
1.08
0.62
0.57
317
325.23
8
87.0
86.4
88.0
-0.36
85
0.0070
0.0223
51.5
29700
0.10
10.11
58
3700
450
3
1.54
0.69
0.44
0.36
323
335.55
12
122.6
116.6
116.1
-0.19
114
0.0097
0.0260
45.0
25300
0.15
9.21
71
4500
387
4
1.29
0.49
0.38
0.38
320
329.64
10
101.0
102.4
100.8
-2.09
98
0.0081
0.0230
37.7
21500
0.11
9.69
68
4300
476
5
1.70
0.83
0.51
0.52
324
332.55
9
111.0
112.5
111.0
-0.93
109
0.0089
0.0252
49.6
28100
0.11
9.37
70
4400
508
6
1.17
0.47
0.40
0.39
320
332.64
13
111.3
NA
111.3
-0.12
109
0.0088
0.0246
34.2
19325
0.16
9.55
59
3700
298
7
1.31
0.49
0.37
0.37
321
330.85
10
105.0
106.8
104.5
-2.06
102
0.0084
0.0244
38.3
21700
0.12
9.58
69
4400
462
8*
1.93
1.39
0.72
0.72
322
325.55
4
84.6
NA
NA
0.01
NA
0.0071
0.0231
56.4
32500
0.05
9.85
47
3000
742
9
1.59
1.11
0.69
0.63
328
334.25
6
113.0
NA
NA
NA
NA
0.0094
0.0265
46.5
26200
0.08
9.07
40
2500
418
10
2.12
1.37
0.64
0.64
320
327.85
8
91.4
NA
NA
-0.10
NA
0.0076
0.0234
62.0
35500
0.10
9.83
64
4100
503
11
1.30
0.45
0.35
0.38
321
329.29
8
99.8
100.2
99.7
-1.62
97
0.0080
0.0240
38.0
21700
0.10
9.70
72
4600
537
20
(J/s)
(W/m2) (W/m2 K)
(g/s)
Table 2: Experimentally measured data and some key calculated variables for steady states
achieved for category I (specified exit condition) partial condensation flows.
Run
M
in
M
V
Ze-exp
Tw
Tsat
T
pin
px6
pT"
(K)
(K)
p
pD2
(K)
(kPa)
(kPa)
(kPa)
(kPa)
(kPa)
2/ 1
2/ 1
G
Re
Ja
Pr1
Qout
(kg/m2 s)
h
(W/m2) (W/m2 K)
(g/s)
(g/s)
±0.05
±0.04
±0.04
±1
±0.15
±1
±0.6
±0.6
±0.7
±0.05
±2
±0.0001
±0.0001
±1.5
±900
±0.02
±0.02
±5
±800
±80
1
1.44
0.52
0.36
320
331.91
12
108.8
110.9
108.0
-1.82
106
0.0087
0.0244
42.1
23800
0.15
9.59
78
4900
416
2
1.31
0.50
0.38
320
329.16
9
99.4
100.3
99.5
-0.43
98
0.0080
0.0238
38.4
21900
0.11
9.72
69
4400
505
3
1.34
0.59
0.44
320
328.55
9
97.4
98.4
97.3
-0.32
96
0.0078
0.0236
39.1
22400
0.10
9.76
64
4100
491
4*
1.80
0.64
0.36
323
330.53
8
103.9
105.9
104.4
-0.59
103
0.0083
0.0246
52.6
29900
0.09
9.50
97
6200
851
5*
1.92
1.00
0.52
323
329.64
7
101.0
103.3
101.7
-0.40
100
0.0081
0.0244
56.1
32000
0.08
9.55
78
4900
771
6
1.20
0.00
0.00
320
332.23
12
110.5
113.2
110.5
-0.80
108
0.0088
0.0245
35.0
19800
0.16
9.57
101
6400
514
7
1.22
0.44
0.35
320
331.71
12
108.0
110.6
108.0
-0.62
106
0.0086
0.0243
36.6
20200
0.15
9.62
66
4200
351
8*
1.93
1.15
0.60
322
325.21
3
87.15
NA
NA
-1.30
NA
0.0071
0.0231
56.4
32600
0.04
9.87
67
4300
1174
9
1.61
1.14
0.70
322
332.57
11
111.4
104.6
107.0
-0.87
103
0.0089
0.0250
47.1
26630
0.13
9.46
35
2200
212
10
1.67
1.37
0.82
322
331.45
10
105.8
98.6
100.8
-0.74
97
0.0085
0.0245
49.0
27800
0.12
9.57
16
1000
108
11
1.81
1.02
0.56
317
331.48
14
106.8
102.1
NA
-0.58
99
0.0085
0.0237
52.9
30000
0.18
9.80
54
3400
236
12
1.66
0.60
0.36
318
333.15
15
113.5
108.6
NA
-1.46
106
0.0090
0.0243
48.5
27400
0.20
9.66
88
5600
363
13
1.71
0.49
0.28
316
331.14
15
108.0
107.9
108.1
-2.10
105
0.0085
0.0237
50.1
28400
0.19
9.81
103
6500
441
14
1.30
0.65
0.50
320
326.34
6
90.5
92.1
90.5
-1.15
88
0.0073
0.0230
38.0
21800
0.08
9.90
56
3500
545
15
1.29
0.55
0.43
321
331.63
11
107.7
113.2
110.1
-1.60
107
0.0086
0.0245
37.7
21400
0.14
9.57
62
3900
366
16
1.31
0.76
0.58
321
331.07
11
105.8
109.1
105.8
-0.95
104
0.0084
0.0243
38.3
21700
0.13
9.62
46
2900
279
17
1.14
0.88
0.77
319
326.73
8
91.6
92.8
91.9
-0.14
90.4
0.0074
0.0230
33.5
19300
0.09
9.92
22
1400
186
18
1.39
0.64
0.46
318
329.01
11
98.9
101.3
98.4
-0.65
97.1
0.0079
0.0234
40.6
23200
0.14
9.85
64
4100
377
No.
(J/s)
q′′
The experimental runs reported in the next section were taken after ensuring that: (i) representative
runs were repeatable, (ii) the mass flow rates for partial condensation cases added up to satisfy mass
.
balance, (iii) overall energy balance for the test section was satisfied i.e. M in hfg
.
M w Cpw
Tw ,
where, Tw is the rise in the water temperature in the annulus surrounding the test-section, and (iv)
various data were reasonable (based on simulation estimates) and consistent with one another. The
experimentally obtained partial condensation cases in category II (unspecified exit) are listed in Table 1
with all the essential details including exit vapor quality Ze (fourth column) and its value obtained from
simulations (fifth column). Note that the fourth and fifth columns are in good agreement with one
another. The corresponding partial condensation flow cases under category I (specified exit condition
cases) are listed in Table 2. The discussion for these partial condensation flow results for exit condition
categories I and II are given next.
21
Fig. 9a: Time history depiction of Min , MV and ∆ T values for multiple steady states of partial
condensation cases viz. Natural-1 (run 1 from Table 1), Specified-1 (run 1 from Table 2), Specified2 (run 18 from Table 2), and Specified-1 Approx.
Test Section (Sub-System ) Pressures
px6
Px6
pin
Pin
px9
Px9
Pp
120
5
116
4
Natural-1
Specified-1
Inlet Pressure
Specified-1
Approx.
Specified-2
3
108
2
Inlet Disturbance
104
1
Inlet Disturbance
Exit Pressure
100
0
.
P ( kPA)
Pressures: pin, px6, px9 (kPa)
112
-1
96
P
92
-2
Transient D
88
-3
Transient D
84
-4
76 min.
20 min.
80
0t 1
20
20 minutes
40
60
t2
80
30 min.
t3
100
20 min.
120
t4
-5
140
Time
Fig. 9b: Time history of pressures (along the test section) and p values (across the test-section)
for the cases shown in Fig. 9a.
22
Test Section (Sub-System ) Temperatures
Tx1
Tx4
Tx9
Saturation Temperature
60
Inlet Disturbance
58
Inlet Disturbance
Transient D
56
54
Temperatures (°C)
Transient D
52
Wall Temperatures
Specified-1
Natural-1
Specified-1
Approx.
Specified-2
50
.
48
46
44
42
76 min.
20 min.
30 min.
20 min.
40
0t 1
20
20 minutes
40
60
t2
80
t3
100
Time
120
t4
140
Fig. 9c: Time history of temperature values along the test section (sub-system ) and Tsat (pin) for
the cases shown in Fig. 9a.
Partial Condensation Flows
Specified Exit Condition (Category I) and Unspecified Exit Condition (Category II) Cases
The results over time interval t4
t (min)
t4 + 20, termed “Specified-1 Approx.” show the
experiment’s ability to approximately repeat the data for a case that is approximately the same as the one
for t2 t t2 + 20 termed Specified-1. Following the method described in section 3, Figs. 9a-9b also
show, over the time interval t1 t (min) t1 + 76, the attainment of a corresponding “natural” steady exit
condition and associated steady flow variables for an unspecified exit condition (category II) case. This
case corresponds to run 1 in Table 1. The Natural-1, Specified-1 and Specified-2 steady states (in Fig 9a
– 9c) have the same values of M in 1.44 ± 0.05 g/s and ∆ T 11 ± 1oC but different values of M L and
MV that satisfy M L |1 + M V |1 = M L |2 + M V |2 = M in . The differences between the Specified-1 and Specified-2
cases are: (i) they have different heat transfer rates (since, energy balance gives: Q out ≈ M Lh fg ), the two
cases respectively have approximate heat transfer rates of 78 ± 4 W and 64 ± 4 W and average heat
transfer coefficients of 416 ± 40 W/m2-K and 377 ± 40 W/m2-K, and (ii) different hydrodynamics – the
signature of which is clear through corresponding computational simulations and, also, through the
difference between experimentally obtained mean values of p for the two cases (they are, in Table 2,
respectively, -1.82 kPa and -0.65 kPa). Furthermore, specified (category I) and unspecified (category II)
flows have different dynamic responses to a disturbance (in Figs. 9a-9c, a disturbance was induced by
momentarily shutting or decreasing the opening in the valve V1 shown in Figs. 6-7). The difference in
dynamic response is seen by comparing Specified-2 and Natural-1 cases for transients’ decay time Dτ
23
associated with exit vapor flow rate M V in Fig. 9a or Dτ associated with p in Fig. 9b. In Fig. 9a, the
rapid shutting or closing of valve V1 caused the indicated responses in Min time history. For the Natural1 case, this response is as shown in Fig. 9a but a more rapid response for Specified-2 case is not captured
by the resolution of the figure and it is indicated by a dotted line. With regard to dynamic responses to a
disturbance – it is clear that Specified-2 case of Fig. 9a (though it is farther from a “natural” case) is
more stable than the case for Natural-1 because its transients decay time Dτ is much shorter. In other
words, “natural” steady states for unspecified exit conditions (category II) are generally more noisesensitive because the exit, in category II cases is not as isolated from flow variations further downstream
of it as is the exit for specified exit condition cases (category I) - and this causes additional lingering
impact of noise arising from the flow variables in the exit zone. Time histories of pressure and
temperature values in sub-system , , and for the cases in Figs. 9a-9c are not shown here for brevity
but are available in Figs. 7d-7e of Narain et al. [34].
The cases shown in Fig. 9a-9c are representative runs taken from a set of partial condensation runs
for specified (category I) and unspecified (category II) exit condition cases in Tables 1-2. The data
matrix associated with these partial condensation category I and category II cases is best represented by
Fig. 10. The test matrix for all partial condensation (including both the categories I and II) cases is
limited by the system limits and flow regime boundaries indicated on the plane marked by inlet mass
flow rate M in and temperature difference ∆ T values. Figure 10 shows all the partial condensation cases
plotted on the two dimensional plane formed by M in and ∆ T . These parameters were found to be the
key variables controlling the dynamics of the condensing flows in the test section.
Test Matrix for Category I and Category II Partially Condensing Flows
15
The region
encompassing
data matrix
13
Zone of Full
Condensation
11
Curve - C
Curve - D with computationally
obtained two points on it
__
T ,°C
9
Curve - A with experimentally
obtained two points on it
Oscillatory or unsteady flow
7
(System induced)
5
Zone of wispy
Annular Flow
3
1
0.69
Curve - B with experimentally obtained three points on
it
0.89
1.09
1.29
.
1.49
Dry patches/No annular Flow
1.69
1.89
2.09
2.29
Min, g/s
Fig. 10: Two-dimensional test data matrix for category II (unspecified exit condition) partial
condensation cases’ points and different bounding curves represented on Min - ∆ T plane.
24
.
M in , Tsat , pin
Fig. 11: The schematic of full condensation showing “point of full condensation” and its
downstream region with zero interfacial mass and heat transfer.
The typical values for lower and upper limits for inlet mass flow rate were found to be 1 g/s and 2
g/s respectively and that for the ∆ T were recorded to be 2 °C and 12 °C respectively. The interior
shaded zone in Fig. 10 represent M in and ∆ T values for which steady flows were attained for both
specified (category -I) and unspecified (category II) exit condition cases. The bounding curve-B in Fig.
10 indicates lower threshold of ∆ T such that steady condensing flows attained below that curve (see
points marked in Fig. 10) were drop wise patchy – i.e. not annular – on the condensing surface near the
inlet. Below this curve, the condensation – as observed from the inlet boroscope – indicates that the flow
is no more film annular near the point of onset of condensation as there are wet and dry patches
associated with drop-wise condensation. This happens because ∆ T value is below a lower threshold.
The bounding curve-B is partly experimental and curve-C on the right in Fig. 10 is, at present, entirely
schematic (i.e. not fully explored by experiments). Curve-C represents expected transition to wispyannular flows (see Fig. 10.3 in Carey [36]) at very high Min at any ∆ T . The dotted curve-A on the left
bottom has been experimentally noticed. This is a system-instability curve and does not represent a flow
regime boundary for the test-section, as it is a result of the exit pressure oscillations or unsteadiness in
test section imposed by oscillatory or other plug/slug instabilities occurring elsewhere in the system (in
this case, in the auxiliary condenser downstream of the test section). An example of such an instability
case is discussed in section 5. The bounding curve in the upper left corner of Fig. 10 is marked as curveD. This curve represents transition from partial condensation to full condensation. If M in is reduced and
∆ T is increased further, computations show that the left side of curve-D represent the zone for which the
entire vapor coming in condenses inside the test section (i.e., for category II flows, fc in the schematic
25
of Fig. 11 starts satisfying fc L on the left side of curve-D as opposed to fc > L on the right side of
curve-D). Note that Fig. 11 suggests, in accord with computations as well as remarks of Rabas and
Arman [11], that this point of full condensation does not necessarily imply an all liquid phase for >
> fc is such that average vapor mass flow rate MV is zero and so
fc. It simply means that the zone
are the interfacial mass and heat transfer rates at these locations.
For a few data points in Fig. 10, the rotameter F2 data was corrupted by the float’s occasional
stickiness to the rotameter walls. These cases are marked by unfilled circles in Fig. 10 and all the rest of
the good cases (also based on comparisons with computational simulations) are marked by dark filled
circles. These dark filled circles representing good partial condensation cases in Fig. 10 are actually the
projections on the Min - ∆T plane of the points reported in the three dimensional data matrix which has
M in , ∆ T
and Ze ( MV / Min ) as three axes. This three-dimensional data matrix is not shown here for
brevity but is shown in Fig. 8b of Narain et al. [34] where the figure is able to depict all the cases of
category II (unspecified exit) as well as category I (specified exit) partial condensing flows. Each point
in Fig. 10 represents, for given values of M in and ∆ T , a set of data consisting both category I cases and
its associated unique, “natural” category II case. There are, however, as seen in Fig. 8b of Narain et al.
[34], some data sets representing only category I or category II cases. For each data set in Fig. 10
consisting of projections of category I and associated category II cases, category I flows were found to
be more robust and stable as compared to their associated category II counterparts.
Comparisons with Relevant Computational Results (Partial Condensation)
Figure 12 shows the computationally obtained (employing the tools reported in [6]) details of local
film thickness and heat flux variations for the specified and unspecified “natural” cases (under noise-free
conditions) marked as Specified-2 (run no. 18 in Table 2) and Natural-1 (run no. 1 in Table 1) in Figs.
⋅
9a-9c. The variation of vapor quality Z (
⋅
M v / Min at any
LC . x) along the test section can be easily
obtained from the graphical results in Fig. 12 from the relation Z( ) ≡ 1 − {( ⋅ D)/(M in ⋅ h fg )} q′′w d .
0
The exit vapor quality for Specified-2 case was greater than that of the associated Natural-1 case. As
.
a result, higher amount of vapor condenses for Natural-1 case and this makes heat transfer rate Qout for
natural case to be on the higher side (see Table 1 and 2 for details). For all other conditions remaining
the same, as observed from the computational results in Fig. 12, this makes Natural-1 case’s liquid film
thickness to be lower and wall heat flux to be higher than the values for Specified-2 case.
Such details of representative local variations in film thickness and heat-flux are very important and
should be more extensively synthesized with experimental results before heat-transfer correlations are
developed for suitable categories and sub-categories of internal condensing flows. However, reliable
experimental information on “local” spatial variations of these quantities is not expected until
forthcoming incorporation of film thickness sensors and heat-flux sensors in these experiments. Observe
that the computationally obtained prediction of “natural” exit vapor quality Ze|Na Comp ( 0.33) for
category II flow in Fig. 9a-9c is in a very good agreement with the experimentally obtained Ze|Na Expt (
0.33) value (see Table 1). In fact, a very good agreement between Ze|Na Comp and Ze|Na Expt values was
found for all category II cases in Fig. 10 and this is clear from their numerical values in Table-1. Note
that good agreement between experimental and theoretical Ze values have also been obtained and
26
reported (see [5]) for shear driven condensing flow in a channel (category II experiments of Lu and
Suryanarayana [37]) of gap height h and length L are such that L/h xe < x*, where x* has the same
meaning as in Fig. 3.
Fig. 12: For the “Natural-1” and “Specified-2” flow cases in Figs 9a-9c, this figure shows the
computationally obtained representative film thickness and wall heat-flux variations along the test
section. The film thickness and heat flux values shown have been obtained for smooth interface
conditions. In reality, they are modulated by waves in presence of noise (see [6]).
The values of pressure drop p (given by pin - pexit) obtained from simulations for all the category II
partial condensation cases were negative and below 50 Pa, indicating pexit was greater than pin for all the
condensation cases in given M range. Thus the pressure rises across the test-section. The pressure rise
is needed to decelerate the vapor, feeding the gravitationally accelerated condensate at the rate needed
for condensate motion under unspecified exit pressure condition. This is confirmed by the experimental
values of p (see Tables 1 and 2) which are also all negative (except a very few cases). However, as
expected, the magnitudes for experimental values of p were found to be greater than those from
simulations. The reason behind this is that the simulations assume laminar vapor/laminar liquid flows
while, in reality, the vapor Reynolds number are in the higher range (20000-30000); and this makes
in
27
vapor flows significantly turbulent in the core (see Tables 1-2), but still laminar near the interface. The
turbulence in the vapor core does not affect the mass transfer across the interface by much because
condensate motion is gravity driven and the flow of the condensate and near interface vapor is laminar
in nature. However, turbulent vapor core significantly increases needed p values for the vapor domain
while still keeping its impact on the gravity driven condensate motion negligible. Because of this, the
values of vapor quality obtained from the simulation are in good agreement with the experiments but the
values of pressure drop p obtained from experiments are higher in magnitudes than those predicted by
the assumption of fully laminar vapor flows. However, predicted pressure drop p values do become
comparable to experimental values if the predicted interface location is retained and k-ε model (see
Narain et al. [34]) is used for the turbulent interior of the vapor while retaining the laminar shear stress
estimates at the interface. The modifications in computational procedure needed to account for vapor
turbulence are briefly discussed in Narain et al. [34].
()
*
Q out
M in
+
∆p
Natural $
∆p∆p==∆∆pp Natural
∆∆p = ∆∆ppff1
$
,
%&
%'
M in
ML
pP
Qin
ML
!
M L = M in" #
"
Fig. 13a: Specially designed ideal experimental arrangement. This is a specialized arrangement for
a condenser (experiencing complete condensation) in a “thermal system” that transfers heat from
a high temperature location to a low temperature location. The specialty of this arrangement is
that it can be used to keep exit pressure of the condenser unspecified (by not using the
accumulator option for A) or to specify it with precise control (by using the accumulator option
for A). It should be noted that the pressure controlling accumulators, whether they are used for
the evaporator B or the L/V separator A, accomplish their task by adjusting the amounts of vapor
that is exchanged (come in and go out) between the accumulator and the liquid reservoir of
interest (A or B).
28
Q out
M in
p
p ≡ p*
p1 ≡ p B -pin
ML
p 2 ≡ p exit − p p
p Total
System
≡ p1 + p+ p 2
Q in
Fig. 13b: Example of an experimental arrangement for which condenser exit conditions are not
directly controlled. This system, a variation of the system in Fig.13a, is again for a condenser
(experiencing complete condensation) in a “thermal system” that transfers heat from a high
temperature location to a low temperature location. This arrangement does not allow for the
condensing flow to seek a natural pressure drop as the system decides this pressure drop. This is
essentially a specified exit condition flow with the specification being decided by the system (see
section 6).
Fully Condensing Flows
The experimental data and discussions for category II fully condensing flows are not reported here
for brevity but can be found in Narain et al. [1]. The experimental investigations of fully condensing
category I flows are currently in progress. However the essential ideas for these experiments are
discussed here. For fully condensing flow cases the temperature difference ∆ T for a given inlet mass
flow rate Min in the experimental set up of Fig. 7 is such that that there is zero mass flow through the
depicted auxiliary condenser and, therefore, valve V3 can be considered closed. For the case of
unspecified exit pressures (category II flows), the flow loop in Fig. 7 makes available a range of exit
pressures for the flow in the test-section by allowing a range of values for temperature TD1 in the L/V
separator at point D1 in the flow loop of Fig. 7. As a result, the pressure p = psat (TD1) in the L/V
separator can vary over a range because temperature TD1 can take any value between the ambient
temperature and the temperature at the exit of the test-section. This special arrangement for fully
condensing flows under unspecified exit pressure conditions at fixed boiler pressure pB is best indicated
by the simplified flow-loop in Fig. 13a where the L/V separator location is now marked by point A
(instead of the marker D1 used in Fig. 7). For the case discussed above, the L/V separator at point A in
Fig. 13a is not connected to an active pressure controlling accumulator. In practice, many condensers
employ this L/V separator by making it an integral part of the bottom portion (which is not cooled) of
29
the test-section condenser – this is effectively and practically achieved by using an orifice-plate (i.e. a
significantly smaller diameter tube than the test-section) at the exit of the test-section. For specified exit
pressure fully condensing flow cases (category I), besides the boiler in Fig. 13a, the L/V separator at
point A is also connected to an “active” pressure controlling accumulator which holds the pressure fixed
at a specified value.
For the complete condensation condenser in the system of Fig. 13a, an inadvertent design – which
does not consciously keep the exit pressure specified or unspecified – is indicated in Fig. 13b.
5. RESULTS/COMMENTS ON SYSTEM INSTABILITIES AND OSCILLATORY FLOWS
While seeking the “natural” exit condition for some unspecified exit condition (category II) partial
condensation flows, system instabilities – involving oscillatory flows – of the type shown in Figs. 14a 14b are observed. The origin of these oscillatory flows appears to be the auxiliary condenser, which sees
an approximate category III flow for which, in Fig. 6, the L/V separator at point D1 is at approximately
fixed stagnation pressure and the L/V separator downstream of the auxiliary condenser at point D3 is at
another approximately fixed stagnation pressure. These pressures respectively correspond to pressures
pTank-in and pTank-out that appear in the definition given in section-1 for category III flows. Recall that
these fully condensing flows in the auxiliary condenser, unlike the ones studied for the test section, are
known (as in [14]) to become oscillatory under certain choices of pressures pTank-in and pTank-out. Since
complete auxiliary condenser flow data were not obtained (because this component was not the focus of
the reported investigations), relating this auxiliary condenser instability to the type of stability
boundaries discussed in [14] and [16] is outside the scope of this study.
It is clear from Figs. 14a-14b that oscillations in vapor mass flow rate at the exit of the test section
impose oscillations on the exit pressure and on the pressure drop ∆p across the test section while the
inlet vapor mass flow rates remain relatively unaffected. Figure 14b shows the oscillations in other
pressure values and the temperature at the rotameter F2 (which is nearer to the auxiliary condenser).
This, along with the known fact (see Fig. 9 in [4]) that there is one to one relation between exit vapor
quality and exit pressure, indicate imposition of oscillatory pressures at the exit of the test-section.
It suffices here to note that the flow oscillations in the auxiliary condenser can induce an oscillatory
exit pressure at the exit of the test-section condenser and this is the cause, in Figs. 14a-14b, of somewhat
reduced level of oscillatory behavior of other test-section flow variables. As a result of the instability in
the auxiliary condenser, the dotted curve-A in Fig. 10 is merely suggestive of the possible presence of
system instabilities. This is because the actual onset of oscillatory conditions has only a very indirect
and incomplete relation to test-section M in and ∆T values used in Fig. 10.
The issues regarding start-up time for the flow loop are discussed in Narain et al. [34] and Kurita
[33].
30
Oscillatory Category II Partial Condensation Flow
.
.
.
Exit Vapor Mass Rate MV
Inlet Vapor Mass Rate Min
Exit Liquid Flow Rate ML
Test Section -P
1.7
0.5
in
1.5
0
1.3
-0.5
P
1.1
-1
0.9
-1.5
.
MV
.
0.7
-2
Test Section P (kPa)
Mass Flow Rate (g/s)
.
M
ML
0.5
-2.5
30 min.
0.3
-3
3 min.
Time
Fig. 14a: Time history depiction of M , M , M L and p values for an oscillatory partial unspecified
exit condition case (category II). In Fig. 10, this flows’ appearance is indicated by crossing of the
dotted curve-A.
in
V
Oscillatory Category II Partial Condensation Flow
Inlet Vapor Pressure (pin)
Pressure(px6)
Exit Pressure (pexit)
Temperature at Rotameter(TR)
111
50
TR
Pressure (Kpa)
107
49.5
49
pexit
105
48.5
103
48
101
Temperature (°C)
109
47.5
pin
px6
99
47
30 min.
97
46.5
Time
3min.
Fig. 14b: For the case in Fig. 14a, this figure shows the time history depiction of rotameter
temperature TR and the following pressures: inlet pressure pin, pressure (px6) at location 6 in Fig.
7, and exit pressure pexit.
31
6. EXIT CONDITION PARAMETER AND SUMMARIZED GENERALIZATIONS OF
RESULTS FOR THIS AND OTHER INTERNAL TWO-PHASE FLOWS
As discussed above, pressure difference ∆p = pin – pexit in Fig. 13a, is typically not well defined for
condensing duct flows just by the virtue of specified inlet and wall conditions alone - though it is well
defined for single phase fluid flows. This is because outlet pressure often needs to be prescribed as a
boundary condition to determine the flow field in the interior and the associated pattern of gas-liquid
interfacial configurations in phase-change flows. The above stated conclusion for condensing flows, is
also true, for many boiling flows - particularly if changes in interface location near the heat transfer
surfaces (at specified wall temperatures) can be easily accomodated by significantly changing the
interfacial mass-flux (and heat-flux) conditions. The reason behind the significance of exit pressure for
these flows is that changes in exit pressure pexit tries to change the gas-phase squeezing pressure levels
(pin + pexit)/2 and, concurrently, vapor motion due to different ∆p = pin - pexit values that are being
imposed. Even small changes in these two factors (i.e. mean pressure level and pressure difference from
inlet to exit of two-phase component) can be accomodated by phase-change flows because of the extra
degree of freedom associated with various liquid-vapor interface configurations (flow morphologies)
that are easily altered by changing the interfacial mass-flux (and associated wall heat-flux) assuming
cooling/heating is done at fixed wall temperatures (not fixed heat-fluxes).
However, historically, exit pressure specification has not been considered because of an incorrect
and untested assumption that all two phase flow components behave similar to single phase flow
components with regard to pressure drop across the component. As a result of this misunderstanding, for
many “systems” in application, the pressure difference ∆p across a two-phase component gets
inadvertantly specified or is subjected to large system imposed variations in ∆p. This leads to
difficulties in experimentally characterizing two-phase flow behavior. Because of this, characterization
of two-phase flow behavior of a condensing or flow boiling component in a system cannot easily be
decoupled from the system and, therefore, due consideration should be given to the manner in which the
component is integrated in the system hardware. All this requires a paradigm change from seeking a
correlation for pressure difference ∆p to first assessing the value or range of externally imposed ∆p
values. Only after this assessment, one can meaningfully develop and use suitable heat-transfer
correlations.
In the context of the results reported in section-1, when exit pressure is left unspecified (category II
flow) through specially designed experimetnal arrangements, the flows further fall in the following two
sub-categories, namely: (A) the ones for which a well defined steady “natural” pressure-difference
“∆p|Natural” and associated “natural” interfacial configuration exist, and (B) the flows for which well
defined steady “natural” pressure-difference “∆p|Natural” and associated “natural” interfacial
configuration do not exist. An example of “natural” flow of type (A) is gravity dominated condensing
flows or some upstream portions (x < x*) of shear driven flows. However, in practice, type (A) flows are
seldom realized in gas-liquid flow components/test-sections unless the system is especially designed to
offer freedom to choose exit pressures. This is because, if the condenser is not integrated in a specially
designed system, an example of which is the system shown in Fig. 13b, then the flow is either unable to
seek “∆p = ∆p|Natural” value due to exit pressure constraints or it gets inadvertently specified. Examples
32
of flow type (B) are fully condensing flows in micro-gravity or certain condensing flows in horizontal
test sections (e.g. flows of the type in Fig. 3 with xe > x** or x* 0 because M in < M in − cr ). According to
first principle based theory and supporting computational results (see [1]), these condensing flows,
simply, cannot occur (over a range of inlet flow rates and duct lengths) in micro-gravity by self-selection
of a steady exit pressure.
p
pf
p= p
p= p
p
f
Natural
Natural
M in
M in >M in-cr
M=Min
M
Fig. 15a: This figure shows ‘ p|Natural vs M ’ curve for the condenser in Fig. 13a which is operating
at a given value of temperature difference function T( ). The figure shows the operating vertical
straight line AB that fixes the mass flow rate M = M in due to fixed values of heat load Q and fixed
pressure pB for the boiler in Fig. 13a. The value p = p|Natural is achieved at intersection point D
only for unspecified exit pressure condition when operating under the procedure O-1. Any other
p = pf value can be fixed (see pf corresponding to point E above) with the help of operating
procedure O-2. This is because for the specified exit condition cases (category I flows in [1]) pf
value can lie anywhere in the shaded zone and is not required to lie on any particular curve.
Operation of Specially Designed Experimental Arrangement for Condensing Flows
Fig. 13a shows an special experimental arrangement designed and employed by us towards realizing
both specified (category I) as well as unspecified (category II) exit conditions for condensing flow test
section. The sytem and sub-system curves for the arrangement shown in Fig. 13a can be represented by
the schematic shown in Fig. 15a. The undefined ∆p values for the condenser test section (at a specified
value of vapor to condensing-surface temperature difference function ∆T( ), where is the coordinate
that specifies the downstream locations of the condensing surface) in Fig. 13a are represented by the
shaded zones in Fig. 15a. Within these shaded zones, there typically exists a zone (bounded by curves C1
and C2) within which the vapor flows have density variations within 1% of their inlet density values.
33
The mass flow rate M in = Q /hfg(pB) (where hfg is the heat of vaporization of the pure fluid employed in
the system) is held fixed along line AB of Fig. 15a by the fixed heat load Q and a fixed pressure pB for
the evaporator/boiler. In many practical applications, however, the line AB is a band which accounts for
changing heat loads Q as well as changes in steady boiler pressure pB (which is often not held fixed by
an accumulator).
The unspecified pressure pexit case – procedure O-1
As stated earlier, pressure pA psat(TA) for nearly stagnant liquid in the accumulator A (in Fig. 13a)
is allowed to vary in a range determined by the range of allowed temperatures TA. This degree of
freedom for pressure pA is typically sufficient –except for some cases (see an example in [1]) - for the
two-phase condensing flow to be able to vary its exit pressure “pexit” and seek its “natural” value
“pexit|Natural.” As shown in Fig. 15a, this attainment of p|Natural is achieved at the intersection of the line
AB and the well defined curve C that exists for type (A) flows in this category. The curve C represents a
function p|Natural = f ( M in ) for a given inlet pressure pin and vapor to condensing-surface temperature
difference ∆T( ). For each M in , the system design allows the flow in test section to have sufficient
freedom in selection of p values and this is indicated by converging arrows at the intersection point D
in Fig. 15a. Note that these curve C flows do not exist for many type (B) flows common to fully
condensing microgravity situations (with xe > x** in Fig. 3) as no steady flow can be realized.
The specified exit pressure pexit case – procedure O-2
Different pressure levels pA in the L-V separator of Fig. 13a are fixed by connecting it to the
accumulator which employs active heating/cooling. This active heating/cooling of the accumulator is
accomplished by a feedback control process using thermo-electric coolers and electric heaters controlled
by suitable electronics, software, and a computer. These different fixed values of pA correspond to
different fixed values of pexit and p
p|Natural – and this is indicated by pf in Fig. 15a. Therefore
resulting flows are in the specified exit pressure category (category I). This operational procedure allows
us the choice to fix any value of pf leading to essentially incompressible vapor flow (if pf is fixed
within the 1% compressibility band) or a flow with significant vapor compressibility (as shown in Fig.
15a). The compressible flows in this category require identification of another curve C4 in Fig. 15a –
because flows to the top and right of C4 are likely to exhibit compressibility induced instabilities (such
as oscillatory flows).
As indicated earlier, for a 0g or horizontal category II flow in a condenser of a given length L, there
exists M in-cr and hence a curve C3 in Fig. 15a needs to be demarcated. Only for flows to the top and right
of this curve one has M in > M in-cr . Only under these conditions, one can ensure xe < x* for category II
flows or suitably specify p for category I flows.
Definition and Importance of Exit Pressure Parameter
It is clear from the above discussion that for unspecified exit condition cases, type (A) condensing
flows can be realized with pactual = p|Natural under a suitable experimental arrangement whereas type
(B) flows will remain ill defined. On the other hand, for many condensing flows in the systems, p may
be intentionally (as for Fig. 13a) or inadvertently (as for the system in Fig. 13b) specified. To
34
assess/predict condensing flows in this specified p category it is important to know the relation
between pactual and a certain reference pressure p|Ref. The relationship between pactual and p|Ref. is
characterized by a non-dimensional exit pressure parameter:
pactual / p|Ref.
(3)
If the corresponding flow under unspecified exit condition is type (A) – that is p|Natural exists – we
set p|Ref. = p|Natural. If the corresponding flow under unspecified exit condition is type (B) – that is
p|Natural does not exist - we set p|Ref. = fknown ( Min ) to be a fixed number (e.g. 1 kPa) through a predefined function fknown of M in . It is recommended that the definition of this function fknown be chosen
such that the resulting curve also lies (as does curve C) within the feasible curves (between curves C3
and C4) and furthermore, nearly incompressible zone bounded by the curves C1 and C2 in Fig. 15a.
The above definition means that, for “natural” flows defined by type (A), the value of =1. Also,
with the above choice of p|Ref.,
1 signifies nearly incompressible flows for all specified exit
pressure flow categories as well.
pTotal
p
Pump
p Total *
p2
p2*
p1*
p1
p
Natural
p = p*
p*
M in
M
Fig. 15b: This figure shows the system and subsystem curves for condensing flow arrangement in
Fig. 13b. The values pTotal*, p1*, p2* are decided by intersection of the line AB and various
curves. p|Natural vs M curve for the condenser is the same curve shown in Fig. 15a that would have
been obtained if the flow was realized under arrangement shown in Fig. 13a. The pressure drop
across the condenser p* could be anywhere - its value being fixed by the system through the
equation p* = pTotal* - ( p1*+ p2*). The band around p|Natural vs M curve shows that if p* falls
within that band vapor compressibility effects will be small.
35
Determination of the Range of
in Other Arrangements for Condensing Flows
Fig 13b shows a representative arrangement for condensers employed in different applications where
the exit condition is unknown and inadvertently specified. Instead of a displacement pump, a centrifugal
pump is used to pump liquid flow back to the evaporator. The main difference between arrangements in
Figs. 13a and 13b is that, in Fig. 13b, the accumulator-cum-L/V separator is missing in the liquid line
between the condenser exit and the pump . This means as opposed to the arrangement in Fig. 13a, the
two-phase flow in the condenser is not decoupled from single phase heavier fluid flow. The system and
sub-system curves for the arrangement in Fig. 13b are shown by the schematic in Fig. 15b. Similar to the
arragement in Fig. 13a, the mass flow rate Min in the system is assumed to be fixed along line AB shown
in Fig. 15b (or in a narrow band around it) for fixed heat load Q and a fixed pressure pB for the boiler.
As shown in Fig. 13b, the pressure difference between points B and P (as the flow goes through points I
and E) is given by ∆pTotal System . This can be split in to pressure drops across sub-systems identified in Fig.
13b as:
∆pTotal
System
= p1 + p + p2
(4)
where, as shown in Fig. 13b, p1 = pB –pin, p = pin –pexit , and p2= pexit - pP. Also, in Fig. 15b, for
single phase fluid flows, p1 and p2 values are well defined functions of M because they are pressure
differences for single–phase flow regions. For the two-phase condenser, pressure difference p is not a
well defined function of M - and this fact is being represented by a shaded zone in Fig. 15b. Furthermore,
in Fig. 15b, ∆pTotal Pump is the value of the pressure difference “pB -pP” as the flow goes across the pump.
The
∆pTotal
Pump
curve in Fig. 15b is, however, obtained for the centrifugal pump by a separate
characterization experiment in which the pressure difference “pB-pP” is measured as the liquid flows
through the pump at different mass flow rates M . The steady mass flow rate M = Min achieved by this
system is given by the intersection of line AB and the pump characterization curve ∆pTotal Pump in Fig. 15b.
At this intersection point in Fig. 15b,
∆pTotal
Pump
= pTotal*. Furthermore, the system requires that at this
Min ,
the pressure difference between points P and B be the same whether it is obtained by the flow (along B,
I, E, and P) through the system or through the pump from P to B (as given by the pump curve). This
means p = p* value actually achieved must be such that it satisfies
(5)
∆pTotal Pump = ∆pTotal System
leading to p*= pTotal* - p1* - p2* where p1* and p2* are also as shown in Fig. 15b. This means for
the arrangement shown in Fig. 13b, the condition at the exit of the condenser is essentially a specified
exit pressure flow condition (category I). In these kinds of practical arrangements, the value of is at
the mercy of the system and is generally not known unless the above type of analysis is done and
p*/ p|Ref is computed.
If the condenser in Fig. 13b is properly sized to ensure that, over the range of values involved, the
condenser actually achieves complete condensation then the remaining parts of the system in Fig. 13b
can be made insensitive to the complexities of various flow regimes inside the condenser. For this
special design, the curves represeting p1, p2 vs M in Fig. 15b must have high slopes and satisfy: | p1|
~ | p2| >> | p|. However, for proper designs of all such systems, one still needs to have good estimate of
the range of values involved.
Addition of an L/V separator (without the accumulator shown in Fig. 13a) at the exit of the
36
condenser section of Fig. 13b has also been used by others (see [37]). While this may allow attainment
of natural unspecified exit condition steady flows, the start-up time is significantly larger (about 12
hours in [37] as opposed to about 20 minutes in [1]) than those associated with the system in Fig. 13a
that employs a displacement pump. This is because the centrifugal pump in Fig. 13b seeks the correct
mass flow rate Min while strongly affecting the pressure in L/V separator (without the accumulator).
Impact of
on Flow Regime Transition Boundaries for Condensing and Other Two-phase Flows
It should be noted that the results described above are mainly applicable to condensing flows as
these new ideas were developed by our group after extensive experimental/computational research on
condensing flows. Even then, relevant ideas may also be extended to other two phase flows namely,
boiling, adiabatic gas-liquid flows, etc.
For a typical operating point P, in Fig. 15a and Fig. 15b, the entire condenser in the system is
characterized by p, M , T( ) values. However, for flow within the condenser, for same set of these
three parameters, there exist different flow regimes viz. annular, plug/slug, bubbly etc. Within the
condenser, transition curves for these important flow regime boundaries (see [36]-[42]) – for given
function T( ) - are generally defined as curves in “ M in - Z( )” space where vapor quality Z( ) is
defined as a ratio of cross sectional vapor mass flow rate of M V to M in and takes a value of one at the
inlet and zero at the exit. In the light of the results presented above, these transtion boundaries are
significantly affected by the parameter and must be sought - for given function T( ) – as surfaces in
“ M in - – Z( )” space.
The above remarks about the impact of variable also clearly apply to flow boiling with sufficiently
large exit vapor qualities.
For adiabtic two-phase flows, though Z( ) is constant along in a two-phase test section, the flow
regime boundaries are expected to vary because is expected to vary (in a small neighborhood of = 1)
in “ M in - – Z( )” space. Though the variations in may be small from one experiment to another, the
variations are expected to have a larger impact, particularly, on the location of the boundaries for the
plug/slug and bubbly regimes. This may explain why ([42]) the available empirical correlations for p
as a function of M in and Z( ) are good in the interior of flow regime boundaries but not so good on the
more poorly defined flow regime boundaries.
7. CONCLUSIONS
•
•
•
This paper experimentally confirms the significance of exit conditions on the nature of quasi-steady
internal condensing flows and proposes a novel and necessary exit-condition based categorization of
these flows.
In particular, for gravity driven condensate flows, experimental procedures and results for achieving
steady and stable partial condensing flows under specified and unspecified exit conditions is
presented here. In section 6, corresponding procedures for fully condensing flows are described.
The experiments reinforce simulation results that, for partial condensation, multiple steady states,
with quite different local and average heat transfer rates, are often achieved under different exit
condition specifications (category I flows). Therefore, correlations for heat transfer coefficient
(though not developed here) are only meaningful if flow regimes are clearly defined and developed
in the framework of proposed exit-condition based categories.
37
•
•
•
•
•
The existing simulation tool’s ability to be quantitatively correct in identifying “natural” exit
conditions for gravity driven partial condensation cases under unspecified exit conditions (category
II) is very good as this is supported by the reported experiments. This agreement adds credibility to
the experimental results, simulation tool, and the proposed exit condition based categorizations.
The transients in partial condensation experiments establish that the steady flows are definitely more
robust under specified exit condition (category I) operation of condensers. This lends credibility to
the simulation result that steady operation of shear driven condensers (in zero gravity and horizontal
configurations) are much more difficult to achieve under unspecified exit conditions (category II). In
general, these results suggest that specified exit-condition flows (category I flows) – for both partial
and complete condensation cases - with properly selected exit pressure values are likely to be more
robust and more readily realized than flows under other arrangements (category II or category III
under arbitrarily fixed pressures and valve settings).
Some flow regime and system boundaries for annular category I and category II flows are observed
and reported here. Though these reported preliminary identifications of flow regime boundaries need
to be made more definitive, their identifications are clearly important for attaining or ascertaining
steady performances of condensers.
This paper generalizes the above results and identifies a parameter , whose proper accounting is
important for characterizing most forced internal duct phase-change flows.
The experiments clearly demonstrate the difference between flow regime boundaries and systeminstability boundaries and the importance of identifying system-instability boundaries that are
specific to individual systems. For example, curve A in Fig. 10 represents, system instability
boundaries that arise from restrictions imposed on the exit pressure by phenomenon occurring in
components downstream of the test section.
8. APPENDIX
Channel Flow Governing Equations
The liquid and vapor phases in the flow (e.g. see Fig. 1) are denoted by a subscript I: I = 1 for
liquid and I = 2 for vapor. The fluid properties (density ρ, viscosity µ, specific heat Cp, and thermal
conductivity k) with subscript I are assumed to take their representative constant values for each phase (I
= 1 or 2). Let TI be the temperature fields, pI be the pressure fields, Ts (p) be the saturation temperature
of the vapor as a function of local pressure p, ∆ be the film thickness, be the local interfacial mass
flux, Tw ( ) (< Ts (p)) be a known temperature variation of the cooled bottom plate, and v = ˆi + ˆj
I
I
I
be the velocity fields. The flow fields are defined at every point x (a 3-D Euclidean position vector) and
and be the components of gravity along x and y
time t. Furthermore, let h be the channel height,
axes, p0 be the inlet pressure, ∆T ≡ Ts (p0) - Tw (0) be a representative controlling temperature difference
between the vapor and the bottom plate, hfg be the heat of vaporization at temperature Ts (p), and U be
the average inlet vapor speed determined by the inlet mass flux. With t representing the actual time and
( , ) representing physical distances of a point with respect to the axes for the channel flow ( = 0 is
at the inlet, = 0 is at the condensing surface, and = h is an isothermal slightly superheated noncondensing surface), we introduce a new list of fundamental non-dimensional variables – viz.
(x,y, , ,u I ,v I , I , I ,m) - through the following definitions:
38
{ , , ∆, I , } ≡ {h ⋅ x, h ⋅ y, h ⋅ , U ⋅ uI , 1 ⋅ U ⋅ m}
{ I , TI , pI , t} ≡ {U ⋅ vI , (∆ ) ⋅ I , p0 + I U2 ⋅ I , (h U) ⋅τ}.
(A.1)
Interior Equations
The non-dimensional differential forms of mass, momentum (x and y components), and energy
equations for incompressible flow in the interior of either of the phases are the well-known equations:
∂u I ∂v I
+
=0
∂x
∂y
∂u I
∂u I
∂u I
∂π I
1 ∂ 2u I ∂ 2u I
−1
+ uI
+ vI
=−
+ Frx +
+
∂τ
∂x
∂y
∂x
Re I ∂x 2 ∂y 2
1 ∂ 2 vI ∂ 2 vI
∂v I
∂v I
∂v I
∂ I
−1
+ uI
+ vI
=−
+ Fry +
+
∂τ
∂x
∂y
∂y
Re I ∂x 2 ∂y 2
∂ I
∂
∂
1
∂2 I ∂2 I
+ u I I + vI I ≈
+
∂τ
∂x
∂y Re I PrI ∂x 2 ∂y 2
where Re I ≡ I Uh /
−1
I
, PrI ≡ I C pI / k I , Frx ≡
−1
x
h / U 2 and Fry ≡
y
(A.2)
,
h / U2.
Interface Conditions
The nearly exact interface conditions (see Delhaye [43]) for condensing flows are used. Utilizing a
superscript “i” for values of flow variables at the interface (x,t) = − ∆ ( , t) = 0, the unit normal at
any point on the interface, directed from the liquid towards the vapor, is denoted by n̂ and is equal to
∇ ∇ . Here, “ ∇ ” and “ ∇ S ” respectively denote 3-D (in Euclidean space) and surface 2-D (on
curved surface of the interface) gradient/divergence operators depending on whether they are acting on
scalar/vector fields. The unit tangent at any point on the interface, directed towards increasing , is
denoted by t̂ . The velocity vs of a point on the dynamic interface is such that it satisfies the well known
kinematic restriction: vs • n̂ = (−∂ / ∂t) / | ∇ | . Each phase is modeled as a viscous and incompressible
T = −p I 1 + S I
where
viscous
stress
tensor
Newtonian
fluid
with
stress
tensor
T
S I = I {(grad ⋅ v I ) + (grad ⋅ v I ) } / 2 and 1 is the identity tensor. The interface conditions and their nondimensional forms are given below.
i.e.
The tangential component of the vapor and liquid velocities at the interface must be continuous,
v1i • tˆ = v i2 • tˆ .
(A.3)
The non-dimensional Cartesian form of the above requirement for the channel flow of interest becomes:
39
u i2 = u 1i − δ x ( v i2 − v1i ),
(A.4)
where δ x ≡ ∂δ / ∂x.
Ignoring normal component of ∇ s σ and viscous stresses in the first equality, the second
equality given below models the normal component of momentum balance at a point on the interface.
That is:
i
i
p1 = p 2 +
2
i
(1/
2
≅ p2 +
(1/
2
− 1/ 1) + ∇ s • nˆ − ∇ sσ • nˆ + (S1i − S i2)nˆ • nˆ
2
− 1/ 1) − ( ∆ ) / [1+ ∆ 2 ]
(A.5)
2/3
.
The non-dimensional Cartesian form of the above requirement for the channel flow of interest becomes:
i
1
=
2
i
2
−
1
1
We [1 +
xx
2 3/ 2
x
]
+ m2
1
−1 ,
(A.6)
2
2
where We
1U h/σ, and surface tension σ is assumed to be nearly constant because of the nearly
constant interface temperature. Because of smallness of surface tension force components and viscous
stress components relative to the interfacial pressures, Eq. (A.6) with constant
σ = σ (Ts (pi2 )) = σ (Ts (p )) is adequate.
The tangential component of momentum balance at any point on the interface reduces to:
Si = S1inˆ • tˆ = Si2nˆ • tˆ + ∇sσ• tˆ
(A.7)
The tangential component of momentum balance at the interface, as given by Eq. (A.7), becomes:
∂u1
∂y
i
=
2
1
∂u 2 i
+ [ term ] ,
∂y
(A.8)
where the term [t] in Eq. (A.8) is defined as
i
[term] = {
2
1
i
∂ v 2 ∂ v1
2
−
}+
[1 −
∂x
∂x
i
x
i
∂u
∂v
2
{ 1 − 1 }−
2
[1 −
∂y
x ] ∂x
i
x
i
i
∂u
∂v
∂
1
{ 2 − 2 } + Ma 2 ⋅
2
∂y
∂x
x ] 1 ∂x
1+
2
2
. (A.9)
x
where the Marangoni number Ma ρ2Uc1d1/µ1 represents the surface tension contribution to tangential
stress under the notation c1 ≡ dTs/dp and d1 ≡ -dσ/dT . For the cases considered here, a representative
(for R113) set of constant values of c1 ≈ 0.0003 K/Pa, d1 ≈ 0.1046 N/(m-K), and Ma = 0.196 in Eq. (A.9)
gives the same results as simulations without the Marangoni term. Hence the last Marangoni term on the
right side of Eq. (A.9) can be dropped.
The mass-fluxes V K and L K as determined by kinematic restrictions imposed by interfacial
values of vapor and liquid velocities are:
40
VK
The interfacial mass flux
≡−
2
Energy
( vi2 − vis ) • nˆ and
LK
(A.10)
as obtained from energy balance at a point on the interface is given by:
i
E nergy
≡ − 1 ( v1i − vis ) • nˆ .
i
= 1/ h fg[{k1∇T1 • nˆ − k 2∇T 2 • nˆ } +
2
2
dσ
1
+
{ v1i − v is − v i2 − v is }
dt s 2
+ {S1inˆ • ( v1i − v si ) − S i2nˆ • ( v i2 − v is)}]
i
(A.11)
i
∂
∂
≅ 1/ h fg[k1 T1 − k 2 T 2 ] .
∂n
∂n
The non-dimensional form of kinematic mass fluxes
i
1
VK
and
LK
in Eq. (A.10) become:
i
1
m LK ≡ u (∂ /∂x) − (v − ∂ /∂τ) / 1 + (∂ /∂x) 2 , and
(A.12)
m VK ≡ ( 2 / 1 ) u i2 (∂ /∂x) − (vi2 − ∂ /∂τ) / 1 + (∂ /∂x) 2
The non-dimensional form of
Energy
in Eq. (A.11) becomes:
m Energy ≡ Ja /(Re1 Pr1 ){∂ 1 / ∂n i −(k 2 / k 1 )∂
where Ja ≡ Cp1
i
2
/ ∂n } ,
(A.13)
/ h 0 fg , and h 0 fg ≡ h fg (Ts (p )) .
The dimensional and non-dimensional forms of interfacial mass balance requirements respectively
become:
LK
=
VK
=
Energy
≡
.
(A.14)
m LK = m VK = m Energy ≡ m.
(A.15)
Under the assumption of negligible non-equilibrium thermodynamic effects and, therefore,
negligible interfacial resistance (a valid assumption for all away from ~ 0 because interfacial massflux
rapidly drops to modest values), the interfacial temperature values satisfy:
i
i
i
(A.16)
T1 ≅ T 2 = T s(p 2) .
The non-dimensional thermodynamic restriction on interfacial temperatures, as given by Eq. (A.16),
becomes:
i
1
≅
i
2
=
S
(p )
i
2
≡
s
( ).
i
2
(A.17)
Within the vapor phase, for the refrigerants considered here, changes in absolute pressure relative to the
inlet pressure are typically so small that s ( i2 ) ≅ s (0) constant is a good approximation.
Boundary Conditions
The problem posed by Eqs. (A.2), (A.4), (A.6), (A.8), (A.15), and (A.17) are computationally
solved subject to boundary conditions that are
41
At the inlet (x = 0, 0
y
1) at any time τ:
u 2 (0,y,τ) = 1
2
(0,y,τ) = 0
At the bottom wall (y = 0, 0
x
v 2 (0,y,τ) = 0
2
(0,y,τ) =
s
(A.18)
(0) .
xe) at any time τ:
u1 (x, 0, τ) = 0 , v1 (x, 0, τ) = 0 ,
1
(x, 0, τ) =
,
w
(A.19)
where w ≡ w ( )
is a constant unless it is otherwise specified. In case of flow in Fig. 2, this
situation arises whenever, for a given heat load, the coolant flow rate is high enough to make the coolant
temperature rise negligible as it flows past the test-section.
At the top wall (y = 1, 0
x
xe) at any time τ:
u 2 (x,1, τ) = 0 , v 2 (x,1, τ) = 0 ,
2
(x,1, τ) =
s
(0) .
(A.20)
Furthermore, because of the nature of boundary conditions in Eqs. (A.18) – (A.20), 2 (x, y, τ) ≅ s (0) is
assumed/prescribed to limit the discussions in this paper to the flow of saturated vapor. This is done
because, for the pure vapor flows considered here, it is easy to verify the well-known fact that the effects
of superheat (commonly in the 5 – 10 oC range) are negligible.
Exit Conditions
Any condenser section of the type shown in Fig. 1 is typically a part of a closed flow loop. A
flow loop which maintains a steady condensing surface temperature Tw( ), a constant flow rate, and a
constant pressure p (i.e. π2 = 0) at the inlet, may also be designed to provide: (a) specified steady exit
pressure (category-I flows) which is equivalent to, for partial condensation flows, a specified value of
steady exit quality Ze (the ratio of vapor mass flow rate at exit, x = xe, to vapor mass flow rate at inlet),
and (b) unspecified exit pressure (category-II flows) which is equivalent to, for partial condensation
flows, unspecified values of exit quality Ze (t).
Category-I flows: For this case of specified exit conditions, it is assumed that the exit pressure or the exit
quality Ze specifications are still such that vapor compressibility effects can be ignored. The boundary
value problem (BVP) constituting of the steady governing equations and boundary conditions –
including specified steady exit conditions (steady Ze or pexit) – constitutes a steady “well-posed”
category I flow problem. Steady category-I flow solutions of the governing equations, obtained by
dropping all time dependencies in all the relevant equations, and solving the resulting steady equations
(which are elliptic) for steady prescribed values of Ze are presented here. The prescription of Ze within 0
< Ze< 1 is arbitrary except that it should be such that a steady computational solution in the assumed
stratified/annular regime be feasible.
The initial boundary value problem (IBVP) for the unsteady governing equations with arbitrary
initial guesses under specified steady exit conditions are also “well-posed.” These solutions are
supposed to be such that they get attracted, as
, to the corresponding steady solution. If not, the
unsteady solutions’ behavior at large τ indicates the stability of the corresponding steady solution. Such
stability analyses are reported, for h
external flow problems of Nusselt [21] and Koh [29], in Phan
42
and Narain ([28]) and Kulkarni et al. ([2]-[3]). Such stability analyses for steady internal category I
flows, however, are not currently available and are the subjects of forthcoming papers.
An inspection of all the non-dimensional governing equations, interface conditions, and
boundary conditions reveal the fact that, for category I flows, the flow is affected by the following set of
non-dimensional parameters:
{Rein , Ja, Fr −x1 ,
2
,
2
1
, Pr1 , x e , Z e , We, Fr −y1},
(A.21)
1
where Rein ρ2Uh/µ2 Re2. Here Rein, Frx-1, and Ja are control parameters associated with inlet speed
U, channel inclination α that determines the components of gravitational acceleration, and temperature
difference T. Clearly the value of Ze or exit pressure is important for this problem. The density ratio
ρ2/ρ1, viscosity ratio µ2/µ1, and Prandtl number Pr1 are passive fluid parameters. Also, for unsteady or
quasi-steady wavy-interface situations, the above equations imply additional dependences on a surface
-1
2
2
tension parameter, Weber number We
1U h/σ, and a transverse gravity parameter Fry
yh/U . For
superheated vapors, there is a very weak dependence, through Eq. (A.13), on the thermal conductivity
ratio k2/k1.
Category-II flows: For these flows, no exit conditions are prescribed. The steady governing equations
and the remaining boundary conditions constitute an “ill-posed” boundary value problem (BVP) and no
unique solution is possible. However, when unsteady category-II flows are considered for suitable initial
condition choices, the initial boundary value problems (IBVP) are ‘well-posed.” However, in general, as
, no well-defined steady solution may exist. If they do, as is the case for gravity driven flows, one
obtains a unique well defined “attracting” solution that is independent of initial guesses and one also
obtains a corresponding unique exit condition Ze (τ) . Ze|Na as τ . . The stability of these attracting
flows to initial disturbances has been extensively considered in [4] – [6]. For shear driven internal flows
in this category, the existence or non-existence of “attractors” follow the pattern discussed for Fig. 3.
Note τ = 0 cannot be chosen to be the time when saturated vapor first comes in contact and
condenses on a dry sub-cooled (Tw ( ) < Ts (p0)) bottom plate. This is because the above described
continuum equations will not apply at early times (τ ~ 0) because inter-molecular forces will be
important for very thin (approximately over 10 - 100 nm of film thickness) condensate film δ(x, τ) .
Because of the above modeling limitations, the strategy here is to start at τ = 0, with any sufficiently
thick guess and look at the long time τ
behavior of the solution. Most of the time, for τ = 0 initial
guess, we choose steady solution of a category-I flow. That is, if φ(x,y,τ) is any variable (such as uI, vI,
πI, θI, etc.), the initial values of φ and film thickness δ(x,τ) are such that:
φ( x , y,0) = φsteady ( x , y)
and
δ( x ,0) = δsteady ( x ) ,
(A.22)
where φsteady and δ steady are solutions of the category-I problem for a certain exit condition (exit quality
Ze or exit pressure).
43
9. ACKNOWLEDGMENT
This work was supported by the NSF grant CTS-0086988 and NASA grant NNC04GB52G.
10. REFERENCES
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
Narain, A., J. H. Kurita, M. Kivisalu, A. Siemionko, S. Kulkarni, T. Ng, N. Kim, and L. Phan, Oct.
2007, “Internal Condensing Flows inside a Vertical Pipe – Experimental/Computational
Investigations of the Effects of Specified and Unspecified (Free) Conditions at Exit,” Journal of
Heat Transfer, Vol. 129, pp. 1352-1372.
S. D. Kulkarni, A. Narain, S. Mitra, and L. Phan, 2008, “Forced Flow of Vapor Condensing over a
Horizontal Plate (Problem of Cess and Koh*) Part I – Steady Solutions of the Full 2D Governing
Equations,” Being submitted by Oct. 10, ‘07 (after good IMECE07 reviews) for publication in
ASME Journal of Heat Transfer.
S. D. Kulkarni, A. Narain, and S. Mitra, 2008, “Forced Flow of Vapor Condensing over a Horizontal
Plate (Problem of Cess and Koh*) Part II – Unsteady Solutions of the Full 2D Governing Equations
Yielding Steady “Attractors” and their Unsteady Responses to Initial Disturbances and Noise,”
Being submitted by Oct. 10, ‘07 (after good IMECE07 reviews) for publication in ASME Journal of
Heat Transfer.
Narain, A., Q. Liang, G. Yu, and X. Wang, Jan. 2004, “Direct Computational Simulations for Internal
Condensing Flows and Results on Attainability/Stability of Steady Solutions, their Intrinsic
Waviness, and Their Noise-sensitivity,” Journal of Applied Mechanics, Vol. 71, pp. 69-88.
Liang, Q., X. Wang, and A. Narain, October 2004, “Effect of Gravity, Shear and Surface Tension in
Internal Condensing Flows - Results from Direct Computational Simulations.” ASME Journal of
Heat Transfer, 126 (5), pp. 676-686.
Phan, L., X. Wang, and A. Narain, 2006, “Exit condition, Gravity and Surface-Tension Effects on
Stability and Noise Sensitivity Issues for Steady Condensing Flows inside Tubes and Channels,”
International Journal of Heat and Mass Transfer, Vol. 49, Issues 13-14, pp.2058-2076.
Goodykoontz, J. H. and R. G. Dorsch, 1966, “Local heat transfer coefficients for condensation of
steam in vertical down flow within a 5/8-inch-diameter tube,” NASA TN D-3326.
Goodykoontz, J. H. and R. G. Dorsch, 1967, “Local heat transfer coefficients and static pressures for
condensation of high-velocity steam within a tube,” NASA TN D-3953.
Carpenter, F. G., 1948, “Heat transfer and pressure drop for condensing pure vapors inside vertical
tubes at high vapor velocities,” Ph. D. thesis, University of Delaware.
Yu, G., 1999, “Development of a CFD Code for Computational Simulations and Flow Physics of
Annular/Stratified Film Condensation Flows,” Ph.D. Thesis, ME-EM Department, Michigan
Technological University.
Rabas, T. J., and B. Arman, 2000, “Effects of the Exit Condition on the Perofromance of In-Tube
Condensers,” Heat Tranfer Engineering, 21(1), pp. 4-14.
Wedekind, G. L. and B. L. Bhatt, 1977, “An Experimental and Theorotical Investigation in to
Thermally Governed Transient Flow Surges in Two-Phase Condensing Flow,” ASME Journal of
Heat Transfer 99 (4), pp. 561 – 567.
Bhatt, B. L. and G. L. Wedekind, 1980, “Transient and Frequency Characteristics of Two-Phase
Condensing Flows: With and Without Compressibilty,” ASME Journal of Heat Transfer 102 (3),
495-500.
Bhatt, B. L. and G. L.Wedekind, 1980b, “ A Self Sustained Oscillatory Flow Phenomenon in Two-
44
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.
Phase Condensing Flow Systems,” ASME Journal of Heat Transfer 102 (4), 695-700.
Wedekind, G. L. and B. L. Bhatt, 1989, “ Modeling the Thermally Governed transient Flow Surges
in Multitube Condensing Flow Systems with Thermal Flow Distribution Asymmetry,” ASME
Journal of Heat Transfer 111 (3), pp. 786 –791.
Bhatt, B. L., G. L. Wedekind, and K. Jung, 1989, “ Effects of Two-phase pressure Drop on the SelfSustained Oscillatory Instabilty in Condensing Flow,” ASME Journal of Heat Transfer 111, 538545.
Boyer, D. B., G. E. Robinson, and T. G. Hughes, 1995, “Experimental Investigation of Flow
Regimes and Oscillatory Phenomena of Condensing Steam in a Single Vertical Annular Passage,”
International Journal of Multiphase Flow, Vol. 21, No. 1, pp. 61 – 74.
Wedekind, G. L., C. J. Kobus, and B. L. Bhatt, 1997, “Modeling the Characteristics of Thermally
Governed Transient Flow Surges in Multitube Two-Phase Condensing Flow Systems with
Compressibility and Thermal and Flow Distribution Asymmetry,” ASME Journal of Heat Transfer,
Vol. 119, No. 3, pp. 534 – 543.
Kobus, C. J., G. L. Wedekind, and B. L. Bhatt, 2000, “Predicting the Influence of Compressibility
and Thermal and Flow Distribution Asymmetry on the Frequency-Response Characteristics of
Multitube Two-Phase Condensing Flow Systems,” ASME Journal of Heat Transfer, Vol. 122, No.
1, pp. 196 – 200.
Patankar, S. V., 1980, Numerical Heat Transfer and Fluid Flow, Hemisphere, Washington D. C.
Journal of Multiphase Flow, Vol. 16, pp. 139 – 151.
Nusselt, W., 1916, “Die Oberflächenkondesation des Wasserdampfes,” Z. Ver. Dt. Ing. 60 (27), pp.
541-546.
Narain, A. and Y. Kizilyalli, 1991, “Pressure driven flow of pure vapor undergoing laminar film
condensation between parallel plates,” International Journal of Non-Linear Mechanics, Vol. 26, No.
5, pp. 501-520.
Narain, A., G. Yu, and Q. Liu, 1997, “Interfacial Shear Models and Their Required Asymptotic
Form for Annular Film Condensation Flows in Inclined Channels and Vertical Pipes,” International
Journal of Heat and Mass Transfer, Vol. 40, N0. 15, pp. 3559 - 3575.
Garimella, S., J. D. Killion, and J. W. Coleman, 2002 “An Experimentally Validated Model for
Two-Phase Pressure Drop in the Intermittent Flow Regime for Circular Microchannels,” Journal. of
Fluids Engineering, Vol. 124, pp. 205-214.
Garimella, S., J. D. Killion, and J. W. Coleman, 2003 “An Experimentally Validated Model for
Two-Phase Pressure Drop in the Intermittent Flow Regime for Non-Circular Microchannels,”
Journal of Fluids Engineering., Vol. 125, pp. 887-894
Liao, N. S., C. C. Wang, and C. L. Tien, 1988, "Analysis of Transient Flow Surge Phenomena in a
Single-Tube Condenser,” International Communications in Heat and Mass Transfer, Vol. 15, pp.
257 – 268.
Liao, N. S. and C. C. Wang, 1990, “Transient response Characteristics of Two-Phase Condensing
Flows,” International Journal of Multiphase Flow, Vol. 16, 139-151.
Phan, L. and A. Narain, 2007, “Non-linear Stability of the Classical Nusselt problem of Film
Condensation and Wave Effects,” ASME Journal of Applied Mechanics, Vol.74, No.2, pp. 279-290.
Koh, J. C. Y., 1962, “Film Condensation in a Forced-Convection Boundary-Layer Flow,”
International Journal of Heat and Mass Transfer, Vol. 5, pp. 941-954.
Palen, J. W., R. S. Kistler, and Y. Z. Frank, 1993, “What We Still Don’t Know About Condensation
in Tubes,” In Condensation and Condenser Design (Edited by J. Taborek, J. Rose and I. Tanasawa),
Pub.: United Engineering Trustees, Inc. for Engineering Foundation and ASME, New York, pp. 19–
45
31.
32.
33.
34.
35.
36.
37.
38.
39.
40.
41.
42.
43.
53.
Ng, T. W., 2006, “Development and Calibration of a Fluorescence and Fiber-Optics Based RealTime Thickness Sensor for Dynamic Liquid Films,” Ph.D. Thesis, MEEM, Michigan Technological
University.
Siemonko, A., 2006, “Design, Fabrication, and Operation of a System to Control FC-72
Condensation Inside a Vertical Tube,” Ph.D. Thesis, Chem Eng, Michigan Technological
University.
Kurita, J. H., 2006, “Experimental Investigation of Fully Condensing Downward Vapor Flows in a
Vertical Tube - Unspecified (Free) Exit Condition Cases,” M. S. Thesis, MEEM, Michigan
Technological University.
Narain, A., J. H. Kurita, M. Kivisalu, A. Siemionko, S. Kulkarni. T. Ng., N. Kim, and L. Phan, Jan
2007, “Internal Condensing Flows inside a Vertical Pipe – Experimental/Computational
Investigations of the Effects of Specified and Unspecified (Free) Exit Conditions at Exit,” A
Technical Report which is an extended version of ref. [1] and is available at:
http://www.me.mtu.edu/~narain/narainpublications1.htm
Parratt, L. G., 1961, Probability and Experimental Errors in Science – An Elementary Survey,
Wiley, New York.
Carey, V. P., 1992, Liquid-Vapor Phase-Change Phenomena, Series in Chemical and Mechanical
Engineering, Hemisphere Publishing Corporation, New York.
Lu, Q. and N. V. Suryanarayana, 1995, “Condensation of a Vapor Flowing Inside a Horizontal
Rectangular Duct,” Journal of Heat Transfer, 117, pp. 418-424.
Hewitt, G. F., Shires, G. L., and Bott, T. R., 1994, Process Heat Transfer, CRC Press, Inc., Begall
House, Boca Raton, Florida.
Lockhart, R. W. and Martinelli, R.C., 1949, “Proposed Correlation of Data for Isothermal TwoPhase two-component flow in a pipe,” Chem. Eng. Prog., 45, No. 1, pp. 39–48.
Taitel, Y. and Dukler, A. E., 1976, “A Model for predicting Flow regime transition in horizontal and
near horizontal gas-liquid flow,” AIChE Journal, 22, pp. 47-55.
Chen, J. C., 1966, “Correlation for boiling heat transfer to saturated fluids in convective flow,” Ind.
Chem. Eng. Proc. Design and Dev., 5, pp. 322-339.
McDermott, B. and Timothy Shedd, “Detailed Pressure Drop and Flow Regime Characteristics of
Adiabatic Two-Phase Flow of Pure Refrigerant R-123,” Paper no. IMECE2007-43829, Proceedings
(CD ROM) of ASME, IMECE 2007, Seattle, Nov. 11-15, 2007.
Delhaye, J. M., 1974, “Jump Conditions and Entropy Sources in Two-phase Systems; Local Instant
Formulation,” Int. J. of Multiphase Flow, 1, pp. 395-409.
46