You are on page 1of 8

Applied Energy 87 (2010) 3322–3329

Contents lists available at ScienceDirect

Applied Energy
journal homepage: www.elsevier.com/locate/apenergy

Molten-salt thermal energy storage in thermoclines under different


environmental boundary conditions
Zhen Yang, Suresh V. Garimella *
Cooling Technologies Research Center, an NSF I/UCRC, School of Mechanical Engineering, 585 Purdue Mall, Purdue University, West Lafayette, IN 47907-2088, USA

a r t i c l e i n f o a b s t r a c t

Article history: Operation during the charge and discharge cycles of molten-salt thermoclines used for solar thermal
Received 8 December 2009 energy storage depends strongly on the environmental boundary conditions to which the tanks are
Received in revised form 11 February 2010 exposed. A comprehensive model which accounts for thermal transport in the molten-salt heat transfer
Accepted 25 April 2010
fluid and the filler material in the tank is developed for exploring the effects of boundary conditions on
Available online 23 June 2010
thermocline performance. Heat loss from the tank under non-adiabatic boundary conditions is found to
distort the temperature and salt flow distributions relative to the uniform conditions found in adiabatic
Keywords:
thermoclines; as a result, the outflow temperature drops more rapidly in the former case. Such effects of
Solar thermal energy
Energy storage
non-adiabatic boundaries become insignificant at large salt-flow Reynolds numbers. As the Reynolds
Thermocline number increases beyond 250, the discharge efficiency of non-adiabatic thermoclines approaches that
Molten salt of the adiabatic counterparts. In the case of significant heat loss at the walls, the discharge efficiency
Concentrating solar plants of thermoclines increases with increasing Reynolds number, a trend that is opposite to that in adiabatic
Sustainability thermoclines.
Ó 2010 Elsevier Ltd. All rights reserved.

1. Introduction such as quartzite rock [12], is used to fill most of the volume in the
thermocline and acts as the primary thermal storage material. An
Thermal energy storage (TES) is a critical element in solar en- experimental demonstration of such a thermocline on a pilot scale
ergy applications, including in the increase of building thermal (2.3 MWh) was provided in Ref. [13]; the results of this demonstra-
capacity [1,2], solar water heating systems [3–5] for domestic tion showed the feasibility of using a molten-salt thermocline sys-
use, and Concentrated Solar Thermal (CST) power plants for elec- tem for thermal storage in a parabolic trough plant.
tricity generation. CST power plants, such as the Solar Energy Gen- While a few studies have explored molten-salt thermocline
erating Systems (SEGS) [6–8], are believed to be one promising thermal energy storage for parabolic-trough solar thermal plants,
approach for converting solar energy into electricity on a large a number of aspects of their thermal behavior, such as the effect
scale. Molten-salt thermocline TES for CST power plants can: (1) of the distributors at either end of the filler bed, remain poorly
offer power plants the potential to continuously deliver electricity understood. Recent studies by the authors [14,15] investigated
without fossil-fuel backup; (2) meet peak demand independent of the discharge and cycle efficiencies of thermal energy storage in
weather conditions; (3) increase the storage temperature above thermoclines for different melt flowrates, filler particle diameters
450 °C to raise the Rankine cycle efficiency above 40% [9]; and and tank heights. The thermocline tank boundary, however, was
(4) save 35% of cost compared to a two-tank storage system only considered to be adiabatic in these studies. The effects on
[10,11]. In a molten-salt thermocline, a molten salt (e.g., Hitec or the heat transfer and fluid flow of a non-adiabatic tank wall, as
Hitec XL [9]) is used as the heat transfer fluid (HTF) that transports would be found in actual applications were not explored, and form
the thermal energy between the storage unit and the other sec- the focus of the present study.
tions of the power system such as the collector field and the steam
generator. Separation between the hot and cold zones of the mol-
ten salt is naturally ensured by buoyancy forces; stable thermal
2. Theory and numerical modeling
stratification is thus maintained in the fluid in a single tank. To re-
duce the inventory of relatively expensive molten salt in the stor-
2.1. Problem description
age system, a low-cost filler material compatible with molten salts,
The details of a TES thermocline are depicted in Fig. 1. The tank
* Corresponding author. Tel.: +1 (765) 494 5621. of inner diameter d is filled with a porous filler bed of quartzite
E-mail address: sureshg@purdue.edu (S.V. Garimella). rock to a height h. A molten salt (HITEC [16]) serves as the heat

0306-2619/$ - see front matter Ó 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.apenergy.2010.04.024
Z. Yang, S.V. Garimella / Applied Energy 87 (2010) 3322–3329 3323

Nomenclature

Cp specific heat, J kg1 K1 u velocity vector, m/s


d diameter of thermocline tank, m um mean velocity magnitude at the inlet of filler region, m/s
d0 diameter of the inlet and outlet ports, m
ds diameter of filler particles, m Greek
er unit vector in the r direction, – a thermal diffusivity of liquid salt, m2/s
eh unit vector in the angular direction, – e porosity, –
ex unit vector in the x direction, – l viscosity of liquid salt, kg m1 s1
F inertial coefficient, F ¼ p1:75
ffiffiffiffiffiffiffiffiffi3ffi [20] m kinematic viscosity of liquid salt, m2/s
150e
g acceleration due to gravity, m/s2 h polar angle, rad
h thermocline tank height, m q density, kg/m3
h0 distributor region height, m
hi interstitial heat transfer coefficient, W m3 K1 Subscripts
2
K permeability, K ¼
ds e3
[21], m2 c at the inlet low temperature
175ð1eÞ2 e effective
k thermal conductivity, W m1 K1 h at the outlet high temperature
p pressure, Pa l liquid salt phase
T temperature, K s solid filler phase
t time, s

transfer fluid (HTF) and flows through the filler bed to exchange As a result, only a part of the initially stored thermal energy can
heat with the quartzite filler. The dimensions d and h are set to be retrieved as useful heat.
be equal in this work although a wide combination of values is pos-
sible for these dimensions. On both the top and bottom sides of the 2.2. Governing equations
porous bed are distributors of height h0 (h0 = 0.05h), which are free
of quartzite rock. These distributor regions serve to maintain a uni- 2.2.1. Molten-salt flow in the filler bed
form flow condition at both ends of the filler bed in order to Non-dimensional governing equations in the axisymmetric
achieve good thermal stratification inside the filler bed. The dis- coordinate system (x–r) shown in Fig. 1 for the laminar fluid flow
tributors are connected with tubes of diameter d0 (d0 = 0.05d) at and heat transfer in the filler bed are provided in [14] as follows:
the top and bottom ports. Initially, the whole tank, including the
molten salt and the filler bed, is held at a high temperature level a. Continuity equation:
Th of 450 °C. An inflow of cold molten salt is pumped into the ther-
mocline tank through the bottom port, and the thermal energy @ðUq Þ
e þ r  ðUq UÞ ¼ 0 ð1Þ
stored in the thermocline is thus discharged from the tank by the @s
hot outflow at the top port. The inflow temperature at the bottom
b. Momentum equation:
port is set to Tc = 250 °C. During the first part of the discharge per-
iod, molten-salt outflow through the top port is maintained at the  
@ðUq UÞ Uq UU
desired high temperature level, and is delivered for generating ReW þ ReWr 
superheated steam for electricity production. As the discharge pro- @s e
 
cess proceeds, the outflow decreases in temperature, eventually Ul U FReW
¼ erP þ r  T þ eUq Grex  e þ Uq jUjU ð2Þ
falling to a value that is no longer suitable for generating steam. Da2 Da

Fig. 1. Schematic illustration of an axisymmetric TES thermocline. The filler bed consists of quartzite rocks as shown in the inset (image from http://www.concretema-
terialscompany.com/products/5eighthx.jpg).
3324 Z. Yang, S.V. Garimella / Applied Energy 87 (2010) 3322–3329

c. Energy equation for the molten salt: At the bottom port

@ *
PrRe ðeUq UCpl Hl Þ þ PrRer  ðUq UCpl Hl UÞ 0
U ¼ ðd=d Þ2 ex ; Hl ¼ 0 ð5Þ
@s
1
¼ r  ðUke rHl Þ þ 2PrAReUl ½S  S0 þ trðSÞtrðS0 Þ
W At the top port
þ Ukl Nui WðHs  Hl Þ ð3Þ
*
@ U =@X ¼ 0; @ Hl =@X ¼ 0 ð6Þ
d. Energy equation for the filler bed:

@ At the wall of the tank


PrRe ½ð1  eÞXUqs UCps Hs  ¼ Ukl Nui WðHs  Hl Þ ð4Þ
@s
*
T T
where T ¼ 2S  23 trðSÞ, S ¼ rUþð2rUÞ and S0 ¼ rðU=eÞþ½2rðU=eÞ , and the U ¼ 0; @ Hl =@n ¼ ak NuW ðHW  H0 Þ ð7Þ
other terms are defined elsewhere under Nomenclature.
where Nuw = hconvd/kair, and ak is a ratio of the thermal conductivity
The non-dimensional parameters included in Eqs. (1)–(4) are
of the surrounding air (0.0242 W/m K) to the effective thermal con-
defined as follows:
ductivity for turbulent flow (at the distributor walls) or laminar
tum u @ @ eh @ x flow (at the filler bed walls) of molten salt. n is the directional unit
s¼ ; U¼ ; r ¼ ex þ er þ ; X¼ ;
h um @X @R R @h h normal to the wall and points into the tank. The parameter hconv in
2 Nuw is an effective heat transfer coefficient, which is expressed as
r um ds ph T  Tc gh
R ¼ ; Re ¼ ; P¼ ; H¼ ; Gr ¼ ; 00
hconv ¼ ðT wqT 0 Þ, where T0 is the surrounding air temperature and Tw
h mc lc u m Th  Tc mc um
pffiffiffiffi 2 is the molten-salt temperature at the tank wall. The conductive
K mc u2m hi ds h
Da ¼ ; Pr ¼ ; A ¼ ; Nui ¼ ; W¼ ; resistance of tank wall is included in hconv.
h ac Cpl;c ðT h  T c Þ kl;c ds Eq. (5) indicates that a uniform flow at the low-temperature le-
q Cps;c q l Cpl kl vel Tc is present at the bottom port, while Eq. (6) indicates that the
X ¼ s;c ; Uq ¼ l ; Ul ¼ ; UCpl ¼ ; Ukl ¼ ;
ql;c Cpl;c ql;c lc Cpl;c kl;c outflow of molten salt at the top port has no upstream influence.
k q Cps Eq. (7) represents a non-slip boundary condition on the tank wall
Uke ¼ e ; Uqs ¼ s ; UCps ¼ : for the molten-salt flow, and a heat transfer boundary.
kl;c qs;c Cps;c
At the beginning of a discharge process, the temperatures of the
The interstitial Nusselt number Nui thermally couples the molten filler bed and molten salt in the thermocline are set to a high tem-
salt with the filler bed. The parameter W represents the ratio of perature level of Th, resulting in the initial conditions:
the tank height to the diameter of the filler particles; this parameter *
is hereafter referred to as length ratio. Coefficients Uq, Ul, UCpl, and U ¼ 0; Hl ¼ 1; Hs ¼ 1 ð8Þ
Ukl, account for the temperature dependence of the density, viscos-
ity, specific heat, and thermal conductivity of the molten salt,
respectively; Uke, Uqs and UCps are the corresponding coefficients 2.3. Solution procedure
for the effective thermal conductivity of the molten salt-filler com-
bination, and the density and specific heat of the filler material, The computational domain is discretized into finite volumes.
respectively. Expressions for these coefficients, which closely match Values of all the variables at the center of the square mesh cells
the experimental data in Ref. [16], are provided in [14]. are stored. A second-order upwind scheme is used for the convec-
tive fluxes, while a central-differencing scheme is used for discret-
2.2.2. Flow in the distributors izing the diffusion fluxes, as explained in [14]. A second-order
In practice, flow distributors are generally employed to ensure a implicit scheme is used for time discretization. Pressure–velocity
uniform flow condition at the inlet/outlet of the filler region. The coupling is implemented through the PISO algorithm [19]. Itera-
characteristic length of the distributors (represented by diameter tions at each time step are terminated when the dimensionless
d or height h0 as shown in Fig. 1) is much larger than the particle residuals for all equations drop below 104. The computations
size in the filler bed. Also, the molten salt is pumped into/out of are performed using the commercial software package, FLUENT
the distributors through the ports at a velocity much larger than 6.1 [18]. User-defined functions (UDFs) are developed to account
that in the filler region because of the small cross-sectional area for Eqs. (3) and (4). Grid and time-step independence is verified
of the ports relative to the open frontal area of the filler region. by evaluating the results obtained with various grid densities
For these two reasons, a high-Reynolds-number turbulent flow is and time steps. Settings of DX = DR = 0.01 and Ds = 1.0  104
present in the distributor region; the flow in the filler bed, in con- are chosen for the work presented here, as this case resulted in a
trast, is laminar, at a much lower local Reynolds number defined temperature along the line r = 0 throughout the whole discharge
based on the particle size. The turbulent flow in the distributor re- process which differs by less than 1% relative to results from a finer
gion is modeled using the standard k–e model with a standard wall mesh and time step of DX = DY = 0.005 and Ds = 5  105. This
function [17]. Details of the model are available in [18]. At the numerical model and solution method was previously validated
interface between the distributor and the filler bed, the scalars, [14] against experimental results from a pilot-scale thermocline
i.e., pressure and temperature, and the fluxes, i.e., mass, momen- [13].
tum and energy, are all held continuous; the turbulent energy kt
and its rate of dissipation et are set in the usual way for standard 3. Results and discussion
1:5
wall functions: okt/on = 0 and et ¼ 0:090:75 kt =ð0:42DlÞ, where o/
on is the gradient in the wall normal direction and Dl the distance The discharge phase of a thermocline is considered in the pres-
from the wall to its adjacent cell center. ent work. At the start of the discharge process, the molten salt and
filler bed inside the thermocline are held at a high temperature
2.2.3. Boundary and initial conditions (Hl = Hs = 1) and a cold molten-salt inflow (Hl = 0) is initiated at
The boundary conditions are specified as: the bottom port. As a result of heating by the filler bed during its
Z. Yang, S.V. Garimella / Applied Energy 87 (2010) 3322–3329 3325

upward flow, hot molten salt (Hl = 1) is available at the top exit. As Thermal stratification is well established in the filler bed for this
cold inflow continues into the tank, the low-temperature zone in adiabatic-wall case.
the lower part of the thermocline expands and the high-tempera- When the tank wall is non-adiabatic, such as in the example for
ture zone in the upper part shrinks. a wall Nusselt number, NuW of 1.6  105 considered in Fig. 3, flow
streamlines are no longer straight lines in the filler bed: instead, a
3.1. Flow and temperature fields vortex is formed in the top-left corner of the filler bed, as shown in
the figure. The wall Nusselt number chosen here represents a wall
Typical flow streamlines and temperature distributions at dif- exposed to an effective heat transfer coefficient hconv = 100 W/m2 K
ferent stages during the discharge process of the thermocline tank for a tank of diameter 40 m and surrounding air temperature of
are presented in Fig. 2 for an adiabatic tank wall and Fig. 3 for a 25 °C (H0 = 1.125, for Th = 450 °C and Tc = 250 °C). The vortex in
non-adiabatic tank wall. When the tank wall is adiabatic, no the top-left corner of the filler bed is formed by a downward flow
heat-exchange takes place between the molten salt inside the tank adjacent to the wall, where the molten salt is cooled by the outside
and the surrounding air outside the tank. The flow streamlines are air and becomes heavier. The downward flow is large early in the
then straight and uniform in the filler bed (the region from 0 to 1 in discharge process when the driving temperature difference be-
the vertical direction) due to the presence of effective distributors tween the molten flow in the bulk of the tank and that near the
present at the top and bottom of the filler bed. In the distributor, wall is large; as the discharge process proceeds, the vortex shrinks
the molten-salt flow is turbulent and have a complex structure rel- and finally vanishes (as seen in the change in flow field from
ative to the laminar flow inside the filler bed, as shown in Fig. 2. s = 0.069 to 0.69 in Fig. 3) due to the approach of cooler flows from

Fig. 2. Flow streamlines and temperature contours in the thermocline tank (only half of the tank is shown here due to the axial symmetry) with an adiabatic boundary on the
tank wall (NuW = 0). The filler region extends from 0 to 1 in the vertical direction, with the distributors (outlined as dashed boxes) and ports shown above and below the filler
region.
3326 Z. Yang, S.V. Garimella / Applied Energy 87 (2010) 3322–3329

Fig. 3. Flow streamlines and temperature contours in the thermocline tank (only half of the tank is shown here due to the axial symmetry) with a non-adiabatic boundary at
the tank wall (NuW = 1.6  105). The filler region extends from 0 to 1 in the vertical direction, with the distributors (outlined as dashed boxes) and ports shown above and
below the filler region.

the bottom and from the side wall (the start of a low-temperature the uniform flow streamlines inside the filler bed and the absence
zone near the side wall is noted in the left-top corner at s = 0.33 in of heat transfer from the tank wall to the fluid. Even when the tank
Fig. 3). A cooler region (Hl < 0) is also seen to grow in the bottom- wall is non-adiabatic, however, a reasonable degree of thermal
left corner of the tank (s = 0.33 and 0.69 in Fig. 3). This cooler re- stratification is maintained as shown by the temperature field in
gion is a direct consequence of interaction with the cooling at Fig. 3. In thermoclines which exhibit thermal stratification, the
the wall (outside air temperature H0 = 1.125). For the tank with temperature distribution in the tank is well represented by profiles
an adiabatic wall, there is no influence of the outside air, and there- along the vertical centerline. Fig. 4 presents temperature profiles in
fore, the lowest temperature throughout the domain is that of the the filler bed along the vertical axis of the thermocline tank. The
cold molten-salt inflow, Hl = 0, as seen in Fig. 2. The non-adiabatic typical temperature profiles in an adiabatic thermocline, taking
wall also causes the high-temperature zone in Fig. 3 to shrink more the solid thick line (NuW = 0, Re = 15) at s = 0.41 as an example,
rapidly than it does in Fig. 2. For instance, the region with Hl > 0.9 consist of three zones: a constant low-temperature zone (Hl = 0),
has completely vanished at s = 0.69 in Fig. 3, while it persists in a constant high-temperature zone (Hl = 1) and a heat-exchange
Fig. 2. zone (0 < Hl < 1) [14,15]. In the constant high- and low-tempera-
ture zones, the molten-salt and filler bed are in thermal equilib-
3.2. Axial temperature profiles rium, while in the heat-exchange zone, heat is transferred from
the filler bed to the molten salt in a discharge process (the heat
When the tank wall is adiabatic, good thermal stratification is flow is in the reverse direction during a charge process). Increasing
achieved in the filler bed, as was shown in Fig. 2, as indicated by the Reynolds number moderates the slopes of the temperature
Z. Yang, S.V. Garimella / Applied Energy 87 (2010) 3322–3329 3327

Fig. 4. Temperature profiles along the vertical axis of the thermocline in the filler bed. The bottom inlet of the filler bed is at X = 0 while the outlet at the top is at X = 1.

profiles in the heat-exchange zone, as seen by comparing the solid


thick lines (NuW = 0, Re = 15) to the dashed thick lines (NuW = 0,
Re = 240) in Fig. 4. The heat-exchange zone is thus extended at
higher Reynolds numbers, shortening the constant high-tempera-
ture zone and ultimately reducing the discharge efficiency [14].
The temperature profiles under non-adiabatic conditions are
similar in nature to those for adiabatic thermoclines, except that
the entire profile shifts to lower temperature values as time pro-
gresses. For example, the temperature of the constant low-temper-
ature zone for profiles at NuW = 1.6  105 and Re = 15 (solid thin
lines in Fig. 4) is 0.05 at s = 0.14 and decreases to 0.1 by
s = 0.69. Towards the top of the thermocline, the cooler molten-salt
near the wall flows towards the top port and towards the tank axis
(see Fig. 3 at s = 0.33), with a detrimental effect on the constant
high-temperature zone of the axial temperature profiles, as clearly
illustrated in the profiles (NuW = 1.6  105 and Re = 15) of Fig. 4. In
the lower portion of the thermocline, in contrast, the molten-salt
flow direction is from the axis towards the side wall (Fig. 3) which
leads to a smaller influence of the cold wall on the axial tempera-
ture profile. The constant low-temperature zone, therefore, is lar-
gely unchanged by the presence of the cold wall as evident from
the temperature profiles (NuW = 1.6  105 and Re = 15) in Fig. 4, ex-
Fig. 5. Temperature histories of molten-salt outflows at the top port of the tank
cept that the magnitude of the temperature shifts to lower values during a discharge process. Solid lines are for a non-adiabatic wall (NuW = 1.6  105)
as time progresses. This shift to lower temperatures is a direct re- while dashed line indicate an adiabatic wall (NuW = 0).
sult of the heat loss from the lower distributor region.
Interestingly, an increase in the Reynolds number significantly
reduces the influence of the cold wall. For example, the tempera-
ture profiles for NuW = 1.6  105 at the higher Reynolds number maintain a high temperature level for some initial part of the dis-
of Re = 240 (thin dashed lines in Fig. 4) are quite comparable to charge process and then drop rapidly in value. In an adiabatic-wall
those at NuW = 0 (thick dashed lines), and the temperature differ- thermocline, the temperature of the outflow is maintained at a
ence between the profiles at s = 0.69 is only about 0.05 in the constant high level for much of the discharge process, dropping
heat-exchange zone. This is because the discharge period is greatly only when the heat-exchange zone approaches the top port of
reduced at a higher Reynolds number (higher salt flow rate), the tank [14]. The molten-salt flow at a low Reynolds number,
resulting in an insufficient time for significant heat loss at the tank e.g., 15, shows a long constant-high-temperature period while that
wall. at a high Reynolds number, e.g., 240, has a shortened period of high
temperature discharge. The extent of the heat-exchange zone in
3.3. Outflow temperature history the vertical direction is small at a small Reynolds number and, cor-
respondingly, the constant-high-temperature zone is long [14], as
It is instructive to examine the temperature of molten-salt out- also shown by a comparison between the profiles for Re = 15 and
flow from the top port of the thermocline. Fig. 5 shows the temper- 240 at NuW = 0 in Fig. 4.
ature histories under adiabatic and non-adiabatic tank wall In a thermocline with a non-adiabatic wall, the high-tempera-
conditions for different Reynolds numbers. In general, all profiles ture period of outflow is no longer held at a constant level. At
3328 Z. Yang, S.V. Garimella / Applied Energy 87 (2010) 3322–3329

the beginning of the discharge process, the temperature of outflow


first drops gradually, and then more rapidly. The gradual decrease
in the early part of the discharge period is caused by the heat loss
at the tank wall, while the rapid decrease later in the process is due
to the arrival of the heat-exchange zone at the top port. The fluctu-
ations in temperature, such as those visible in the solid lines at
Re = 15 and 30, are most likely induced by the recirculation cells
of cold molten salt exiting the top port which are caused by the
heat loss at the top wall of the thermocline. These recirculation
patterns are demonstrated in animations created from the com-
puted results and provided as supplementary material. Molten salt
is seen to be cooled by the top wall and flows downward to form
local vortices; these vortices are periodically flushed by the main
flow stream out of the top port and thus cause fluctuations in
the outflow temperature.
The influence of Reynolds number shows opposing trends in the
cases of the adiabatic and non-adiabatic thermoclines. In an adia-
batic thermocline, increasing the Reynolds number reduces the
constant high-temperature period and the drop in temperature to-
wards the end of the discharge process occurs earlier as previously
discussed. However, in a non-adiabatic thermocline, the period of
high-temperature outflow increases with increasing Reynolds
Fig. 6. Discharge efficiency g as a function of Reynolds number under different wall
number, as shown by the solid lines in Fig. 5. For instance, the per-
heat transfer rates represented by NuW.
iod with an outflow temperature higher than 0.9 occurs over
0 < s < 0.66 at Re = 240, while it is reduced to 0 < s < 0.3 at Z Th
Re = 15. A longer period of high-temperature outflow indicates that El ¼ ml Cpl dT
Tc
a larger portion of the thermal energy stored in the thermocline
can be retrieved as high-temperature energy that can be used for where Es is the energy initially stored in the filler material, El is the
generating steam. In a non-adiabatic thermocline, since the mol- energy stored in the molten salt, ms is the filler mass, and ml is the
ten-salt flow rate decreases with decreasing Reynolds number, molten salt mass initially stored in the tank.
the discharge time (dimensional) increases, in turn increasing the From Eq. (9), it is seen that thermal energy retrieved at a tem-
energy loss at the tank wall and causing a general drop in temper- perature above Ht contributes the discharge efficiency.
ature (NuW = 1.6  105 in Fig. 4). This not only decreases the out- Fig. 6 shows the discharge efficiency for thermoclines with dif-
flow temperature but also causes a more rapid decrease in the ferent environmental boundary conditions with respect to heat
outflow temperature at small Reynolds numbers (Re = 15 and loss at the tank wall. In an adiabatic thermocline (NuW = 0), the dis-
NuW = 1.6  105 in Fig. 5). On the contrary, increasing the Reynolds charge efficiency decreases as the Reynolds number increases,
number shortens the (dimensional) discharge time and minimizes which is caused by the expanded heat-exchange zone at large
the influence of heat loss at the tank wall, which renders the out- Reynolds numbers [14]. In a non-adiabatic thermocline, the dis-
flow temperature history to be more similar to that for an adiabatic charge efficiency increases with Reynolds number in contrast to
thermocline, as evident from the proximity of the solid and dashed the behavior of an adiabatic thermocline, as discussed earlier. It
lines at Re = 240 in Fig. 5. is interesting to note that at a modest wall Nusselt number such
as NuW = 1.6  104 (corresponding to an effective heat transfer
3.4. Discharge efficiency coefficient hconv = 10 W/m2 K for a tank of diameter 40 m), the dis-
charge efficiency first increases and then decreases as the Reynolds
During the discharge process, only the thermal energy retrieved number increases. The initial increase indicates that the increased
above a certain temperature level is capable of generating super- discharge time (dimensional) has a dominant influence on the dis-
heating steam, and therefore is termed ‘‘useful” energy hereafter. charge efficiency, while the subsequent decrease shows that the
In order to study the capability of a thermocline in providing useful expansion of the heat-exchange zone caused by the increase in
energy, a discharge efficiency is defined as follows [14]: Reynolds number has a more important effect on the efficiency.
As the Reynolds number increases to Re = 240 (Fig. 6), the dis-
Output energy with Hl > Ht charge time is severely reduced, which minimizes the effects of
g¼ ð9Þ
Total energy initially stored in the thermocline tank the heat loss on the tank wall and causes the values of efficiency
where Ht is a threshold value determined by the application, and is at the two non-adiabatic boundaries to approach that for an adia-
chosen as 0.9 in this study. The output energy is calculated by batic thermocline, as evident form the proximity of the three lines
Z at Re = 250 in Fig. 6.
T
Eflow ¼ mflow Cpl dT T > Ht ðT h  T c Þ þ T c
Tc
4. Conclusions
where Eflow is the energy output by the molten salt with
T > Ht ðT h  T c Þ þ T c and mflow is the molten-salt outflow mass at The discharge process of thermoclines for thermal energy stor-
temperature T. The initial total energy consists of that in the filler age subject to adiabatic and non-adiabatic wall boundaries is
material and in the molten salt in the pore region, calculated investigated using a comprehensive computational heat transfer
respectively using model.
Z The flow field in adiabatic thermoclines is characterized by uni-
Th
Es ¼ ms Cps dT formly distributed, straight lines, whereas that in non-adiabatic
Tc thermoclines shows distorted streamlines influenced by the
Z. Yang, S.V. Garimella / Applied Energy 87 (2010) 3322–3329 3329

downward cold flow near the tank wall. The presence of the flow [4] Sutthivirode K, Namprakai P, Roonprasang N. A new version of a solar water
heating system coupled with a solar water pump. Appl Energy
distributors above and below the filler bed continues to ensure
2009;86:1423–30.
thermal stratification for both heat transfer boundaries (for [5] Cruz JMS, Hammond GP, Reis AJPS. Thermal performance of a trapezoidal-
NuW < 1.6  105). shaped solar collector/energy store. Appl Energy 2002;73:195–212.
Overall temperatures are lower in non-adiabatic thermoclines [6] Pacheco JE, Gilbert R. Overview of recent results of the solar two test and
evaluations program. In: Hogan R, Kim Y, Kleis S, O’Neal D, Tanaka T, editors.
as expected. The differences in temperature profiles between adia- Renewable and advanced energy systems for the 21st century. NewYork: ASME
batic and non-adiabatic thermoclines are greatly reduced as the Int.; 1999.
molten-salt flowrate is increased. At small Reynolds numbers, the [7] Price H. A parabolic trough solar power plant simulation model; 2003. <http://
www.nrel.gov/docs/fy03osti/33209.pdf>.
outflow temperature during the discharge process is severely de- [8] Kearney and Associates. Engineering evaluation of a molten salt HTF in a
creased in non-adiabatic thermoclines, compared to that in adia- parabolic through solar field, NREL Contract No. NAA-1-30441-04; 2001.
batic ones. <http://www.nrel.gov/csp/troughnet/pdfs/ulf_herrmann_salt.pdf>.
[9] Kearney D, Herrmann U, Nava P. Assessment of a molten salt heat transfer fluid
At a large wall Nusselt number (NuW = 1.6  105), the discharge in a parabolic through solar field. J Sol Energy – Trans ASME 2003;125:170–6.
efficiency of thermoclines increases with increasing Reynolds [10] Price H. A parabolic trough solar power plant simulation model. <http://
number, in contrast to the opposite effect of Reynolds number on www.nrel.gov/docs/fy03osti/33209.pdf>.
[11] Kearney D. Engineering evaluation of a molten salt HTF in a parabolic through
adiabatic thermoclines (NuW = 0). At an intermediate wall Nusselt solar field, NREL Contract No. NAA-1-30441-04; 2001. <http://www.nrel.gov/
number (NuW = 1.6  104) the discharge efficiency first increases csp/troughnet/pdfs/ulf_herrmann_salt.pdf>.
and then decreases as the Reynolds number increases, indicating [12] Brosseau D, Kelton JW, Ray D, Edgar M. Testing of thermocline filler materials
and molten-salt heat transfer fluids for thermal energy storage systems in
a discharge process first dominated by heat loss at the tank walls
parabolic trough power plants. J Sol Energy – Trans ASME 2005;127:109–16.
at small Reynolds numbers and then by the expansion of the [13] Pacheco JE, Showalter SK, Kolb WJ. Development of a molten-salt thermocline
heat-exchange zone at large Reynolds numbers. thermal storage system for parabolic trough plants. J Sol Energy – Trans ASME
2002;124:153–9.
[14] Yang Z, Garimella SV. Comprehensive analysis of solar thermal energy storage
Appendix A. Supplementary data in a molten-salt thermocline. Sol Energy 2010;84:974–85.
[15] HITEC heat transfer salt. Costal Chemical Co., L.L.C., Brenntag Company.
<http://www.coastalchem.com/>.
Supplementary data associated with this article can be found in [16] Wakao N, Kaguei S. Heat and mass transfer in packed beds. New York: Gordon
the online version, at doi:10.1016/j.apenergy.2010.04.024. and Beach; 1982.
[17] Launder BE, Spalding DB. Lectures in mathematical models of
turbulence. London, England: Academic Press; 1972.
References
[18] FLUENT 6.1 Documentation. <http://www.fluent.com/>.
[19] Issa RI. Solution of implicitly discretized fluid flow equations by operator
[1] Xiao W, Wang X, Zhang YP. Analytical optimization of interior PCM for energy splitting. J Comput Phys 1986;62:40–65.
storage in a lightweight passive solar room. Appl Energy 2009;86:2013–8. [20] Krishnan S, Murthy JY, Garimella SV. A two-temperature model for analysis of
[2] Kuznik F, Virgone J. Experimental assessment of a phase change material for passive thermal control systems. ASME J Heat Transfer 2004;126:628–37.
wall building use. Appl Energy 2009;86:2038–46. [21] Beckermann C, Viskanta R. Natural convection solid/liquid phase change in
[3] Garnier C, Currie J, Muneer T. Integrated collector storage solar water heater: porous media. Int J Heat Mass Transfer 1998;31:35–46.
temperature stratification. Appl Energy 2009;86:1465–9.

You might also like