Sei sulla pagina 1di 20

VOL. 5, NO.

3 HVAC&R RESEARCH JULY 1999


209
Modeling Partial Charging and Discharging of
Area-Constrained Ice Storage Tanks
Jonathan West James E. Braun, Ph.D, P.E.
Member ASHRAE
Although several models have been developed and validated for predicting the performance of
indirect, area-constrained, ice-on-pipe storage tanks, none have fully considered operation
involving partial charging or discharging. In this case, multiple layers of ice and water may
coexist around the heat exchanger tubes. This paper presents two models for predicting the
behavior of tanks with partial charging and discharging processes. One model is based upon
physical principles and is useful for studying design and control issues through simulation. The
second model involves the use of an empirical correlation and is useful for simulating an exist-
ing design or could be appropriate for on-line control. Both models were validated using exper-
imental results for a range of charging and discharging scenarios.
INTRODUCTION
Figure 1 shows one of the more popular types of ice storage tanks in use for commercial cool-
ing. Tightly spaced, small diameter tubes are arranged in a parallel geometry and are surrounded
by water and/or ice in a cylindrical tank. During either charging (ice making) or discharging (ice
melting) processes, a secondary fluid mixture of water and glycol (sometimes termed a brine
solution) is pumped through the tubes. Each individual tube circuit forms a spiral located within
a horizontal plane. Adjacent circuits are attached in an alternating sequence between two sepa-
rate supply and return manifolds. This counterflow arrangement promotes uniform ice growth
and melting throughout the storage tank. The use of multiple flow paths also reduces the pres-
sure drop associated with the flow of the secondary fluid through the tank.
The design of Figure 1 can be characterized as an indirect, area-constrained, ice-on-pipe stor-
age tank. The term indirect denotes the use of a secondary fluid, whereas the term area-con-
strained results from the constraint imposed on ice making and melting when tubes are spaced
closely together. The heat transfer area between ice and water is reduced whenever ice forma-
tions intersect during charging or water formations intersect during discharging.
Numerous models have been developed to predict the behavior of area-constrained ice stor-
age tanks for the geometry shown in Figure 1. Models based on physical principles were devel-
oped by Jekel et al. (1993), Strand et al. (1994), Drees and Braun (1995), Vick et al. (1996),
Nelson et al. (1996), and Neto and Krarti (1997a, 1997b). In addition, Drees and Braun (1996)
developed an empirical model for correlating measured performance data for ice storage tanks.
A physical model may be useful in developing a tank design or studying the relationship
between design and control issues through simulation, whereas an empirical model is useful for
simulating an existing design or could be appropriate for on-line control.
Typically, ice storage tanks are fully charged each day during the off-peak, unoccupied period
and partially or fully discharged during the occupied period. The existing models have no trou-
ble predicting storage performance in response to these charging and discharging processes.
Jonathan West is an engineer with Johnson Controls and James E. Braun is an associate professor at the Ray W. Herrick
Laboratories, Purdue University, West Lafayette, Indiana.
210 HVAC&R RESEARCH
However, most of the models do not handle arbitrary partial and discharging scenarios. It
appears that only Vick et al. (1996) accounted for partial charging and discharging. Nelson et al.
(1996) provided some limited validation of this model, but did not consider successive partial
charging and discharging processes. Partial charging and discharging of ice storage may occur
with modern control strategies in the presence of real-time pricing of electricity. In this case,
multiple layers of ice and water may coexist around the heat exchanger tubes.
The objective of the work described in this paper was to modify existing models to predict the
performance of ice storage tanks during partial charging and discharging processes and to vali-
date these models with measurements. Two models were developed: (1) a physical model based
upon the work of Drees and Braun (1995) and (2) an empirical approach based upon the model
of Drees and Braun (1996). Both models were validated using experimental results for a range
of charging and discharging scenarios.
BACKGROUND AND DEFINITIONS
Figure 2 shows an idealized representation of the growth of ice and water formations during
full charging and discharging processes for adjacent tubes of an area-constrained, ice-on-pipe
storage tank. Ice initially forms in a cylindrical geometry around each tube exterior as cold sec-
ondary fluid [e.g., 25F or 3.9C) flows through the tubes (Figure 2a). Eventually, the ice for-
mations intersect, causing a rapid loss in surface area and heat transfer rate (Figure 2b). During
discharge, a warm secondary fluid solution (e.g., 50 to 60F or 10 to 15.5C) flows through the
inside of the tubes, is cooled by melting the ice, and is returned to the building. Figure 2 also
shows the progression of water layer formation during ice melting conditions for adjacent tubes.
The water initially forms in a cylindrical geometry around each tube exterior (Figure 2c). When
the water formations intersect (Figure 2), there is not only a rapid loss in surface area and heat
transfer rate, but the ice may float freely making the geometry difficult to predict.
Figure 1. Area-constrained, internal-melt ice storage tank
VOLUME 5, NUMBER 3, JULY 1999 211
Ice
Brine
(a) Unconstrained Ice Build
(b) Area-constrained Ice Build
(c) Unconstrained Ice Melting (d) Area-constrained Ice Melting
Water Water
Water
Brine
Ice
Ice
Figure 2. Ice and water formations during full charging and discharging processes
Figure 3. Ice and water formations during partial charging and discharging processes
(a) Ice Build after Full Discharge (b) Partial Discharge After Initial Charge
(c) Initial Ice Build after Partial Discharge
(d) Final Ice Build
Trapped Water Layer
Trapped Ice Layer
Ice
Brine
Water
Brine
Water
Water
Water
212 HVAC&R RESEARCH
Figure 3 depicts idealized ice and water formations during a particular sequence of partial
charging and discharging processes. In Figure 3a, ice is growing on the outside of the tubes fol-
lowing a complete discharge. This is identical to the situation of Figure 2a, except discharge
begins before the tank is fully charged. Figure 3b shows the formation of water layers due to dis-
charging of the partially charged tank. In this case, a thin layer of ice is trapped between water
layers within the tank. During this process, the heat transfer performance is similar to that for
discharging of a fully charged tank. However, the time required to discharge the tank (i.e., the
storage capacity) is much less. If the discharge were allowed to continue, the trapped ice forma-
tion would eventually disappear and the internal water layer would intersect the external water
formation. In Figure 3c, the tank is recharged prior to complete discharge of the previously, par-
tially-charged tank. This leads to a trapped water layer between the growing ice formation and
the ice formation leftover from the previous charging process. Again, the heat transfer perfor-
mance during this process is similar to that for a tank that was fully discharged initially. At some
point, the growing ice formations intersect the existing ice formation and the heat transfer per-
formance changes dramatically. After this point, the ice grows at the outer boundary of the pre-
viously existing ice formation, resulting in a greater overall resistance to heat transfer between
the brine and the water. Eventually, the ice formations on adjacent tubes intersect and the sur-
face area for heat transfer is reduced (Figure 3d). Similar scenarios could be described for dis-
charging of a tank that was initially fully charged.
This paper describes a physical and empirical model for the performance of ice storage tanks
that are subjected to partial charging and discharging processes. Both models modified existing
models to keep track of the ice and water layers, such as those depicted in Figure 3. The result-
ing models were validated with experimental results for a variety of charging and discharging
processes. Experimental and modeling results are presented in terms of heat transfer effective-
ness as a function of a relative state of charge. The heat transfer effectiveness is the ratio of the
actual heat transfer to the maximum possible heat transfer rate and was estimated as
(1)
where is the heat transfer effectiveness, T
f,i
and T
f,o
are the secondary fluid inlet and outlet
temperatures, and T
s
is the storage phase change temperature.
State of charge X is defined as the change in internal energy of the tank between its current
state and initial state (32F or 0C water) divided by the maximum internal energy change if the
tank were completely charged (solid ice at 32F or 0C) or
(2)
where U is the internal energy of the tank referenced to 32F (0C) with no ice, M
ice
is the mass
of usable water (i.e., water that can be frozen), and u
sl
is the specific internal energy required to
freeze water initially at 32F (0C).
EXPERIMENTAL TESTING
The experimental setup documented by Drees and Braun (1995) and West (1997) was used to
run partial charging and discharging tests for model validation. The measurements were used to
estimate the heat transfer rates between the secondary fluid and the tank and the state of charge
for storage. The uncertainty of the measurements, as analyzed by Drees and Braun (1995),

T
f i ,
T
f o ,

T
f i ,
T
s

---------------------- =
X
U
M
ice
u
sl

---------------------- =
VOLUME 5, NUMBER 3, JULY 1999 213
indicate that errors in the heat transfer rates are less than 3% for both charging and discharging.
The storage tank used in this study has a nominal capacity of 90 ton-hours (317 kWh) and a
latent capacity of 75 ton-hours (264 kWh).
Figures 4 and 5 depict two sets of partial charging and discharging tests that were performed
for validation of the physical model. Prior to the start of each test, the tank was equalized to
32F (0C) with no ice. The figures show the state of charge as a function of time. The tests were
comprised of a sequence of partial charging and discharging processes separated by 2 h of idle
time to allow the ice and water in the tank to come into equilibrium. The two tests were both
designed to examine tank performance when charging a partially discharged tank. However, the
second test incorporated faster discharge rates and included sequential discharge stages without
an intermediate charging stage.
Figures 4 and 5 also depict idealized formations of ice and water layers that would exist for
these processes. Each discharge process creates a layer of water between the secondary fluid
tubes and an annular ice formation. The subsequent charging process rebuilds ice in the region
directly around the tubes until it intersects the existing ice formation, at which point ice is added
at the outside of the ice formations. Although the ice formations in Figures 4 and 5 are depicted
as concentric circles, the formations intersect with the surrounding formations at a state of
charge above about 70%.
Some additional partial charging and discharging tests were performed for evaluating the
accuracy of the empirical storage model. Figures 6 and 7 depict these test sequences.
PHYSICAL MODEL
Model Description
The physical model of Drees and Braun (1995) is based upon an earlier model of Jekel (1993)
and uses mass and energy conservation equations applied to different elements distributed along
the axial length of the parallel tubes. Drees and Braun found that five elements were sufficient to
provide good modeling accuracy. The primary assumptions incorporated within the model are:
Negligible heat transfer between adjacent tank elements,
Negligible heat gain from the environment to the shell of the tank,
No transients associated with secondary fluid and piping,
Uniform temperature of ice at 32F (0C) within each element (no subcooling),
Uniform temperature of water within each element,
No flow of water between adjacent elements,
Negligible gravitational effects (symmetric ice formations).
The last assumption implies that the ice and water formations grow symmetrically about both
vertical and horizontal axes. As depicted in Figure 2, the heat transfer problem for each element
is one-dimensional prior to intersection of ice or water formations and two-dimensional after
their intersection. However, the model uses a one-dimensional model in both cases, but uses cor-
rection factors to account for two-dimensional conduction and the loss of surface area after the
formations intersect.
Drees and Braun (1995) compared the model predictions with experimental results for full
charging and discharging processes that were obtained using the experimental setup described
earlier. The model predictions for the heat transfer rates were within about 6% of the measured
results during charging and within about 3% during discharging for relative states of charge
greater than 0.1. The model assumptions break down at lower states of charge when the ice
floats free and the geometry becomes indeterminate. Fortunately, the ice mass is small at this
point and has less influence on the tanks internal energy. The greater error during charging was
214 HVAC&R RESEARCH
Figure 4. Test sequence for Test One
Charging: inlet temperature = 20F (6.7C), flowrate = 27 gpm (1.7 L/s)
Discharging: inlet temperature = 50F (10C), flowrate = 19 gpm (1.1 L/s)
Figure 5. Test sequence for Test Two
Charging: inlet temperature = 22F (5.6C), flowrate = 33 gpm (2.1 L/s)
Discharging: inlet temperature = 50F (10C), flowrate = 33 gpm (2.1 L/s)
VOLUME 5, NUMBER 3, JULY 1999 215
Figure 6. Test sequence for Test Three
Charging: inlet temperature = 22F (5.6C), flowrate = 33 gpm (2.1 L/s)
Discharging: inlet temperature = 50F (10C), flowrate = 33 gpm (2.1 L/s)
Figure 7. Test sequence for Test Four
Charging: inlet temperature = 22F (5.6C), flowrate = 33 gpm (2.1 L/s)
Discharging: inlet temperature = 50F (10C), flowrate = 33 gpm (2.1 L/s)
216 HVAC&R RESEARCH
probably due to the larger experimental uncertainty associated with the smaller secondary fluid
temperature differences during charging.
The model of Drees and Braun (1995) was modified in this study to predict the storage perfor-
mance for incomplete charging and discharging processes. All assumptions of the original
model were retained, but additional variables were added to keep track of trapped ice and water
formations, such as those depicted in Figure 3. Theoretically, any number of ice and water for-
mations could be created with carefully controlled charging and discharging processes. How-
ever, in the implementation considered in this study, the physical model was only modified to
consider two ice and water formations at any given time. In addition, at the end of each dis-
charge stage, the sensibly heated water adjacent to the tubes was assumed to instantaneously
achieve equilibrium with any remaining ice within each element (i.e., cooled back to the freez-
ing point while melting additional ice).
Model Validation
Figures 8 and 9 show storage heat transfer efrfectiveness during charging phases of tests one
and two, respectively, determined from the model and measurements. Overall, the model cap-
tures the trends associated with the experimental data. At the beginning of each charging stage,
the effectiveness is nearly constant (~ 0.9), independent of the past history. Following a dis-
charge, ice is built through water directly next to the heat transfer tubes and the effectiveness is
at a maximum. However, once the ice formations intersect existing formations, the effective-
ness falls off quickly since additional ice must be built at the outside of the existing formations.
The point where the new and old ice formations intersect depends upon the previous history of
charging and discharging. For the results of Figures 8 and 9, each successive charging stage was
initiated with ice formations that were further from tubes. As a result, the effectiveness is higher
at the same state of charge for each successive charging stage of the test. The rapid falloff in
effectiveness at high states of charge is due to a reduction in surface area between the ice and
water after ice formations from adjacent tubes intersect. These effects are apparent in both the
experimental and simulated results.
Although the model predicts the experimental trends, it tends to under-predict the effective-
ness, particularly after new ice formations intersect existing ones. In addition, the errors are
more pronounced for the charging results that follow discharge at higher rates (Figure 9). The
Figure 8. Charging effectiveness for Test One
VOLUME 5, NUMBER 3, JULY 1999 217
errors may be due to the assumptions of rigid ice structures and uniform melting employed by
the model. In reality, the ice formations will deform, break up, and eventually float away from
the tubes near the end of a complete discharge. This phenomenon may occur earlier for high dis-
charge rates. In general, the breakup of the ice formations should lead to higher heat transfer
effectiveness since new ice builds are less likely to be built on the old ones. This is consistent
with the experimental results. It would be extremely difficult to model the effects of the break-
down in ice formations. Overall, the existing model appears to be adequate for studying the
impact of design parameters on charging performance since it captures the performance trends
and is conservative in its performance predictions.
Figure 10 gives effectiveness results for the discharging stages of test one. The model and
experimental trends and magnitudes are in very close agreement. In general, heat transfer effec-
tiveness decreases during discharge (decreasing state of charge) as the water layers in contact
with the tubes grow. The effectiveness was nearly the same at the beginning of each discharging
stage since each discharge stage followed a charging stage and began with ice in direct contact
with tubes. For the results of Figure 10, each successive discharging stage began with a higher
initial state of charge. Therefore, for the same state of charge, the water layer thickness next to
the tubes was greater and heat transfer effectiveness was lower for the higher initial states of
charge.
Figure 11 shows the last three discharge stages for test two. These three discharge stages
were not separated by charging stages, but were initiated following two hours with no charg-
ing or discharging. During these rest periods, the warm water in contact with the tubes melts
additional ice and cools. The experimental and model results agree quite closely, except at the
start of each discharge stage. The model does not consider the transients associated with the
secondary fluid. In reality, the secondary fluid approaches 32F (0C) during the off period
such that the exit temperature from storage at the start of the discharge stage is also 32F
(0C). This condition exists until the entering fluid has flushed out the secondary fluid stored
in the tank. This is clearly exhibited in the experiment results where the effectiveness
approaches one at the start of each discharge stage. Since the model estimates the exit second-
ary fluid temperature using a steady-state energy balance on the secondary fluid, it can not
predict this transient effect.
Figure 9. Charging effectiveness for Test Two
218 HVAC&R RESEARCH
EMPIRICAL MODEL
Drees and Braun (1996) developed an empirical model for ice storage tanks that is useful for
system simulations, but does not consider arbitrary partial charging and discharging processes.
At any time, the state of storage is updated from its previous state according to
(3)
where k denotes a stage in time, u
k
is the rate at which energy is removed from storage (charg-
ing rate) for the current time stage, t is the time increment, and Cap
s
is the maximum change
in internal energy of the storage tank that can occur during a complete charging or discharging
process (e.g., the latent capacity).
Figure 10. Discharging effectiveness for Test One
Figure 11. Discharging effectiveness for Test Two
X
k
X
k 1
u
k
t
Cap
s
------------ + =
VOLUME 5, NUMBER 3, JULY 1999 219
The limit on the rate at which energy may be removed from storage depends on the inlet
fluid temperature, maximum secondary fluid flow rate, and heat transfer effectiveness accord-
ing to
(4)
where
c
is a heat transfer effectiveness for charging at the current storage condition if the sec-
ondary fluid flow rate were at its design value of , c
f
is the secondary fluid specific heat,
T
f,i
is the temperature of secondary fluid inlet to the tank, and T
s
is the temperature at which the
storage medium melts or freezes.
The minimum charging rate is actually the negative of the maximum discharging rate and is
given by
(5)
where
d
is the heat transfer effectiveness for discharging at the current storage condition with
the design secondary fluid flow rate.
Using the design secondary flow and for full charging and discharging processes, Drees and
Braun (1995, 1996) demonstrated that heat transfer effectiveness is primarily a function of state
of charge and then developed separate correlations for charging and discharging. Furthermore,
they suggested a method for applying these correlations for the following specialized partial
charging and discharging processes: (1) full discharge following partial charging and (2) full
charging following partial discharging. However, the model was not validated for these cases
and would not be appropriate for other partial charging and discharging scenarios.
Characterizing Heat Transfer Effectiveness during
Partial Charging and Discharging
In general, storage heat transfer effectiveness is a function of both the state of charge and the
time history of charging and discharging. Consider the ice formations illustrated in Figure 12.
Although both conditions are depicted with the same state of charge, they would have very dif-
ferent heat transfer effectiveness for charging and discharging. In particular, the formation with
the ice in contact with the tube (Case A) would have a higher effectiveness for discharging and a
lower effectiveness for charging than the formation with water surrounding the tube (Case B).
The effectiveness model for partial charging is based upon the assumption that effectiveness
only depends upon the amount of the ice that is directly in contact with the tube. This implies
u
max

c
m

f des ,
c
f
T
s
T
f i ,
( ) =
m

f des ,
u
min

d
m

f des ,
c
f
T
s
T
f i ,
( ) =
Figure 12. Example of different ice formations with same state of charge
220 HVAC&R RESEARCH
that the charging effectiveness for the condition depicted in Case B of Figure 12 is nearly the
same as that for a fully discharged tank. After charging is initiated, the ice formation grows from
the tube and the effectiveness decreases as it would for a completely discharged tank until it
intersects the existing ice formation. At this point, there is a step change reduction in the effec-
tiveness since ice must now grow at the outside of the combined formation. The effectiveness is
assumed to be the value that would have resulted if ice had been built continuously to this point
from a completely discharged tank.
In general, the charging effectiveness at any storage condition is related to the effectiveness
determined from a full charging test (starting from a complete discharge) according to
(6)
where
c,des
is the effectiveness determined from a full charging test and evaluated at a charging
state associated with the ice formations that are in direct contact with the tubes (X
inner
). Follow-
ing a discharge process, X
inner
is zero and all of the energy removal works to increase this state
according to Equation (3). At the point where the growing ice formation intersects an existing
formation, there is a step change in X
inner
as the state of charge associated with the trapped ice
formation is added to the existing one. In the general case of arbitrary partial charging and
discharging processes, it is necessary to keep track of all of the charge states that began each
charging and discharging process in order to determine the necessary information associated
with the trapped ice formations.
Consider the special case where there is no more than one trapped ice formation, such as the
situations depicted in Cases A or B of Figure 12. In Case A, ice will grow on the existing forma-
tion and X
inner
is equal to the state of charge. For Case B, ice will grow around the tube and
X
inner
is the difference between total state of charge and the state of charge associated with the
trapped ice formation. Furthermore, the state of charge associated with the trapped formation is
simply the state of charge that existed at the end of the discharge process when it was created.
With these points in mind, X
inner
can be determined according to
(7)
where X
max
is the maximum state of charge that has occurred since the last complete dis-
charge and X
prev
is the state of charge at the end of the previous dissimilar process (i.e., dis-
charge process for this case). In this special case, X
max
and X
prev
represent the charge states
that existed when the outer and inner boundaries of the trapped ice formation were formed,
respectively. The growing ice formation intersects the trapped ice formation when X equals
X
max
. To accommodate additional trapped ice formations, it would be necessary to keep track
of the charge states associated with the creation of the outer and inner boundaries of each
trapped formation.
Similar to the charging effectiveness, the effectiveness model for discharging is based upon
the assumption that effectiveness only depends upon the amount of the water that is directly in
contact with the tube. Thus, the discharging effectiveness for the condition depicted in Case A
of Figure 12 is considered to be the same as that for a fully charged tank. After discharging is
initiated, the water formation grows from the tube and the effectiveness decreases as it would for

c

c des ,
X
inner
( ) =
if X X
max
<
X
inner
X X
prev
=
otherwise
X
inner
X =
VOLUME 5, NUMBER 3, JULY 1999 221
a completely charged tank until it intersects the existing water formation. At this point, there is a
step change reduction in the effectiveness since water must now grow at the outside of the com-
bined formation. The effectiveness is assumed to be the value that would have resulted if ice had
melted continuously to this point from a completely charged tank.
It is convenient to define a state of discharge for storage as
(8)
The state of discharge is essentially the fraction of the water and ice that is water (whereas the
state of charge is the fraction that is ice). With this definition, a zero state of discharge is associ-
ated with a fully charged tank (solid ice at 32F or 0C) and a discharge state of one is associated
with a fully discharged tank (water at 32F or 0C). In practice, the discharge state is always
greater than zero, since the tank is never fully charged. The method used for estimating dis-
charge effectiveness is analogous to the approach for charging effectiveness, except that state of
discharge is used in place of state of charge since water layers are growing and intersecting pre-
viously trapped water layers. The discharging effectiveness at any storage condition is related to
the effectiveness determined from a full discharging test (starting from a complete charge)
according to
(9)
where
d,des
is the effectiveness determined from a full discharging test and evaluated at a
charge state associated with a fully charged tank surrounding the inner water formation that is
growing directly around the tubes (1 X
d,inner
).
Consider the special case where there is no more than one trapped water formation, such as
the situations depicted in Cases A and B of Figure 13. Case B differs from that of a single
trapped ice formation depicted in Case B of Figure 12, in that three water formations exist after
the initiation of discharge (inner, trapped, and outer water layers). The outer water formation
always exists because the tank is never fully charged. In Case A, the inner water formation
grows during discharge and X
d,inner
is equal to the difference between the total state of discharge
and the state of discharge associated with the outer water formation. For Case B, the inner water
layer grows and X
d,inner
is the difference between the total state of discharge and the discharge
state associated with both the trapped and outer water formations. Furthermore, the state of dis-
charge associated with both the trapped and outer water formations is simply the state of dis-
charge that existed at the end of the charging process when the trapped water formation was
created. During a full discharge, the growing inner water formation will eventually intersect the
outer water formation. At this point, the inner state of discharge is equal to the total discharge
state. This occurs at a discharge state of unity. With these points in mind, X
d,inner
can be deter-
mined according to
(10)
X
d
1 X =

d

d des ,
1 X
d inner ,
( ) =
if X
d
X
d max ,
<
X
d inner ,
X
d
X
d prev ,
=
elseif X
d
1 <
X
d inner ,
X
d
X
d min ,
=
otherwise
X
d inner ,
X
d
=
222 HVAC&R RESEARCH
where X
d, max
is the maximum state of discharge (minimum state of charge) that has occurred
since the tank was charged to its most recent maximum state, X
d,prev
is the state of discharge at
the end of the previous dissimilar stage (i.e., charging stage for this case), and X
d, min
is the min-
imum state of discharge since the last complete discharge (maximum charge state). In this spe-
cial case, X
d, max
and X
d,prev
are the discharge states when the outer and inner boundaries of the
trapped water formation were formed, respectively. In addition, X
d, min
is the discharge storage
state associated with the formation of the outer boundary of the outermost ice formation (or
inner boundary of the outermost water formation). The growing water formation intersects the
trapped water formation when X
d
equals X
d,max
and intersects the outer formation when X
d
is
unity (fully discharged). To accommodate additional trapped water formations, it would be nec-
essary to keep track of the discharge states associated with the creation of the outer and inner
boundaries of each trapped formation.
The expressions for X
d,inner
in Equation (10) can be rewritten in terms of states of charge
using the definition for state of discharge (1 X) given in Equation (8), so that
(11)
where X
min
is the minimum state of charge since the last complete discharge, X
prev
is the state of
charge at the end of the previous dissimilar process (i.e., charging process for this case), and
X
max
is the maximum state of charge since the last complete discharge.
When the state of charge falls below about 10 to 15% during discharge, the ice geometry would
likely be very irregular. For this reason, X
max
and X
min
are set to zero (or X
d, min
and X
d,max
are
set to one) at the end of any discharge stage that finishes with a state of charge below 15%.
Figure 13. Water layer formations during discharge
if X X
min
>
X
d inner ,
X
prev
X =
elseif X X
max
<
X
d inner ,
X
max
X =
otherwise
X
d inner ,
1 X =
VOLUME 5, NUMBER 3, JULY 1999 223
Validation for Partial Charging and Discharging Processes
Typically, manufacturers provide data that could be used to determine full charging and dis-
charging effectivenesses for design flow rates. This data can then be correlated and used in con-
junction with the model of the previous section to estimate performance for partial charging and
discharging processes. For this study, the experimental system described previously was used to
develop the necessary heat transfer effectiveness correlations for full charging and discharging
tests and then used to evaluate performance for partial charging and discharging scenarios.
Table 1 gives the conditions for the full charging and discharging tests used to develop the
effectiveness correlations. All of the tests were run with the maximum secondary fluid flow rate
(33 gpm or 2.1 L/s) through the tank. As with the tests discussed earlier, the tank was equalized
to near zero state of charge (32F or 0C with zero ice) prior to the first test stage and sat idle for
2 hours between stages to allow the ice and water in the tank to come into equilibrium.
Figure 14 shows effectiveness determined from the data for the full charging and discharging
tests as a function of relative state of charge X
r
. Relative state of charge is defined as
(12)
Equations (7) and (11) were used to determine X
inner
and X
d,inner
in order to account for the
fact that the tank was not fully charged and discharged during the tests.
Table 1. Full Charging and Discharging Test Specifications
Stage Mode X
start
X
end
T
f,i
1 Charge 0.0 0.95 24F (4.4C)
2 Discharge 0.95 0.05 45F (7.2C)
3 Charge 0.05 0.95 20F (6.7C)
4 Discharge 0.95 0.05 55F (12.8C)
Figure 14. Heat transfer effectiveness data from full charging and discharging tests
Conditions of Table 1; flowrate = 33 gpm (2.1 L/s)
X
r
X
inner
for charging
1 X
d inner ,
for discharging

=
224 HVAC&R RESEARCH
The results of Figure 14 indicate that most of the charging and discharging effectiveness data
correlate very well with state of charge, independent of secondary fluid inlet temperature. How-
ever, there is a large fluctuation in charging effectiveness during the first thirty minutes of
charging (low charge states) that occurs because of meta-stable supercooling of the water in the
tank below the freezing point. The supercooling occurs because the plastic tubing does not pro-
vide adequate nucleation sites for freezing at 32F (0C). At the onset of freezing, approxi-
mately 250 lb. (113 kg) of ice formed within a 120 s interval, raising the water temperature from
27F (2.8C) to the phase change temperature of 32F (0C).
The effectiveness data of Figure 14 were curve fit using polynomial functions, resulting in the
following correlations
(13)
(14)
where X
r
is the relative state of charge defined in Equation (12). Figure 14 shows that these
functions provide very good correlations to the data. The coefficients in Equations (13) and (14)
are specific to the tank model examined in this study, but could be estimated for any tank using
manufacturers data.
The tests depicted in Figures 5, 6, and 7 were used to evaluate the method for estimating heat
transfer effectiveness for partial charging and discharging operation using the design correla-
tions. These tests incorporated charging and discharging stages that produced trapped ice and
water layers of different thickness and locations relative to the tubes. The tests also included
multiple charging and discharging stages in sequence and separated by periods of tank inactivity
to allow water and ice layers at different temperatures to approach equilibrium. The complete set
of test stages provided a representative set of operating scenarios for an ice storage system.
Figure 15 shows comparisons between the correlation and experimental results for discharge
effectiveness determined from the discharging stages of test four. The data and discharge corre-
lation are both presented as a function of relative state of charge, with X
r
determined with Equa-
tion (12). Overall, the agreement is excellent. The discrepancies are higher at the beginning of
each stage due to transients occurring within the secondary fluid when the flow is initiated. This
is particularly noticeable for Stage 5 because of the low steady-state effectiveness as compared
with the start of the other stages. On average, the effectiveness predictions were about 4% differ-
ent than the values determined from measurements. However, if the actual state of charge were
used to evaluate heat transfer effectiveness, the average difference would be about 16% overall
and about 30% for Stage 5.
Figure 16 shows similar results for the discharge stages of test two. Again, the model does an
excellent job of predicting discharge effectiveness except at the beginning of each stage when
the transients associated with initiating the secondary fluid flow are important. The average
error was about 4%, while the error associated with using the actual state of charge in the corre-
lation would be about 20%.
Figure 17 shows results from the discharging stages of the third test. Stage four of this test
involved the melting of ice near the tube and a trapped ice layer away from the tube. Although
the heat transfer effectiveness is slightly different for the two discharge stages, the model per-
formed well in predicting performance for both cases. On average, the model predictions are
within about 3% of the values determined from measurements.
Figures 18 and 19 show comparisons between the correlation and experimental results for
charging effectiveness determined from the charging stages of tests two and four as a function of

c des ,
0.92 0.62 X
r
4.93 + X
r
2
17.05 X
r
3
24.02 + X
r
4
12.12 X
r
5
. =

d des ,
0.49 0.81 + X
r
0.98 X
r
2
0.67 + X
r
3
=
VOLUME 5, NUMBER 3, JULY 1999 225
relative state of charge [Equation (12)]. Again, the agreement is excellent. The largest discrep-
ancies occur at very low states of charge where supercooling of the tank water occurs prior to
the initiation of freezing. Overall, the predicted charging effectiveness are within about 4% of
the values determined from measurements.
Figure 15. Discharging effectiveness for Test Four
Figure 16. Discharging effectiveness for Test Two
226 HVAC&R RESEARCH
CONCLUSIONS
This paper described two models for predicting the performance of ice storage tanks subjected
to partial charging and discharging processes. One model is useful for investigating the effects of
tank design parameters, while the other is appropriate for system simulations. The physical model
is based on the model described by Drees and Braun (1995). This model was modified to consider
Figure 17. Discharging effectiveness for Test Three
Figure 18. Charging effectiveness for Test Two
VOLUME 5, NUMBER 3, JULY 1999 227
trapped ice formations that may be created during partial charging and discharging processes.
Overall, the model predicted the experimental trends. However, it does tend to under-predict
charging effectiveness, particularly after new ice formations intersect existing ones.
The empirical model uses effectiveness correlations developed from full charging and dis-
charging tests to predict performance for partial charging and discharging. These correlations
are evaluated at modified states of charge that account for the ice (charging) or water (discharg-
ing) in contact with the tubes. On average, predictions of charging and discharging effectiveness
were within about 4% of those determined from measurements. The errors associated with using
actual state of charge to evaluate effectiveness from the correlations were between 10 and 30%.
ACKNOWLEDGEMENTS
Financial support for this research was provided by Johnson Controls, Inc. and the Electric
Power Research Institute, Inc. The authors appreciate this support and the guidance provided by
Kirk Drees.
NOMENCLATURE
c specific heat
heat transfer effectiveness
mass flow rate
M
ice
mass of ice corresponding to design latent
capacity of ice storage
T temperature value or measurement
U internal energy of the tank referenced to
32F (0C) with no ice
u storage charging rate (rate of energy
removal from storage)
u
sl
specific internal energy required to freeze
water initially at 32F (0C)
X state of charge
X
prev
state of charge at end of previous dissimi-
lar stage (charging or discharging)
X
d
state of discharge (1 X)
X
inner
state of charge associated with ice in direct
contact with tubes
X
d,inner
state of discharge associated with water in
direct contact with tubes
X
max
maximum state of charge since the last
complete discharge
X
min
minimum state of charge since the last
complete discharge
Figure 19. Charging effectiveness for Test Four
m

228 HVAC&R RESEARCH


X
d,max
maximum state of discharge (minimum
state of charge) that has occurred since the
tank was charged to its most recent maxi-
mum (1 X
min
)
X
d,min
minimum state of discharge (maximum
charge state) since the last complete dis-
charge (1 X
max
)
X
d,prev
state of discharge at the end of the previous
dissimilar stage (1 X
prev
)
X
r
relative state of charge (X
inner
for charging
and 1 X
d,inner
for discharging)
Additional Subscripts
c charging
d discharging
f secondary fluid
i inlet conditions
o outlet conditions
s storage conditions or properties
REFERENCES
Drees, K.H., and J.E. Braun. 1995. Modeling of Area-Constrained Ice Storage Tanks. International Jour-
nal of HVAC&R Research 1(2): 143-159.
Drees, K.H., and J.E. Braun. 1996. Development and Evaluation of a Rule-Based Control Strategy for Ice
Storage Systems. International Journal of HVAC&R Research 2(4): 312-336.
Jekel, T.B., J.W. Mitchell, and S.A. Klein. 1993. Modeling of Ice Storage Tanks. ASHRAE Transactions
99(1): 1016-1023.
Nelson, D.J., B. Vick, and X. Yu. 1996. Validation of the Algorithm for Ice-on-Pipe Brine Thermal Stor-
age Systems. ASHRAE Transactions 102(1): 55-62.
Neto, J.H.M., and M. Krarti. 1997a. Deterministic Model for an Internal Melt Ice-on-Coil Thermal Storage
Tank. ASHRAE Transactions 103(1).
Neto, J.H.M., and M. Krarti. 1997b. Experimental Validation of a Numerical Model for an Internal Melt
Ice-on-Coil Thermal Storage Tank. ASHRAE Transactions 103(1).
Strand, R.K., C.O. Pedersen, and G.N. Coleman. 1994. Development of Direct and Indirect Ice-Storage
Models for Energy Analysis Calculations. ASHRAE Transactions 100(1): 1230-1244.
Vick, B., D.J. Nelson, and X. Yu. 1996. Model of an Ice-on-Pipe Brine Thermal Storage Component. ASH-
RAE Transactions 102(1): 45-54.
West, J.D. 1997. Evaluation of Methods for Controlling Ice Storage Systems. M.S. thesis, Purdue Univer-
sity, W. Lafayette, IN.

Potrebbero piacerti anche