Nuclear Engineering and Design: An Analysis of Interacting Instability Modes, in A Phase Change System

Nuclear Engineering and Design 240 (2010) 31783201
Contents lists available at ScienceDirect

Nuclear Engineering and Design
j our nal homepage: www. el sevi er . com/ l ocat e/ nucengdes
An analysis of interacting instability modes, in a phase change system
William R. Schlichting, Richard T. Lahey Jr.
, Michael Z. Podowski
Center for Multiphase Research, Rensselaer Polytechnic Institute, Troy, NY 12180-3590, USA
a r t i c l e i n f o
Article history:
Received 2 March 2010
Received in revised form 22 April 2010
Accepted 25 April 2010
a b s t r a c t
This paper presents an analysis of the interaction of pressure-drop oscillations (PDO) and density-wave
oscillations (DWO) for a typical NASA type phase change system. A transient lumped parameter model is
developed for use in the analysis of the dynamics of this type of system. A compressible volume (e.g., an
accumulator vessel) dynamics model was also developed and PDO/DWO interactions are investigated.
2010 Elsevier B.V. All rights reserved.
1. Introduction
The purpose of this paper was to analyze interacting instability modes in phase change systems, such as those used in nuclear reactors.
This research has been motivated by the requirements of space exploration. In particular, an ambitious long range research program has
been formulated by NASAfor the human exploration and development of space (HEDS). This programincludes the possible development of
a lunar base and subsequent manned-missions to Mars. It has been found that one of the key enabling technologies which will be required
to support the power, propulsion and life support aspects of these missions is the use of a phase change system in space (NRC, 2000). In
particular, it appears that an on-board nuclear ssion reactor, and an associated Rankine cycle energy conversion system, may be required
to supply reliable power and propulsion for NASAs missions (NRC, 2000). This, in turn, implies that one must be able to reliably predict
the performance of two-phase ows and phase change systems (e.g., boilers and condensers) in microgravity (i.e., 10
6
g) environments.
There is a need to develop a better understanding of the effect of gravity on many important multiphase phenomena, such as ow
regimes, critical heat ux (CHF), pressure drop and phase change system stability (NRC, 2000).
This paper is focused on the development of the analytical capability required to analyze the effect of gravity on phase change system
stability and the possible interaction of various system instability modes in typical phase change systems. In particular, we focus here on
the analysis of the interactionof density-wave oscillations (DWO) andpressure-droposcillations (PDO) ina typical experiment whichcould
be performed both on earth and aboard the International Space Station (ISS). These types of interactions are of concern in NASAs proposed
nuclear ssion reactors, since any associated Rankine cycle energy conversion system will necessary be operating at low pressures to
minimize weight.
Fig. 1 shows a typical test loop which was designed for use by NASA. The size of this test loop is appreciably smaller than a typical
boiling system so that it can be tested aboard the ISS. Nevertheless, when properly scaled, the system instabilities are representative of
much larger phase change systems. The geometry and operating parameters of the test section shown in Fig. 1 are given in Table 1.
Various thermalhydraulic system instabilities have been observed in phase change systems. Typical instabilities are density-wave
oscillations (DWO), ow excursions (i.e., Ledinegg instabilities), and pressure-drop oscillations (PDO). Both density-wave oscillations and
pressure-drop oscillations are classied as dynamic instabilities whereas an excursive instability is normally classied as a static instability
(Lahey and Moody, 1993). Kakac and Bon (2008) have done a comprehensive reviewof dynamic systeminstabilities, including when there
may be interaction between DWO and PDO.
A density-wave instability can produce oscillations in the ow, which may either increase or decay in amplitude. These oscillations
occur when a uctuation in the subcooled inlet owcreates an enthalpy perturbation in the single-phase region of the heated channel, such
that the boiling boundary begins to oscillate due to these enthalpy perturbations. As a consequence, these enthalpy perturbations induce
quality and void fraction perturbations in the two-phase region of the test section. These perturbations cause, in turn, a perturbation in the
two-phase pressure drop. If the total pressure dropacross the heatedchannel is imposed(e.g., due toa parallel channel boundary condition),
a perturbation in the two-phase pressure drop induces a feedback perturbation in the pressure drop of the subcooled (single-phase) region,
which can either reinforce or attenuate the initial ow perturbation, and if in phase, may lead to a self-sustained DWO.
Corresponding author.
E-mail address: Laheyr@rpi.edu (R.T. Lahey Jr.).
0029-5493/$ see front matter 2010 Elsevier B.V. All rights reserved.
doi:10.1016/j.nucengdes.2010.05.057
W.R. Schlichting et al. / Nuclear Engineering and Design 240 (2010) 31783201 3179
Fig. 1. Typical NASA test loop.
Non-periodic ow excursions arise from the interaction between the systems impressed head ow characteristics and the hydraulic
characteristics of the heated channel. An excursive instability may occur when trying to operate where the slope of the steady-state
pressure drop versus ow curve of the heated channel is negative (e.g., region B to D in Fig. 2). We can only operate on this branch of the
systems pressure-drop curve when the ow is forced by a suitable pump (e.g., a positive displacement pump). In general, when the slope
of the pressure drop versus ow curve is negative, we may have an unstable xed point, point A, which is a repeller (Strogatz, 1994). For
an impressed constant pressure drop, uctuations in the channels ow will cause the ow to be repelled from this unstable xed point
to one of the two stable xed points, or attractors (e.g., A
or A
when having the same pressure drop, as for a parallel channel boundary
condition), depending on the sign of the perturbation. If the owreaches the single-phase attractor (e.g., the stable xed point, A
, in which
the ow is single-phase liquid throughout the heated section) then the ow will be stable. In contrast, if the ow is sent to the two-phase
attractor, A
, then the ow may operate at this xed point or it may oscillate about it in the form of a DWO (Kakac and Bon, 2008; Yin et
al., 2006).
A pressure-drop instability can produce oscillations which are somewhat similar to those found in density-wave oscillations, but at a
different frequency, with the frequency based on the dynamics of the compressibility of the system. In particular, a compressible volume
will result fromthe use of an accumulator, such as that shown in Fig. 1. During a pressure-drop oscillation (PDO), the owmay be diverted
from entering the test section and will enter the accumulator. The increase of uid in the accumulator volume compresses the gas in the
accumulator, which after an over-compression, will later expand (actually over-expand), pushing uid out of the accumulator and back
into the inlet of the test section. These pressure-drop oscillations only occur for the case when, if the accumulator was coupled to the inlet
of the test section, an excursive instability can occur.
Stenning et al. (1967) veried that a negative slope in the heaters pressure-drop characteristics was required for the pressure-drop
oscillations (PDOs) to occur. These oscillations were associated with the presence of a compressible volume at the inlet of the heater. Kakac
and Bon (2008) described a PDO as a ow oscillation that roughly traces the limit cycle EBCDE in Fig. 2. Stenning and Verizoglu (1965) and
Maulbetsch and Grifth (1965) observed density-wave oscillations (DWOs) when operating on the EB branch of the systems operating
Table 1
Proposed ISS test section.
Fluid FC-72
System pressure 300kPa
Heated channel type Annular
Hydraulic diameter 0.00127m
Heated perimeter 0.03591m
Channel area 2.40710
5
m
2
Heated length 0.3048m
3180 W.R. Schlichting et al. / Nuclear Engineering and Design 240 (2010) 31783201
Fig. 2. Pressure drop vs. mass ux curve.
curve. They also observed that PDOs occurred only on part of the negative slope region, instead of the entire region between B and D(Fig. 2).
In related analyses, Ozawa et al. (1979) and Dogan et al. (1983) made the restrictive assumption that the inow into the system, G
in
, prior
to the accumulator, was constant and that the exit pressure of the heated channel was also constant. These are not valid assumptions for
the test loop shown in Fig. 1, and, as will be discussed subsequently, a different formulation is required for the analysis of this test loop.
2. Discussion
The simplest model which can describe the dynamics of a boiling channel are the one-dimensional continuity, momentum, and energy
conservation equations based on a homogenous equilibrium model (HEM), in which phasic slip and subcooled boiling are neglected. The
assumption normally made is that the system pressure is constant, which implies the pressure dependent uid properties are given at
the system pressure. As the test section has a subcooled liquid and a two-phase region, each region has different conservation equations
associated with them.
The two-phase HEM conservation equations are, for a constant ow area and a constant, uniform, axial heat ux ( q
):
t
+

z
C = 0 (1)
t
C +

z
C
2
=

z
p g
j
2D
H
C
2

N
2
n=1
K
n
2
C
2
(z z
n
) (2)

t
h +C

z
h =
q
P
H
/
XS
(3)
and the equations of state are:
h = h
j
+xh
jg
(4a)
=
1
[v
j
+xv
jg
]
(4b)
The mixture energy equation, Eq. (3), can be recast into the formof a quality propagation equation by inserting the HEMstate equations,
Eqs. (4a) and (4b), into Eq. (3) (Lahey and Moody, 1993) to yield:
t
x +j

z
x x =
v
j
v
jg
(5)
where the characteristic frequency of phase change, , and the supercial velocity of the two-phase mixture, j, are respectively:
=
q
P
H
/
XS
h
jg
v
jg
(6)
j = j
in
+(z z) (7)
In Eq. (7), j
in
is the inlet velocity to the heater and z is the location of the boiling boundary (i.e., where bulk boiling begins).
The single-phase conservation equations are a special case of the two-phase conservation equations. Due to the assumption that the
uid properties are constant (i.e. evaluated at a constant system pressure), the single-phase uid may be treated as an incompressible
uid. For single-phase ow, Eqs. (1)(3) simplify to:
z
C = 0 (8)
t
C =

z
p g
j

j
2D
H
C
2
N
1
n=1
K
n
2
C
2
j
(z z
n
) (9)
t
h
l
+C

z
h
l
=
q
P
H
/
XS
(10)
Eq. (8) implies that the mass ux through the single-phase liquid region of the heated channel is equal to the inlet mass ux. Eq. (10)
can be used to determine the location of the boiling boundary, z, where the boiling boundary is dened to be the axial position in the
heater in which the liquid enthalpy is equal to the saturation enthalpy of the liquid:
h
l
(z = z) = h
j
(11)
When the heated channel is subjected to a perturbation in the ow, for an imposed pressure drop, the oscillations in the owcan either
increase in amplitude or decay back to the initial steady-state value, based on the operating conditions being perturbed. A systemis stable
if the oscillations decay back to the steady-state and is unstable if the oscillations grow. A system in which small amplitude oscillations
neither grow nor decay is said to be neutrally stable. Whether the ow oscillations will increase, decrease, or remain at a very small
amplitude, can be determined without running transients in the time-domain by analyzing the stability of the linearized conservation
equations about a steady-state operating condition.
The operating conditions of the test section can be described using non-dimensional numbers. By introducing these non-dimensional
numbers, the conservation equations can be simplied by removing the majority of the uid-dependent properties and channel geometry.
The inlet enthalpy is quantied by a subcooling number, N
sub
, Eq. (12), and the phase change number, N
pch
, Eq. (13), which describes
the ratio of the power to the ow. The friction number, /, Eq. (14) denotes the relationship between the friction factor (f), the hydraulic
diameter (D
H
), and the length of the heated channel (L
H
). For the geometry given in Table 1, the friction number has a value of 2.436. Finally,
the Froude number, Fr, Eq. (15), quanties the effect of gravity.
N
sub
=
h
j
h
in
h
jg
v
jg
v
j
(12)
N
pch
=
q
0
P
H
L
H
C
in0
/
XS
h
jg
v
jg
v
j
(13)
/=
j
2D
H
L
H
(14)
Fr =
j
2
in0
gL
H
=
C
2
in0
2
j
gL
H
(15)
It is convenient to remove the spatial dependence in the conservation equations by integrating them along the single-phase and two-
phase regions of the heated channel and by introducing appropriately spatially-averaged terms into these equations. Integrating Eq. (8)
along the subcooled length implies, as noted previously, that the mass ux throughout the single-phase region is equal to that of the inlet
mass ux. Integrating Eq. (10) along the single-phase region results in a boiling boundary (z) dynamics equation:
d
dt
z = 2
C
in0
j
_
C
in
C
in0
z
z
0
_
(16)
in which the steady-state boiling boundary, z
0
, can be expressed as:
z
0
=
N
sub
N
pch
L
H
(17)
The average two-phase density dynamics equation is determined from the two-phase continuity equation, Eq. (1):
d
dt
2
= 2
C
in0
z
0
_
1
j
_
z
L
H
z

2
(18)
where the average density two-phase is dened as:
2
=
1
L
H
z
_
L
H
z
dz (19)
The nal state variable equation in the nodal model is the lumped parameter momentum equation for the heated channel, which is
comprised of the combined single-phase and two-phase momentum equations:
_
z +(L
H
z)
j
_
d
dt
C
in
= ^p 2C
in0
z
z
0
_
C
in
j
+

2
(L
H
z)
__
1
j
_
_
K
in
2
1
_
C
2
in
_
K
cut
2
+1
_
_
C
in
_
2
j
1
_
+(L
H
z)
2
__
C
in
j
+(L
H
z)
_
g
j
_
z +(L
H
z)
j
_
2
C
in0
j
_
C
in
_
1
j
_
(L
H
z)
2
_
_
C
in
C
in0
z
z
0
_
+(L
H
z)
2
_
C
in
j
+

2
(L
H
z)
_
/
_
C
2
in
j
z
L
H
+
2
_
C
in
j
+

2
(L
H
z)
_
2
_
1
z
L
H
_
_
(20)
These nodal HEM conservation equations are derived in Appendix A. It should be noted that similar, but more detailed, nodal models
can be derived using a drift-ux model, and taking into account subcooled boiling (Schlichting, 2009).
3. Linear analysis
Eqs. (16), (18), and (20) can be linearized to investigate the neutral stability boundary of the nodal model, for a uniformly heated parallel
channel. The resulting linear system can be written as:
d
dt
j = /j (21)
where
j
-
j(t) j
0
= [ C
in
z
2
]
1
(22)
and
/ =
_
u
11
u
12
u
13
u
21
u
22
0
0 u
32
u
33
_
(23)
The terms in the Jacobian matrix, /, are:
u
11
=
[(K
cut
]2)(1 (
20
/
j
)) +1](L
H
z
0
) (K
in
+K
cut
+2/)(C
in0
/
j
)
z
0
+(L
H
z
0
)(
20
]
j
)
(24a)
u
12
=
(K
cut
]2 +1)(u
20
+u
cut0
)
20
C
in0
((L
H
z
0
)]z
0
)(
20
]
j
) +3]2/C
in0
(1 (z
0
]L
H
))
z
0
+(L
H
z
0
)
20
]
j
(
j

20
)
2(C
in0
]
j
)(u
20
](L
H
z
0
))(L
H
]z
0
) +g
z
0
+(L
H
z
0
)(
20
]
j
)
(24b)
u
13
=
K
cut
u
cut0
u
20
+u
20
(L
H
z
0
) +/u
20
2
(1 (z
0
]L
H
)) +g(L
H
z
0
)
z
0
+(L
H
z
0
)
20
]
j
(24c)
u
21
=
2
j
(24d)
u
22
=
2
z
0
C
in0
j
(24e)
u
23
= 0 (24f)
u
31
= 0 (24g)
u
32
= 2
C
in0
z
0
L
H
(L
H
z
0
)
2
_
1
20
j
_
(24h)
u
33
= 2
u
20
L
H
z
0
(24i)
where the average two-phase velocity and the two-phase exit velocity are, respectively:
u
20
=
C
in0
j
+

2
(L
H
z
0
) (25)
Fig. 3. Linearized nodal model neutral stability boundary.
u
cut0
=
C
in0
j
+(L
H
z
0
) (26)
Assuming a modal solution of the form,
j
-
= 4
-
exp(t) (27)
Eqs. (21) and (27) yield a cubic characteristic equation, where are the eigenvalues of the system:
2
(u
11
+u
22
+u
33
) +(u
22
u
33
+u
11
u
33
+u
11
u
22
u
12
u
21
) +u
12
u
21
u
33
u
13
u
21
u
32
u
11
u
22
u
33
= 0 (28)
At least one of the eigenvalues will be real and the other two will either be real or complex conjugates. For the complex conjugate
eigenvalues, theoperatingconditions that will generatetheneutral stabilityboundaryarethoseinwhichthecomplexconjugateeigenvalues
are purely imaginary (i.e., the real part is zero). Fig. 3 shows a typical neutral stability boundary for the heated channel in the NASA test
loop (Fig. 1) using the nodal model.
A comparison between an exact analytical solution (Lahey and Moody, 1993; Lahey and Podowski, 1989; Yin et al., 2006) and the nodal
HEM models prediction of the neutral stability boundary is given in Fig. 4 for the NASA test section. For the same subcooling number,
the lumped parameter model yields a larger stability margin than that for the exact analytical solution, and thus it is somewhat non-
conservative. This discrepancy is due to the formulation of the simple, two node, lumped parameter model. Increasing the number of
nodes in the lumped parameter model yields a neutral stability boundary which agrees with the exact analytical model (Garea et al., 1999).
Nevertheless, the basic phenomena are well captured by the simple two node lumped parameter model.
Fig. 5 shows that for the NASA test section shown in Fig. 1 the effect of gravity implied by the nodal model is not great. However, one
must be careful in generalizing these results since other analysis (Achard et al., 1981) has shown that the effect of gravity strongly depends
on heater geometry, orientation and operating conditions.
Fig. 4. Neutral stability model comparisons.
Fig. 5. Effect of gravity on the neutral stability boundary.
Fig. 6. Nodal model: transient for a stable point.
Fig. 7. Nodal model: transient for operating conditions near the neutral stability boundary.
4. Non-linear analysis
Figs. 68 show typical transients for the NASA test section using the non-linear nodal model, given by Eqs. (16), (18) and (20). The
initial perturbation used in these gures was an initial mass ux perturbation which was one percent above the steady-state inlet mass
Fig. 8. Nodal model: transient envelope for an unstable xed point leading to a limit cycle (with an insert of the oscillations).
Fig. 9. Three transient limit cycle envelopes.
ux. The heater power was different in these three gures, as shown by the increase in the phase change number, N
pch
, for a constant
steady-state inlet mass ux (G
in0
). Fig. 6 shows an inlet mass ux transient decaying sinusoidally to the steady-state, signifying that the
systemis damped. Fig. 7 shows a systemthat is essentially neutrally stable, with the oscillations neither growing nor decaying (i.e., a small
amplitude limit cycle).
Fig. 8 plots the envelope (the values of the maximum and minimum of the oscillations during limit cycle oscillations) for a system that
is unstable, with the amplitude of the mass ux oscillations increasing until reaching a maximum (i.e., as the system reaches a relatively
large amplitude limit cycle). The insert in Fig. 8 shows the oscillations of this limit cycle, which has a period, T
DWO
, that is approximately
twice the uids transient time in the heater (Lahey and Podowski, 1989).
It is interesting to analyze the envelopes of the limit cycles. By using the envelopes, multiple limit cycles can be conveniently viewed
in one gure. Fig. 9 shows three such inlet mass ux envelopes, the values of the maximum and minimum of the oscillations during the
limit cycles, for different phase change numbers for the NASA test section. Those transients with a higher power level, and thus a higher
phase change number (since the steady-state inlet mass ux was the same for all transients), will have larger oscillations and will reach a
limit cycle more quickly than those closer to the neutral stability boundary.
The increase in the amplitude of the oscillations for higher phase change numbers implies that our non-linear nodal model might be
experiencing a supercritical Hopf bifurcation (Achard et al., 1985). For example, consider the control parameter, , given by Eq. (29) below,
where N
pch,C
is the critical phase change number on the neutral stability boundary:
=
N
pch
N
pch,C
N
pch,C
(29)
For negativevalues of , theowis stable, andthus it is expectedtodecaytothesteady-stateinresponsetoa small external perturbation.
For positive values of , the ow is unstable, and for a supercritical Hopf bifurcation, the amplitude of the limit cycle oscillations will be
proportional to the square root of . Fig. 10 shows that, for the NASA test section, this is indeed the case.
Fig. 11 shows a pressure-drop vs. mass ux curve for a heated channel in which the operating conditions are at point A (i.e., see Fig. 2).
Recalling the discussionof Fig. 2, a perturbationoff of the steady-state curve will induce anexcursive (Ledinegg) instability to occur, causing
the ow to move to either point A
or A
, depending on the direction of the initial perturbation.

Fig. 10. Hopf bifurcation diagram.
Fig. 11. Pressure drop vs. mass ux curve.
Fig. 12. Excursive (Ledinegg) instabilities leading to either a single-phase ow or a two-phase ow with density-wave oscillations (DWO).
In Fig. 11, if a parallel heated channel is operating on the negative slope region (e.g., point A) of the steady-state operating curve and the
ow is perturbed to point A
, the ow will become a single-phase liquid throughout the channel and thus be stable. A ow going to point
A
may experience DWO after the initial Ledinegg excursion, depending on the stability of point A
. Fig. 12 shows these two transients

for a given Ledinegg instability. The dashed curve is the transient to point A
and the solid curve is the transient to point A
, which shows
Fig. 13. Heated channel and an inline accumulator geometry for a test loop with a constant inlet mass ux.
a Ledinegg-type excursive instability that leads to a DWO. The initial steady-state was a repeller, and the ow was sent to an attractor,
which, in the case of point A
, was unstable, leading to a relatively large amplitude DWO limit cycle.

5. Accumulator dynamics
By adding an accumulator at the inlet of the boiling channel the dynamics of the system can change and there is the possibility of
pressure-drop oscillations (PDOs), where the occurrence of these oscillations depends on the geometry of the accumulator which may
have a surge line connecting the accumulator to the inlet of the heated channel. The test section and type of accumulator analyzed by
Ozawa et al. (1979) and Dogan et al. (1983) is shown schematically in Fig. 13. The inlet mass ux prior to the accumulator junction
(G
in0
) and the exit pressure (P
out
) were specied as being constant, and the imposed pressure drop on the heated channel is between
the junction (p
J
), which may vary in time, and the outlet of the heated channel; or conversely, between the accumulators gas pressure
(p
gas
) and the exit pressure. These are rather non-physical boundary conditions, but such a systemcan be investigated experimentally and
analyzed.
By introducing the accumulator dynamics, the system of conservation equations are modied. In particular, Eq. (16) is modied to use
the inlet mass ux to the heated channel, G
HC
, rather than G
in
:
d
dt
z = 2
C
in0
j
_
C
HC
C
in0
z
z
0
_
(30)
and the corresponding modied momentum equation, Eq. (20), becomes:
_
z +(L
H
z)
j
_
d
dt
C
HC
= p
j
p
cut
2C
in0
z
z
0
_
C
HC
j
+

2
(L
H
z)
__
1
j
_
+
C
2
HC
_
K
cut
2
+1
_
_
C
HC
_
2
j
1
_
+(L
H
z)
2
__
C
HC
j
+(L
H
z)
_
g
j
_
z +(L
H
z)
j
_
2
C
in0
j
_
C
HC
_
1
j
_
(L
H
z)
2
_
_
C
HC
C
in0
z
z
0
_
+(L
H
z)
2
_
C
HC
j
+

2
(L
H
z)
_
/
_
C
2
HC
j
z
L
H
+
2
_
C
HC
j
+

2
(L
H
z)
_
2
_
1
z
L
H
_
_
g
j
_
z +(L
H
z)
j
_
(31)
where the driving pressure drop is between the so-called junction (J) and the impressed outlet pressure. The average two-phase density
dynamics equation, Eq. (18), does not have to be modied.
Let us now consider the accumulator dynamics. The continuity of the liquid in the accumulator vessel (for a positive ow rate, w
ucc
)
species an inow that increases the height of the liquid level in the accumulator vessel:
j
/
ucc
d
dt
H
ucc
= w
ucc
(32)
and the corresponding momentum equation in the accumulator vessel is:
H
ucc
/
ucc
d
dt
w
ucc
= p
j
p
gus
j
gH
ucc
8
l
w
ucc
H
ucc
/
2
ucc
K
in,ucl
|w
ucc
|
2
j
w
ucc
/
2
ucc
(33)
Fig. 14. Nodal model: test section with an accumulator transient inlet mass ux of the heated channel.
Fig. 15. Nodal model: test section with an accumulator transient boiling boundary.
The pressure of the gas in the accumulator is related to the height of the liquid level by, noting that the cross-sectional area of the
accumulator is constant,
p
gus
= p
gus,i
_
H
ucc,m
H
ucc,i
H
ucc,m
H
ucc
_
n
(34)
where n can either be unity, for an isothermal process, or =c
p
/c
v
=1.4, for an isentropic process. The parameter H

acc,m
is the maximum
height of the liquid level (i.e., the height of the vessel), while H
acc,i
is the initial liquid level height. For most purposes, the compressible
volume dynamics can be treated as an isothermal process (i.e. n=1.0). For a constant mass ux into the system, the accumulator ow rate
is given by the continuity equation at the junction:
w
ucc
= (C
in0
C
HC
)/
XS
(35)
Substituting Eqs. (34) and (35) into Eq. (32), and allowing the accumulator gas pressure to replace the liquid level height as a state
variable, yields:
d
dt
p
gus
=
n/
XS
j
/
ucc
p
gus
1+1]n
p
1]n
gus,i
(H
ucc,m
H
ucc,i
)
(C
in0
C
HC
) (36)
For an isothermal process, Eq. (36) simplies to the state variable equation given by Gurgenci et al. (1983) and Akyuzlu et al. (1980),
neglecting liquid volume evaporation in the accumulator.
Figs. 1417 show transients for a typical heated channel/accumulator system (i.e., see Fig. 1). The heated channel operating con-
ditions are those in Fig. 11, with the initial conditions corresponding to point A. The exit pressure was imposed to be the same
pressure as given in the transients in Fig. 12. The initial and maximum heights of the accumulator vessel liquid level were given as
was the diameter of the accumulator vessel. These three geometrical values can be varied to determine their effects on system dynam-
ics.
Fig. 16. Nodal model: test section with an accumulator transient accumulator gas pressure.
Fig. 17. ^p
HC
G
HC
phase plane plot of a PDO superimposed on a steady-state operating curve.
Fig. 14 shows the transient of the mass ux entering the inlet of the heated channel, in which the maximum and minimum values
over/under shoot the values of the mass ux at points A
and A
, respectively, thus showing that the inlet mass ux is oscillating between

the two attractors as a typical PDO.
Fig. 15 shows the corresponding transient of the non-dimensional boiling boundary in the test section, dened by:
z
=
z
L
H
. (37)
For this transient, the owthroughout the heated channel becomes single-phase as the boiling boundary reaches, or exceeds, the length
of the channel (i.e., z* =1), before returning to two-phase ow at the exit.
A change in the accumulator gas pressure (about a 2kPa difference between the maximum and minimum values) is shown in Fig. 16.
Fig. 17 is a phase plot of the transient superimposed with the corresponding steady-state pressure drop vs. mass ux curve (the dashed
curve) for the pressure drop between the junction and exit pressures:
^p
HC
= p
j
p
cut
(38)
The essential difference between the junction pressure and the accumulator gas pressure is the hydrostatic head of the uid, as the
friction and local losses are negligible compared to the gravity head. As expected, the change in the pressure drop between the exit and
junction pressures is about 2kPa, similar to the change in pressure in accumulator gas pressure.
The pressure-drop oscillation (PDO) shown in Fig. 17 has a period of T
PDO
and is centered about an initial mass ux of 1200kg/m
2
s. It
is similar in shape to the experimental limit cycles presented by Stenning et al. (1967) and Ozawa et al. (1979). Fig. 18 shows the effect of
the accumulator vessel diameter on the corresponding limit cycle. As can be seen, the smaller the accumulator diameter, the smaller the
range of mass ux variation in the PDO transient. While these phenomena are interesting, they are not typical of the transients that may
occur in the NASA test loop shown in Fig. 1.
Fig. 19 shows a schematic of the heated channel and the interconnected accumulator for the NASA test loop shown in Fig. 1. The total
pressure drop along the channel (^p=p
in
p
out
) may be impressed as a parallel channel boundary condition, however, as in the previous
case, the pressure at the junction, p
J
, between the heated channel and the accumulator line, can vary with time. With that pressure varying,
Fig. 18. ^p
HC
G
HC
phase plots of PDOs superimposed with the steady-state operating curve showing the effect of accumulator vessel diameter.
Fig. 19. Heated channel and accumulator geometry based on NASA test loop (Fig. 1).
the imposed pressure drop on the heated channel, ^p
HC
=p
J
p
out
, also varies. Part of the mass ux at the inlet to the junction, G
in
, can have
a portion diverted into or out of the accumulator line (given by the ow rate, w
ucc
; a positive ow dened as ow into the accumulator),
while the rest of the mass ux goes into the heated channel, G
HC
.
The heated channel dynamics equations, Eqs. (18), (30) and (31) and the accumulator vessel continuity equation, Eq. (32) still apply for
this new accumulator model geometry. However, the previous accumulator momentum equation, Eq. (33), must be modied to include
the effect of the accumulator line:
_
L
ucl
/
ucl
+
H
ucc
/
ucc
_
d
dt
w
ucc
= p
j
p
gus
_
/
ucl
/
ucc
1
_
|w
ucc
| w
ucc
j
/
2
ucl
g
j
(H
ucl
+H
ucc
) 8
j
_
L
ucl
/
2
ucl
+
H
ucc
/
2
ucc
_
w
ucc

K
ucl
|w
ucc
| w
ucc
2
j
/
2
ucl
(39)
where
K
ucl
= K
in,ucl
+K
ucl
+K
cut,ucl
(40)
We note that by setting L
acl
=H
acl
=0 and A
acl
=A
acc
, Eq. (39) reduces to Eq. (33).
Closure comes fromcontinuity at the junction, Eq. (41a) (where nowG
in
can vary with time), the pressure drop at the inlet local loss, Eq.
(41b), and the impressed constant pressure drop from the inlet to the exit pressure, Eq. (41c), for a parallel channel boundary condition:
w
ucc
= (C
in
C
HC
)/
XS
(41a)
^p
in
= p
in
p
j
=
K
in
2
C
2
in
j
(41b)
^p = ^p
in
+^p
HC
(41c)
Combining and linearizing Eqs. (32), (34) and (39), we obtain a second-order system for the accumulators dynamics:
d
2
dt
2
H
ucc
+2
n
d
dt
H
ucc
+
2
n
H
ucc
=
1
((L
ucl
//
ucl
) +(H
ucc,i
//
ucc
))
j
/
ucc
p
j
(42)
Fig. 21. Nodal model: test section with an accumulator transient accumulator ow rate.
where the angular frequency (and thus the period of oscillation) is:
n
=
_
n(p
gus,i
](H
ucc,m
H
ucc
)) +g
j
j
/
ucc
((L
ucl
]/
ucl
) +(H
ucc,i
]/
ucc
))
=
2
1
PDO
(43a)
and the damping factor is:
=
4
j
n
(L
ucl
]/
2
ucl
) +(H
ucc,i
]/
2
ucc
)
(L
ucl
]/
ucl
) +(H
ucc,i
]/
ucc
)
= 4
j
((L
ucl
]/
2
ucl
) +(H
ucc,i
]/
2
ucc
))
_
j
/
ucc
_
(n(p
gus,i
]H
ucc,m
H
ucc
) +g
j
)((L
ucl
]/
ucl
) +(H
ucc,i
]/
ucc
))
(43b)
The basic geometry of the accumulator vessel and surge line is taken to have the length of the surge line and the liquid level height
comparable to the length of the heated channel, and for the ow area of the surge line to be on the same order of magnitude as the ow
area of the heated channel. The geometry of the accumulator vessel is that of the model given for Fig. 13, for consistency between the two
different accumulator models.
Whereas the accumulator shown in Fig. 13 oscillates about the xed inlet mass ux, the accumulator shown in Fig. 19 will initially
undergo an excursion to either point A
or A
(Fig. 11). An excursive transient frompoint A to A
is shown in Fig. 20, with the only difference

in this transient and the classical Ledinegg excursion, shown in Fig. 12, being the damped oscillations about point A
, due to effect of the

damped accumulator dynamics.
The oscillations in Fig. 20 occur due to the change of the operating pressure of the gas in the accumulator vessel. When the system is
operating at a steady-state (i.e., point A
in Fig. 11), the mass ux at the inlet of the system, G

in
, is equal to the mass ux at the inlet to
the heated channel, G
HC
. An increase of the mass ux into the heated channel causes the junction pressure, p
J
, to decrease. At steady-state
operating conditions, the difference between the junction pressure and the gas pressure is just the hydrostatic head in the accumulator;
therefore the pressure of the gas would also decrease during the excursive transient. To reduce the gas pressure, the height of the liquid
level must decrease, in accordance with Eq. (34). Flow out of the accumulator, a negative ow, increases the ow into the heated channel,
causing the accumulator outow to overshoot and then oscillate to the new steady-state operating point (point A
) (Fig. 21).
Fig. 23. z*G
HC
phase plane plot.
A transient with an initial perturbation opposite that in Fig. 20; that is, from point A to A
(see Fig. 11), is given as Fig. 22. The effect

of the accumulator dynamics adds PDO interaction to the DWO expected for such a system without an accumulator vessel (Fig. 12). As
can be seen for the case analyzed the PDO and DWO have about the same frequency and their interaction yields a beating-type periodic
oscillation. It is interesting to note that similar PDO/DWO interactions have been found using linear frequency domain techniques (Yin et
al., 2006).
Fig. 23 is the corresponding phase plane plot of the non-dimensional boiling boundary, Eq. (37), and the mass ux into the heated
channel. This phase plot shows the results of the two interacting limit cycles. Interestingly, the owat the exit of the heated channel never
becomes single-phase and Fig. 24 gives the dynamics of the heated channels pressure drop, ^p
HC
, which indicates the transient junction
pressure, p
J
.
The corresponding phase plane plot of the inlet mass ux to the heated channel, G
HC
, and the heated channels pressure drop, ^p
HC
,
superimposed with the steady-state operating curve, is shown in Fig. 25. Note that we have a period two DWO, which oscillates about the
two-phase attractor, and is superimposed with a periodic PDO.
As before (see Fig. 18), changing the accumulator vessel diameter, D
acc
, will change the resulting transient. However, for
the coupled system shown in Fig. 19, such a change can completely change the ensuing transient. For instance, increasing the
accumulator diameter from 10 to 13cm causes the oscillations to increase in amplitude until the ow becomes single-phase
throughout the heated channel. Once the ow is single-phase, it overshoots the single-phase attractor (i.e., point A
in Fig. 11)
and eventually reaches the steady-state operating conditions, as seen in Fig. 26. The corresponding transient is given as a phase
plane plot in Fig. 27, with the ow becoming single-phase at the exit of the channel when the inlet mass ux is sufciently
large.
By decreasing the accumulator vessel diameter to, for example, 0.07m, the system no longer experiences interacting oscilla-
tions, but instead exhibits a pure DWO. Fig. 28 shows such a transient, with the amplitude of the oscillations comparable to those
seen in Fig. 12, after the initial transient has decayed. However, before the initial transient disappears, the amplitude of the oscilla-
tions in Fig. 28 varies, due to an interaction of damped accumulator dynamics with the DWO associated with the heated channel.
This interaction should be physically possible by arranging the accumulator line in a helical fashion (i.e., to achieve the desired
surge line length). Note that the corresponding envelope (i.e., the dashed lines) shows the interactions of these oscillations. Even-
Fig. 24. Nodal model: test section with an accumulator transient heated channels pressure drop.
Fig. 25. ^p
HC
G
HC
phase plane plot superimposed with the steady-state operating curve.
tually, the accumulator dynamics decay away, leaving just a DWO. These interactions are also readily shown in the transient of
the accumulator ow rate, Fig. 29, and in the z*G
HC
phase plane plot, Fig. 30. We note that the theoretical approximations for
the periods of these oscillations (T
PDO
=4.4s and T
DWO
=0.88s) is consistent with the non-linear time-domain predictions shown in

Figs. 28 and 29.
Fig. 27. Nodal model: test section with an accumulator phase plane transient.
Fig. 28. Nodal model: test section with an accumulator transient heated channel inlet mass ux with the corresponding envelope.
Fig. 29. Nodal model: test section with an accumulator transient accumulator ow rate.
Fig. 30. z*G
HC
phase plane plot.
Fig. 31. ^p
HC
G
HC
phase plane plot superimposed with the steady-state operating curve.
Also, as can be seen in Fig. 31, the ^p
HC
G
HC
phase plane plot, which is superimposed with the dashed steady-state operating curve,
clearly shows a DWO about the two-phase attractor (i.e., at G
HC
=850kg/m
2
s).
Finally, it should be noted that the results using a more detailed multinode drift-ux model with subcooled boiling are qualitatively the
same as those presented herein (Schlichting, 2009). However, these models are necessarily more complicated to derive and evaluate.
6. Summary and conclusions
This paper presents an analysis the interactions of pressure-drop oscillations (PDO) and density-wave oscillations (DWO) for a typical
NASA test loop. It is shown that very interesting non-linear interactions are possible. Indeed, these type of interactions are much richer
than has been previously discussed (Kakac and Bon, 2008).
Acknowledgements
The authors would like to acknowledge the nancial support given this research project by NASA-GRC. Moreover, the support given by
Rensselaer Polytechnic Institute (RPI) is gratefully acknowledged.
Appendix A. Derivation of non-linear and linear nodal models
Let us assume that the uniform axial heat ux can be treated as a time dependent variable for purposes of perturbing the system. In all
other aspects, including linearization of the model, the heat ux was treated as a constant.
Single-phase continuity equation
z
C(z, t) = 0 (A.1)
Due to Eq. (A.1), the incompressible single-phase mass ux in a constant area duct does not have spatial dependence, thus:
C
1
(t) = C
in
(t) (A.2)
Single-phase energy equation
Since the system pressure is assumed to be constant:
t
h(z, t) +C
in
(t)

z
h(z, t) =
q
P
H
/
H1
(A.3)
Integrating Eq. (A.3) from the inlet to the boiling boundary, z(t):
j
_
z(t)
0
t
h(z, t)dz +C
in
(t)
_
z(t)
0
z
h(z, t)dz =
_
z(t)
0
q
P
H
/
H1
dz (A.4)
Leibnizs rule is dened in Eq. (A.5):
_
b(t)
u(t)
t
j (z, t)dz =
d
dt
_
b(t)
u(t)
j (z, t)dz +j (u(t), t)
d
dt
u(t) j (b(t), t)
d
dt
b(t) (A.5)
Applying Leibnizs rule to Eq. (A.4):
t
_
z(t)
0
h(z, t)dz
j
h(z(t), t)
d
dt
z(t) +C
in
(t)
_
z(t)
0
z
h(z, t)dz =
_
z(t)
0
q
P
H
/
H1
dz (A.6)
t
_
z(t)
0
h(z, t)dz
j
h(z(t), t)
d
dt
z(t) +C
in
(t)[h(z(t), t) h(0, t)] =
q
P
H
/
H1
z(t) (A.7)
Noting that the inlet and boiling enthalpies are given below as Eqs. (A.8a) and (A.8b). The inlet enthalpy is assumed to be constant and
liquid saturation enthalpy is constant for a constant system pressure:
h(0, t) = h
in
(A.8a)
h(z(t), t) = h
j
(A.8b)
Now let:
h
1
(t) =
1
z(t)
_
z(t)
0
h(z, t)dz (A.9)
Substituting Eqs. (A.8a), (A.8b) and (A.9) into Eq. (A.7):
j
d
dt
[h
1
(t)z(t)]
j
h
j
d
dt
z(t) +C
in
(t)(h
j
h
in
) =
q
P
H
/
H1
z(t) (A.10)
j
z(t)
d
dt
h
1
(t) +
j
(h
1
(t) h
j
)
d
dt
z(t) +C
in
(t)(h
j
h
in
) =
q
P
H
/
H1
z(t) (A.11)
Let the single-phase average liquid enthalpy be the numerical average of the inlet and liquid saturation enthalpies. The single-phase
average enthalpy would then be a constant:
h
1
(t) = h
1

h
j
+h
in
2
(A.12)
Substituting Eq. (A.12) into Eq. (A.11):
j
_
h
j
+h
in
2
h
j
_
d
dt
z(t) +C
in
(t)(h
j
h
in
) =
q
P
H
/
H1
z(t) (A.13)
d
dt
z(t) = 2
C
in
(t)
j
2
q
P
H
/
H1
j
(h
j
h
in
)
z(t) (A.14)
Let:
=
q
P
H
/
H1
v
jg
h
jg
(A.15a)
N
sub
=
h
j
h
in
h
jg
v
jg
v
j
(A.15b)
N
pch
=
q
P
H
L
H
C
in0
/
H1
h
jg
v
jg
v
j
(A.15c)
Substituting Eqs. (A.15a), (A.15b) and (A.15c) into Eq. (A.14):
d
dt
z(t) = 2
C
in0
j
_
C
in
(t)
C
in0
N
pch
N
sub
z(t)
L
H
_
(A.16)
Note that the steady-state of Eq. (A.16) yields:
z
0
=
N
sub
N
pch
L
H
(A.17)
Substituting Eq. (17) into Eq. (16):
d
dt
z(t) = 2
C
in0
j
_
C
in
(t)
C
in0
z(t)
z
0
_
(A.18)
Two-phase continuity equation:
t
(z, t) +

z
C(z, t) = 0 (A.19)
Integrating Eq. (A.19) from the boiling boundary, z(t), to the exit:
_
L
H
z(t)
t
(z, t)dz +
_
L
H
z(t)
z
C(z, t)dz = 0 (A.20)
_
L
H
z(t)
t
(z, t)dz +C(L
H
, t) C(z(t), t) = 0 (A.21)
Again using Leibnizs rule:
t
_
L
H
z(t)
(z, t)dz +(z(t), t)
d
dt
z(t) +C(L
H
, t) C(z(t), t) = 0 (A.22)
The density and mass ux at the boiling boundary (and at the inlet) are just the single-phase values for each, Eqs. (A.23a) and (A.23b),
and the mass ux at the exit is given by Eq. (A.23c):
(z(t), t) = (0, t) =
j
(A.23a)
C(z(t), t) = C(0, t) = C
in
(t) (A.23b)
C(L
H
, t) = C
cut
(t) (A.23c)
Let the two-phase average density be dened as:
2
(t) =
1
L
H
z(t)
_
L
H
z(t)
(z, t)dz (A.24)
Substituting Eqs. (A.23a), (A.23b), (A.23c) and (A.24) into Eq. (A.22):
d
dt
[
2
(t)(L
H
z(t))] +
j
d
dt
z(t) +C
cut
(t) C
in
(t) = 0 (A.25)
(L
H
z(t))
d
dt
2
(t) +(
j

2
(t))
d
dt
z(t) +C
cut
(t) C
in
(t) = 0 (A.26)
or,
(L
H
z(t))
d
dt
2
(t) = C
in
(t) C
cut
(t) (
j

2
(t))
d
dt
z(t) (A.27)
The two-phase supercial velocity (Lahey and Moody, 1993) is:
u(z, t) =
C
in
(t)
j
+(z z(t)) (A.28a)
The average two-phase supercial velocity is thus:
u
2
(t) =
1
L
H
z(t)
_
L
H
z(t)
_
C
in
(t)
j
+(z z(t))
_
dz =
C
in
(t)
j
+

2
(L
H
z(t)) (A.28b)
Similarly the exit two-phase supercial velocity:
u
cut
(t) =
C
in
(t)
j
+(L
H
z(t)) (A.28c)
The two-phase mass ux, dened as the spatial average mass ux in the two-phase region, is the product of the average two-phase
supercial velocity and the average two-phase density:
C
2
(t) =
1
(L
H
z(t))
_
L
H
z(t)
C(z, t)dz =
2
(t)u
2
(t) (A.28d)
The exit mass ux is thus:
C
cut
(t) = 2C
2
(t) C
in
(t) =
cut
(t)u
cut
(t) (A.28e)
Substituting Eqs. (A.18), (A.28b), (A.28d), and (A.28e) into Eq. (A.27):
d
dt
2
(t) = 2
C
in0
z
0
_
1
2
(t)
j
_
z(t)
L
H
z(t)

2
(t) (A.29)
The momentum equation:
t
C(z, t) +

z
_
C
2
(z, t)
(z, t)
_
=

z
p(z, t)
j
2D
H
C
2
(z, t)
(z, t)
g(z, t) K
in
C
2
(z, t)
2(z, t)
(z) K
cut
C
2
(z, t)
2(z, t)
(z L
H
)
(A.30)
Integrating Eq. (A.30) from the inlet to the exit of the heated channel:
_
L
H
0
t
C(z, t)dz +
_
L
H
0
z
_
C
2
(z, t)
(z, t)
_
dz =
_
L
H
0
z
p(z, t)dz
_
L
H
0
j
2D
H
C
2
(z, t)
(z, t)
dz
_
L
H
0
g(z, t)dz
_
L
H
0
K
in
C
2
(z, t)
2(z, t)
(z)dz
_
L
H
0
K
cut
C
2
(z, t)
2(z, t)
(z L
H
)dz
(A.31)
The interval union property of integrals is:
_
b
u
j (z)dz =
_
c
u
j (z)dz +
_
b
c
j (z)dz if u - c - b (A.32)
Applying this property of integrals to Eq. (A.31):
_
z(t)
0
t
C(z, t)dz +
_
L
H
z(t)
t
C(z, t)dz +
_
L
H
0
z
_
C
2
(z, t)
(z, t)
_
dz =
_
L
H
0
z
p(z, t)dz
j
2D
H
_
z(t)
0
C
2
(z, t)
(z, t)
dz
j
2D
H
_
L
H
z(t)
C
2
(z, t)
(z, t)
dz g
_
z(t)
0
(z, t)dz g
_
L
H
z(t)
(z, t)dz
_
L
H
0
K
in
C
2
(z, t)
2(z, t)
(z)dz
_
L
H
0
K
cut
C
2
(z, t)
2(z, t)
(z L
H
)dz
(A.33)
Substituting in the single-phase values for the density and the mass ux:
_
z(t)
0
d
dt
C
in
(t)dz +
_
L
H
z(t)
t
C(z, t)dz +
_
L
H
0
z
_
C
2
(z, t)
(z, t)
_
dz =
_
L
H
0
z
p(z, t)dz
j
2D
H
C
2
in
(t)
j
_
z(t)
0
dz
j
2D
H
_
L
H
z(t)
C
2
(z, t)
(z, t)
dz g
j
_
z(t)
0
dz g
_
L
H
z(t)
(z, t)dz
_
L
H
0
K
in
C
2
(z, t)
2(z, t)
(z)dz
_
L
H
0
K
cut
C
2
(z, t)
2(z, t)
(z L
H
)dz
(A.34)
Using Leibnizs rule on Eq. (A.34):
d
dt
_
z(t)
0
d
dt
C
in
(t)dz C
in
(t)
d
dt
z(t) +C(z(t), t)
d
dt
z(t) +
d
dt
_
L
H
z(t)
C(z, t)dz +
_
L
H
0
z
_
C
2
(z, t)
(z, t)
_
dz =
_
L
H
0
z
p(z, t)dz
j
2D
H
C
2
in
(t)
j
_
z(t)
0
dz
j
2D
H
_
L
H
z(t)
C
2
(z, t)
(z, t)
dz g
j
_
z(t)
0
dz g
_
L
H
z(t)
(z, t)dz
_
L
H
0
K
in
C
2
(z, t)
2(z, t)
(z)dz
_
L
H
0
K
cut
C
2
(z, t)
2(z, t)
(z L
H
)dz
(A.35)
The friction number is dened as:
/=
j
2D
H
L
H
(A.36)
Substituting Eqs. (A.23b) and (A.36) into Eq. (A.35) and simplifying:
d
dt
C
in
(t)
_
z(t)
0
dz +
d
dt
_
L
H
z(t)
C(z, t)dz +
_
L
H
0
z
_
C
2
(z, t)
(z, t)
_
dz =
_
L
H
0
z
p(z, t)dz
/
L
H
C
2
in
(t)
j
_
z(t)
0
dz
/
L
H
_
L
H
z(t)
C
2
(z, t)
(z, t)
dz g
j
_
z(t)
0
dz g
_
L
H
z(t)
(z, t)dz
_
L
H
0
K
in
C
2
(z, t)
2(z, t)
(z)dz
_
L
H
0
K
cut
C
2
(z, t)
2(z, t)
(z L
H
)dz
(A.37)
Let the average two-phase dynamic pressure be dened as:
C
2
2
(t)
2
(t)
=
1
L
H
z(t)
_
L
H
z(t)
C
2
(z, t)
(z, t)
dz (A.38)
Substituting Eqs. (A.24), (A.28d) and (A.36) into Eq. (A.37):
d
dt
C
in
(t)
_
z(t)
0
dz +
d
dt
[C
2
(t)(L
H
z(t))] +
_
L
H
0
z
_
C
2
(z, t)
(z, t)
_
dz =
_
L
H
0
z
p(z, t)dz
/
L
H
C
2
in
(t)
j
_
z(t)
0
dz /
C
2
2
(t)
2
(t)
_
1
z(t)
L
H
_
g
j
_
z(t)
0
dz g
2
(t)(L
H
z(t))
_
L
H
0
K
in
C
2
(z, t)
2(z, t)
(z)dz
_
L
H
0
K
cut
C
2
(z, t)
2(z, t)
(z L
H
)dz
(A.39)
d
dt
[C
in
(t)z(t)] +
d
dt
[C
2
(t)(L
H
z(t))] +
C
2
(L
H
, t)
(L
H
, t)

C
2
(0, t)
(0, t)
= p(0, t) p(L
H
, t)
/
C
in
2
(t)
j
z(t)
L
H
/
C
2
2
(t)
2
(t)
_
1
z(t)
L
H
_
g
j
z(t) g
2
(t)(L
H
z(t))
K
in
C
2
(0, t)
2(0, t)
K
cut
C
2
(L
H
, t)
2(L
H
, t)
(A.40)
For a parallel channel boundary condition, a constant pressure drop is imposed across the heated channel:
^p = p(0) p(L
H
) = p(0, t) p(L
H
, t) (A.41a)
Let:
cut
(t) = (L
H
, t) (A.41b)
Substituting Eqs. (A.23a), (A.23b), (A.23c), (A.41a) and (A.41b) into Eq. (A.40):
d
dt
[C
in
(t)z(t)] +
d
dt
[C
2
(t)(L
H
z(t))] = ^p
_
K
in
2
1
_
C
2
in
(t)
_
K
cut
2
+1
_
C
2
cut
(t)
cut
(t)
/
_
C
in
2
(t)
j
z(t)
L
H
+
C
2
2
(t)
2
(t)
_
1
z(t)
L
H
_
_
g(
j
z(t) +
2
(t)(L
H
z(t)))
(A.42)
or,
z(t)
d
dt
C
in
(t) +(C
in
(t) C
2
(t))
d
dt
z(t) +(L
H
z(t))
d
dt
C
2
(t) = ^p
_
K
in
2
1
_
C
2
in
(t)
_
K
cut
2
+1
_
C
2
cut
(t)
cut
(t)
/
_
C
2
in
(t)
j
z(t)
L
H
+
C
2
2
(t)
2
(t)
_
1
z(t)
L
H
_
_
g(
j
z(t) +
2
(t)(L
H
z(t)))
(A.43)
Substituting Eq. (A.28b) into Eq. (A.28d):
C
2
(t) = C
in
(t)
2
(t)
j
+

2
(L z(t))
2
(t) (A.44)
Substituting Eqs. (A.28d), (A.28e) and (A.44) into Eq. (A.43):
z(t)
d
dt
C
in
(t) +(C
in
(t) C
2
(t))
d
dt
z(t) +(L
H
z(t))
d
dt
_
C
in
(t)
2
(t)
j
+

2
(L z(t))
2
(t)
_
= ^p
_
K
in
2
1
_
C
2
in
(t)
_
K
cut
2
+1
_
C
cut
(t)u
cut
(t) /
_
C
in
2
(t)
j
z(t)
L
H
+
C
2
2
(t)
2
(t)
_
1
z(t)
L
H
_
_
g(
j
z(t) +
2
(t)(L
H
z(t)))
(A.45)
or,
_
z(t) +(L
H
z(t))
2
(t)
j
_
d
dt
C
in
(t) +
_
C
in
(t) C
2
(t)

2
(L
H
z(t))
2
(t)
_
d
dt
z(t)
+(L
H
z(t))u
2
(t)
d
dt
2
(t) = ^p
_
K
in
2
1
_
C
2
in
(t)
_
K
cut
2
+1
_
C
cut
(t)u
cut
(t)
/
_
C
2
in
(t)
j
z(t)
L
H
+
C
2
2
(t)
2
(t)
_
1
z(t)
L
H
_
_
g
j
_
z(t) +(L
H
z(t))
2
(t)
j
_
(A.46)
Substituting Eqs. (A.18), (A.28d) and (A.29) into Eq. (A.46):
_
z(t) +(L
H
z(t))
2
(t)
j
_
d
dt
C
in
(t) = ^p 2C
in0
u
2
(t)
z(t)
z
0
_
1
2
(t)
j
_
+(L
H
z(t))C
2
(t)
2
C
in0
j
_
C
in
(t) C
2
(t)

2
(L
H
z(t))
2
(t)
_
_
C
in
(t)
C
in0
z(t)
z
0
_
g
j
_
z(t) +(L
H
z(t))
2
(t)
j
_
_
K
in
2
1
_
C
2
in
(t)
_
K
cut
2
+1
_
C
cut
(t)u
cut
(t) /
_
C
2
in
(t)
j
z(t)
L
H
+C
2
(t)u
2
(t)
_
1
z(t)
L
H
_
_
(A.47)
Finally, substituting Eqs. (A.28b), (A.28c), (A.28d), and (A.28e) into Eq. (A.47) we obtain:
_
z(t) +(L
H
z(t))
2
(t)
j
_
d
dt
C
in
(t) = ^p 2C
in0
z(t)
z
0
_
C
in
(t)
j
+

2
(L
H
z(t))
__
1
2
(t)
j
_
_
K
in
2
1
_
C
2
in
(t)
_
K
cut
2
+1
_
_
C
in
(t)
_
2
2
(t)
j
1
_
+(L
H
z(t))
2
(t)
__
C
in
(t)
j
+(L
H
z(t))
_
g
j
_
z(t) +(L
H
z(t))
2
(t)
j
_
2
C
in0
j
_
C
in
(t)
_
1
2
(t)
j
_
(L
H
z(t))
2
(t)
_
_
C
in
(t)
C
in0
z(t)
z
0
_
+(L
H
z(t))
2
(t)
_
C
in
(t)
j
+

2
(L
H
z(t))
_
/
_
C
2
in
(t)
j
z(t)
L
H
+
2
(t)
_
C
in
(t)
j
+

2
(L
H
z(t))
_
2
_
1
z(t)
L
H
_
_
(A.48)
Using a similar approach more involved multimode models can be derived based on a drift-ux model with subcooled boiling
(Schlichting, 2009), however due to the complexity of these models they have not been presented in this paper.
References
Achard, J.L., Drew, D.A., Lahey Jr., R.T., 1985. The analysis of nonlinear density-wave oscillations in boiling channels. J. Fluid Mech. 155, 213232.
Achard, J.L., Drew, D.A., Lahey Jr., R.T., 1981. The effect of gravity and friction on the stability of boiling ow in a channel. Chem. Eng. Commun. 11, 5979.
Akyuzlu, K., Veziroglu, T.N., Kakac, S., Dogan, T., 1980. Finite difference analysis of two-phase owpressure-drop and density-wave oscillations. Warme-und Stoffubertragung
26, 365376.
Dogan, T., Kakac, S., Vezirglu, T.N., 1983. Analysis of forced-convection boiling ow instabilities in a single-channel upow system. Int. J. Heat Fluid Flow 4, 145
156.
Garea, V.B., Drew, D.A., Lahey Jr., R.T., 1999. A moving-boundary nodal model for the analysis of the stability of boiling channels. Int. J. Heat Mass Transfer 42 (19), 3575
3584.
Gurgenci, H., Veziroglu, T.N., Kakac, S., 1983. Simplied nonlinear descriptions of two-phase ow instabilities in vertical boiling channel. Int. J. Heat Mass Transfer 26, 671
679.
Kakac, S., Bon, B., 2008. A review of two-phase ow dynamic instabilities in tube boiling systems. Int. J. Heat Mass Transfer 51, 399433.
Lahey Jr., R.T., Podowski, M.Z., 1989. On the analysis of various instabilities in two-phase ows. Multiphase Science & Technology, vol. 4. Hemisphere Press, pp. 183370.
Lahey Jr., R.T., Moody, F.J., 1993. The thermal-hydraulics of a boiling water nuclear reactor. ANS Monogr..
Maulbetsch, J.S., Grifth, P., 1965. A study of system-induced instabilities in forced-convection ows with subcooled boiling. MIT Eng. Lab. Rep., 53355382.
NRC Space Studies Board, 2000. Microgravity research in support of technologies for the human exploration and development of space and planetary bodies. Natl. Res.
Council (NAS/NAE) Rep..
Ozawa, M., Nakanishi, S., Ishigai, S., Mizuta, Y., Tarui, H., 1979. Flow instabilities in boiling channels. Bull. JSME 22, 11131117.
Schlichting, W.R., 2009. An analysis of the effect of gravity on interacting DWO/PDO instability modes. PhD Thesis. Rensselaer Polytechnic Institute.
Stenning, A.H., Verizoglu, T.N., 1965. Flow oscillation modes in forced-convection boiling. In: Proceedings of the Heat Transfer & Fluid Mechanics Institute, pp. 301
316.
Stenning, A.H., Verizoglu, T.N., Callahan, G.M., 1967. Pressure-drop oscillations in forced convection ow with boiling. In: Proceedings of the Symposium on Two-Phase Flow
Dynamics, Eindhoven, pp. 405427.
Strogatz, S.H., 1994. Nonlinear Dynamics and Chaos, with Applications to Physics, Biology, Chemistry, and Engineering. Perseus Books Publishing.
Yin, J., Lahey Jr., R.T., Podowski, M.Z., Jensen, M.K., 2006. An analysis of interacting stability modes. J. Multiphase Sci. Technol. 18 (4), 359385.

Nuclear Engineering and Design: An Analysis of Interacting Instability Modes, in A Phase Change System

Diunggah oleh

Informasi Dokumen

Deskripsi Asli:

Judul Asli

Hak Cipta

Format Tersedia

Bagikan dokumen Ini

Bagikan atau Tanam Dokumen

Opsi Berbagi

Apakah menurut Anda dokumen ini bermanfaat?

Apakah konten ini tidak pantas?

Hak Cipta:

Format Tersedia

Nuclear Engineering and Design: An Analysis of Interacting Instability Modes, in A Phase Change System

Diunggah oleh

Hak Cipta:

Format Tersedia

Nuclear Engineering and Design 240 (2010) 31783201

Contents lists available at ScienceDirect

, depending on the direction of the initial perturbation.

. Fig. 12 shows these two transients

and the solid curve is the transient to point A

, was unstable, leading to a relatively large amplitude DWO limit cycle.

=1.4, for an isentropic process. The parameter H

, respectively, thus showing that the inlet mass ux is oscillating between

(Fig. 11). An excursive transient frompoint A to A

is shown in Fig. 20, with the only difference

, due to effect of the

in Fig. 11), the mass ux at the inlet of the system, G

(see Fig. 11), is given as Fig. 22. The effect

=0.88s) is consistent with the non-linear time-domain predictions shown in

Anda mungkin juga menyukai