Professional Documents
Culture Documents
Can Xu 2022
Can Xu 2022
Can Xu 2022
a r t i c l e i n f o a b s t r a c t
Article history: The thermocline storage tank (TCST) is a cost-effective device that can be integrated into district heat-
Received 21 July 2021 ing networks for peak shaving. The formation and variation in the thermocline thickness with time in
Revised 8 December 2021
the TCST are key parameters to evaluate the performance of the TCST. Therefore, the TCST with a radial
Accepted 22 December 2021
plate-type diffuser is studied experimentally and theoretically in this study. The stratification of the stor-
Available online 4 January 2022
age tank is divided into the formation and developmental stages, and a modified 1D analytical model is
Keywords: developed and compared with the experimental data. The correction coefficient ε is defined to quantify
Thermocline storage tank the enhancement in the thermal diffusion of the thermocline. This parameter can improve the accuracy
Analytical model of the 1D analytical model. The average percentage error of the analytical model is reduced from 48.7% to
Thermal diffusion enhancement 18.5%. The quantitative relationship between ε and the tank Reynolds number is obtained. For the forma-
Initial thermocline thickness tion stage of the thermocline, the correlation equation between the thickness of the initial thermocline
and the inlet Reynolds and inlet Froude numbers is determined experimentally. Then, with ε and the
thickness of the initial thermocline, the algebraic relationship of the thickness of the thermocline with
time is investigated. The prediction error of the 1D analytical model for the variation in the thermocline
thickness with time is 13.9%.
© 2021 Elsevier Ltd. All rights reserved.
https://doi.org/10.1016/j.ijheatmasstransfer.2021.122478
0017-9310/© 2021 Elsevier Ltd. All rights reserved.
C. Xu, M. Liu, S. Jiao et al. International Journal of Heat and Mass Transfer 186 (2022) 122478
fuser [13], tank geometry structure [16], porosity [17], particle di-
Nomenclature ameter [18] and effective thermal transfer coefficient [19] on the
thermal storage efficiency and thermocline thickness [20] in TC-
Acronyms STs. Yang et al. [18] studied the discharging behavior of TCSTs by
TCST Thermocline storage tank using a finite-volume approach to solve numerically the momen-
1D One-dimensional tum and volume-averaged mass equations. The result showed that
small Re numbers and large tank heights can enhance discharg-
List of symbols
ing efficiency. Xu et al. [15] developed a transient two-dimensional
A The cross sectional area of the TCST, m2
model and numerically investigated the thermal performance of
Q Water mass flow rate during charging, kg•s − 1
a molten salt thermocline heat storage system and evaluated the
T Temperature of water, K
coefficients related to the thermal storage performance. Flueckiger
Th Inlet temperature of charging, K
et al. [21] developed a computational fluid dynamic model to sim-
Tc Initial temperature in the tank, K
ulate molten-salt TCSTs to analyze the temperature distribution
h Thickness of the thermocline, m
and heat transfer in the tank. Van et al. [22] established a new nu-
xh The height when θ =0.85, m
merical method to solve the governing equations of heat transfer.
xc The height when θ =0.15, m
The developed model was validated using experimental data, and
c Specific heat capacity of water, kJ kg−1 K − 1
the method can handle varying inlet boundary conditions and ar-
T0 Initial temperature in the tank, K
bitrary initial conditions. Bayon and Rojas [20] developed a single-
Ti Inlet temperature of charging, K
phase 1D model to analyze the thermal performance of the TCSTs,
s Laplace Operator
and the design equation, which can improve the thermal storage
E Error
efficiency, was established. Bonanos and Votyakov [23] proposed a
x Axial coordinate measured along the vertical axis of
1D thermocline thickness model and carried out sensitivity anal-
the tank, m
ysis to show the relevant parameters that affect the thermal ef-
v Velocity of the downward movement of the thermo-
ficiency. Hoffmann et al. [14] proposed a 1D numerical model to
cline, m•s − 1
predict the characteristics of the thermocline. The result showed
H Height of tank, m
that the 1D dual-phase model can predict accurately and quickly
u Dimensionless velocity
the thermocline behaviors. Altuntop et al. [24] numerically studied
Fo Fourier number
stratification characteristics in a cylindrical TCST with different ob-
Pe Peclet number
stacles and found that they will affect the thermocline evolution.
t Normalized time of charging
Kursun and Okten [16] investigated numerically the effect of the
D Hydraulic diameter of the tank, m
aspect ratio and the position of the rectangular tank to the ther-
P ecor Correction Peclet number
mocline, and the results could be used to guide the TCST design.
Res Tank Reynolds number
Hanchen et al. [19] solved numerically and verified experimentally
Rei Inlet Reynolds number
the equations for 1D two-phase energy conservation. Moreover, the
Fri Inlet Froude number
charging/discharging characteristics and thermal storage efficiency
Greek symbols of the tank were analyzed. Numerical studies can provide substan-
θ Dimensionless temperature tial information for the TCSTs, and they can guide the design of
ξ Dimensionless height of tank TCSTs without considerable cost.
λ Thermal conductivity, W m − 1 K − 1 Given that numerical simulations are time-consuming and chal-
δ Dimensionless thickness of thermocline lenging for engineers, many comprehensive studies have been con-
ν Kinematic viscosity, m2 •s − 1 ducted on the analytical modeling of TCSTs that can reveal the
τ Dimensionless time relationship between thermal performance and dimensionless pa-
ρ Mass density of water, kg•m − 3 rameters. One-dimensional single-phase analytical models [25] and
ε Correction coefficient one-dimensional double-phase analytical models [26] have been
δ0 Dimensionless thickness of initial thermocline developed, analytical models were used to describe the tempera-
ture distribution of thermocline storage tank. In order to obtain
Subscripts the analytical solutions for the corresponding differential energy
0 Initial balance equation describing the TCSTs, various researchers adopted
s Thermocline storage tank different mathematical methods, such as Laplace transforms tech-
c Cold nology [27], the Greens’ function methodology [28], Normal Cu-
a Average mulative Distribution Function (Normal-CDF) [29] and Logistic Cu-
exp experiment mulative Distribution Function (Logistic-CDF) [25]. Bayon et al.
i Inlet [29] developed an algebraic sigmoid as the solution for the energy
h hot balance equation, and key parameters were fitted with the experi-
cor Correction mental data. The results showed that the analytical model can ac-
ana analytical curately predict the efficiency of the heat storage tank. Kuznetsov
developed an analytical solution for heating a two-dimensional
porous packed bed by using a nonthermal equilibrium flow of in-
Given that industrial-scale TCSTs are very large, detailed ex- compressible fluid [30]. Nelson et al. [31] developed a 1D con-
perimental studies are very expensive. Therefore, some numerical jugate heat conduction model to analyze the stratification of the
studies have been performed on the thermal efficiency and the de- TCST. The results showed that the heat capacity and Biot number
velopment of thermocline to optimize the structure of the TCSTs. can be used to define the stratification of the tank. Votyakov et al.
Researchers developed three-dimensional models of hot water [32,33] developed a new single-phase perturbation model by using
storage tank [13] and one-dimensional two-phase energy equa- an algebraic approximation to predict the temperature distribution
tion models of molten salt thermal storage tank [14], respec- in the TCST. Marafie and Vafai [34] analytically studied the forced
tively. The effects of mass flow rate [15], different types of dif- convection flow through a channel filled with a porous medium
2
C. Xu, M. Liu, S. Jiao et al. International Journal of Heat and Mass Transfer 186 (2022) 122478
3
C. Xu, M. Liu, S. Jiao et al. International Journal of Heat and Mass Transfer 186 (2022) 122478
The main data from the measurement are mass flow rate and tem- in the TCSTs. In this study, 0.15 ≤ θ ≤ 0.85 is defined as the tem-
perate data. perature range corresponding to the thermocline [36]. As shown in
Fig. 5, 70% of the change in temperature in the hot water storage
2.2. Thermocline thickness tanks can be reflected in the thermocline. The following formula
can be used to calculate the thickness of the thermocline:
Thermocline thickness is an important indicator to evaluate the h = xh −xc (2)
stratification efficiency of the TCSTs. Thermocline thickness is as-
sessed by the dimensionless temperature. The dimensionless tem- where h is the thickness of the thermocline;xh is the height
perature relation is defined as follows [35,36] when θ =0.85; and xc is the height when the θ =0.15.
T (x ) − Tc
θ (x ) = (1) 3. Comparison of experimental results with the 1D analytical
Th − Tc
model
where θ is the dimensionless temperature; T (x ) is the tempera-
ture at a certain point in the TCST during charging, K; Th is the in- 3.1. 1D analytical model
let temperature of charging, K; Tc is the initial temperatures of the
TCST, K. The denominator represents the temperature difference of The 3D flow and heat transfer in the TCST can be simplified to
charging, and the numerator represents the difference value be- a 1D problem. The 1D analytical model to analyze the variation in
tween the temperature at any point and the cold water. the thermocline along with time has been developed by Yoo and
During charging, the thermal storage tank is filled with cold Pak [37,38]. The following assumptions are made in the 1D analyt-
water at temperature Tc , and the hot water at temperature Th flows ical model:
into the tank through the diffuser. As shown in Fig. 2, due to buoy-
ancy, the thermocline is formed during charging. The value of θ • The flow and heat conduction of water in the tank are both
increases from 0 at the cold water side to 1 at the hot water side one-dimensional.
4
C. Xu, M. Liu, S. Jiao et al. International Journal of Heat and Mass Transfer 186 (2022) 122478
5
C. Xu, M. Liu, S. Jiao et al. International Journal of Heat and Mass Transfer 186 (2022) 122478
ξ = x/H (8)
τ = F o = ta/H 2 (9)
vH
u=Pe= (10)
a
θ = (T − T0 )/(Ti − T0 ) (11)
where H is the tank height, m; a is the thermal diffusivity, t m2 •s;
is the charging time, s; F o is the Fourier number for the time scale;
Fig. 6. schematic of 1D flow during charging. Pe is the Peclet number for the velocity scale; θ is the dimension-
less temperature; Pe Fc is the normalized time of charging.
The dimensionless form of the initial and boundary condition is
constant during charging, so the averaged velocity of the cross sec-
tion is also unchanged. Thus, the thermocline is gradually formed θ (ξ , 0 )=0 (12)
and completes the exchange of heat with hot and cold water as it
moves downward at a constant rate.
Thus, the following unsteady 1D energy equation can be used
θ (0, τ )=1 (13)
to describe the transient temperature inside the tank as illustrated
in Fig. 5. θ ( ∞ , τ )= 0 (14)
∂T ∂T ∂ T 2
The Laplace transform technique has been applied to solve the gov-
ρc + ρ cv =λ 2 (3)
∂t ∂x ∂x erning equation, the transformed equations are
where x is the axial coordinate measured along the vertical axis d 2 θ (ξ , s ) d θ (ξ , s )
of the tank, m; v = Q/A is the convection velocity, which repre- −u − sθ (ξ , s )+θ (ξ , 0 )=0 (15)
dξ 2 dξ
sents the velocity of the downward movement of the thermocline,
m•s − 1 ; Q is the water mass flow rate during charging, kg•s − 1 ;
θ (0, s)=1/s (16)
A is the cross-sectional area of the TCST, m2 ; T is the temperature
of the working fluid, K, ρ is the working fluid density, kg•m − 3 ;
c is the specific heat capacity of the working fluid, kJ•kg−1 •K − 1 ; θ (∞, s)= 0 (17)
λis the thermal conductivity, W•m − 1 •K − 1 .
where θ (ξ , s ) is the transformation of θ (ξ , τ ).
To obtain the analytical solution for Eq. (3), certain initial con-
Solving the differential Eqs. (15–17) results in
ditions and boundary conditions should be provided. At first, the
TCST is filled with cold water, so the initial condition of the charg- 1 2
θ (ξ , s )=e 2 ξ e−ξ s+ u4
u
6
C. Xu, M. Liu, S. Jiao et al. International Journal of Heat and Mass Transfer 186 (2022) 122478
and 1.2 t/h. Calculation results with the analytical model under the
same operating conditions are also obtained.
7
C. Xu, M. Liu, S. Jiao et al. International Journal of Heat and Mass Transfer 186 (2022) 122478
Fig. 10. Effect of the Pe number on the axial temperature distribution at half of the
charging time.
8
C. Xu, M. Liu, S. Jiao et al. International Journal of Heat and Mass Transfer 186 (2022) 122478
Fig. 12. Effect of the Pe number on the time-wise temperature variations at mid-
height of the tank.
Fig. 16. Temperature variations at different positions for Pecor =1270 with respect to
the normalized time.
Fig. 17. Temperature variations at different positions for Pecor =1100 with respect to
the normalized time.
Fig. 15. Error analysis of the empirical coefficient ε . thermal conduction near the thermocline under different charging
mass flow rates. The use of ε increases the accuracy of the ana-
where ν is kinematic viscosity, and D is characteristic length, lytical solution model in predicting the temperature distribution in
which is the perimeter of the TCST. The physical parameters of wa- the thermal storage tank.
ter are obtained at the average temperature. However, the 1D analytical model with ε still has obvious er-
As shown in Fig. 14, with the increase in the Res value, the cor- rors at the formation stage because of the closeness of the initial
rection coefficient ε also increases. This result suggests that with thermocline to the upper diffuser such that the thermocline is con-
the increase in the charging mass flow rate, the thermal conduc- siderably affected by the diffuser. The formation of the initial ther-
tion of the thermocline is enhanced. The following experimental mocline will be discussed in Section 4.2.
correlation is obtained through fitting: The experimental results of Castell et al. [11] are used to vali-
date the developed 1D analytical model. As shown in Fig. 18, the
ε = 1.22 × 10−6 e2.29 ln(Res ) (26) developed 1D analytical model reduces the average percentage er-
As shown in Fig. 15, the error range of the experimental correlation ror between the results from 58.2% to 33.8%, with the evaluation
is within 10%, which is within a reasonable range. The Eq. (26) is indicator of the thermocline thickness. This result verifies the fea-
obtained by fitting the experimental results, where the maximum sibility of the developed 1D analytical model.
mass flow rate of the experiment is 2.25 t/h. Therefore, the formula
is applicable to the tank Reynolds number (Res ) less than 1708. 4.2. Considering the initial thermocline thickness
Figs. 16 and 17 show the comparison between the results from
the experimental and modified 1D analytical model at mass flow To further modify the 1D analytical model of the forma-
rates of 1.0 and 1.2 t/h, in which the correction numbers P ecor = tion stage of the thermocline, many experiments are performed.
1270 and P ecor = 1100 were used. The average percentage error Flow rates during these tests are varied between 0.6 and 2.25
compared with the developed 1D analytical model dropped to ap- t/h. The results of some selected tests are presented in Fig. 20.
proximately 18.5%. This result indicates that the addition of the Fig. 19 shows the temperature distribution with time with charging
correction coefficient ε improves the accuracy of the 1D analyt- at flow rates of 1. 0 and 1.2 t/h. The formation of the thermoclines
ical model. This coefficient theoretically quantifies the degree of during charging is clearly observed in Fig. 19.
10
C. Xu, M. Liu, S. Jiao et al. International Journal of Heat and Mass Transfer 186 (2022) 122478
is functionally the same when the mass flow rate is less than
1.6 t/h, so the development stage has a larger influence on the
final thickness of the thermocline. The thermocline becomes
thicker with a larger inlet mass flow rate, when the mass flow
rate is greater than 1.9 t/h. Therefore, the inlet disturbance has
a more remarkable influence on the final thickness of the ther-
mocline.
As shown in Fig. 20, the formation of the thermocline is mainly
affected by the diffuser. The inlet Reynolds (Rei ) and inlet Froude
(Fri ) numbers are important dimensionless numbers in the design
of the diffuser. The Rei number represents the ratio of the charac-
teristic inlet inertial and viscous forces and is customarily defined
as follows:
q
Rei = (27)
ν
where ν is the kinematic viscosity, and q is volumetric flow per
unit diffuser length which is the perimeter of the diffuser plate.
The Fri number, which represents the ratio of the characteristic
inertial and body forces, is defined as follows:
q
Fri = (28)
h3 g ρρ
11
C. Xu, M. Liu, S. Jiao et al. International Journal of Heat and Mass Transfer 186 (2022) 122478
Fig. 19. Variation in the temperature distribution with time at different mass flow rates.
As shown in Fig. 22, Eqs. (30) and (31) are used to calculate the
variation curve of the thickness of the thermocline with time, and
the findings are compared with the experimental data. The results
of the two equations are considerably different from the experi-
mental data. However, the growth trend of the results from the
Eq. (31) is similar to the experimental results. The formation of
the thermocline in the tank is complex and related to many fac-
tors, such as the diffuser type and charging mass flow rate. Thus,
this process is difficult to describe using the analytical model.
Therefore, the variation in the thickness of the thermocline
with time cannot be accurately described by Eq. (31). To remedy
this inadequacy, the fitting formula of the thickness of the initial
thermocline obtained by the experiments is added to the 1D ana-
lytical model. The following equation is obtained:
Fig. 20. Variation in the thickness of the thermocline during charging under differ- t
ent mass flows. δ= + δ0 (32)
P ecor
12
C. Xu, M. Liu, S. Jiao et al. International Journal of Heat and Mass Transfer 186 (2022) 122478
Fig. 21. Error analysis of the experimental fitting correlation formula for the initial
thermocline thickness.
5. Conclusions
13
C. Xu, M. Liu, S. Jiao et al. International Journal of Heat and Mass Transfer 186 (2022) 122478
Acknowledgement [16] B. Kurşun, K. Ökten, Effect of rectangular hot water tank position and as-
pect ratio on thermal stratification enhancement, Renew. Energ. 116 (2018)
639–646.
This work was supported by the Natural Science Foundation of [17] C. Xu, Z. Wang, Y. He, Parametric study and standby behavior of a packed-bed
China (Grant Number 52022079) and Energy Security Technology molten salt thermocline thermal storage system, Renew. Energ. 48 (2012) 1–9.
Research Project of Huaneng Group Science and Technology Foun- [18] Z. Yang, S.V. Garimella, Thermal analysis of solar thermal energy storage in a
molten-salt thermocline, Sol. Energy 84 (2010) 974–985.
dation (Grant Number HNKJ20-H87). [19] M. Hänchen, S. Brückner, A. Steinfeld, High-temperature thermal storage using
a packed bed of rocks – Heat transfer analysis and experimental validation,
References Appl. Therm. Eng. 31 (2011) 1798–1806.
[20] R. Bayón, E. Rojas, Simulation of thermocline storage for solar thermal power
[1] M. Liu, S. Wang, J. Yan, Operation scheduling of a coal-fired CHP station in- plants: from dimensionless results to prototypes and real-size tanks, Int. J.
tegrated with power-to-heat devices with detail CHP unit models by particle Heat Mass Tran. 60 (2013) 713–721.
swarm optimization algorithm, Energy 214 (2021) 119022. [21] S.M. Flueckiger, S.V. Garimella, Second-law analysis of molten-salt thermal en-
[2] A.Q. Huang, J. Baliga, FREEDM System: role of power electronics and power ergy storage in thermoclines, Sol. Energy 86 (2012) 1621–1631.
semiconductors in developing an energy internet, in: 21st International Sym- [22] T. VL Jon, Analysis of heat storage and delivery of a thermocline tank having
posium on Power Semiconductor Devices & ICs. Barcelona, SPAIN: IEEE, 2009, solid filler material, J. Sol. Energy Eng. (2011) 133.
pp. 9–12. [23] A.M. Bonanos, E.V. Votyakov, Sensitivity analysis for thermocline thermal stor-
[3] P. Favre-Perrod. A Vision of Future Energy Networks. 1st IEEE-Power- age tank design, Renew. Energ. 99 (2016) 764–771.
Engineering-Society Conference and Exposition in Africa. Univ KwaZulu Natla, [24] N. Altuntop, M. Arslan, V. Ozceyhan, Effect of obstacles on thermal stratifica-
Durban, SOUTH AFRICA2005. p. 13–17. tion in hot water storage tanks, Appl. Therm. Eng. 25 (2005) 2285–2298.
[4] H.O. Njoku, O.V. Ekechukwu, S.O. Onyegegbu, Analysis of stratified thermal [25] R. Bayón, E. Rojas, Analytical function describing the behaviour of a ther-
storage systems: an overview, Heat Mass Transfer 50 (2014) 1017–1030. mocline storage tank: a requirement for annual simulations of solar thermal
[5] K. Fahad, Plate diffuser performance in spherical tank thermocline storage sys- power plants, Int. J. Heat Mass Tran. 68 (2014) 641–648.
tem, J. Energy Resour. Technol. (2016). [26] L. Dae-Young, Analytical characterization and conceptual assessment of solid
[6] A. Bruch, J.F. Fourmigué, R. Couturier, Experimental and numerical investiga- and fluid temperature differentials in porous media, Int. J. Heat Mass Tran. 42
tion of a pilot-scale thermal oil packed bed thermal storage system for CSP (1998) 423–435.
power plant, Sol. Energy 105 (2014) 116–125. [27] Z. Chang, X. Li, Q. Falcoz, Approximate analytical characterization and multi-
[7] H. Mehling, L.F. Cabeza, S. Hippeli, PCM-module to improve hot water heat -parametric optimization design of single-tank thermocline heat storage sys-
stores with stratification, Renew. Energ. 28 (2003) 699–711. tem, Appl. Therm. Eng. 181 (2020) 116010.
[8] D.L. Savicki, H.A. Vielmo, A. Krenzinger, Three-dimensional analysis and in- [28] M.W. Jack, J. Wrobel, Thermodynamic optimization of a stratified thermal stor-
vestigation of the thermal and hydrodynamic behaviors of cylindrical storage age device, Appl. Therm. Eng. 29 (2009) 2344–2349.
tanks, Renew. Energ. 36 (2011) 1364–1373. [29] R. Bayón, E. Rojas, R. Mancilla, Analysis of packed-bed thermocline storage
[9] M.A. Karim, Experimental investigation of a stratified chilled-water thermal tank performance by means of a new analytical function, AIP Conf. Proc. (2018)
storage system, Appl. Therm. Eng. 31 (2011) 1853–1860. 2033.
[10] Z. Wang, H. Zhang, B. Dou, Influence of inlet structure on thermal stratification [30] A.V. K, An analytical solution for heating a two-dimensional porous packed
in a heat storage tank with PCMs: CFD and experimental study, Appl. Therm. bed by a non-thermal equilibrium fluid flow, Appl. Sci. Res. 55 (1995) 83–93.
Eng. 162 (2019) 114151. [31] J.E.B. N, Parametric studies on thermally stratified chilled water storage sys-
[11] A. Castell, M. Medrano, C. Solé, Dimensionless numbers used to characterize tems, Appl. Therm. Eng. 19 (1999) 89–115.
stratification in water tanks for discharging at low flow rates, Renew. Energ. [32] E.V. Votyakov, A.M. Bonanos, A perturbation model for stratified thermal en-
35 (2010) 2192–2199. ergy storage tanks, Int. J. Heat Mass Tran. 75 (2014) 218–223.
[12] Z. He, Y. Qian, C. Xu, Static and dynamic thermocline evolution in the water [33] E.V. Votyakov, A.M. Bonanos, Algebraic model for thermocline thermal storage
thermocline storage tank, Energy Procedia 158 (2019) 4471–4476. tank with filler material, Sol. Energy 122 (2015) 1154–1157.
[13] M.R. Assari, H. Basirat Tabrizi, M. Savadkohy, Numerical and experimental [34] A. Marafie, K. Vafai, Analysis of non-Darcian effects on temperature differen-
study of inlet-outlet locations effect in horizontal storage tank of solar water tials in porous media, Int. J. Heat. Mass Tran. 44 (2001) 4401–4411.
heater, Sustain. Energy Technol. Assess. 25 (2018) 181–190. [35] E. Kaloudis, D.G.E. Grigoriadis, E. Papanicolaou, Large eddy simulation of ther-
[14] J.F. Hoffmann, T. Fasquelle, V. Goetz, A thermocline thermal energy storage mocline flow phenomena and mixing during discharging of an initially homo-
system with filler materials for concentrated solar power plants: experimen- geneous or stratified storage tank, Eur. J. Mech. B. Fluids 48 (2014) 94–114.
tal data and numerical model sensitivity to different experimental tank scales, [36] A. Musser, W.P. Bahnfleth, Parametric study of charging inlet diffuser per-
Appl. Therm. Eng. 100 (2016) 753–761. formance in stratified chilled water storage tanks with radial diffusers: part
[15] C. Xu, Z. Wang, Y. He, Sensitivity analysis of the numerical study on the ther- 1-model development and validation, HVAC&R Res. 7 (2001) 31–49.
mal performance of a packed-bed molten salt thermocline thermal storage [37] H. Yoo, E. Pak, Analytical solutions to a one-dimensional finite-domain model
system, Appl. Energ. 92 (2012) 65–75. for stratified thermal storage tank, Sol. Energy 56 (1996) 315–322.
[38] H. Yoo, E. Pak, Theoretical model of the charging process for stratified thermal
storage tanks, Sol. Energy 51 (1993) 513–519.
14