Energy 170 (2019) 967e977

Contents lists available at ScienceDirect

Energy
journal homepage: www.elsevier.com/locate/energy

Numerical and experimental study on the thermal performance of the

concrete accumulator for solar heating systems
Jacek Sacharczuk*, Dawid Taler

w, Poland
Cracow University of Technology, Faculty of Environmental Engineering, Ul. Warszawska 24 31-155 Krako

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

Article history: The article presents the concept of the solar facility operating in the domestic hot water and space
Received 3 October 2018 heating system, equipped with thermal energy storage based on a water tank and an auxiliary storage
Received in revised form module made of concrete elements. A mathematical model of transient heat transfer in the concrete
16 December 2018
structure using the Control Volume Finite Element Method (CVFEM) has been proposed. The method
Accepted 18 December 2018
allows modelling the transient heat conduction based on a rare mesh of nodes in a relatively short time,
Available online 27 December 2018
with accuracy comparable to traditional Finite Element Method (FEM) results including CFD modelling. A
laboratory stand has been made for the verification of the model, and a set of comparative measurements
Keywords:
Concrete
has been carried out.
CVFEM © 2018 Elsevier Ltd. All rights reserved.
Heat storage
Modelling
Solar energy
Space heating

1. Introduction accumulating structure, thermal properties of the storage material

Water tanks are the most popular method of thermal energy
storage for solar facilities. The alternative method for heat storage is
the use of the thermal capacity of solid materials. In spite of lower
thermal capacity which characterizes solid materials, there are
numerous undeniable advantages of this method. The accumula-
tion material is relatively cheap and can form a part of a building's
construction, or it may be incorporated into its structure. This
particular solution creates many innovative possibilities. However,
the idea of heat storage which uses the thermal capacity of solid
materials is nowhere near new. Quite contrary, it has been known
since ancient times (heating systems based on hypocaust). In solar
heating systems, accumulators which are made of solids are often
used in ACPS (active collection - passive storage) systems [1]. In
ACPS systems, heat is obtained from air solar collectors and is
transferred directly to heated objects characterised by high thermal
capacity afterwards. In systems with active charging and dis-
charging, the air flow is forced through the air channels Fig. 1 or
around solid accumulating elements.
Adequate selection of geometric parameters of the
* Corresponding author.
E-mail addresses: jsacharczuk@pk.edu.pl (J. Sacharczuk), dtaler@pk.edu.pl
(D. Taler).
and convective heat transfer conditions allow previously obtained
solar energy utilisation to spread over the whole day. Regular
concrete or concrete with improved accumulation properties,
firebricks, stone, ceramics, and even sand can be used to build the
accumulator. The accumulator can be structured in the form of a
closed space (A), horizontal floor slab (B) or internal wall (C). Ex-
amples of locations of the heat accumulators in buildings are pre-
sented in Fig. 2.
2. Analysis and modelling
2.1. Literature survey
A precise description of heat storage systems based on solid
materials can be found in the basic solar engineering literature [1].
Solids are commonly used in systems with air collectors [2]. The
model solution for solar energy storage in solid materials is thermal
storage wall - an air collector integrated with a concrete accumu-
lation structure patented by F. Trombe in 1972. The results of the
most recent investigation on Trombe wall are described in
Refs. [3,4]. Popular solutions are packed pebble beds [5]. Their
thermal performance/capacity is usually described by lumped
models [6]. Operation of packed bed heat storage system for air
heaters was described thoroughly in Ref. [7]. Currently, most

https://doi.org/10.1016/j.energy.2018.12.142
0360-5442/© 2018 Elsevier Ltd. All rights reserved.
968 J. Sacharczuk, D. Taler / Energy 170 (2019) 967e977

Nomenclature Greek symbols

a heat transfer coefficient, W/m2K
a thermal diffusivity, m2/s l thermal conductivity, W/mK
A duct cross-section area, m2 m dynamic viscosity, Pa$s
Ai,j,k finite element surface area, m2 r density, kg/m3
bi,j,ci,j,ei matrix coefficients x Darcy-Weisbach friction factor
c specific heat of solid material, J/kgK
cp specific heat of the air, J/kgK Subscripts
d diameter of the tube or air duct, m 1,2,3 nodes numbers
L length of the tube or air duct, m a air, air duct, accumulated
Nu Nusselt number a,b,c midpoints of triangle sides
Pr Prandtl number h hydraulic
q_ heat flux, W/m2 hx heat exchanger
Q heat accumulated in concrete, J in inlet
Q_ heat flow, W o initial
Re Reynolds number out outlet
s length, m O gravity centre of the triangle
t time, s s surface
T temperature, oC Sw inner surface of the air duct
Uw duct cross-section perimeter, m T constant temperature
w airflow velocity, m/s w wall, wetted, water
x,y,z Cartesian coordinates, m

units with ducted airflow are not as thoroughly described in the

literature as systems based on phase change materials and packed
beds [11]. However, they can be used as alternative energy storage
for both water and air solar systems.
A pioneer who applied hollow masonry structures to store
thermal energy is James Kachadorian. In 1978 he patented a heat-
ing system which used solar energy from air collectors or the
direct-gain system and stored it in a gravel-filled or hollowed sub-
floor space with air distributing system by natural or forced cir-
culation [12].
Short and Kutscher [13] investigated a hybrid building heating
system with a concrete accumulator with a channel structure in the
form of a concrete floor slab with an area of 140 m2, with channels
of 150
150 mm square cross-section. The air heated in the col-
Fig. 1. The internal structure of the concrete accumulator.
lector was directed to the air-water heat exchanger, which, as a
result, heated domestic hot water.
The paper by Yuxiang et al. [14], describes the design, modelling,
investigations on energy storage supplied from solar installations
and performance assessment based on monitored data of a
cover systems which use phase change materials (PCM). Analysis of
building-integrated photovoltaic-thermal system thermally
different PCM storage concepts and their thermal performance
coupled with a ventilated concrete slab. In the second paper by the
models are described in Ref. [8]. Analysis of external wall structures
same authors [15] an in-depth study of a ventilated concrete slab
with PCM materials are described in Ref. [9]. PCM enables efficient
used in a near net-zero energy solar house was carried out. Inte-
thermal energy storage but are rather high-priced when compared
gration of hollow-core masonry walls and precast slabs as heat
to concrete or ceramic materials. Ceramic honeycomb elements for
storage units is also presented in Ref. [16]. Fraisse et al. [17] describe
thermal energy storage in a solar thermal power plant which use
a heat storage system with an internal heavy wall with an inner air
air were described in Ref. [10].
gap. The use of ceramic elements for storing heat in energy systems
Heat storage systems which use hollow concrete or ceramic

Fig. 2. Possible accumulator locations in the building.

 J. Sacharczuk, D. Taler / Energy 170 (2019) 967e977 969

is described by Cisek and Taler [18].

2.2. The concept

The concept of the analysed heating system is based on the

thermal energy storage in a water tank and simultaneously in
concrete or ceramic accumulator. The heat used to preheat hot
water and support the heating of a building is obtained from
traditional solar collectors. First of all, the water tank is heated, and
the surplus of acquired energy is accumulated in the structure of a
concrete or ceramic exchanger e an accumulator which is charged
and discharged by the air flowing in internal channels. During the
charging phase, the air circulates in a closed circuit and is heated in
a plate-fin heat exchanger. Afterwards, during the night, the accu-
mulator is discharged by the air blown directly into the heated
space (see Fig. 3).
In favourable climatic conditions, which include high insolation
and high amplitudes of ambient air temperature, the system can be Fig. 5. Experimental facility real view.
used to heat the entire building. In temperate climatic conditions, it
can be used to, for instance, reheat the bathroom space, where
thermal comfort requires a higher space temperature, and solar boiler and an electric stabilising heater was also used. Accumu-
energy obtained for DHW production can easily meet heat demand. lating material fills a rectangular prism measuring
2240 mm
560 mm x 280 mm, placed in a wooden case insulated
by 100 mm EPS. Concrete units are made of heavy concrete of high
2.3. Experimental facility
thermal capacity. The density of accumulating material is
r ¼ 2820 kg/m3, specific heat capacity cp ¼ 0.94 kJ/kgK and thermal
The model installation was constructed in a laboratory at the
conductivity l ¼ 1.7 W/mK. An internal air channel's hydraulic
Faculty of Environmental Engineering, CUT (see Figs. 4 and 5). The
diameter is 178 mm, and the heat exchange surface is 2.75 m2. In-
main parts of the facility are the accumulator made of 16 concrete
lets and outlets from air ducts were profiled with steel sheet
units forming two parallel air channels, air circulation system, two-
fittings.
speed coil plate-fin heat exchanger, and a water tank heated by a
solar collector. During the experiments, alternative heating by a gas
2.4. Mathematical model

The mathematical model of the heating system was based on

Fig. 3. The concept of a solar heating system with auxiliary concrete accumulator.
the assumption that the heat for the storage system was supplied in
a stream of water taken from the domestic hot water tank heated
by the solar facility. The supply water temperature can, therefore,
change according to the scheme of obtaining heat from the solar
collectors and disassembling the hot water in the installation. The
developed and verified the mathematical model of the plate-fin
heat exchanger [19] had been used to calculate accumulator air
inlet temperature on the basis of the water temperatures and flow
between a tank and a heat exchanger. For modelling transient heat
transfer in the accumulator structure, it has been assumed that it is
sufficient to analyse the temperature distribution in the repeatable
modules limited by the symmetry axes of the cross-section [20].

Fig. 4. Scheme of the experimental facility.

970 J. Sacharczuk, D. Taler / Energy 170 (2019) 967e977

Fig. 6. Diagram of the accumulator model structure and accumulation module geometry.

The concrete blocks which were used for the verification of the
mathematical model are shown in Fig. 6. The cross-section of the Tjt¼0 ¼ Ta jt¼0 ¼ T0 (4)
heat storage unit has eight symmetry planes. An element corre-
sponding to 1/8 of the cross-section area (shaded area) can be The air temperature at the accumulator inlet Ta,in is also known
adopted for modelling the temperature field.
As the air temperature in the duct changes in flow-through di- Ta jz¼0 ¼ Ta;in ðtÞ (5)
rection, the temperature distribution should be analysed in several
cross-sections corresponding to the concrete blocks. The inlet temperature Ta,in is equal the air temperature down-
It was assumed that the temperature gradient in the material stream a plate fin-and-tube heat exchanger Ta,hx2.
structure in the z-axis direction within individual modules is The initial-boundary value problem was solved using Control-
negligible. The heat conduction equation for steady-state, two- Volume Finite-Element Method (CVFEM) for determining the
dimensional temperature field in the part of an accumulator cross- temperature in concrete segments and finite difference method
section determined by the symmetry planes Fig. (6) has the (FDM) for determining the air temperature.
following form The finite difference grid for modelling air flow is defined as
follows:



vT v vT v vT
cðTÞrðTÞ ¼ lðTÞ þ lðTÞ (1) zj ¼ ðj  1ÞDz; j ¼ 1; …; 9 (6)
vt vx vx vy vy
The outer surface of the air duct is thermally insulated. The
Dz
convective boundary condition at the inner surface of a duct is as zi ¼ þ ði  1ÞDz; i ¼ 1; …; 8 (7)
follows: 2

 where Dz is equal the length of a concrete unit Fig. 6, and,7. The

vT    coordinate zi refers to the centre of the unit.
lðTÞ  ¼ a Ta  TjSw (2)
vn Sw The following assumptions were adopted to solve the problem
of transient heat transfer in the solid accumulator:
The energy conservation equation for the air flowing through
the internal channel is given by  the channel was divided into n ¼ 8 segments, with a length
corresponding to the length of a single accumulation module.
   the temperature of the storage mass was computed at the cross-
vTa vTa a T w  Ta Uw
ra cp þ wa ¼ (3) section located in the middle of the length of each module
vt vz Aa
(indices i ¼ 1, …,n)
The initial temperature of concrete units and the air is uniform  the staggered grid was used in the air domain

Fig. 7. Channel division with balance planes and location of measurement points.
 J. Sacharczuk, D. Taler / Energy 170 (2019) 967e977 971

 section j ¼ 1 corresponds to the inlet plane to the duct (inlet air Next, the determination of the temperature of the concrete unit
temperature) will be discussed.
 the cross-sections j, and jþ1 correspond to the inlet and outlet of The control volume finite element method (CVFEM) [21] is used
air and coincide with contact planes of the concrete modules to solve the problem of two-dimensional transient heat transfer in
 the cross-section j ¼ nþ1 corresponds to the outlet plane from the planes of the accumulator cross-section, This method allows
the duct. modelling the temperature field in a repeatable section of the
accumulator section using a triangle mesh based on a relatively
Equation (3) was solved using an explicit finite difference small number of nodal points.
method. The time derivative in Eq. (3) was approximated by for- In the CVFEM method, an energy balance is created for the
ward difference quotient and space derivative by the backward control area surrounding each node. The control area around the
finite difference node 1 is formed by one-third of all of the triangles with sides
  coming out of the joint node 1. The energy balance equation is
k
T kþ1 k
a;jþ1  T a;jþ1 T ka;jþ1  T ka;j akj T w;i¼j  T ka:i Uw written for the 1-a-O-c area, in which the point O is the centre of
þ wa ¼ (8) gravity of the triangle 123, and the points a and c are respectively
Dt Dz Aa ra cp
the centres of the sides 1e2 and 1e3 (Fig. 8). The surface area of a
The air temperature at the point zi was assumed to be equal the Fig. 1-a-O-c is one-third of a surface area of a triangle 123.
arithmetic mean of the temperatures Ta,j and Ta,jþ1 (Fig. 7) In the case of a single triangular element 123, the energy balance
equation for the control volume associated with node 1 can be
Ta;j þ Ta;jþ1 written as follows:
Ta;i ¼ (9)
2

A123 dT1 yc  ya
cðT1 ÞrðT1 Þ ¼ lx ðT0 Þ ½ðy2  y3 ÞT1 þ ðy3  y1 ÞT2 þ ðy1  y2 ÞT3 þ
3 dt 2A123
(13)
x  xa
ly ðT0 Þ c ½ðx2  x3 ÞT1 þ ðx3  x1 ÞT2 þ ðx1  x2 ÞT3 
2A123

The superscript kþ1 in Eq. (8) refers to the new time step while k
to the old one If convection occurs on the edge surface 1-a, and the heat flux q_s
is set on the surface 1-c, Eq. (13) takes the form

A123 dT1 yc  ya
cðT1 ÞrðT1 Þ ¼ lx ðT0 Þ ½ðy2  y3 ÞT1 þ ðy3  y1 ÞT2 þ ðy1  y2 ÞT3 þ
3 dt 2A123

  (14)
x  xa 3T1 T2
ly ðT0 Þ c ½ðx2  x3 ÞT1 þ ðx3  x1 ÞT2 þ ðx1  x2 ÞT3  þ a Ta  þ s1a þ q_s s1c
2A123 4 4

tkþ1 ¼ tk þ Dt; k ¼ 0…; t0 ¼ 0 (10)

Energy balance equations written for all control areas form a
vTa system of ordinary differential equations for node temperatures.
The time derivative in Eq. (3) can be neglected since specific
vt
thermal capacity of air ra cp is very small. The finite difference The number of mesh nodes and the corresponding energy conser-
equation Eq. (8) reduces to vation equations in the CVFEM method can be relatively small with
good accuracy of calculations.
 k
 In the case of the analysed concrete accumulator, the symmet-
T ka;jþ1  T ka;j akj T w;i¼j  T ka:i Uw
wa ¼ (11) rical fragment of the cross-section was first divided into triangular
Dz Aa ra cp elements, and then 16 control areas were created. Because of the
symmetry, six fictitious.
Solving Eq. (11) concerning T ka;jþ1 gives: The derivative of temperature in the normal direction (heat flux)
    to the symmetry planes 1-2-3-4 and 13-14-16 are equal zero. If the
D za k U w k D za k U w
T ka;j 1  2Aa r jcp wa þ T w;i¼j 1  2Aa r jcp wa outer surface of a concrete unit was insulated perfectly then, zero
T ka;jþ1 ¼
a a
(12) heat flux was assumed. In this case, the entire external surface of
Dzak Uw the analysed area 1-2-3-4-5-10-11-15-16-14-13 is subject to the
1 þ 2Aa r jcp wa
a
boundary condition

j ¼ 1; …; 9; k ¼ 0; 1; …
972 J. Sacharczuk, D. Taler / Energy 170 (2019) 967e977

Fig. 8. Energy conservation equation in CVFEM for the area 1-a-c; a) inner cell; b) boundary cell with convective boundary condition on the side 1-a; c) boundary cell with heat flux
on the side 1-c.

modules forming the accumulator channel. The use of the explicit

vT
l ¼0
vn
Calculations were also carried out for the case of imperfect
insulation when there were heat losses from the outer accumulator
surface to the environment (nodes 4-5-10-11-15-16).
The system of ordinary differential equations of the first order
for the node temperatures at solid and air domains was solved by
the explicit finite difference method. The number of equations for
the whole accumulator is 16
(n-1), where n ¼ 8 is the number of
(15) method is a significant advantage in online control systems based
on the mathematical model of the accumulator. The explicit
method does not require iteration or solving a system of algebraic
equations at each time step when controlling the air temperature at
the accumulator exit.
Equations (16)e(18) show energy balances for internal, internal
symmetrical and edge nodes of the analysed cross-section.
 Internal node 6

Dt
T6 ðt þ DtÞ ¼ T6 ðtÞ þ
rcðA6105 þ A6910 þ A679 þ A627 þ A632 þ A653 Þ
y65  y610
$ lx $½ðy10  y5 ÞT6 ðtÞ þ ðy5  y6 ÞT10 ðtÞ þ ðy6  y10 ÞT5 ðtÞ
2A6105
x  x610
þly 65 $½ðx10  x5 ÞT6 ðtÞ þ ðx5  x6 ÞT10 ðtÞ þ ðx6  x10 ÞT5 ðtÞ
2A6105
y  y69
þlx 610 $½ðy9  y10 ÞT6 ðtÞ þ ðy10  y6 ÞT9 ðtÞ þ ðy6  y9 ÞT10 ðtÞ
2A6910
x  x69
þly 610 $½ðx9  x10 ÞT6 ðtÞ þ ðx10  x6 ÞT9 ðtÞ þ ðx6  x9 ÞT10 ðtÞ
2A6910
y  y67
þlx 69 $½ðy7  y9 ÞT6 ðtÞ þ ðy9  y6 ÞT7 ðtÞ þ ðy6  y7 ÞT9 ðtÞ
2A679
x  x67
þly 69 $½ðx7  x9 ÞT6 ðtÞ þ ðx9  x6 ÞT7 ðtÞ þ ðx6  x7 ÞT9 ðtÞ (16)
2A679
y  y62
þlx 67 $½ðy2  y7 ÞT6 ðtÞ þ ðy7  y6 ÞT2 ðtÞ þ ðy6  y2 ÞT7 ðtÞ
2A627
x  x62
þly 67 $½ðx2  x7 ÞT6 ðtÞ þ ðx7  x6 ÞT2 ðtÞ þ ðx6  x2 ÞT7 ðtÞ
2A627
y  y63
þlx 62 $½ðy3  y2 ÞT6 ðtÞ þ ðy2  y6 ÞT3 ðtÞ þ ðy6  y3 ÞT2 ðtÞ
2A632
x  x63
þly 62 $½ðx3  x2 ÞT6 ðtÞ þ ðx2  x6 ÞT3 ðtÞ þ ðx6  x3 ÞT2 ðtÞ
2A632
y  y65
þlx 63 $½ðy5  y3 ÞT6 ðtÞ þ ðy3  y6 ÞT5 ðtÞ þ ðy6  y5 ÞT3 ðtÞ
2A653
x63  x65
þly $½ðx5  x3 ÞT6 ðtÞ þ ðx3  x6 ÞT5 ðtÞ þ ðx6  x5 ÞT3 ðtÞ
2A653
 J. Sacharczuk, D. Taler / Energy 170 (2019) 967e977 973

Fig. 9. Division of 1/8 of a cross-section of the accumulating module into triangular

finite elements and control volumes around the nodes.

 internal node 3 on the symmetry axis of the cross section

Dt
T3 ðt þ DtÞ ¼ T3 ðtÞ þ
rcðA354 þ A365 þ A326 þ A3182 þ A31718 þ A3417 Þ
y34  y35
$ lx $½ðy5  y4 ÞT3 ðtÞ þ ðy4  y3 ÞT5 ðtÞ þ ðy3  y5 ÞT4 ðtÞ
2A354
x  x35
þly 34 $½ðx5  x4 ÞT3 ðtÞ þ ðx4  x3 ÞT5 ðtÞ þ ðx3  x5 ÞT4 ðtÞ
2A354
y  y36
þlx 35 $½ðy6  y5 ÞT3 ðtÞ þ ðy5  y3 ÞT6 ðtÞ þ ðy3  y6 ÞT5 ðtÞ
2A365
x  x36
þly 35 $½ðx6  x5 ÞT3 ðtÞ þ ðx5  x3 ÞT6 ðtÞ þ ðx3  x6 ÞT5 ðtÞ
2A365
y  y32
þlx 36 $½ðy2  y6 ÞT3 ðtÞ þ ðy6  y3 ÞT2 ðtÞ þ ðy3  y2 ÞT6 ðtÞ
2A326
x  x32
þly 36 $½ðx2  x6 ÞT3 ðtÞ þ ðx6  x3 ÞT2 ðtÞ þ ðx3  x2 ÞT6 ðtÞ (17)
2A326
y  y318
þlx 32 $½ðy18  y2 ÞT3 ðtÞ þ ðy2  y3 ÞT18 ðtÞ þ ðy3  y18 ÞT2 ðtÞ
2A3182
x  x318
þly 32 $½ðx18  x2 ÞT3 ðtÞ þ ðx2  x3 ÞT18 ðtÞ þ ðx3  x18 ÞT2 ðtÞ
2A3182
y  y317
þlx 318 $½ðy17  y18 ÞT3 ðtÞ þ ðy18  y3 ÞT17 ðtÞ þ ðy3  y17 ÞT18 ðtÞ
2A31718
x  x317
þly 318 $½ðy17  y18 ÞT3 ðtÞ þ ðx18  x3 ÞT17 ðtÞ þ ðx3  x17 ÞT18 ðtÞ
2A31718
y  y34
þlx 317 $½ðy4  y17 ÞT3 ðtÞ þ ðy17  y3 ÞT4 ðtÞ þ ðy3  y4 ÞT17 ðtÞ
2A3417
x317  x34
þly $½ðx4  x17 ÞT3 ðtÞ þ ðx17  x3 ÞT4 ðtÞ þ ðx3  x4 ÞT17 ðtÞ
2A3417

Because of a symmetry T17 ¼ T5 and T18 ¼ T6 .

974 J. Sacharczuk, D. Taler / Energy 170 (2019) 967e977

 edge node 7 with convective heat transfer

2 3
dT1
6 dt 7 2 3
6 7 2 3 T1
Dt 6 7 b1;1 / b1;m b1;mþ1
6 dT2 7 6 T2 7
T7 ðt þ DtÞ ¼ T7 ðtÞ þ 6 7 6 7 6 7
rcðA789 þ A796 þ A762 þ A721 Þ 6 dt 7$½A1 /An  ¼ 1 6 b2;1 / b2;m b2;mþ1 7$6 « 7
6 7 rc 4 « 1 « « 5 6 7
6 « 7 4 Tm 5
y79  y78 6 7 b /b b
$ lx $½ðy8  y9 ÞT7 ðtÞ þ ðy9  y7 ÞT8 ðtÞ þ ðy7  y8 ÞT9 ðtÞ 6 7 n;1 n;m n;mþ1
Ta
2A789 4 dTn 5
x x dt
ly 79 78 $½ðx8  x9 ÞT7 ðtÞ þ ðx9  x7 ÞT8 ðtÞ þ ðx7  x8 ÞT9 ðtÞ 2 3
2A789 2 3 T1
c1;1 / c1;m c1;mþ1 6 T2 7
y y 1 6 c2;1 / c2;m c2;mþ1 7 6 7
lx 76 79 $½ðy9  y6 ÞT7 ðtÞ þ ðy6  y7 ÞT9 ðtÞ þ ðy7  y9 ÞT6 ðtÞ þ 6 7$6 « 7
2A796 4
rc « 1 « « 5 6 7
4 Tm 5
x  x79 cn;1 /cn;m cn;mþ1
þly 76 $½ðx9  x6 ÞT7 ðtÞ þ ðx6  x7 ÞT9 ðtÞ þ ðx7  x9 ÞT6 ðtÞ Ta
2A796 2 3
e1
y  y76
þlx 72 $½ðy6  y2 ÞT7 ðtÞ þ ðy2  y7 ÞT6 ðtÞ þ ðy7  y6 ÞT2 ðtÞ 1 6 e2 7
2A762 þ 6 7$q_ (22)
rc 4 « 5 s
x  x76 em
þly 72 $½ðx6  x2 ÞT7 ðtÞ þ ðx2  x7 ÞT6 ðtÞ þ ðx7  x6 ÞT2 ðtÞ
2A762
The matrices A, B, C, and E, can be generated easily for any shape
y  y12
þlx 71 $½ðy2  y1 ÞT7 ðtÞ þ ðy1  y7 ÞT2 ðtÞ þ ðy7  y2 ÞT1 ðtÞ using, e.g., Excel sheet entering the coordinates of individual nodes.
2A721 It is necessary to determine the heat transfer coefficient on the
x  x12 inner surface of the channel to calculate the air temperature and
þlx 71 $½ðx2  x1 ÞT7 ðtÞ þ ðx1  x7 ÞT2 ðtÞ þ ðx7  x2 ÞT1 ðtÞ
2A721 the temperature of the channel wall. Air and water heat transfer
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi correlations were also used in modelling of the plate fin-and-tube

   2  2
3T7 T1 x7 þ x1 y þ y1
þai Ta;i  þ  x7 þ 7  y7 heat exchanger.
4 4 2 2 In the range of turbulent or transitional flow regimes
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 9 (Re > 2300), which occur in the air flow through the accumulator

   2  2 =
3T7 T8 x7 þ x8 y þ y8 channel, and in water flow through the heat exchanger, various
þai Ta;i  þ  x7 þ 7  y7
4 4 2 2 ; heat transfer correlations were used. In turbulent flow range, the
Petukhov-Kirillov correlation, recommended by Kays and Crawford
(18)
[22].
The computation results are stable if the Fourier and Courant-
Friedrichs-Lewy CFL criteria are satisfied:

aDt 1 aDt 1 wa Dt
< < 1 (19)
ðDxÞ2 2 ðDyÞ2 2 Dz
In a numerical solution, it is convenient to set a system of bal-
ance equations in a matrix form. For perfect insulation of the outer
accumulator surface, the equation system is

1
D$A ¼ ½ðB þ CÞ$T (20)
rc
When the thermal insulation is not perfect, then heat loss occurs
on the outer accumulator surface, and equation system has the
following form

1
D$A ¼ ½ðB þ CÞ$T þ E$q_ s  (21)
rc

where: D column matrix of temperature-time derivatives at nodes

1 … n, A matrix containing surface areas of elements around nodes,
B matrix of coefficients in terms including conduction, C matrix of
coefficients regarding convection, E matrix of coefficients defining
the length of the element boundary section with the heat flux set
on the boundary, T column matrix of temperatures at the nodes and Fig. 10. Theoretical modelling charge-discharge cycle at different air flow rates a)
the air temperature in the channel. mean temperature of accumulator b) energy accumulated in concrete.
 J. Sacharczuk, D. Taler / Energy 170 (2019) 967e977 975

 n correlations for the Nusselt number calculation for the flow con-
ðx=8ÞRePr m ditions found in the concrete accumulator channels (Re < 30,000,
Nu ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi (23)
1:07 þ 12:7 x=8 Pr 2=3 1 mw Pr ¼ 0.7), gives noticeable differences between the values of Nusselt
numbers that reach up to 17%.
Finally, the Gnielinski formula (24) was applied in the analysis of
0:6 < Pr < 1000; 10000 < Re < 106 ; n ¼ 0:11 for heating and 0:25 for heat exchange in concrete accumulator channels. For this correla-
cooling or its form modified by Gnielinski [23]. tion, the best agreement with the experimental results was ob-
  " tained. The correlation (28) was used to modify the air-cooled heat
x
ðRe  1000ÞPr  23 #
8 dh exchanger model when transitional or turbulent flow occurred on
Nu ¼  1  2  1 þ (24) the water side.
x 2 L
1 þ 12:7 8 Pr 3  1
3. Results and discussion
0 < dh =L < 1;
3.1. Theoretical calculations
6
0:6 < Pr < 1000; 4000 < Re < 10 ; The mathematical model developed in the paper was used to
The friction factor x in the relationships (23) and (24) was analyse the process of heating and cooling the accumulator under
calculated using the Filonienko formula valid for tubes or ducts test conditions (Fortran software). The constant temperature of
with the smooth inner wall surface 60  C (corresponding to the maximum real temperature obtained
from the solar source) was maintained to charge the accumulator.
x ¼ ½1:82logðReÞ  1:642 (25) The analysis was performed for three different air velocities of
1.1 m/s, 2.2 m/s, and 4.4 m/s (Fig. 10).
The latest modification of the Petukhov-Kirillov formula can be
found in the study [24]. 3.2. Experimental validation of the model
"  2=3 #
ðx=8ÞðRe  1000ÞPr dh The software allows to simulate the system operation using
Nu ¼ 1þ
1:1252 þ 11:7826ðx=8Þ1=2 Pr 2=3  1 L input data (calculated or measured), i.e., air and water tempera-
tures, air flow velocity upstream the air-cooled heat exchanger. In
(26)
the tests concerning the full charging and discharging cycle of the
The VDI correlation [25] for laminar flow in a duct with constant accumulator, a comparison between measured heat flow rates on
wall temperature has the following form the water and air side of the heat exchanger and the accumulator

2 !3 313
 13  1  12 !3
d 2 6 d
NuT ¼ 43:658 þ 0:7 þ
3 3
1:615 RePr h  0:7 þ RePr h 5 (27)
L 1 þ 22Pr L

Re < 2300
The heat transfer correlation for transitional and turbulent duct
flow, proposed by Taler [24] can be used to calculate the heat
transfer coefficient in the concrete accumulator as well as in air-
cooled heat exchanger.
were done. Also, the energy taken from the air and transferred to
the concrete accumulator during a given period were compared. In
addition, the results of temperature measurements at selected
points of the solid storage mass were compared with the results of
modelling.
The heat flux was measured and compared to computations in
order to determine the heat flux transferred from the insulation to

2 !3 313
 1  1  12 !3
dh 3 2 6 d
NuT ¼ 43:6583 þ 0:73 þ 1:615 2300Pr  0:7 þ 2300Pr h 5
L 1 þ 22Pr L
x " (28)
ðRe  2300ÞPr 1:008  23 #
8 dh d
þ  12  1þ 0 < h < 1; 0:6 < Pr < 1000 Re  2300
x 2
 L L
1:08 þ 12:399 Pr 3  1
8

the environment in the charging period. The overall heat transfer

Several correlations have been used for the calculating of heat coefficient between the outer surface of the concrete duct and
transfer coefficient in the concrete accumulator. Comparing environment was estimated based on experimental tests. This
976 J. Sacharczuk, D. Taler / Energy 170 (2019) 967e977

Fig. 11. Comparison of calculated and measured air temperature (charge phase e
different water inlet temperatures and charging periods). Fig. 12. a) Air temperatures and b) energy balance of storage material (charge-
discharge cycle), model and the experimental course.

allowed taking into account heat losses by insulation to the envi-

ronment in the accumulator mathematical model.
The time variations of air temperatures and accumulated energy
obtained by the mathematical model show good compliance with
the measurements. A comparison of the temperature of the air,
temperature of the concrete accumulator at representative points
and energy accumulated in the concrete is depicted in Figs. 11e13
for two different charge phases (wa ¼ 1.7 m/s Re ¼ 18,200 and
wa ¼ 1.4 m/s Re ¼ 14,600) and also for charge-discharge cycle
(wa ¼ 2.0 m/s Re ¼ 21,600).
3.3. The uncertainty analysis
To assess the uncertainty of results obtained using the method
proposed, test calculations simulating the impact of random mea-
surement errors on the computation results were carried out. The
calculations were conducted for a charge-discharge cycle. The air
temperature was constant and equal to 60  C for the first 12 h
(charging period), and then the air temperature was decreased
to þ 20  C and kept for the next 12 h (discharging period). Then,
simulations of accumulator charge-discharge cycle were performed
with air velocity and air temperature burdened with random er-
rors. The random temperature measurement errors were within
the range of ±0.5 K. The random errors of air flow velocity mea-
surement were in the range of ±10% of the measured value
(2.2 ± 0,2 m/s). Mean air temperature during charge phase was
equal to Ta,in ¼ 60.002  C and the standard deviation was
s ¼ 0.297 K. During the discharge phase, the mean value of per-
turbed air inlet temperatures was Ta,in ¼ 19.996  C and s ¼ 0.289 K
while the mean air velocity was equal to wa ¼ 2.196 m/s, and the
standard deviation amounted to s ¼ 0.144 m/s. The calculation
program used measurement data recorded every 10s. The calcula-
tions were carried out with a time step less than 0.28s.
The calculations were conducted for three cases, separately for
the disturbed air temperature, independently for the perturbed air
velocity and air velocity and temperature disturbed simultaneously
with random errors. The impact of disturbances on both the air
outlet temperature of the accumulator and the mean temperature
of the storage mass was analysed. The most significant deviations
from the “exact” results obtained for undisturbed measurement
data occurred in simulations with disturbances both in the inlet air
temperature and air velocity. The average air temperature differ-
ence at the outlet from the accumulator obtained for the “exact”
and disturbed measurement data was 0.226 K with a standard
deviation of 0.13 K. For the storage mass temperature, the average
difference was 0.15 K with a standard deviation of 0.08 K. Small
temperature differences shown in simulations testify to the lack of
sensitivity of the applied model to the random disturbances of
measurement data.
The test calculations carried out indicate low sensitivity of the
calculation results to accidental errors in the measurement of
temperature and air velocity at the accumulator inlet. Also, the
calculations carried out using actual measurement data confirmed
the high effectiveness of the developed modelling method, which is
characterised by high accuracy of the prediction, even with mea-
surement data loaded with random errors.
 J. Sacharczuk, D. Taler / Energy 170 (2019) 967e977
Fig. 13. Comparison of calculated and measured temperatures at selected points of the
storage mass for two different charge-discharge cycles; location of points is shown in
Figs. 7 and 9; a) constant inlet water temperature Tw1 (Fig. 7); b) increasing inlet water
temperature Tw,1 (Fig. 7).
4. Conclusions
The proposed method of modelling a heating system with a duct
concrete heat accumulator allowed simulation of any charging and
discharging of the accumulator for any time changes of heating
water and air supply parameters. A new method of numerical
modelling of duct storage units was presented. The method is
appropriate for modelling of heat storage units with ceramic or
concrete elements of complex cross-section shape, e.g. hexagonal
shape. It was also considered that the airflow could be laminar,
transient or turbulent. In the literature, it is usually assumed that
the air flow in the heat accumulator is turbulent.
Experimental validation of the new mathematical model of the
heat storage unit confirmed the high accuracy of a developed
model. The advantage of the proposed mathematical model is also
the short time of computer calculations.
The analysed and modelled heating system can be used in
conditions of a large daily amplitude of ambient air temperature. In
the practical implementation of the accumulator, it is
977
recommended to use more extended heat transfer surfaces in the
accumulator channels.
References
[1] Duffie JA, Beckmann WA. Solar engineering of thermal processes. second ed.
New York: J.Wiley and Sons Inc.; 1991.
[2] Tyagi VV, Panwar NL, Rahim NA, Kothari Richa. Review on solar air heating
system with and without thermal energy storage system. Renew Sustain
Energy Rev 2012;16:2289e303.
[3] Rabani M, Kalantar V, Rabani M. Heat transfer analysis of a Trombe wall with a
projecting channel design. Energy 2017;134:943e50.
[4] Herna
ndez-Lopez I, Xama
n J, Ch

avez Y, Herna
ndez-Perez I, Alvarado-Jua
rez R.
Thermal energy storage and losses in a room-Trombe wall system located in
Mexico. Energy 2016;109:512e24.
[5] Kumar A, Man-Hoe Kim. Solar air-heating system with packed-bed energy-
storage systems. Renew Sustain Energy Rev 2017;72:215e27.
[6] Esence T, Bruch A, Molina S, Stutzb B, Fourmigue
JF. A review on experience
feedback and numerical modeling of packed-bed thermal energy storage
systems. Sol Energy 2017;153:628e54.
[7] Singh PL, Deshpandey SD, Jena PC. Thermal performance of packed bed heat
storage system for solar air heaters. Energy Sustain Dev 2015;29:112e7.
[8] Li Pei-Wen, Cho Lik Chan. Thermal energy storage analyses and designs.
London: Elsevier Academic Press; 2017.
[9] Chwieduk DA. Dynamics of external wall structures with a PCM (phase change
materials) in high latitude countries. Energy 2013;59:301e13.
[10] Qing Li, Fengwu Bai, Bei Yang, Yan Wang, Mingxu Han. Dynamic simulations
of a honeycomb ceramic thermal energy storage in a solar thermal power
plant using air as the heat transfer fluid. Appl Therm Eng 2018;129:636e45.
[11] Singh H, Saini RP, Saini JS. A review on packed bed solar energy storage
systems. Renew Sustain Energy Rev 2010;14:1059e69.
[12] Kachadorian, J., Solar heated concrete slab building structure, US patent
4127973, 1978.
[13] Short W, Kutscher CF. Analysis of an active charge/passive discharge solar
space conditioning system. Anaheim: American Solar Energy Society; 1984.
[14] Yuxiang Chen, Athienitis AK, Galal Khaled. Modeling design and thermal
performance of a BIPV/T system thermally coupled with a ventilated concrete
slab in a low energy solar house: Part 2 ventilated concrete slab BIPV/T system
and house energy concept. Sol Energy 2010;84:1908e19.
[15] Yuxiang Chen, Galal Khaled, Athientis AK. Modeling design and thermal
performance of a BIPV/T system thermally coupled with a ventilated concrete
slab in a low energy solar house: Part 1 BIPV/T system and house energy
concept. Sol Energy 2010;84:1892e907.
[16] Yuxiang Chen, Galal Khaled, Athienitis AK. Integrating hollow-core masonry
walls and precast slabs into building space heating and cooling. J Build Eng
2016;5:277e87.
[17] Fraisse G, Johannes K, Trilat-Berdal V, Achard G. The use of a heavy internal
wall with a ventilated air gap to store solar energy and improve summer
comfort in timber frame houses. Energy Build 2006;38:293e302.
[18] Cisek P, Taler D. Numerical and experimental study of a solid matrix Electric
Thermal Storage unit dedicated to environmentally friendly residential
heating system. Energy Build 2016;130:747e60.
[19] Taler D. Calculations and experimental tests of heat exchangers. first ed.
Cracow: Publishing House of Cracow University of Technology; 2016 [in
Polish].
[20] Sacharczuk J, Taler D. A concrete heat accumulator for use in solar heating
systems e a mathematical model and experimental verification. Arch Therm
2014;35:281e95.
[21] Baliga BR, Atabaki N. Control-volume based finite-difference and finite-
element methods. In: Minkowycz WJ, Sparrow EM, Murthy JY, editors.
Handbook of numerical heat transfer. New York: J. Wiley and Sons Inc.; 2006.
[22] Kays W, Crawford M. Convective heat and mass transfer. Boston: McGraw-
Hill; 1993.
[23] Incropera F. Fundamentals of heat and mass transfer. sixth ed. New York: John
Wiley and Sons; 2007.
[24] Taler D. Energy Mathematical modeling and experimental study of heat
transfer in a low-duty air-cooled heat exchanger. Energy Convers Manag
2018;159:232e43.
[25] Gnielinski V. Heat transfer in pipe flow. In: VDI heat atlas. Berlin, Heidelberg:
Springer Verlag; 2010. p. 693e9. Chapter G1.