Available online at www.sciencedirect.

com

ScienceDirect
Solar Energy 105 (2014) 71–81
www.elsevier.com/locate/solener

Verification of a model of thermal storage incorporated

with an extended lumped capacitance method for various
solid–fluid structural combinations
Peiwen Li a,⇑, Ben Xu a, Jingxiao Han a,b, Yongping Yang b
a
Department of Aerospace and Mechanical Engineering, The University of Arizona, Tucson, AZ 85721, USA
b
School of Energy, Power and Mechanical Engineering, North China Electric Power University, Chang Ping District, Beijing 102206, China

Received 21 August 2013; received in revised form 18 February 2014; accepted 28 March 2014
Available online 4 May 2014

Communicated by: Associate Editor Halime Paksoy

Abstract

With an extended lumped capacitance method applied to account for the internal heat conduction resistance in a solid (for Biot num-
ber larger than 0.1), a general model of thermal energy storage with various solid–fluid structural combinations is presented and verified
using numerical results. The thermal energy storage system has a heat transfer fluid (HTF) flowing through a packed bed of solid mate-
rials structured in different configurations, such as in the form of solid pebbles, parallel plates, solid rod-bundles, or solids with fluid
tubes imbedded through them. The model of energy conservation in the liquid and solid is transient, one-dimensional in nature, due
to the introduction of a modified lumped capacitance method that counts for the effect of three-dimensional heat conduction in the solid
structures. The computational workload using this modified model is significantly less compared to that of a comprehensive CFD anal-
ysis. Numerical results obtained from a CFD analysis of the thermal energy storage in the solid and liquid are used to verify the model.
The CFD simulated results of temperatures of HTF are compared with the 1D model results, and they show excellent agreement. In
conclusion, the 1D model is recommended as a convenient and accurate tool for general analysis and sizing of thermal energy storage
containers that have various solid–fluid structural combinations.
Ó 2014 Published by Elsevier Ltd.

Keywords: Thermal energy storage; Extended lumped capacitance; Effective heat transfer coefficient; CFD results; Verification

1. Introduction past decade (Pavlović et al., 2012). With thermal energy

The recent fast-paced development of solar energy tech-
nologies has imposed a great demand for large-quantity
energy storage. With the storage of a large amount of solar
thermal energy, concentrated solar thermal power (CSP)
generation is expected to contribute to the world’s energy
supply significantly in the future. Technical subjects related
to CSP have already drawn a lot of attention during the
⇑ Corresponding author.
E-mail address: peiwen@email.arizona.edu (P. Li).
storage, most CSP systems can smooth out the short-term
transients (e.g., collector shading from passing clouds) and
extend the daily operation of solar power plants during the
late afternoon and evening hours at a relatively low cost of
energy storage (Müller-Steinhagen and Trieb, 2004). Ther-
mal energy storage in CSP systems is the key element that
will largely guarantee achievement of the target set by the
U.S. Department of Energy (DOE) –that by 2030, retail
electricity rates will be as low as 6 cents/kilowatt-hour
(kW h) (SunShot Vision Study, 2012). A CSP system uses
a solar tower, parabolic troughs, or linear Fresnel lenses

http://dx.doi.org/10.1016/j.solener.2014.03.038
0038-092X/Ó 2014 Published by Elsevier Ltd.
72 P. Li et al. / Solar Energy 105 (2014) 71–81

to concentrate sunlight and produce intense heat to trans- R

fer to heat transfer fluid (HTF) that carries heat for
thermal energy storage as well as for thermal cycles in con-
ventional power plants (Py et al., 2013). r
H
After its temperature rise, the HTF from a concentrated
solar collection system can be used directly for thermal
energy storage, or can be sent to a solid-packed thermal
storage system, which is a dual-media system. The dual-
media thermal storage system can significantly reduce the
storage cost by introducing low-cost solid thermal-storage (a) Packed solid spherical particles with HTF passing around
materials, but sacrifices efficiency. The heat exchange
between the HTF and solid thermal energy storage medium z
x2
during both energy charge and discharge processes causes a x1
loss in energy storage efficiency. Specifically, a degradation
of the temperature of the HTF typically occurs as it is dis-
charged. In order to better understand and design the size
of dual-media thermal energy storage systems as shown in H
Fig. 1, the heat transfer or energy exchange between HTF
and a solid medium has to be analyzed thoroughly (Zhang
et al., 2013; Zanganeh et al., 2012).
A convenient and effective analytical tool was provided x

by Li et al. (2011a,b) and Van Lew et al. (2009) who devel- 2Δx
oped a numerical model to solve a group of simplified 1D (b) Packed solid flat plates with HTF passing through
transient energy equations proposed by Schumann (1929), the channels in a tank
describing the heat transfer and energy storage/extraction
between the HTF and the packed-bed solid material. The z r
numerical model is robust and efficient in carrying out a
great number of calculations (for a large number of charg-
ing and discharging cycles) for sizing the thermal storage
tanks of various levels of energy storage demand. The
Schumann equations were constructed without considering R
H
the internal heat conduction resistance in the solid thermal
storage material, which is commonly known as the lumped
capacitance assumption, and are only accurate when the
Biot number is sufficiently small, typically less than 0.1. r
In practical applications, where the characteristic scale
of the solid thermal storage material is large, the internal (c) Packed solid cylinders with HTF passing along in a tank
thermal resistance becomes significant. If one still wants
to take the advantage of the 1D model represented by
z
the Schumann equations and the corresponding numerical
method of solution as mentioned above, a correction to the
lumped capacitance method is inevitable.
Earlier in the 1970s, Bradshaw et al. (1970) and Jeffreson
(1972) conducted a pioneering work to extend the use of
the lumped capacitance method for spherical solid material H
to accommodate the circumstances when the temperature
distribution within the solid material cannot be ignored
(at Biot number larger than 0.1). The key idea of their work
is the introduction of a corrected, or an effective heat trans- r
fer coefficient, heff, to replace the intrinsic heat transfer
a b
coefficient h between the solid material and the HTF. A
very important extension of this method has been con- (d) Solid thermal storage material with HTF tubes
ducted by Xu et al. (2012) for solid objects like flat plates, passing through in a tank
infinite long cylinders, and infinite long tubes for Biot Fig. 1. Four typical solid–fluid structural combinations of thermal storage
numbers up to 100 with sufficient accuracy. Combining systems ( solid thermal storage material; HTF).
 P. Li et al. / Solar Energy 105 (2014) 71–81
Table 1
The effective heat transfer coefficient of solid filler material of different structures (Xu et al., 2012).
Sphere
Plate
Cylinder
Tube with HTF inside, and outer boundary insulated
the corrections by Bradshaw et al. (1970) and Jeffreson
(1972), as well as by Xu et al. (2012), the heat transfer of
all typical shaped solid objects (as shown in Fig. 1) with
heat transfer fluid can be considered. For convenient refer-
ence, Table 1 lists the formulas of the effective heat transfer
coefficient heff, on the basis of the intrinsic heat transfer
coefficients h, for the four fluid–solid structural combina-
tions shown in Fig. 1.
It is important to note that the four typical configura-
tions of dual-media thermal energy storage system illus-
trated in Fig. 1 can be generally described as a system of
HTF flowing through a porous medium, where the physical
relationship between HTF and the porous thermal storage
material becomes a general phenomenon. In fact, the so-
called solid medium can also be another fluid for energy
storage, as long as it is separated from the HTF. In practi-
cal application, a large-sized solid material structure can
make the Biot number for the heat conduction in the solid
material above 0.1. Therefore, one has to consider inclu-
sion of the effective heat transfer coefficient heff, so that
the Schumann equations still might be used to obtain accu-
rate solutions to the problem.
Before recommending incorporation of the effective heat
transfer coefficient heff into the 1D transient model
(described by Schumann equations) for general and accu-
rate analysis of the four types of thermal energy storage
scenarios, a validation by experimental data or verification
using data from a comprehensive numerical analysis has to
be conducted (ASME V&V, 2009; Oberkampf and
Trucano, 2002). For thermal storage with solid spherical
pebbles, the 1D transient model with the effective heat
transfer coefficient heff has been compared to and validated
by Valmiki et al. (2012) with experimental data. In the cur-
rent work, a comprehensive CFD-based analysis of the
heat transfer and energy storage of the other three solid–
fluid structural combinations will be conducted and the
numerical results will be used to verify the 1D transient
model in which the Schumann equations and effective heat
transfer coefficient heff are generally applied.
2. The 1D transient mathematical modeling for general
thermal storage systems
All four solid–fluid configurations in Fig. 1 are generally
viewed as systems with HTF flowing through a porous
medium, each should have porosity, or equivalent porosity
73
Characteristic length for Biot number Effective heat transfer coefficient heff
1
r/3 1 r
hþ5k r
1
Dx = x2  x1 1 Dz
hþ3k r
See Fig. 1 (b) for definition of x1, x2
1
r/2 1þ r
h 4k r
2 2 1
(b  a )/2a 2 4
1 1 a3 ð4b a2 Þþab ½4Lnðb=aÞ3
hþk r
See Fig. 1(d) for definition of a, b 4ðb2 a2 Þ
2
for cases in Fig. 1(b–d), defined as the ratio of the volume
of fluid over the total volume of the storage tank. For the
case of packed spherical pebbles shown in Fig. 1(a), Van
Lew et al. (2011) gave the governing equations of the
energy balance for the fluid and solid as
@T f @T f hS filler
þU ¼ ðT LM  T f Þ ð1Þ
@t @z qf C f epR2
@T LM hS filler
¼ ðT LM  T f Þ ð2Þ
@t qr C r ð1  eÞpR2
where Tf denotes the fluid temperature; TLM denotes a
lumped temperature of the solid particles; h is the intrinsic
heat transfer coefficient between the solid filler material and
HTF; qf and Cf are the density and specific heat of HTF,
respectively; qr and Cr are the density and specific heat
for solid (typically rocks) filler material; R is the radius
of the storage tank; and e is the porosity. The hot HTF
flows downward through the packed bed during heat
charging, and the cold HTF flows upward during heat dis-
charging. The fluid inlet (at the top during charging and at
the bottom during discharging) always has z = 0. The flow
velocities are assumed to have a uniform radial distribu-
tion, which Yang and Garimella (2010) have proven to
be reasonable for packed beds. Other assumptions include:
no heat conduction between the spheres/particles, negligent
axial heat conduction in HTF, and no heat loss from the
walls of the tank.
In cases where the solid material is not in the form of
spheres packed as shown in Fig. 1(a) but in a form such
as those shown in Fig. 1(b–d), equivalent porosity must
be used. With equivalent porosity, the same governing
equations can be used, however, an effective heat transfer
coefficient heff, as given in Table 1, must be used to replace
the original (or intrinsic) heat transfer coefficient h in Eqs.
(1) and (2). This treatment is an innovative approach pro-
posed by the authors and will be verified in the following
work.
The parameter Sfiller in the governing equations, Eqs. (1)
and (2), is the total surface area of solid thermal storage
material in contact with HTF per unit length of the thermal
storage tank (see Li et al., 2011a for details on obtaining
this parameter). As a consequence, the heat transfer and
energy balance in the solid–fluid structural combinations
in Fig. 1(b–d) can all be analyzed by only considering
one typical volume, which includes a typical solid and fluid
74 P. Li et al. / Solar Energy 105 (2014) 71–81

Table 2 The method of characteristics is applied to solve the

The total heat transfer surface area of a solid per unit height of a typical dimensionless equations presented above. For details of
volume as shown in Fig. 1(b)–(d).
the process of an effective numerical solution to the Schu-
Total heat transfer surface area (m2) Height (m) Sfiller (m) mann equations, one can refer to Van Lew et al. (2011)
Plate 2(1  H) H 2 and Karaki et al. (2010).
Cylinder 2prH H 2pr
Tube 2paH H 2pa
3.2. Initial and boundary conditions

region as indicated on the right-side of each of the config- Under the assumption of no heat loss from the thermal
urations in Fig. 1. For the case of plates, the length in the storage tank, it is reasonable that the equilibrium tempera-
direction normal to the paper in Fig. 1(b) is chosen to be a ture between HTF and the solid filler material at the end of
unit length of L ¼ 1:0 m. As a result, the flow channel will one charge or discharge will necessarily be the initial condi-
have a ratio of L/Df > 20, where Df is the width of the flow tion of the next discharge or change process in the thermal
channel. Table 2 gives the obtained Sfiller for the other three storage cycle. This connects the discharge and charge pro-
cases in Fig. 1. cesses so that results of a number of periodic charges and
discharges can be obtained.
3. Incorporating the extended lumped capacitance method in For the initial condition of fluid and filler material in
the model any charging or discharging process, t* = 0; hr = hf, which
is the equilibrium state after settling down from the last
In this section, dimensionless governing equations will process.
be introduced in a standard format. The extended lumped For the inlet condition, z* = 0 and hf = 1 for a charging
capacitance method will also be incorporated. The solu- process; otherwise hf = 0 for a discharging process.
tions of the temperature distribution of a thermal storage The hr at the inlet boundary can be directly calculated
tank for different solid–fluid structural combinations will using Eq. (6) from the known inlet fluid temperature.
be provided in the following section.
3.3. Operation conditions for current simulations
3.1. Dimensionless governing equations introduced with
corrected heat transfer coefficient A pair of heat transfer fluid and thermal storage mate-
rial has been chosen for the 1D model as well as the
For the general situation, the heat transfer coefficient h CFD study for the purpose of comparison and verification.
in Schumann equations (Eqs. (1) and (2)) must be replaced The high temperature of fluid in charge process is 390 °C
by the corrected heat transfer coefficient heff. (663.15 K) and the low temperature of fluid flowing into
Considering the following dimensionless variables, the tank during a discharge is 310 °C (583.15 K). The heat
hf ¼ ðT f  T L Þ=ðT H  T L Þ ð3:aÞ transfer fluid is Hitec molten salt (Becker, 1980), having
properties of kinetic viscosity mf ¼ 1:17  106 m2 =s, heat
hr ¼ ðT LM  T L Þ=ðT H  T L Þ ð3:bÞ capacity Cpf = 1549.12 J/kg K, density qf = 1794.07 kg/

z ¼ z=H ð3:cÞ m3, and thermal conductivity kf = 0.57 W/m K. The
 thermal storage material is another molten salt with
t ¼ t=ðH =U Þ ð3:dÞ
properties of qr ¼ 1680 kg=m3 , C r ¼ 1560 J=ðkg KÞ, and
the modified dimensionless Schumann equation for HTF is k r ¼ 0:61 W=ðm KÞ. The required time period for energy
obtained as charge and discharge is 4 h each.
The three cases of solid–fluid structural combinations,
@hf @hf 1
þ ¼ ðhr  hf Þ ð4Þ as shown in Fig. 1(b–d), were calculated using the method
@t @z sr
of characteristics based on the 1D model. For the conve-
where nience of defining the dimensions of the solid and liquid
area, as well as the computational domain for CFD analy-
U qf C f epR2 sis, an equivalent control volume was defined. For the
sr ¼ ð5Þ
H heff S filler bundle of solid rods or the fluid tubes in Fig. 1(c) and
Furthermore, the modified dimensionless Schumann (d), we may consider that each rod or tube has a control
equation for filler material is in the form area enclosed by a hexagon as shown in Fig. 2. The outer
boundary surface is thermally insulated. In the CFD
@hr H CR analysis, the hexagon is equivalent to a circle in the same

¼ ðhr  hf Þ ð6Þ
@t sr area. A comprehensive 3D simulation using the commer-
where cial software package ANSYS Fluent6.3@ proved that
the computational results of energy storage for the control
qf C f e volume in Fig. 2(a) agree with that from the equivalent
H CR ¼ ð7Þ
qr C r ð1  eÞ control volume in Fig. 2(b) (the results and comparison
 P. Li et al. / Solar Energy 105 (2014) 71–81 75

R eq can be easily satisfied due to the very long channels. The

R assumption of a constant wall heat flux heat transfer is
r ilar to that of a countercurrent flow heat exchanger—hot
r
(a) (b)
Fig. 2. (a) Hexagonal control volume. (b) Equivalent circular control
volume.
are presented in the next section). This is a significant con-
venience as a 2D axisymmetric solution can be simply
applied to the investigated geometries.
As shown in Fig. 2, the equivalent radius for the control
area should satisfy the relationship
Ahexagon ¼ Aeq ¼ pR2eq
The equivalent diameters of the control area for the
solid rod and the fluid tube are given in Table 3, which
essentially define the dimensions of flow channels and solid
thermal storage materials. The height of the tank is 10 m.
The related intrinsic heat transfer coefficient of laminar
flow and the corrected heat transfer coefficient in the 1D
model simulations are listed in Table 4.
The laminar flow heat transfer Nusselt numbers listed in
Table 4 are based on heat transfer cases with constant wall
heat flux (Kays and Crawford, 1993) for fully developed
internal flows. The full thermal and hydraulic development
because the temperature difference between the solid mate-
rials and the fluid along the height of a storage tank is sim-
fluid charges into the tank from the top, where there is a
high temperature, and during a discharge cold fluid flows
into the tank from bottom of a tank, where there is a
low temperature. A detailed description to this variation
of temperatures of fluid and solid during energy charge
and discharge is given by Li et al. (2012). The Biot numbers
of all cases of heat conduction in solid materials in this
analysis are above 1.0.
The charging and discharging cyclic operations start
with a charge to a cold tank. After several cyclic runs,
the temperature distribution in a tank after a discharge
becomes independent of the initial temperature distribu-
tion. This state is called a cyclic steady state, and it is the
real situation in a solar power plant practical operation.
4. Simulation results from the 1D model
The simulation results using the 1D transient model,
incorporated with the corrected lumped capacitance
ð8Þ method, are presented in this section. The velocity of
HTF flowing into the tank for the following cases is
V inlet ¼ 1:36  103 m=s.
The average fluid temperatures at height locations along
the tank after a charging are shown in Fig. 3 for the num-
ber of cycles of charging/discharging as indicated by the
legend. The results become the same with no more changes
in charge/discharge cycles after a certain number of cycles
for all the cases. For the case of channels formed by flat
plates, 11 cycles are required to reach the cyclic steady
state; this number for the cylinder case and tube case is
10 and 7, respectively. Because the 1D transient model is
rather convenient, it can simulate many cycles of

Table 3
Dimensions of the fluid channels and solid thermal storage structure.
Equivalent Equivalent outer diameter Deq Diameter d Sr Fluid channela 2x1 Solid platea 2(x2  x1)
porosity e (mm) (mm) (mm) (mm)
Plate case 0.33 No No 2.0 4.56 9.25
Cylinder case 0.33 43.5 35.6 0.1118 No No
Tube case 0.33 43.5 25.0 0.0785 No No
a
See definition of x1, x2 in Fig. 1(b).

Table 4
Parameters in the simulation of 1D transient model.
Hydraulic diameter for Nu number (constant heat Intrinsic heat transfer coefficient Effective heat transfer coefficient sr HCR
Re and Nu flux) h(W/m2 K) heff(W/m2 K)
Plate case (Dh = 4x1) 8.24 305 221 0.0039 0.5223
Cylinder case 5.18 (inner wall) (outer wall 352 99 0.0087 0.5223
(Dh = Deq  d) no heat flux)
Tube case (Dh = d) 4.36 94 64 0.0366 0.5223
76 P. Li et al. / Solar Energy 105 (2014) 71–81

velocities of HTF flow will also be compared to CFD

results. Details of the CFD work are presented in the fol-
lowing sections.

5. CFD analysis and comparison of results from 1D transient

modeling

A comprehensive CFD study was employed to analyze

the energy storage process and verify the 1D simplified
transient model. The commercial software ANSYS
Fluent@ 6.3 was chosen for this analysis, and GAMBIT
2.4 was used to generate the computational domain and
grid system.
(a) Flat plate case
5.1. Computational specifications

The flow and heat transfer in the thermal storage tank

are incompressible, and transient with constant properties.
The fluid flow Reynolds numbers are usually in the laminar
region. In the current three configurations of solid–fluid in
storage tanks, the velocity of HTF all equal V inlet ¼ 1:36
103 m=s, according to the desired mass flow rate. Corre-
spondingly, the Reynolds numbers are: Replate = 10.6,
Retube = 28.9, and Recylinder = 9.2. The hydraulic diameters
defined in Table 4 for the flow channels were used to calcu-
late these Reynolds numbers.
The continuity, momentum, and energy equations in
differential form for laminar and incompressible flow are:
(b) Cylinder case r  V~ ¼ 0 ð9Þ

@ V~ ~ ¼  1 rp þ mr2 V~
þ ðV~  rÞV ð10Þ
@t q
@T
þ ðV~  rÞT ¼ kr2 T ð11Þ
@t
Because of the thermocline effect (hot fluid on top of
cold fluid) in thermal storage tanks, typically there is no
natural convection to consider. The computational
domains and boundary conditions of the three cases are
described in the following:

(1) For the solid plates and the 2D channels, as shown in

Fig. 1(b), half of the flow channel and half of the
(c) Tube case plate were included in the computational domain.
At the symmetric line of the flow channel, ux = 0,
Fig. 3. The temperature of HTF at locations from bottom to top in a tank @uz
¼ 0 , and @T ¼ 0. At the symmetric line of the solid
@x @x
after heat charging processes.
plate, uz = 0, ux = 0, and @T
@x
¼ 0. At the solid walls of
charging/discharging in a relatively short computational z = 0 and =H, uz = 0, uz = 0, ux = 0, and @T @z
¼ 0.
time. This is the advantage of the 1D transient model over Fully development conditions at the outflow bound-
a CFD analysis, which would take a significant amount of ary are used, as the flow channels are sufficiently long.
computational time for the analysis of cyclic charge and This gives ux = 0, @u
@z
z
¼ 0, @T
@z
¼ 0.
discharges. (2) For the solid rod with fluid flowing along the length,
Nevertheless, the results from the 1D transient modeling as shown in Fig. 1(c), the rod and the equivalent
will be compared with the results from a comprehensive circular area around the rod are included in the
CFD analysis in order to fully verify the model. More computational domain. The boundary conditions
results from the 1D transient modeling with different include: at the centerline of the rod, uz = 0, ur = 0,
 P. Li et al. / Solar Energy 105 (2014) 71–81
and @T
@r
¼ 0. At the outer boundary, in the r direction,
ur = 0, @u z
¼ 0, and @T ¼ 0. At the solid walls of z = 0
@r @r
and z = H, uz = 0, ur = 0, and @T @z
¼ 0. At the outflow
boundary, fully developed boundary conditions are:
ur = 0, @u z @t
¼ 0, and @z ¼ 0.
@z
(3) For the tubes surrounded by solid areas, as shown in
Fig. 1(d), a tube and the equivalent circular solid area
around the tube are included in the computational
uz
domain. At the centerline of the tube, ur = 0, @r ¼ 0,
@T
and @r ¼ 0. At the outer boundary, in the r direction,
uz = 0, ur = 0 and @T@r
¼ 0. At the solid walls of z = 0
and z = H, uz = 0, ur = 0 and @T @z
¼ 0. At the outflow
boundary, fully developed boundary conditions are:
ur = 0, @u z
¼ 0, and @T ¼ 0. The fluid and solid inter-
@z @z
faces inside the computational domain have conju-
gated heat transfer, which can be typically treated
in the software package ANSYS Fluent@.
5.2. Computational procedures
To verify the reasonable simplification of the hexagonal
control volume into a circular control volume as shown in
Fig. 2, a 3D comprehensive CFD analysis for the heat
transfer and thermal storage was conducted. For this anal-
ysis, the thermal storage tank is assumed to have been fully
charged to a high temperature of 663.15 K, and the cold
HTF with constant temperature 583.15 K flows into the
tank with a constant velocity of V inlet ¼ 1:36  103 m=s
to extract the heat for 4 h. The average temperature of
HTF at 1.0 m downstream of the inlet has been chosen
for comparison between the two computational domains.
As shown in Fig. 4, the results of temperature versus time
in the simulated period match very well between the two
cases. This firmly supports that the hexagonal domain
can be simplified to an equivalent circular domain with suf-
ficient accuracy and, therefore, the CFD analysis can use a
Fig. 4. Comparison of the average HTF temperature at 1.0 m down-
stream of inlet for hexagonal control volume and its equivalent circular
control volume, as shown in Fig. 2.
77
2D axisymmetric coordinate to study the flow and energy
storage.
5.3. Grid and time step independent study
A grid-independent study was conducted to choose a
grid number that ensured high accuracy of computational
results at a reasonable computational load. The flow and
heat transfer fields were computed in 2D with cell numbers
of 10000, 20,000, 40,000, 60,000, and 80,000. For this
study, we also assume the initial condition that the thermal
storage tank is fully charged to a high temperature of
663.15 K, and the cold HTF with constant temperature
583.15 K flows into the tank with a constant velocity of
V inlet ¼ 1:36  103 m=s to extract the heat for 4 h. The
temperature of HTF at 1.0 m downstream of the inlet
was examined for the transient process. It was found that
the relative difference of the temperature at this location
and any time instance during 4 h has a variation of no
more than 10% when increasing the cell number from
10,000 to 20,000. The variation went down to 2.5% from
20,000 to 40,000, and 0.86% from 40,000 to 60,000, and
0.33% from 60,000 to 80,000. Consequently, 60,000 cells
were adopted for the computational domains in all the for-
mal computations of the study.
The time step for the transient flow field computation
was set as 2 s, based on the time-step independence analy-
sis. Time steps of 0.5 s, 1 s, 2 s, 4 s and 5 s were tested in the
computations. It was found that the relative difference of
the temperature of HTF at the location of 1 m from the
inlet had a variation of 0.16% when increasing the time step
from 0.5 s to 1 s. The variation was 0.83% from 1 s to 2 s,
1.95% from 2 s to 4 s, and 3.67% from 4 s to 5 s.
Consequently, a small time step of 2s was chosen for the
computation of the transient process. At each time step
of computation, convergence is checked to meet a conver-
gence criterion.
5.4. Comparison of CFD results and the results from 1D
modeling
The temperature of the HTF at the exit of a discharge is
an important indication of energy storage efficiency. There-
fore, the HTF temperature at the exit during the 4-h dis-
charging period has been monitored for comparison
between the CFD results and the 1D modeling results.
The results are for the cases after a sufficient number of
cyclic charging and discharging, thus achieving cyclic
steady state. In any 4-h charge process, the entering hot
HTF has a constant temperature of 663.15 K, and in any
discharge the exiting cold HTF has a constant temperature
583.15 K. The inlet flow velocity of the fluid is always set as
V inlet ¼ 1:36  103 m=s.
As shown in Fig. 5, the fluid temperatures in the 4-h
discharge process between the results of CFD and 1D tran-
sient model agree very well. The maximum difference
between them is 0.06% for the plate case, 0.05% for the
78 P. Li et al. / Solar Energy 105 (2014) 71–81
(a) Plate case
(b) Cylinder case
(c) Tube case
Fig. 5. Comparisons of results from CFD and 1D model for the time
variation of exit temperature of HTF.
cylinder case, and 0.01% for the tube case. This comparison
clearly shows that the 1D transient model, which predicted
fluid temperature at the exit of the discharge process during
the time period, is sufficiently accurate. Engineers can use
this method for the analysis of thermal energy storage sys-
tems without using comprehensive CFD computations.
To further compare the results from the 1D transient
modeling with the results from CFD computation, Fig. 6
shows the distribution of temperature of HTF at locations
in the storage tank along the flow direction after 4 h of heat
(a) Plate case
(b) Cylinder case
(c) Tube case
Fig. 6. Comparisons of results from CFD and 1D model for the
temperature distribution of HTF.
discharge. The heat discharge results are from a typical
process after many cyclic charges and discharges so that
a cyclic steady state has been achieved. It is understandable
that after discharge, the temperature of fluid at the lower
part of the tank is low. The temperature distribution from
the 1D transient model and from the CFD computation
agree very well, which further verifies the accuracy of the
1D transient modeling.
 P. Li et al. / Solar Energy 105 (2014) 71–81 79

(a) Plate case

(b) Cylinder case

(c) Tube case

Fig. 7. Comparison of exit temperature of HTF from CFD and 1D modeling during a discharge under cyclic steady state condition.

More numerical computations and 1D simulations were
conducted for higher flow velocities of 2V inlet and 3V inlet .
Comparison between CFD and 1D modeling for the tem-
perature of HTF at the exit of the tank during a typical dis-
charging process for cyclic steady state is given in Fig. 7.
The 1D modeling results agree with the CFD results very
well.
Comparing the results from Figs. 7 and 5, it can be seen
that the discharged fluid temperature decreases faster and
earlier at higher fluid velocities. This is because the greater
fluid velocity results in a greater amount of heat being
removed from the solid thermal storage material.
Comparing the discharged HTF temperatures from the
plate, cylinder, and tube in Figs. 5 and 7, it is seen that
80 P. Li et al. / Solar Energy 105 (2014) 71–81
the plate retains higher temperatures for a longer time
before degradation. This phenomenon can be explained
by the plate’s relatively higher effective heat transfer coeffi-
cient as well as larger heat transfer area, as seen in Tables 3
and 4; therefore, the plate can store and withdraw more
energy from the tank compared to the others, which all
have the same solid volume and fluid volume.
6. Concluding remarks
This paper presented and verified a general 1D transient
model for thermal energy storage systems in three typical
solid–fluid structural combinations. In the 1D model, an
effective heat transfer coefficient due to the use of an
extended lumped capacitance method (for Biot number
up to 100) was introduced to account for the internal heat
conduction resistance in solid thermal storage materials.
The model’s verification was based on the results from a
comprehensive CFD analysis. Results of local temperature
distribution of the fluid in the thermal storage system at the
end of an energy discharging process, and the temperature
variation at the exit of a 4-h discharging process using the
1D transient modeling and the CFD simulation agreed
very well. In conclusion, it is evident that the 1D transient
model with the corrected heat transfer coefficient is robust
and highly accurate for analysis of the heat transfer and
energy storage behavior for the three studied typical
fluid–solid structural combinations.
Because the 1D transient model significantly simplifies
analysis, and especially the computational time, it is
expected to be one of the most convenient and accurate
tools available for industrial engineers to analyze the
behavior of thermal energy storage systems of various
fluid–solid configurations.
Finally, it is worth noting that, typically, the Schumann
equations (1D transient model) for thermal energy storage
ignore the axial heat conduction in both the solid and fluid,
which is sufficiently accurate for regular dual-material ther-
mal storage systems. However, if the solid and the liquid
are highly conductive metals and liquid metals respectively,
which is rarely the case, the validity of the model has to be
further examined. In the 1D transient model the laminar
flow heat transfer coefficient in the cases given in
Fig. 1(b–d) are based on thermally and hydraulically fully
developed flow due to typical long flow channels in thermal
storage systems. For short flow channels, though not com-
mon in thermal storage applications, the entrance effect to
the heat transfer coefficient should be considered.
Acknowledgments
The authors are grateful for the support from the U.S.
Department of Energy and National Renewable Energy
Laboratory under DOE Award: DE-FC36-08GO18155.
One of the authors, Jinxiao Han, also would like to thank
the China Scholarship Council for the financial support of
his 15 months’ study at the University of Arizona.
References
ASME V&V 20, 2009. Standard for Verification and Validation in
Computational Fluid Dynamics and Heat Transfer.
Becker, M., 1980. Comparison of heat transfer fluids for use in solar
thermal power stations. Elect. Power Syst. Res. 3 (3), 139–150.
Bradshaw, A.V., Johnson, A., McLachlan, N.H., Chiu, Y.-T., 1970. Heat
transfer between air and nitrogen and packed beds of non-reacting
solids. Trans. Inst. Chem. Eng. 48, T77–T84.
Fluent, ANSYS, 2006. 6.3 User’s Guide. ANSYS. Inc., Houston, TX.
Jeffreson, C.P., 1972. Prediction of breakthrough curves in packed beds: 1.
Applicability of single parameter models. Am. Inst. Chem. Eng. 18 (2),
409–416.
Karaki, W., Van Lew, J., Li, P.-W., Chan, C.-L., Stephens, J., 2010 May.
Heat transfer in thermocline storage system with filler materials:
analytical model. In: Proceedings of the ASME 2010 Fourth Interna-
tional Conference on Energy Sustainability ES2010, Phoenix, AZ, pp.
17–22.
Kays, W.M., Crawford, M.E., 1993. Convective Heat and Mass Transfer,
third ed. McGraw-Hill Inc., New York.
Li, P.-W., Van Lew, J., Karaki, W., Chan, C.-L., Stephens, J., O’Brien,
J.E., 2011a. Transient heat transfer and energy transport in packed bed
thermal storage systems, 2011. In: Marco Aurélio dos Santos Bernar-
des (Eds.), Book “Developments in Heat Transfer”. ISBN 978-953-
307-569-3, September 15. <http://www.intechopen.com>.
Li, P.-W., Van Lew, J., Karaki, W., Chan, C.-L., Stephens, J., Wang, Q.-
W., 2011b. Generalized charts of energy storage effectiveness for
thermocline heat storage tank design and calibration. Solar Energy 85
(9), 2130–2143.
Li, P.-W., Van Lew, J., Chan, C.-L., Karaki, W., Stephens, J., O’Brien,
J.E., 2012. Similarity and generalized analysis of efficiencies of thermal
energy storage systems. Renew. Energy 39, 388–402.
Müller-Steinhagen, H., Trieb, F., 2004. Concentrating solar power: a
review of the technology. Ingenia Inform QR Acad Eng 18, 43–50.
Oberkampf, William L., Trucano, Timothy G., 2002. Verification and
validation in computational fluid dynamics. Prog. Aerospace Sci. 38
(3), 209–272.
Pavlović, T.-M., Radonjić, I.-S., Milosavljević, D.-D., Pantić, L.-S., 2012.
A review of concentrating solar power plants in the world and their
potential use in Serbia. Renew. Sustain. Energy Rev. 16 (6), 3891–
3902.
Py, X., Azoumah, Y., Olives, R., 2013. Concentrated solar power: current
technologies, major innovative issues and applicability to West African
countries. Renew. Sustain. Energy Rev. 18, 306–315.
Schumann, T.E.W., 1929. Heat transfer: a liquid flowing through a porous
prism. J. Franklin Inst. 208 (3), 405–416.
SunShot Vision Study, February 2012. US Department of Energy,
<http://www1.eere.energy.gov/solar/sunshot/>.
Valmiki, M.M., Karaki, W., Li, P.-W., Van Lew, J., Chan, C.-L.,
Stephens, J., 2012. Experimental investigation of thermal storage
processes in a thermocline tank. ASME J. Solar Energy Eng. 134
(November), pp. 041003-1.
Van Lew, J., Li, P.-W., Chan, C.-L., Karaki, W., Stephens, J., 2009,
November. Transient heat delivery and storage process in a thermo-
cline heat storage system. In: Proceedings of ASME 2009 International
Mechanical Congress and Exposition, Lake Buena Vista, FL, pp. 13–
19.
Van Lew, J., Li, P.-W., Chan, C.-L., Karaki, W., Stephens, J., 2011.
Analysis of heat storage and delivery of a thermocline tank having
solid filler material. J. Solar Energy Eng. 133, 021003.
Xu, B., Li, P.-W., Chan, C.-L., 2012. Extending the validity of lumped
capacitance method for large Biot number in thermal storage
application. Solar Energy 86 (6), 1709–1724.
 P. Li et al. / Solar Energy 105 (2014) 71–81 81

Yang, Z., Garimella, S.V., 2010. Molten-salt thermal energy storage in
thermocline under different environmental boundary conditions. Appl.
Energy 87 (11), 3322–3329.
Zanganeh, G., Pedretti, A., Zavattoni, S., Barbato, M., Steinfeld, A.,
2012. Packed-bed thermal storage for concentrated solar power–Pilot-
scale demonstration and industrial-scale design. Solar Energy 86 (10),
3084–3098.
Zhang, H.L., Baeyens, J., Degrève, J., Cacères, G., 2013. Concentrated
solar power plants: review and design methodology. Renew. Sustain.
Energy Rev. 22, 466–481.