Solar Energy 173 (2018) 882–892

Contents lists available at ScienceDirect

Solar Energy
journal homepage: www.elsevier.com/locate/solener

Effects of different thermal storage tank structures on temperature T

stratification and thermal efficiency during charging
⁎
Angui Li , Feifei Cao, Wanqing Zhang, Bingjin Shi, Huang Li
School of Environmental and Municipal Engineering, Xi’an University of Architecture and Technology, Xi’an 710055, China

A R T I C LE I N FO A B S T R A C T

Keywords: In solar water heating systems, the structures of thermal storage devices have played essential roles in the
Solar water heating system improvement of thermal charging efficiency and system performance. This article was focused on the optimi-
Different structures zation of thermal storage tanks, as well as the influences of thermal storage tank structures on the temperature
Radius ratio (n) stratification and heat storage capacity. The thermal characteristics of three shapes of thermal storage tanks were
Temperature stratification
investigated and analyzed through CFD (Computational Fluid Dynamics) simulation, which were: (a) Cylindrical
tank, (b) Circular truncated cone tank and (c) Spherical tank. It was observed that among the three thermal
storage tanks, the spherical tank had significant symmetrical temperature stratification, while the truncated
circular cone tank had the best temperature stratification and thermal charging efficiency. For the circular
truncated cone tank, it was found that when the radius ratio (n) of the water tank was 0.554, it could approach
the excellent characteristics of temperature stratification and thermal storage performance. This study could
provide a theoretical basis for the structure optimization of the solar energy storage devices, as well as proving to
be beneficial to the enhancement of thermal charging efficiency.

1. Introduction has the effect of decreasing operation periods of the auxiliary energy
supply (Kenjo et al., 2007).
With the development and utilization of renewable energies, solar During discharging, the energy release characteristic of the water
energy is becoming increasingly popular (Liu et al., 2018). Against this tank is associated to many factors, such as the internal structure of the
backdrop, the utilization of solar energy in a water heating system is water tank, the energy consumption intensity of the water tank and the
quite common (Liu et al., 2017). A typical solar water heating system structure of the inlet pipe (Bahnfleth and Song, 2005; Altuntop et al.,
with a cylindrical tank is presented in Fig. 1. As a critical part of the 2005; Ghajar and Zurigat, 1991; Berkel, 1996; Al-Najem and El-Refaee,
solar water heating system, the heat storage tank affects the entire 1997; Eames and Norton, 1998; Alizadeh, 1999). Bouhal et al. (2017)
system efficiency owing to the instantaneous operating characteristics. studied a domestic thermal storage water tank with a baffle and ob-
Since 1960s, many studies were conducted regarding the structure served that the location of the baffle along with the title angle had
optimization and thermal storage characteristics of the solar thermal significant effect on the temperature stratification of the water tank.
storage tank. It was observed that the efficiency of solar collector sys- Nizami et al. (2013) analyzed the effects of the water inlet current
tems would be dramatically improved with the stratified tank (Al- speed, the temperature and the pipe diameter for temperature stratifi-
Najem et al., 1993; Rosen et al., 2004; Armstrong et al., 2014). cation. Li et al. (2013) carried out experiments regarding the dischar-
Temperature stratification within the solar water tank stands for the ging performance of solar storage tanks with different inlet structures.
gradient of temperature along the vertical direction of tank. It is driven The results indicated that the perforated and slotted inlets effectively
by the buoyancy motion due to the effect of temperature difference on improved the performance of thermal stratification.
density. Also, the form and arrangement, as well as the heat conduction
Under wide circumstances, plenty of studies have verified the su- characteristic of the immersed heat exchangers were widely studied
periority of temperature stratification in the solar water tank. On the (Cadafalch et al., 2015; Paradis et al., 2018). In the research of
one hand, the temperature stratification is beneficial to decreasing the Haltiwanger and Davidson (2009), it was reported that the annular
temperature at the collector inlet, which is corresponding to a higher baffle led the convective heat transfer strength to increase by 20% at
temperature efficiency of the solar energy system. On the other hand, it the heat storage side of the water tank. Li (2013) concluded that the

⁎
Corresponding author.
E-mail address: liag@xauat.edu.cn (A. Li).

https://doi.org/10.1016/j.solener.2018.08.025
Received 15 May 2018; Received in revised form 3 August 2018; Accepted 11 August 2018
0038-092X/ © 2018 Elsevier Ltd. All rights reserved.
A. Li et al. Solar Energy 173 (2018) 882–892

Nomenclature surface of the tank, mm

τch charging time, s
→
V the velocity vector (u,v,w), m/s τtot total operating time, s
cp the specific heat, J/(kg⋅K) τ dimensionless time τ = τch/τtot [–]
→ T(t) temperature, K (°C)
F the body force vector per unit volume (x, y, z), N
μf the coefficient of viscosity, Pa⋅s Ttop temperature at the top surface of the tank, K
p the local pressure, Pa Tbottom temperature at the bottom of the tank, K
λ the second viscosity coefficient, Pa⋅s μf Φ the viscous dissipation term due to viscous forces [–]
βf the coefficient of thermal expansion, 1/K
Subscripts and superscripts
g acceleration of gravity, m2/s
kf fluid thermal conductivity, W/(m⋅K)
RE relative error
H tank height, m
RMSE root mean square error
ΔT the temperature difference between the top and bottom
in inlet
surface, K
HF heating fluid
v kinematic viscosity, m2/s
n natural convection
vs average velocity of the water on the stratified surface, m/s
ini initial
a thermal diffusivity, m2/s
tot total operating
l characteristic length of the tank, m
ch charging
Ra Rayleigh number [–]
f fluid
Z the distance between measurement point and bottom

Fig. 1. Schematic diagram of solar water heating system.

annual solar energy guarantee rates of the solar water heating system Table 2
with an intermediate coil and an upper coil tank were 6.1% and 14.9% Geometric parameters of coil in the solar water heating system.
higher compared to the system with the lower coil tank. Moreover, Parameters Value Units Remarks
through further design, by Nicodemus et al. (2017), of the annular
baffle, the convective heat transfer strength increased by 27% during Coil Height (Hc) 400 mm –
water tank pre-discharging at the heat storage side. Coil diameter (d) 14 mm –
Coil thickness (δ) 2 mm –
Furthermore, it was found that the ratio of the height and diameter Coil screw diameter (D) 350 mm –
had a certain effect on the improvement of the temperature Number of circles 18 – –
Pitches (s) 22.22 mm –
Coil total length (L) 19.80 m L = k s 2 + π 2D 2
Table 1 Total heating surfaces(F) 0.87 m2 F = Lπd
Different radius ratios in circular truncated cone tanks (in: mm).
n 0.673 0.554 0.450 0.403
stratification of the cylindrical thermal storage tank (Lavan and
H (mm) 1000 1000 1000 1000 Thompson, 1977; Hahne and Chen, 1998; Ismail et al., 1997). As an
2r (mm) 350 310 270 250 example, Lavan and Thompson (1977) discovered that when the height
2R (mm) 520 560 600 620 to diameter ratio of the water tank was in-between 3 and 4, the tem-
perature stratification performance of the water tank and the cost of the
n Radius ratio between the top and bottom surfaces, H height of the tank, r
radius of the top surfaces of the tank, R radius of the bottom surfaces of the
water tank could be taken into account. In addition, the shape of the
tank. water tank was also preliminarily studied. Ismail et al. (1997) proved

883
A. Li et al. Solar Energy 173 (2018) 882–892

Fig. 2. Different water tank shapes: (a) cylindrical, (b) circular truncated cone, (c) spherical.

Fig. 3. Distribution of grids in the simulation, (a) Grid partition of coil heat exchanger, (b) Grid partition of tank.

Table 3
The summary of the grid division of different forms of tank.
Different forms of tank Grid numbers(element) Minimum grid volume (mm3) Maximum grid volume (mm3)

Total Coil Tank

Cylindrical tank 1,517,301 125,484 1,391,817 0.625 444

Circular truncated cone tank n = 0.673 1,548,204 123,867 1,424,337 0.616 426
n = 0.554 1,592,883 127,431 1,465,452 0.598 407
n = 0.450 1,647,522 132,805 1,514,717 0.597 398
n = 0.403 1,534,627 122,781 1,411,846 0.636 415
Spherical tank 1,522,305 123,307 1,398,998 0.619 439

that the structure of the water tank had a certain effect on the tem- Table 5
perature stratification of the water tank. Yang et al. (2016) studied ten Cases setting of charging process in solar water heating system.
different water tank shapes, observing that the spherical and barrel Cases Temperature of antifreeze, THF (K)
tanks had favorable ability of heat storage.
At the same time, researchers analyzed the dimensionless parameter Case 1 323.15
applicability of the water tank temperature stratification with the Case 2 343.15
Case 3 363.15
Richardson number utilization (Cole and Bellinger, 1982; Yee and Lai,

Table 4
2001; Jordan and Furbo, 2005; Kleinbach et al., 1993). Castell et al.
Boundary conditions of numerical simulations. (2010) compared several dimensionless parameters to evaluate the
temperature stratification characteristics of the water tank with a re-
Names Types Remarks
verse funnel type inlet rectifier. As a result, the Richardson number was
Tank surface Insulated wall the best parameter to qualitatively evaluate the temperature stratifi-
Coil wall Coupled wall cation characteristics of the water tank.
Coil inlet Velocity inlet 0.450 m/s A comprehensive analysis of the literature research results demon-
Coil outlet Outflow
strated that most researches in thermal storage tank analysis were
Heated fluid Fluid l Water
Heating fluid Fluid 2 Antifreeze
mainly focused on the coil structure and layout, the inlet pipe structure
and layout, as well as on the water tank height and diameter ratio.

884
A. Li et al. Solar Energy 173 (2018) 882–892

Fig. 4. Independence verification, (a) Grid size independence study, (b) Time-step dependence study.

Adversely, few studies exist on the water tank structure effects on the radius (n) and the spherical tank were built. The structure and geometry
temperature stratification characteristics and storage capacity. In view of different water tanks are presented in Fig. 2.
of this fact, the temperature stratification and thermal storage char-
acteristics with the conditions of different water tank structures were 2.2. Discretization
investigated. Under the ensured consistency condition of the water tank
volume (150 L) and the heating coil layout mode, the optimal water In order to acquire more accurate grids, the fluid regions were di-
storage tank structure was obtained through the heat storage char- vided into three parts, namely the coil part, the main body of tank and
acteristics analysis of the cylindrical tank, the circular truncated cone the part of tank near the coil. To acquire favorable mesh quality and
tanks and the spherical tank. reduce computing time, the hexahedral structured grid was used for the
main body of tank and the coil part. Considering the curved surface and
the irregularity of coupling between coil and tank, the hexahedral
2. Numerical study structured grid was not suitable for the part of tank near the coil, and
the unstructured grid was adopted. The cells closed to the coil wall are
2.1. Physical model small enough to capture the complex flow structure. The distribution of
grids in the simulation is presented in Fig. 3. And the summary of the
In this work, the objective heat storage device was a cylindrical grid division of different forms of tanks is listed in Table 3.
water tank in domestic solar heating water system. The tank was
H = 1000 mm in height, and V = 150 L in volume. In this paper, the
2.3. Mathematical model
radius of cylindrical and spherical tank are 217.5 mm and 330 mm re-
spectively. For the circular truncated cone water tank, different radii
2.3.1. Simplification of physical conditions
ratios of the top and bottom surfaces were considered, as presented in
In this study, several hypotheses of physical conditions were made.
Table 1. The geometry of the spiral coil was presented in Table 2. In
The simplification included the following:
allusion to the cylindrical water tank, the conditions of other tanks,
such as the volume of the tank, the length of the spiral coil and the
(1) The simulation was considered incompressible and with unsteady
heating surface area, were similar. Based on this fact, the circular
flow without internal heat source;
truncated cone tank with different ratio values of the top and bottom
(2) The outer wall of the thermal storage tank was adiabatic;
(3) In the simulation, the wall of the coiled tube was considered a
coupled wall, while the temperature of the tube inlet remained
unchanged;
(4) In the numerical calculation of the closed natural convection heat
transfer, the Boussinesq approximation was used for the con-
venience of the buoyancy force handling, due to the temperature
difference. Boussinesq approximation assumes a linear dependence
of the density with temperature. It can be expressed as
ρf = ρ fi [1−βf (Tf −Tfi )], where Tf is the temperature of water, Tfi is
the initial temperature of water, ρfi is the density of the initial
water, βf is the coefficient of thermal expansion of water and given
by βf = -
1
ρf ( )
∂ρf
∂Tf p
(Fluent Inc, 2006; Chang et al., 2017).

2.3.2. Governing equations

The Navier-Stokes and energy equations in three-dimensional forms
were utilized to solve the transient hydrodynamic as well as the thermal
field equations (Rohsenow et al., 1998). Therefore, the resulting gov-
erning equations with the gravity effect consideration can be written as
Fig. 5. Temperatures of different layers during water tank discharging. follows (Rohsenow et al., 1998; Chang et al., 2017):

885
A. Li et al. Solar Energy 173 (2018) 882–892

n=1.000 n=0.673 n=0.554 n=0.450 n=0.403 Sphere

n=1.000, ¨T=4ºC n=0.673, ¨T=3ºC n=0.554, ¨T=2ºC n=0.450, ¨T=7ºC n=0.403, ¨T=6ºC Sphere, ¨T=1ºC

n=1.000, ¨T=0.5ºC n=0.673, ¨T=0.5ºC n=0.554, ¨T=0.5ºC n=0.450, ¨T=1ºC n=0.403, ¨T=1ºC Sphere, ¨T=0.5ºC
Fig. 6. Temperature fields at different charging time in case 1 for heat storage tanks (THF = 323.15 K), (a) Temperature fields at τch = 3600 s for heat storage devices,
(b) Temperature fields at τch = 7200 s for heat storage devices.

Conservation of mass: ∂u ∂v ∂w
+ + =0
∂x ∂y ∂z (4)
Dρf →
+ ρf ∇ ·V = 0
Dt (1) ∂u ∂u ∂u ∂u ∂p ∂ 2u ∂ 2u ∂ 2u
ρf ⎛
⎜ +u +v +w ⎞=− ⎟ + μf ⎛ 2 + 2 + 2 ⎞
⎜ ⎟

⎝ ∂t ∂x ∂y ∂z ⎠ ∂x ⎝ ∂ x ∂ y ∂z ⎠ (5)
Conservation of momentum:
→ ∂v ∂v ∂v ∂v ∂p ∂ 2v ∂ 2v ∂ 2v
DV → → → ρf ⎛
⎜ +u +v +w ⎞=− ⎟ + μf ⎛ 2 + 2 + 2 ⎞
⎜ ⎟
ρf = −∇p + μ f ∇2 ·V + (μ f + λ ) ∇ (∇ ·V ) + F ⎝ ∂t ∂x ∂y ∂z ⎠ ∂y ⎝ ∂x ∂y ∂z ⎠ (6)
Dt (2)

Conservation of energy: ∂w ∂w ∂w ∂v ∂p ∂ 2w ∂ 2w ∂ 2w ⎞
ρf ⎛
⎜ +u +v +w ⎞=− ⎟ + μf ⎛ 2 + ⎜
2
+ ⎟

⎝ ∂t ∂x ∂y ∂z ⎠ ∂z ⎝ ∂x ∂y ∂z 2 ⎠
DTf ∂T → Dp
ρf cp, f = ρf cp, f ⎡ f + (V ·∇) Tf ⎤ = ∇ (k f ∇Tf ) + βf Tf + μf Φ + ρf gβf (Tf −Tfi) (7)
Dt ⎣ ∂t ⎦ Dt (3)
∂T ∂T ∂T ∂T ∂ 2T ∂2Tf ∂2Tf ⎞
With the Boussinesq approximation and simplification, the gov- ρf cp,f ⎛⎜ f + u f + v f + w f ⎞⎟ = k f ⎛ 2f +
⎜
2
+ + μf Φ
⎟

erning Eq. (1) can be simplified as (4), Eq. (2) can be simplified as (5), ⎝ ∂t ∂x ∂y ∂z ⎠ ⎝ ∂ x ∂y ∂z 2 ⎠
(6) and (7), Eq. (3) can be simplified as (8). (8)

886
A. Li et al. Solar Energy 173 (2018) 882–892

n=1.000, ¨T=6ºC n=0.673, ¨T=5ºC n=0.554, ¨T=4ºC n=0.450, ¨T=11ºC n=0.403, ¨T=12ºC Sphere, ¨T=1ºC

n=1.000, ¨T=0.5ºC n=0.673, ¨T=0.5ºC n=0.554, ¨T=0.5ºC n=0.450, ¨T=1.5ºC n=0.403, ¨T=1.5ºC Sphere, ¨T=0.5ºC
Fig. 7. Temperature fields at different charging time in case 2 for heat storage tanks (THF = 343.15 K), (a) Temperature fields at τch = 3600 s for heat storage devices,
(b) Temperature fields at τch = 7200 s for heat storage devices.

2.3.4. Initial conditions and cases setting

According to the actual situation in the experiment, the initial water
2.3.3. Boundary conditions
temperature in the tank was assumed to 293.15 K, whereas the tem-
GAMBIT software provides over ten types boundary conditions,
perature distribution was uniform without apparent temperature stra-
such as velocity inlet, outflow, pressure inlet and wall. For the coil
tification. During charging, the water velocity at the import and export
outlet, the details of the flow velocity and pressure are not known prior
of tank is 0 m/s. The fluid in the coil was antifreeze fluid (main com-
to solution of the flow problem. Thus, it is set as outflow boundary. For
ponent was ethylene glycol), while its flow velocity was 0.450 m/s.
the tank surface, with effective insulation measures outside the wall of
Three different temperature of antifreeze were set as different charging
tank, there are little heat transfer with the external environment.
cases, as presented in Table 5.
Ignoring the heat transfer between tank and external environment, the
tank surface is regarded as insulated wall which is expressed by Eq. (9).
For the coil wall, the boundary condition is decided by dynamical heat 2.4. Set-ups in simulation
exchange process between fluid inside/outside of coil, so it cannot be
set in advance and is regarded as coupled wall which is expressed by In the actual numerical simulation, only the three basic control
Eqs. (10) and (11). equations were not sufficient. The CFD introduced different computa-
tional models for various flow and heat transfer problems. Therefore,
∂T
|s = 0 the selection and validation of the turbulence model were vital to the
∂n (9)
numerical study. The mathematical models of convective heat transfer
where s stands for the tank surface, n is the coordinate measured normal of the laminar and turbulence models were mainly of two categories.
to the surface. For the flow in the inner coil, the calculation of Re was performed
by Eq. (12).
T| I = T| II (10)
ρHF v HF d
Re =
∂T μ HF (12)
- k s ⎛ ⎞ | I = h f (Ts−Tf )| II
⎝ ∂n ⎠ (11)
where ρHF is the density of the ethylene glycol, vHF is the flow velocity
where Ts is the temperature of the coil, Tf is the temperature of the of the ethylene glycol (0.450 m/s), d is the diameter of the coil
water, I is the coil surface, II is the region of tank near the coil, hf is the (14 mm), μHF is the dynamic viscosity of the ethylene glycol. As a result
heat transfer coefficient of water and ks is the thermal conductivity of of the numerical solutions, it was determined that the value of Re
the coil. number was 446, which was lower than 2300. This result indicates that
Table 4 lists the other boundary conditions and fluid types. the flow in the inner coil pipe is laminar.

887
A. Li et al. Solar Energy 173 (2018) 882–892

n=1.000, ¨T=10ºC n=0.673, ¨T=10ºC n=0.554, ¨T=5ºC n=0.450, ¨T=13ºC n=0.403, ¨T=7ºC Sphere, ¨T=1ºC

n=1.000, ¨T=0.5ºC n=0.673, ¨T=2ºC n=0.554, ¨T=1ºC n=0.450, ¨T=1.5ºC n=0.403, ¨T=1.5ºC Sphere, ¨T=0.5ºC
Fig. 8. Temperature fields at different charging time in case 3 for heat storage tanks (THF = 363.15 K), (a) Temperature fields at τch = 3600 s for heat storage devices,
(b) Temperature fields at τch = 7200 s for heat storage devices.

For the flow in the side of tank storage, the flow state can be judged the First Order Upwind scheme was used for the momentum and energy
by Ra number. terms, and the iterations were maintained until the residual values were
less than 10−6 for the continuity, momentum and energy equations. A
gβf (Ts−Tf ) l3 ν gβ (Ts−Tf ) l3 time step of 0.5 s was used.
R a= Gr·Pr = · = f
ν 2 a aν (13)

where ν is the kinematic viscosity, l is the characteristic length of the

tank, Ts is the temperature of the coil surface, Tf is the temperature of 3. Model validation
the water region near the coil. In this paper, the value of Ra is higher
than 107. Thus, the flow hypothesis of turbulence flow has been con- 3.1. Grid size and time-step independence study
sidered in the side of tank storage.
The commonly used turbulence models are the standard k–ε model In order to improve the economy of the calculation, the independent
(Launder and Spalding, 1990), the RNG k–ε model (Yakhot and Orszag, verification of the computational model was performed in this paper
1986), the low Reynolds number k–ε turbulence model (Amano et al., and the result is presented in Fig. 4. It was discovered that when the
1987) and the realizable k–ε model. For the model established in this grid quantity increased from 1,517,301 to 2,563,638, the temperature
paper, the heat transfer was mainly concentrated on the wall surface of of the coil outlet presented tiny errors, whereas the maximum differ-
the coil heat exchanger. It was necessary to consider the viscous effect ence was only 0.89%. This meant that when the grid number was
close to the wall surface area. Due to the local turbulent Reynolds 1,517,301, the model of the water tank meets the requirement of the
number lower than 150, the standard k–ε model was inappropriate. grid independence. Thus, the summary of the grid division of different
Therefore, the low Reynolds number k–ε turbulence model was highly forms of tank is listed in Table 3.
adaptable (Li, 2013; Shah et al., 2005). Five parameters, namely Cμ, C1, In addition to the grid independence verification, the time-step in-
C2, σk, and σε, are directly related to the numerical accuracy. The values spection was also essential for the unsteady calculation. Three time-
of these parameters have been discussed and optimized, which are steps of 0.5 s, 1.0 s, and 2.0 s were compared, as presented in Fig. 4(b),
while the temperature error of the coil outlet was quite slight and it
Cμ = 0.09, C1 = 1.44, C2 = 1.92, σk = 1.0, σε = 1.3 (Zhao et al., 2017;
Bae et al., 2017). could be ignored. Therefore, all cases were simulated with a time-step
of 1.0 s. In summary, the grid number of the water tank model was
The solver setting was (Patankar, 1980; Yaïci et al., 2013): Nu-
merical solutions were obtained for the transient laminar flow with the approximately 1.52 million and the time-step was set to 1.0 s.
Boussinesq assumption for the buoyancy modeling; the velocity-pres-
sure coupling was treated with the SIMPLE algorithm (Gao et al., 2017);

888
A. Li et al. Solar Energy 173 (2018) 882–892

Fig. 9. Richardson numbers change along with the charging time for various tanks in different charging cases (ttot = 7200 s). (a) Case 1, THF = 323.15 K. (b) Case 2,
THF = 343.15 K. (c) Case 3, THF = 363.15 K.

3.2. Mathematical model validation 4.1. Analysis of temperature field

In order to ensure the accuracy of the calculation results, in this During the charging process, the temperature distribution of the
study, the working process of the discharging and charging of heat heat storage tanks under different working conditions are presented in
storage water tank was introduced from Li (2013) and the result is Figs. 6–8. With the information provided in these figures, high tem-
presented in Fig. 5. The numerical results in all cases are generally perature difference stands for an evident temperature stratification and
consistent with the experimental data with respect to the mean tem- the temperature distribution of different heat storage tanks could be
perature profiles of the testing section. The errors at the same test specifically compared. Under the same working condition, for the cy-
points with different discharging times demonstrated that: the root lindrical tank, the temperature stratification was apparent at the be-
mean square errors of temperature were RMSEZ=50 = 2.77%, ginning of charging, while the later temperature distribution was more
RMSEZ=500 = 2.04% and RMSEZ=950 = 0.74%; the relative errors of uniform; for the circular truncated cone tanks, the temperature strati-
temperature were REZ=50 = 4.95%, REZ=500 = 4.01% and fication increased as the radius ratio (n) decreased, whereas among
REZ=950 = 1.49%. As the experimental uncertainty was not explicitly them, the temperature stratification of n = 0.450 was apparent; for the
provided by the authors of Li (2013), the relative error between their spherical tank, the temperature distribution was more uniform during
numerical results and experimental data was estimated to 5.0%. charging. Furthermore, compared to the cylindrical heat storage tank,
the circular truncated cone heat storage tanks were more likely to ac-
quire a higher temperature. Under different working conditions, as the
4. Results and discussion temperature of the antifreeze liquid increased, the temperature strati-
fication of differently shaped tanks had unique characteristics. When it
For the charging process, through the analysis of the temperature comes to the cylindrical tank, the temperature stratification was slightly
field distribution, the instantaneous characteristic parameters and the apparent, but for the circular truncated cone tanks, the temperature
average temperature characteristics of different devices under different stratification phenomena of n = 0.673 and n = 0.450 became stronger,
working conditions, the temperature stratification and thermal storage while in the other circular truncated cone tanks the opposite phe-
performance of six differently shaped heat storage devices were com- nomena occurred. Adversely, for the spherical water tank, the tem-
pared. perature distribution was always uniform.

889
A. Li et al. Solar Energy 173 (2018) 882–892

Fig. 10. Natural convection Nusselt numbers for different devices at different charging cases, (ttot = 7200 s). (a) Case 1, THF = 323.15 K. (b) case 2, THF = 343.15 K.
(c) Case 3, THF = 363.15 K.

4.2. Characteristic parameter analysis of water tank cone tanks had a better effect on the temperature stratification. Speci-
fically, for the circular truncated cone tanks, when the radius ratio (n)
4.2.1. Richardson number was 0.450, the Richardson number was always higher. By contrast, for
Generally, the main dimensionless parameters describing the effect the spherical tank, when the dimensionless charging time τ ≦ 0.15, the
of the temperature stratification of the water tank were the MIX corresponding Richardson number was the highest compared to the
number, the Richardson number, the ratio of height and diameter, the other structural forms. This meant that at τ ≦ 0.15, the spherical tank
heat release rate, as well as the Reynolds value (Castell et al., 2010). has a better effect on the temperature stratification.
Castell et al. (2010) indicates that the Richardson number could suc-
cessfully embody the temperature stratification of the water tank 4.2.2. Natural convection Nusselt number
compared to the other parameters. The value is adopted to evaluate the During the charging process, for a heat storage water tank with an
proportion of the buoyancy forces in the mixing forces and can be ex- immersed coil heat exchanger, the natural convection plays a dominant
pressed as: role (Su and Davidson, 2007). Based on this phenomenon, the Nusselt
g ·βf ·H ·(Ttop−Tbottom ) number is the evaluation of the natural convection among different
Ri = temperature layers (Haltiwanger and Davidson, 2009; Li, 2013; Su and
vs2 (14)
Davidson, 2007) and can be expressed as:
where vs is the average velocity of the water on the stratified surface. In
Nu n = CRam (15)
different tanks, vs represents the average velocity of water on the
stratified surface caused by the motion of buoyancy. where C and m are both constant which are depended on the flow state.
A lower Richardson number signifies a mixed storage tank, whereas In this paper, C is 0.125, and m is 1/3. As for Ra, it is expressed in Eq.
a higher Richardson number indicates a stratified storage tank (Castell (13).
et al., 2010). Fig. 10 presents the phenomenon of the Nusselt number change
The variation of the Richardson number along with the charging along with the charging time. From Fig. 10(a) and (b), it could be ob-
time is presented in Fig. 9. It was demonstrated that, under the same served that when the dimensionless charging time τ ≦ 0.5, the natural
working condition and with different device shapes, when the di- convection Nusselt number of a cylindrical tank was higher compared
mensionless charging time τ ≧ 0.2, the Richardson numbers of the cir- to the other tanks. The reason for this phenomenon was that, during
cular truncated cone tanks were mostly higher compared to the cy- charging, the cylindrical tank had smaller volume of water at the
lindrical and spherical tanks. This meant that the circular truncated bottom than the other tanks. Consequently, the corresponding

890
A. Li et al. Solar Energy 173 (2018) 882–892

Fig. 11. Average temperature differences of heat storage devices compared to cylindrical tank at different charging cases, (ttot = 7200 s). (a) case 1, THF = 323.15 K.
(b) case 2, THF = 343.15 K. (c) case 3, THF = 363.15 K).

temperature was quickly improved. Adversely, in the latter period of to the cylindrical tank.
charging, the Nusselt number of the cylindrical tank device decreased. Combining the results from the previous sections, in terms of the
This meant that the temperature distribution became uniform inside the temperature stratification, the circular truncated cone tank of
tank. Also, for the circular truncated cone tanks, the Nusselt numbers n = 0.403 is better than the cylindrical tank, and the spherical tank is
increased at the beginning and slightly decreased in the latter process. more uniform than the cylindrical tank. However, in terms of the
This signified that the natural convection of water was stable inside charging efficiency, the circular truncated cone tank of n = 0.403 is
these tanks. Moreover, the Nusselt number of the spherical tank was lower than the cylindrical tank, and the spherical tank is lower than the
lower compared to the other tanks under any working condition, be- cylindrical tank. It was indicated that a better stratified tank was not
cause the shape characteristic caused the diameter of the spherical tank always effective in the charging efficiency increase promotion, while a
to become lower than the height of the other types at the same volume. more uniform temperature distribution was not conducive to the effi-
As a result, the water was more evenly heated in spherical tank, that the ciency increasing of the heat storage devices.
natural convection was weaker.
5. Conclusions

4.3. Average temperature difference of heat storage devices In this study, the charging processes of the thermal storage tanks
with various shapes were numerically studied under different condi-
In this paper, taking the cylindrical tank as the reference object, the tions. Based on the numerical study and analysis results, the following
average temperature differences of the heat storage tanks under various conclusions were drawn:
working conditions are illustrated in Fig. 11. Larger average tempera-
ture differences stand for higher charging efficiency. It could be ob- (1) The temperature stratification of the circular truncated cone tanks
served that at the end of heat storage, when the radius ratios were was better compared to the other tanks, especially, when the radius
0.673 and 0.554 of the circular truncated cone tanks, the average ratio (n) was 0.450, the temperature stratification was excellent.
temperature differences exceeded 0. Therefore, the average tempera- However, the temperature stratification of the spherical tank was
ture of these two tanks, as well as the heat charging efficiency, were the worst.
higher. On the contrary, for the spherical tank, the average temperature (2) The charging efficiency of the circular truncated cone tanks was
difference of the spherical and cylindrical tanks was below 0 °C. Con- always higher compared to the other tanks, especially, when the
sequently, the average temperature of the spherical storage tank did not radius ratios (n) were 0.673 and 0.554, the charging efficiency was
exceed the average temperature in cylindrical tank, for which, the heat better than the other tanks. While the efficiency of the spherical
charging efficiency of the spherical storage tank was inferior compared

891
A. Li et al. Solar Energy 173 (2018) 882–892

tank was the worst. stratification in thermal energy storage. Numer. Heat Transf. 19 (1), 65–83.
(3) Taking circular truncated cone water tanks as the research object, Gao, R., Fang, Z., Li, A., Liu, K., Yang, Z., Cong, B., 2017. A novel low-resistance tee of
ventilation and air conditioning duct based on energy dissipation control. Appl.
as the radius ratio decreased, there was more obvious temperature Therm. Eng. 132.
stratification in the tank, but the charging efficiency was reduced. Hahne, E., Chen, Y., 1998. Numerical study of flow and heat transfer characteristics in hot
Thus, the enhancement of tank temperature stratification did not water stores. Sol. Energy 64 (1–3), 9–18.
Haltiwanger, J.F., Davidson, J.H., 2009. Discharge of a thermal storage tank using an
always improve the heat efficiency of water tank. When the radius immersed heat exchanger with an annular baffle. Sol. Energy 83 (2), 193–201.
ratio of the circular truncated cone tank was 0.554, the heat storage Ismail, K.A.R., Leal, J.F.B., Zanardi, M.A., 1997. Models of liquid storage tanks. Energy 22
efficiency characteristic was excellent. This occurred similarly for (8), 805–815.
Jordan, U., Furbo, S., 2005. Thermal stratification in small solar domestic storage tanks
the temperature stratification. caused by draw-offs. Sol. Energy 78 (2), 291–300.
Kleinbach, E.M., Beckman, W.A., Klein, S.A., 1993. Performance study of one-dimen-
These findings could provide certain references for the design of the sional models for stratified thermal storage tanks. Sol. Energy 50 (2), 155–166.
Kenjo, L., Inard, C., Caccavelli, D., 2007. Experimental and numerical study of thermal
solar heat storage tank, and improve the thermal performance of solar
stratification in a mantle tank of a solar domestic hot water system. Appl. Therm.
water heating system. Eng. 27 (11–12), 1986–1995.
Launder, B.E., Spalding, D.B., 1990. The numerical computation of turbulent flows.
Acknowledgment Comput. Methods Appl. Mech. Eng. 3 (2), 269–289.
Lavan, Z., Thompson, J., 1977. Experimental study of thermally stratified hot water
storage tanks. Sol. Energy 19 (5), 519–524.
This research is supported by the Shaanxi Science & Technology Co- Li, S., Li, Y., Zhang, X., Wen, C., 2013. Experimental study on the discharging perfor-
ordination & Innovation Project (No.2016KTCL01-13). mance of solar storage tanks with different inlet structures. 8,3 (2013-01-09). Int. J.
Low-Carbon Technol. 8 (3), 203–209.
Li, Y., 2013. Study on Performance of Solar Water Heater System with Different Heat
References Exchanger Coil Position. Southeast University, Nanjing.
Liu, Z.J., Li, H., Liu, K.J., Yu, H.C., Cheng, K.W., 2017. Design of high-performance water-
in-glass evacuated tube solar water heaters by a high-throughput screening based on
Amano, R.S., Bagherlee, A., Smith, R.J., Niess, T.G., 1987. Turbulent heat transfer in
machine learning: a combined modeling and experimental study. Sol. Energy 142,
corrugated-wall channel with and without fins. J. Heat Transf. 109 (1), 218–229.
61–67.
Al-Najem, N.M., Al-Marafie, A.M., Ezuddin, K.Y., 1993. Analytical and experimental in-
Liu, Z.J., Wu, D., Yu, H.C., Ma, W.S., Jin, G.Y., 2018. Field measurement and numerical
vestigation of thermal stratification in storage tanks. Int. J. Energy Res. 17 (2), 77–88.
simulation of combined solar heating operation modes for domestic buildings based
Al-Najem, N.M., El-Refaee, M.M., 1997. A numerical study for the prediction of turbulent
on the Qinghai-Tibetan plateau case. Energy Build. 167.
mixing factor in thermal storage tanks. Appl. Therm. Eng. 17 (12), 1173–1181.
Nizami, D.J., Lightstone, M.F., Harrison, S.J., Cruickshank, C.A., 2013. Negative buoyant
Alizadeh, S., 1999. An experimental and numerical study of thermal stratification in a
plume model for solar domestic hot water tank systems incorporating a vertical inlet.
horizontal cylindrical solar storage tank. Sol. Energy 66 (6), 409–421.
Sol. Energy 87 (1), 53–63.
Altuntop, N., Arslan, M., Ozceyhan, V., Kanoglu, M., 2005. Effect of obstacles on thermal
Nicodemus, J.H., Jeffrey, J., Haase, J., Bedding, D., Nicodemus, J.H., Jeffrey, J., 2017.
stratification in hot water storage tanks. Appl. Therm. Eng. 25 (14–15), 2285–2298.
Effect of baffle and shroud designs on discharge of a thermal storage tank using an
Armstrong, P., Ager, D., Thompson, I., Mcculloch, M., 2014. Improving the energy storage
immersed heat exchanger. Sol. Energy 157, 911–919.
capability of hot water tanks through wall material specification. Energy 78,
Patankar, S.V., 1980. Numerical Heat Transfer and Fluid Flow. Hemisphere Pub. Corp.,
128–140.
pp. 125–126.
Berkel, J.V., 1996. Mixing in thermally stratified energy stores. Sol. Energy 58 (4–6),
Paradis, P.-L., Rousse, D.R., Lamarche, L., Nesreddine, H., Talbot, M.-H., 2018. One-di-
203–211.
mensional model of a stratified thermal storage tank with supercritical coiled heat
Bahnfleth, W.P., Song, J., 2005. Constant flow rate charging characteristics of a full-scale
exchanger. Appl. Therm. Eng. 134.
stratified chilled water storage tank with double-ring slotted pipe diffusers. Appl.
Rohsenow, W.M., Hartnett, J.P., Cho, Y.I., 1998. Handbook of Heat Transfer. McGraw-
Therm. Eng. 25 (17–18), 3067–3082.
Hill, New York.
Bae, Y.Y., Kim, E.S., Kim, M., 2017. Assessment of low-Reynolds number k-ε turbulence
Rosen, M.A., Tang, R., Dincer, I., 2004. Effect of stratification on energy and exergy ca-
models against highly buoyant flows. Int. J. Heat Mass Transf. 108, 529–536.
pacities in thermal storage systems. Int. J. Energy Res. 28 (2), 177–193.
Bouhal, T., Fertahi, S., Agrouaz, Y., Rhafiki, T.E., Kousksou, T., Jamil, A., 2017.
Shah, L.J., Andersen, E., Furbo, S., 2005. Theoretical and experimental investigations of
Numerical modeling and optimization of thermal stratification in solar hot water
inlet stratifiers for solar storage tanks. Appl. Therm. Eng. 25 (14), 2086–2099.
storage tanks for domestic applications: CFD study. Sol. Energy 157, 441–455.
Su, Y., Davidson, J.H., 2007. Transient natural convection heat transfer correlations for
Cole, R.L., Bellinger, F.O., 1982. Thermally stratified tanks. Ashrae. Transactions 88,
tube bundles immersed in a thermal storage. J. Sol. Energy Eng. 129 (2), 210–214.
1005–1017.
Yakhot, V., Orszag, S.A., 1986. Renormalization group analysis of turbulence. i. Basic
Castell, A., Medrano, M., Solé, C., Cabeza, L.F., 2010. Dimensionless numbers used to
theory. J. Sci. Comput. 57, (14),1722.
characterize stratification in water tanks for discharging at low flow rates. Renew.
Yee, C.K., Lai, F.C., 2001. Effects of a porous manifold on thermal stratification in a liquid
Energy 35 (10), 2192–2199.
storage tank. Sol. Energy 71 (4), 241–254.
Cadafalch, J., Carbonell, D., Consul, R., Ruiz, R., 2015. Modelling of storage tanks with
Yaïci, W., Ghorab, M., Entchev, E., Hayden, S., 2013. Three-dimensional unsteady CFD
immersed heat exchangers. Sol. Energy 112 (112), 154–162.
simulations of a thermal storage tank performance for optimum design. Appl. Therm.
Chang, C., Wu, Z., Navarro, H., Li, C., Leng, G., Li, X., Yang, M., Wang, Z., Ding, Y., 2017.
Eng. 60 (1–2), 152–163.
Comparative study of the transient natural convection in an underground water pit
Yang, Z., Chen, H., Wang, L., Sheng, Y., Wang, Y., 2016. Comparative study of the in-
thermal storage. Appl. Energy 208.
fluences of different water tank shapes on thermal energy storage capacity and
Eames, P.C., Norton, B., 1998. The effect of tank geometry on thermally stratified sensible
thermal stratification. Renew. Energy 85, 31–44.
heat storage subject to low Reynolds number flows. Int. J. Heat Mass Transf. 41 (14),
Zhao, C.R., Zhang, Z., Jiang, P.X., Bo, H.L., 2017. Influence of various aspects of low
2131–2142.
Reynolds number k-ε turbulence models on predicting in-tube buoyancy affected heat
Fluent Inc, 2006. Fluent 6.3 User’s Guide. Fluent Inc., USA.
transfer to supercritical pressure fluids. Nucl. Eng. Des. 313, 401–413.
Ghajar, A., Zurigat, Y., 1991. Numerical study of the effect of inlet geometry on

892