international journal of refrigeration 80 (2017) 238–251

Available online at www.sciencedirect.com

ScienceDirect

j o u r n a l h o m e p a g e : w w w. e l s e v i e r. c o m / l o c a t e / i j r e f r i g

Novel compact bed design for adsorption cooling

systems: Parametric numerical study

Ramy H. Mohammed a,b,*, Osama Mesalhy a,b, Mohamed L. Elsayed a,b,

Louis C. Chow a
a
Department of Mechanical and Aerospace Engineering, University of Central Florida, Orlando, FL 32816-2450,
USA
b
Department of Mechanical Power Engineering, Zagazig University, Zagazig 44159, Egypt

A R T I C L E I N F O A B S T R A C T

Article history: A new bed configuration consists of two layers of packed beads separated by vapor passage
Received 18 January 2017 is simulated using transient three-dimensional local thermal non-equilibrium model (LTNE).
Received in revised form 29 March Darcy–Brinkman equation is solved in both the porous layers and the vapor passage. Silica-
2017 gel/water is selected as a working pair. Heat and mass diffusion time are calculated from
Accepted 30 April 2017 the scaling analysis of the governing equations. Results show that reducing particle diam-
Available online 4 May 2017 eter and adsorbent bed thickness while enhancing the bed thermal conductivity can lead
to a dramatic improvement in specific cooling power (SCP). Also, the feeding vapor passage
Keywords: is needed for particle size smaller than 0.5 mm but it can be removed for bigger particles.
Adsorption cooling Analysis of results indicates that the adsorption process is controlled by heat diffusion re-
Silica-gel/water sistance when heat diffusion time to mass diffusion time ratio (tth/tm)~O(100) or more. While
Time scale ratio the adsorption is controlled by mass diffusion resistance when (tth/tm)~O(1) or less.
SCP © 2017 Elsevier Ltd and IIR. All rights reserved.
Vapor passage
LTNE

Conception d’un nouveau lit compact pour systèmes de

refroidissement à adsorption : étude numérique
paramétrique
Mots clés : Refroidissement par adsorption ; Gel de silice/eau ; Rapport d’échelle de temps ; SCP ; Passage de la vapeur ; LTNE

Generally, energy is consumed in huge amounts as the economy

1. Introduction of the society and the industry develop swiftly, and a cau-
tious eye has to be kept on environmental pollution and
Sustainable development is a mutual follow up for research- guarding the ozone layer. The main critical problems of the
ers worldwide and wise energy utilization is a key element. vapor compression refrigeration (VCR) and the heat pump

* Corresponding author. Department of Mechanical Power Engineering, Zagazig University, Zagazig, 44159, Egypt.
E-mail address: rhamdy@zu.edu.eg, rhamdy@knights.ucf.edu (R.H. Mohammed).
http://dx.doi.org/10.1016/j.ijrefrig.2017.04.028
0140-7007/© 2017 Elsevier Ltd and IIR. All rights reserved.
 international journal of refrigeration 80 (2017) 238–251 239

systems are energy utilization and the refrigerants used in the

Nomenclature refrigerators may create greenhouse effect (Youssef et al., 2016).
Adsorption cooling system is an alternative technique to me-
asv area of gas–solid interface per unit volume
[m−1] chanical vapor compression refrigeration system. It is powered
CPv cooling capacity per unit volume [Kw m−3] by low-grade heat, solar energy, or waste heat from indus-
Cf drag coefficient of the Forchheimer term trial processes or automotive engines. Moreover, it uses
Cp specific heat [J kg−1 K−1] environmentally friendly refrigerants like water as the working
dp particle diameter [m] fluid (Sah et al., 2016; Youssef et al., 2016). This system con-
D mass diffusivity [m2 s−1] sists of three main parts: adsorber/desorber, condenser, and
g‘ derivative of equilibrium isotherm evaporator (Saha et al., 2016). To produce cooling power, the
Hads adsorption cooling cycle undergoes two main processes:
heat of adsorption [J/kg]
heating–desorption–condensation and cooling–adsorption–
h heat transfer coefficient [W m−2 K−1]
K permeability [m]2 evaporation. The sorption bed is the core of the system in which
k thermal conductivity [W m−1 K−1] the working fluid is adsorbed/desorbed to compensate the work
Nu Nusselt number done in conventional mechanical compression cycle. The re-
P pressure [Pa] frigerant vapor is desorbed by heating the adsorbent material
Pr Prandtl number and condensing in the condenser while the refrigerant liquid
R universal gas constant [8314 J kmol−1 K−1] evaporates in the evaporator and adsorbed by cooling the ad-
Re Reynolds number sorbent. The performance of the cycle is evaluated by using
S source term the specific cooling power (SCP) parameter which is the cooling
SCP specific cooling power [kW kg−1 of adsorbent] power per unit mass of adsorbent.
T time [s] The requirements for a suitable refrigerant are high latent
T temperature [K] heat of vaporization, thermal stability, environmental harm-
t1 bed thickness [m] less and non-flammable (Wang et al., 2014). The common
t2 vapor passage [m] refrigerants for adsorption refrigeration system are ammonia,
U velocity vector [m s−1] water, ethanol and methanol. Water is preferred because it is
W bed width [m] harmless and non-flammable. However, water cannot produce
X water uptake [kgw kg−1 of adsorbent] evaporation temperatures below 0 °C and has low adsorption
X∞ equilibrium adsorption capacity [kgw kg−1 of capacity (Wang et al., 2014). Moreover, the system works under
adsorbent] a low vacuum pressure and sealing could be a problem. Silica
gel has the ability to adsorb water with good stability. Des-
Greek symbols orption temperature should not be higher than 120 °C, and it
α thermal diffusivity [m2 s−1] can be lower than 90 °C (Wang et al., 2014). Such a low des-
ε porosity orption temperature is very suitable for solar energy and waste
μ dynamic viscosity [Pa. s] heat energy utilization.
ρ density [kg m−3] Although this cycle has considerably sparked attention in
τ tortuosity the recent years and is widely used in many applications, the
existing adsorption cooling systems still cannot compete with
Subscripts traditional vapor compression cooling systems due to their
bed bed lower performance. Significant efforts to enhance the perfor-
c cooling, condenser mance of the adsorption bed, which is the core of the cycle,
e evaporator is required to increase the SCP (Solmuş et al., 2015). The overall
performance of thermal adsorption cooling systems is mainly
eff effective
limited by heat and mass transfer inside the adsorbent bed
i initial
because of the low thermal conductivity of the adsorbent par-
K Knudsen
ticles (~0.1–0.4 W/m.K) (Freni et al., 2002; Poyelle et al., 1999;
m mass
Tamainot-Telto and Critoph, 2001) and the low mass diffusivity
M molecular weight
of adsorbent-adsorbate pairs (~10−8 to 10−14 m2/s) (Sharafian and
max maximum
Bahrami, 2013; Sharafian et al., 2015). This results in low heat
mo momentum
and mass transfer rates inside the adsorbent beds and con-
ord ordinary
sequently heavy and bulky systems. Accordingly, many attempts
p particle
have been made to enhance the heat and mass transfer char-
s solid phase, surface
acteristics of the adsorbent beds and to minimize intraparticle
sat saturation
and interparticle mass transfer resistances. However, reduc-
t total
ing heat transfer resistances inside a bed tends to increase the
Th thermal
mass transfer resistances and vice versa (Solmuş et al., 2015).
v vapor phase
Therefore, the heat and mass transfer conditions inside the
w water
adsorbent bed and the dimensions of the bed need to be in-
vestigated well before an optimal, high-performance bed can
be designed.
240 international journal of refrigeration 80 (2017) 238–251
Up to now, numerous investigations on various adsorp-
tion system configurations, experimental investigations and
numerical simulation of the adsorption cycles were reported.
Sapienza et al. (2014) constructed an experimental set up to
investigate the effect of the particle size of adsorbent placed
on a flat plate heat exchanger with monolayer and multi-
layer configurations on the dynamic performance of an
adsorbent/heat exchanger and thus, on the SCP. Miltkau and
Dawoud (2002) presented a one-dimensional model taking into
account combined heat and mass transfers to examine the in-
fluence of layer thickness on the dynamics of a small scale
adsorption heat pump. Freni et al. (2012) simulated non-
isothermal water dynamics for adsorbent-heat exchanger
configurations of one, two and four layers of loose grains of
Fuji silica type RD. It was found that the water adsorption is
noticeably faster for the monolayer of silica gel grains due to
the efficient heat transfer from the adsorbent grains to the
coolant in the heat exchanger. The heat transfer became a
bottleneck for multilayer of grains due to point-to-point contact
between the adsorbent grains. The size of the adsorbent grain
was another crucial effect on the dynamic behavior of the sorp-
tion process as the process is much faster for smaller grains.
In the same way, the SCP and COP (coefficient of perfor-
mance) were reduced with increasing the grain size and the
number of grain layers (Wang et al., 2004).
To enhance the heat transfer inside the bed, Mauran et al.
(1993) and Han et al. (1998) suggested the use of natural graph-
ite as a heat transfer additive in adsorption heat pumps.
Composite and consolidated solid desiccant materials were pro-
posed to enhance the thermal conductivity and adsorption
performance of silica gel bed (Wang et al., 2016; Zheng et al.,
2014). Fins were also used to decrease the heat transfer resis-
tance caused by the adsorbent particles and increase the heat
transfer surface area leading to improved adsorption and de-
sorption processes and a shorter cycle time (El Fadar, 2015; Hong
et al., 2015; Jiang et al., 2015; Li et al., 2016; San and Tsai, 2014).
Lang et al. (1999) proposed a design of the adsorber heat ex-
changer, in which a mono-layer of loose pellets of an adsorbent
is placed on the surface of the fins of a finned tube heat ex-
changer. Stricker (2002) proved that this design is effective
during the investigations of a multi-modular adsorption heat
pump. Finless and finned tube-type adsorbent bed for a thermal
wave adsorption cooling cycle was numerically investigated
using a 2D coupled heat and mass transfer model (Çağlar, 2016).
It was found that doubling the fin thickness does not have a
significant effect on the heat transfer while increasing the
number of fins enhances the heat transfer significantly.
Sharafian et al. (2015) pointed out that the optimum fin spacing
for a finned tube heat exchanger packed with 2–4 mm silica
gel beads is about 6 mm. The contact resistance between par-
ticles and fin surface can be eliminated by coating the metal
surface with a thin layer of adsorbent (Freni et al., 2007).
However, this leads to a low adsorbent/metal mass ratio, thus
affecting the size and weight of the adsorption cooling system.
Therefore, this is not a realistic solution.
Although the previous studies widely investigated the per-
formance of adsorption cycles for different design parameters
and operating conditions, few of the reported adsorber beds
in the literature were specifically designed for commercial ad-
sorption cooling systems. Adsorption beds described in the
literature are either cylindrical annulus (Solmuş et al., 2015;
Wang et al., 2014) or fin and tube heat exchangers filled with
silica-gel beads (Sharafian et al., 2015; Wang et al., 2014). The
cylindrical annulus beds require more volume. On the other
hand, the finned tube beds suffer from low cooling power due
to the high percentage of the metallic material compared to
the adsorbent because few particles should be inserted between
each two adjacent fins.
Although researchers pointed out that the bed perfor-
mance mainly depends on the heat and mass transfer inside
the bed, this information still needs more clarification. Fur-
thermore, little attention has been focused on investigating the
bed pressure drop, the vapor channel feeding the adsorbent,
and the transient heat and mass transfer behavior of the ad-
sorbent bed (Guilleminot et al., 1987; Jemni and Nasrallah, 1995;
Yong and Sumathy, 2004). This study proposes a new modular
bed design consists of an array of adsorbent modules placed
in cooling water. Each model consists of two layers of adsor-
bent separated by a vapor passage. To estimate the importance
of the vapor passage, the governing equations are solved in
both domains (vapor and porous layer). The effect of enhanc-
ing the bed thermal conductivity, changing the bed dimensions,
changing the size of the feeding vapor passage on the SCP is
investigated and analyzed. Also, the present study combines
the bed characteristics (i.e. adsorbent particle diameter, bed
thickness, thermal conductivity, specific heat, and thermal
diffusivity) and operation conditions (i.e. initial and cooling
water temperature) into one parameter called time scaling ratio.
The time scaling ratio specifies under what conditions the bed
performance is controlled by heat transfer or mass transfer or
both. This parameter provides guidelines for the design of a
high performance adsorption bed.
2. Proposed bed geometry
The proposed bed, as shown in Fig. 1A and 1B, consists of re-
peated packed bed cells (modules). Each module consists of
two layers of packed silica-gel beads separated by a narrow
vapor passage (Fig. 1C). A small percentage of metal wires may
be added with the silica-gel beads to enhance the thermal con-
ductivity. The heat transfer fluid passages surround the cell from
all sides except the upper side, which is subjected to the water
vapor from/to the evaporator/condenser as shown in Fig. 1C
and 1D.
2.1. Mathematical model
The flow and heat transfer processes in the proposed adsorber
(Fig. 1D) include: vapor flow in both packed bed layer and the
vapor passage, heat transfer from the cooling fluid to both solid
and vapor phases, and vapor mass transfer to the solid silica-
gel due to adsorption process. The following assumptions are
considered in deriving the 3D governing equations for laminar
flow:
(1) The porosity and effective thermal conductivity of the
porous media are assumed to be constant.
(2) The vapor adsorbed by the adsorbent is assumed to have
the same properties of its liquid phase.
 international journal of refrigeration 80 (2017) 238–251 241

Fig. 1 – A schematic view of (A) compact bed configuration, (B) top view of the entire bed, (C) one cell of adsorber, and (D)
computational domain.

(3) The vapor phase is assumed to behave as an ideal gas.
(4) The convective heat transfer coefficient between cooling
water and outer adsorber surface is constant.
(5) Typical adsorption process involves internal heat gen-
eration in solid phase so, vapor and solid phases are
assumed to be at thermal non-equilibrium condition.
Based on these assumptions, the flow in the vapor channel
is governed by (Solmuş et al., 2015):
• Conservation of mass
∂ρ 
∂t
( )
+ ∇ ⋅ ρU = 0
• Conservation of momentum
in the pure vapor zone. This means that the pressure in the
pure vapor zone and in the pores is continuous over the in-
terface between the two domains. It also means that continuity
is enforced between the fluid velocity in the pure fluid and the
Darcy velocity in the porous domain (Le Bars and Worster, 2006).
The change of water uptake is defined by the Linear Driving
Force (LDF) model (Ali and Chakraborty, 2016):
dX 60D
= 2 ( X∞ − X ) (7)
dt dp
where X∞ is the equilibrium adsorption uptake of the adsor-
(1)
bent which is a function of the adsorbent temperature and
adsorbate pressure, X∞ = f (p, Ts). Its value is evaluated using
the following modified Dubinin–Astakhov (D-A) equation
(Solmuş et al., 2015).


∂ρU   ⎡ ⎛ T ⎞ ⎤
1 .6

∂t
( )
+ ∇ ⋅ ρUU = −∇p + μ∇2U + Smo (2) X∞ = 0.346 exp ⎢ −5.6 ⎜ s − 1⎟ ⎥
⎝ Tsat ⎠ ⎦
(8)
⎣

For the flow in the porous layer, the Darcy–Brinkman equa- where Ts is the solid temperature and Tsat is the saturation tem-
tion is used and can be written as (Mitra et al., 2016): perature. Due to the pressure drop throughout the bed, the
saturation temperature is not constant during the adsorption
∂ρ 
εt
∂t
( )
+ ∇ ⋅ ρU = Sm = − (1 − ε t ) ρs
dX
dt
(3) process. Accordingly, the saturation temperature is calculated
from the following equation which is valid up to 0.2 bar.


∂ρU   μ  cfρ   Tsat = 191.054 ∗ p0.0554 (9)
εb
∂t
( )
+ ∇ ⋅ ρUU = −ε b2∇p + ε b μ∇2U − ε b2 U − ε b2
K K
UU (4)

Inside the pores of the silica-gel particle, the diffusion mainly

where, εb and εt represents the packed bed porosity and the occurs due to the ordinary and Knudsen diffusion mecha-
total porosity. The total porosity can be expressed in terms of nisms. The ordinary diffusion occurs when the molecules of
the bed and particle porosities (εp) as (Solmuş et al., 2015), the vapor collide with one another more often than with the
pore walls, and the Knudsen diffusion occurs when the vapor
ε t = ε b + (1 − ε b ) ε p (5) molecules collide more frequently with pore walls than with
one another (Yong and Sumathy, 2004).
The last two terms in equation (4) represent the viscous and
inertial drag of the porous solid on the flow motion, (also known Tv1.685
Dord = 1.735 ∗ 10−9 (10)
as the Darcy and Forchheimer terms, respectively). Only Darcy p
effect is adopted and K represents the permeability of the
porous bed, which can be expressed as (Solmuş et al., 2015):
DK = 22.86Tv0.5rpore (11)

ε b3d2p
K= (6) The total pore diffusion can be set as:
150 (1 − ε b )2
−1
εp ⎡ 1 1 ⎤
The flow in the whole domain is solved with the Darcy– Dpore = + (12)
τ v ⎢⎣ Dord DK ⎥⎦
Brinkman equation where εb = 1, the Darcy source term vanishes
242 international journal of refrigeration 80 (2017) 238–251

where τv is the tortuosity factor of the vapor phase in silica-

Table 1 – Thermo-physical properties of silica-gel/water
gel. There is another diffusion mechanism, which is called
bed.
surface diffusion. The surface diffusion is the transport of ad-
Property Value Unit
sorbed molecules on the adsorbent surface by hopping, which
takes into account that the vapor molecules move on the Cps 924 J kg−1K−1
surface by jumping from site to site with a specific velocity Cpv 1800 J kg−1K−1
Cpw 4200 J kg−1K−1
(Sladek et al., 1974). Based on the hopping mechanism, Sladek
εb 0.40 –
et al. (1974) proposed the following expression for surface εp 0.42 –
diffusion. ρs 2027 Kg m−3
dp 0.25, 0.5, 1.0, 1.5, 2.0 mm
1.6 ∗ 10−6 ⎛ −0.974Hads ⎞ rpore 1.1 nm
Ds,eff = exp ⎜ ⎟⎠ (13)
τs ⎝ Ts pev 1228 Pa
Tc 313 K
hc 100 W m−2 K−1
where, τs is the solid tortuosity factor and Hads is the heat of
µ 1.5e-5 Pa s
adsorption. The heat of adsorption is a function of adsorbate
kv 0.024 W m−1 k−1
concentration. For RD silica gel, the dependency of Hads on ks,eff 0.15, 0.5, 1.5, 2.0 W m−1 k−1
concentration takes the following form (Pesaran and Mills, τv, τs 2.8 –
1987),

Hads = −1400X + 2950 kJ kg −1 (14) 6 (1 − ε t )

asv = and the interfacial heat transfer coefficient
dp
Since the pore void diffusion and the surface diffusion occur hsv is defined as (Nield and Bejan, 2006):
in parallel, the total diffusivity in the porous structure can be
written as: Nu = (2.0 + 1.1 Pr 0.33 Re0.6 ) (18)

g′
D = Ds,eff + Dpore (15) 1 dp dp
ρp = + (19)
hsv kv Nu 10 kv

where ρp is particle density and g’ is derivative of equilib-

where Pr is the Prandtl number of the vapor and Re is the Reyn-
rium isotherm, which is 0.2 for RD silica gel (Pesaran and Mills,
olds number based on particle diameter.
1987).
All thermo-physical properties of the silica-gel/water pair
Due to the big difference in the thermal diffusivity between
and the main simulation parameters are given in Table 1
the vapor and solid phases, it is expected that the heat will
(Solmuş et al., 2015).
diffuse in a very different rate in each phase. So, two differ-
ent energy equations are used for the vapor and solid phases
2.2. Numerical solution
(Solmuş et al., 2015).
The continuity, momentum, and energy equations are coupled
• For the vapor phase
and solved numerically using the finite volume technique. A
staggered grid is used and hybrid differencing scheme is se-
∂Tv 
(ε b ρCp,v )
∂t
( )
+ ∇ ⋅ ρCp,vUTv = ∇ ⋅ [kv,eff ∇Tv ] − asvhsv (Tv − Ts ) lected to calculate the fluxes at the interfaces. The thermal
(16) conductivity is calculated by harmonic mean expression at the
∂X
− (1 − ε t ) ρsCp,v (Ts − Tv ) control volume interfaces.
∂t

The vapor pressure, temperature, and density are related k1k2 (Δx1 + Δx2 )
ki =
by the equation of state p = ρRTv M k1 Δx2 + k2 Δx1

• For the solid phase

The standard SIMPLE algorithm is used to solve the coupled

∂Ts
(1 − ε t ) ρs [Cp,s + Cp,w X ] = ∇ ⋅ [ks,eff ∇Ts ] + asvhsv (Tv − Ts ) continuity and momentum equations (Patankar, 1980). The
∂t
(17) energy equation is solved separately every iteration after solving
∂X
+ (1 − ε t ) ρsHads the continuity and momentum equations. The adsorption
∂t
source term and the density change with time in the conti-
The effective thermal conductivities of the vapor and solid nuity equation appeared as a source terms in the developed
phases are estimated as: discretized pressure correction equation. A fully implicit for-
kv,eff = ε t kv τ v and ks,eff = (1 − ε t ) ks τ s , where ks and kv are mulation is used for the time dependent terms and the
the thermal conductivity of solid silica and water vapor, combined convection diffusion coefficients are evaluated. The
respectively. The parameter a sv represents the heat resulting discretized equation for a control volume around an
exchange area between the vapor and solid phase, which is arbitrary point P for any variable ϕ (u, v, w, p’, Ts, and Tv) takes
defined as the following form (Mesalhy et al., 2005; Patankar, 1980):
 international journal of refrigeration 80 (2017) 238–251 243

6
Apφ p = ∑ Anφn + Sφ
n =1

where, n refers to the neighbor points of the point P.

A FORTRAN program is developed to solve the discretized
equations iteratively using a line by line solver based on
TDMA (Tri-Diagonal Matrix Algorithm) (Mesalhy et al., 2005).
The convergence is assumed to be achieved when the
residuals in the velocities (u, v, and w) are less than 10−3 and
the residuals in the temperature equations reach less than
10−6.

2.3. Initial and boundary conditions

The solid and gas temperature, pressure, and adsorbate con-

centration distributions inside the adsorbent bed are considered Fig. 2 – Time evolution of the mean bed temperature and
to be uniform initially. The initial temperature and pressure dimensionless water uptake.
of the bed are 350 K and 1,228 Pa, respectively, and the initial
water vapor uptake is 0.198 kg kg−1 of adsorbent.
dTs
−ks (t, x, y, w) = h (Ts − Tc ) ,
p ( 0, x, y, z) = Pe, X ( 0, x, y, z) = Xi, dx
(26)
(20) dTv
Ts ( 0, x, y, z) = Tv ( 0, x, y, z) = Ti − kv (t, x, y, w) = h (Tv − Tc )
dx

At the inlet, it is assumed that the adsorbate vapor pres-

sure and temperature are 1,228 Pa and 283 K, respectively.
3. Results and discussion
p (t, x, H, z) = Pe, Ts (t, x, H, z) = Tv (t, x, H, z) = Te (21)
3.1. Present model validation
At the solid impermeable wall (i.e. the outer surface of the
Validation of the present code is carried out by comparing the
bed), heat transfer occurs between the wall and the cooling fluid
results obtained from the present model with the behavior of
of 313 K with a heat transfer coefficient of 100 W m−2 K−1. A sym-
a simple bed (Aristov et al., 2012; Freni et al., 2012; Glaznev and
metry boundary condition is used for the vapor passage vertical
Aristov, 2010). A good agreement, with maximum tempera-
surface (Fig. 1D).
ture error of 0.5% K, is established as shown in Fig. 2 when the
effective thermal conductivity of the bed is about 0.2 W m−1
∂p ∂Ts ∂T
(t, 0, y, z) = 0, (t, 0, y, z) = v (t, 0, y, z) = 0 (22) K−1. This validation demonstrates the suitability of the model
∂x ∂x ∂x
proposed in this study.

dTs
−ks (t, t1 + t 2, y, z) = h (Ts − Tc ) , 3.2. Base case
dx
(23)
dT
− kv v (t, t 1 + t 2, y, z) = h (Tv − Tc ) Silica-gel particle adsorbs the water vapor that causes a
dx
vapor pressure difference between the inlet evaporator pres-
dTs sure and the pressure inside the bed. This pressure difference
−ks (t, x, 0, z) = h (Ts − Tc ) , drives the vapor to flow in the vapor channel as well as in
dx
(24)
dT the packed bed voids. The base case is 10 × 10 × 1 cm packed
− kv v (t, x, 0, z) = h (Tv − Tc )
dx bed with particle diameter of 0.5 mm and vapor channel of
10 × 10 × 0.1 cm and effective solid thermal conductivity
dTs of 0.15 W m−1 K−1. A grid size of 45 × 100 × 60 and time step
−ks (t, x, y, 0) = h (Ts − Tc ) , of 0.5 s gave a grid independent solution and it is sufficient
dx
(25)
dT to capture the temperature and flow details as shown in
− kv v (t, x, y, 0) = h (Tv − Tc )
dx Table 2.

Table 2 – Effects of the number of grid points and time steps on the uploading ratio.
Grid size 30 × 60 × 40 40 × 80 × 50 45 × 100 × 60
Time step (s) 0.5 0.25 0.5 0.25 0.5 0.25
Bed uploading ratio 0.82287 0.82291 0.82152 0.82153 0.82152 0.82161
244 international journal of refrigeration 80 (2017) 238–251

Fig. 3 – Pressure contours and vapor flow streamlines at different planes after 500 s.

Firstly, Reynolds number based on the bed dimensions high and the maximum temperature in the bed decreases only
and vapor velocity is calculated to check whether the flow is from 350 to 345 K due to the very low thermal diffusivity of
laminar or turbulent. This calculation showed that Reynolds the bed (αeff = 2 × 10−7 m2 s−1) (Mitra et al., 2016).
number is always less than 100 and assumption of laminar
flow is justified. Fig. 3 shows the variations of vapor pressure 3.3. Time scaling ratio
and vapor flow streamlines in the base case bed. This figure
indicates that the pressure decreases gradually in the vapor Scaling analysis for Eqs. 7 and 17 is carried out to derive the
channel due to the adsorption process. But in the porous time scales for the heat and mass diffusion processes during
layer, the pressure drops sharply from the evaporator inlet to the adsorption process. It is difficult to represent the coupled
the bottom cold wall. The lowest pressure zone is generated mass and heat diffusion processes with a single parameter. So,
beside the intersections of the cold walls (bottom, left, and different time scales for mass and heat diffusion are calcu-
front side walls). For this case, it is found that the pressure lated for the adsorption process:
drop through the bed is about 20 Pa (~1.6% of the average The mass diffusion time scale can be obtained from the LDF
bed pressure) which is expected to have a minimal effect on equation as:
the bed upload. The vapor flow stream lines show that most
of the adsorbed mass of the vapor flows vertically in the dX 60D
vapor channel and then from the vapor channel it moves = ( X∞ − X )
dt
 d2p
X (Tc ) − X (Ti )
  
horizontally to feed the bed. This of course is attributed to 60 D(Ti )
tm [X (Tc ) − X (Ti )]
d2p
the large flow resistance that the vapor faces in a packed bed
of particle diameter of 0.5 mm.
Solid phase temperature and concentration contours are por- d2p
tm = (27)
trayed in Figs. 4 and 5, respectively. The figure shows that the 60D (Ti )
adsorption process is driven by the heat transfer process. Even
after 500 s, the gradients in temperature and uptake are still where tm is the mass diffusion time scale.
 international journal of refrigeration 80 (2017) 238–251 245

Fig. 5 – Concentration contours at different planes after

Fig. 4 – Temperature contours at different planes after 500 s.
500 s.

where Ti is the initial bed temperature and Tc is the cooling

fluid temperature.
In order to present the importance of the time scaling ratio
The analysis of the results shows that the difference between
(tth/tm), the effect of the effective solid thermal conductivity of
solid and vapor temperature can be neglected. So, equation 17
the bed, the bed layer thickness, the particle diameter of the
can be rewritten as:
silica-gel beads, the vapor channel passage, and the height of

∂Ts dX
ρbedCp,bed = ∇ ⋅ [ks,eff ∇Ts ] + ρbedHads (28)
∂t dt

∂Ts ⎡ ks,eff ⎤ H dX
= ∇⋅⎢ ∇Ts ⎥ + ads (29)
∂t ρ C
⎣ bed p,bed ⎦ Cp,bed dt

∂Ts H dX
= ∇ ⋅ [α bed∇Ts ] + ads
∂t 

  Cp,bed dt
Ti − Tc
Ti − Tc
α bed  
t12 Hads ΔXmax
tth
Cp ,bed tth

Adsorption process has the effect of altering the thermal

diffusivity, due to the heat of adsorption, from αbed to αeff where
α bed (Ti − Tc ) Cp,bed
(Mitra et al., 2016) α eff =
Hads ΔX max
Accordingly, the thermal time scale (tth) for a packed bed
of t1 thickness may be deduced as:

t 12 Fig. 6 – Temporal variation of uploading ratio at different

t th = (30)
α eff effective solid thermal conductivity.
246 international journal of refrigeration 80 (2017) 238–251

the bed on the adsorption bed performance will be studied in

the following sections. For these studies, all parameters are the
same as the base case unless stated otherwise.

3.4. Effect of effective solid thermal conductivity and bed

thickness

Fig. 7 – Temporal variation of uploading ratio at different
adsorbent thickness.
The total amount of water adsorbed in the bed is the main pa-
rameter that determines the cooling power of the adsorption
cooling system. One of the most important parameters is the
effective solid thermal conductivity. Without any additives, the
solid phase of silica gel packed bed has an effective thermal
conductivity of about 0.15 W m−1 K−1 (Tamainot-Telto and
Critoph, 2001) and thermal conductivity enhancer additives can
be added to the bed to increase its thermal conductivity (Han
et al., 1998; Mauran et al., 1993; Wang et al., 2016). The influ-
ence of effective solid thermal conductivity of the porous media
on the variation of average amount of refrigerant adsorbed
inside the adsorbent bed with time is shown in Fig. 6. As shown

Fig. 8 – Temporal variation of uploading ratio at different effective solid thermal conductivity and particle diameter.
 international journal of refrigeration 80 (2017) 238–251 247

in this figure, a considerable increase in the adsorbed refrig-

erant occurs when the thermal conductivity increases from 0.15
W m−1 K−1 to 0.5 W m−1 K−1. But increasing the thermal con-
ductivity above 1.5 W/m.K produces little improvement. This
is because beside thermal diffusion, the adsorption is also con-
trolled by the effectiveness of mass diffusion. Even if the
thermal conductivity of the bed reaches infinity and the bed
temperature reaches the coolant temperature during the entire
adsorption period, the uploading will be limited by the rate of
the mass diffusion process.
As shown in Fig. 7, decreasing the bed thickness en-
hances the adsorption capability because the heat diffusion
time scale is proportional to the square of the bed thickness.
It is also clear that reducing the bed thickness has a more pro-
nounced effect on uploading than increasing the thermal
conductivity. This is because the heat generated due to ad-
sorption process is also reduced. So, with the same coolant
temperature and heat transfer coefficient, the smaller bed thick-
ness will cool down much faster than the thicker beds. But, it
should be mentioned that reducing the bed thickness is not
always recommended since this decreases the mass ratio of
adsorbent to the total mass of the bed, which in turn in-
creases the cost and reduces the cooling power per unit volume
of the bed.

3.5. Effect of particle diameter

Since the diffusion coefficient of water vapor in silica-gel par-

ticle is an intrinsic property of the silica-gel type, the mass
diffusion process can be enhanced only by decreasing the silica-
gel particle diameter. The manufacturers of silica-gel produce
silica-gel with diameters ranging of 0.075–4.0 mm. The bed po-
rosity is around 0.4 when the particle diameter to bed thickness
ratio is less than 0.25 (Abbas, 2011). Accordingly, the bed po-
rosity is assumed to be constant for the all simulated cases.
Decreasing the particle diameter increases the surface area
subjected to water vapor diffusion, asv∝1/dp and decreases the
time required for vapor to reach the center of the particle. Fig. 8
shows the temporal variation of the uploading ratio for various
silica gel particle diameters at different solid effective thermal
conductivity. In general, it is shown that decreasing the par-
ticle diameter improves the uploading ratio of the bed. But, it
is also clear that decreasing the particle diameter under certain
limits has little effect on the uploading ratio, for example, when
dp is reduced from 0.5 mm to 0.25 mm for ks = 0.15 W m−1 K−1.
This can be explained by the fact that the adsorption process
is an exothermic process. Any increase in adsorption rate is
accompanied by an increase in the heat generated inside the
bed. If the thermal conductivity of the bed is incapable of dis-
sipating this generated heat, the temperature will go up and
the uptake will decrease. In this case, the uploading rate of the
bed is controlled mainly by the rate of heat diffusion.
It is also evident from the figures that particles with diam-
eters of 0.25 and 0.5 mm have nearly the same variation for
the uploading ratio regardless of the thermal conductivity value
till 100 s of the adsorption process. This means that the process
is mainly governed by mass diffusion for the first 100 s. For
longer adsorption time, the increase in thermal conductivity
can enhance the performance of the bed.
Fig. 9 – Temporal variation of uploading ratio at different
vapor channel passage.
3.6. Effect of vapor channel passage
The adsorbed vapor in the bed is compensated by the vapor
that enters the bed from the evaporator. Creating a vapor
passage may help in attracting the vapor to flow into the bed
interior and feed the adsorbent. The main function of the vapor
passage is to reduce the pressure drop in the bed, which affects
its uploading capacity.
Figs. 9 and 10 present the influence of the vapor channel
passage on the uploading ratio. It is found that the presence
of vapor channel passage is not important for the larger par-
ticle diameters because the permeability of the bed is high
enough and the top side of the bed can feed the entire bed with
a pressure drop as low as around 100 Pa. However, the pres-
ence of the vapor channel is crucial for the smaller diameter;
dp = 0.25 mm, because the pressure drop along the bed height
can exceed 700 Pa. Changing the vapor channel passage from
1 to 4 mm does not change the uploading of the bed signifi-
cantly. This means that a 1 mm gap is enough to feed the bed
248 international journal of refrigeration 80 (2017) 238–251

Fig. 10 – Pressure contours at a plane parallel to x-y plane at z = 0.04889.

with sufficient vapor and to keep the pressure nearly uniform 1 mm. This kind of bed will have a high adsorbent/total weight
in the entire bed. So, to make the bed more compact it is rec- ratio and a high SCP.
ommended to not increase the vapor channel width. This is
of course to increase the ratio of the adsorbent mass to the
3.8. Heat transfer vs mass transfer
total mass of the bed.

Analysis of all the cases shows that the adsorption process is

3.7. Effect of bed height
mainly heat diffusion dependent for small particle size and
switches to mass diffusion dependent for small bed thick-
Increasing the height of the bed enables one to build an
ness or bed with high thermal conductivity. It is found that
adsorber with a higher adsorbent/metal mass ratio. But for a
the thermal to mass diffusion time scaling ratio can be cal-
tall bed, the pressure may decrease significantly along the vapor
culated to illustrate which process is controlling the adsorption
channel and consequently in the far side of the bed from evapo-
process. The numerical data in this study show that the ad-
rator pressure inlet. This may adversely affect the uploading
sorption process is mainly thermal diffusion dependent when
of the bed. Different bed heights from 5 cm to 30 cm are
this ratio (tth / tm) is in order of O(100) or more. For O(1) < tth /
simulated.
tm < O(100), the bed performance is controlled by both heat and
Fig. 11 shows that, for the same bed thickness, increasing
mass diffusion rates, while the performance is mainly gov-
the height of the bed above 10 cm does not have an adverse
erned by mass diffusion when tth / tm < O(1).
effect of the bed performance, which means that the pres-
sure drop along the vapor channel has a minimal effect on the
bed uploading. A more significant improvement in uploading 3.9. Bed performance
can be observed for the shortest bed height of 5 cm. This is
mainly attributed to the thermal effect of the bed bottom The performance of the bed can be evaluated by calculating
cooling. This means that tall, compact beds with small par- the specific cooling power which is consumed in the evapo-
ticle size can be designed with a vapor channel passage of rator as following:
 international journal of refrigeration 80 (2017) 238–251 249

Fig. 12 – SCP vs particle diameter for different solid thermal

conductivity after 300 s (solid lines) and 600 s (dashed
lines).

of adsorbent, but it enables to increase the mass of adsorbent

in the bed. Also, it is shown that decreasing the bed thickness
leads to a remarkable improvement on SCP. This is of course
due to its strong effect on decreasing the heat diffusion time
and consequently speeding up the rate of cooling process.
It is worth to mention that the proposed compact bed with
particle size of 0.5 mm could produce SCP of 0.63 kW kg−1, 0.39
kW kg−1, and 0.23 kW kg−1 for layer thickness of 0.25 cm, 0.5 cm,
and 1 cm, respectively at the lowest bed thermal conductiv-
ity. The reported maximum SCP for particle size of 0.5 mm and
cycle time of 11 min was 0.093 kW kg−1 (San and Tsai, 2014),
Fig. 11 – Temporal variation of uploading ratio at different
while Liu et al. (2005) estimated a SCP of 0.23 kW kg−1 for a plate-
bed height.
fin adsorber. Also, the cooling capacity per unit volume (CPv)
of this design is more than 160 kW m−3 while the most beds
(X (t ) − Xi ) ∗ h fg can produce about 60 kW m−3 (Solmuş et al., 2015). This means
SCP = (31)
t ads + t preheating

where tads is the adsorption time, tpreheating is the preheating time

before the regeneration which is about 50 s, and hfg is the latent
heat of evaporation of liquid water taken as 2520 kJ g−1 at
P = 1228 Pa.
Effect of effective solid thermal conductivity on the spe-
cific cooling power after 300 and 600 s of adsorption is presented
in Fig. 12. It is obvious that the SCP produced after 300 s is larger
than that after 600 s because the temperature difference
between the bed and the cooling fluid is larger and the ad-
sorption process is faster at the beginning. Moreover, increasing
the solid thermal conductivity enhances the bed perfor-
mance dramatically specially when it is increased from 0.15
to 0.5 W/m.K. Accordingly, increasing the solid thermal con-
ductivity for small particles by factor of 3 could definitely
increase the SCP by several folds.
Furthermore, influence of the bed height and its thickness
on SCP is illustrated in Fig. 13 for bed width of 10 cm. It is shown
that increasing the height of the bed does not change the SCP Fig. 13 – SCP vs particle diameter for different bed height
of the bed, which represents the cooling power per kilogram and thickness after 300 s.
250 international journal of refrigeration 80 (2017) 238–251
that the present bed is more efficient, has low metal to ad-
sorbent mass ratio, and more compact.
4. Conclusion
In this paper, a new compact bed design using silica-gel/water
pair is proposed and many parameters that can affect its per-
formance are simulated under the operating conditions that
are very similar to those realized during isobaric stages of real
adsorption heat pumps. A mathematical model is developed
to simulate the mechanism of heat and mass transfer inside
the adsorbent bed. The governing conservation equations of
heat and mass transfer processes are solved to study the in-
fluence of the solid phase thermal conductivity, particle diameter,
bed height, and the vapor channel width that feeds the bed on
the bed performance. The following conclusions are reached:
• Scaling analysis of the governing equations shows up a time
scaling ratio parameter and the analysis of the results in-
dicates that the adsorption process is mainly heat diffusion
dependent when the time scaling ratio is more than or
~O(100) and becomes mass diffusion dependent when this
ratio is less than or ~ O(1).
• The cooling capacity of the system increases with decreas-
ing particle diameter for the smallest bed thickness but a
particle diameter of 0.5 mm gives the best performance for
other bed thicknesses.
• The presence of vapor channel of 1 mm passage is crucial
for 0.25 mm particle diameter to avoid significant pres-
sure drop in the bed, but the vapor channel is not needed
for the larger beads.
• The presence of vapor channel can reduce the penalty of
increasing the bed height on the SCP while it can increase
the mass of the adsorbent in the bed.
• Increasing the adsorption period is not preferred and it is
shown that an adsorption period of 300 s can yield a much
higher SCP compared to an adsorption period of 600 s.
• The cooling power per unit volume of the proposed bed is
higher than the commercial beds. This means that the
present bed is more compact.
Acknowledgments
This work is supported by National Science Foundation (NSF)
through Grant No. 1603215 and the Egyptian Ministry of Re-
search and Higher Education.
REFERENCES
Abbas, M.N., 2011. Modeling of porosity equation for water flow
through packed bed of monosize spherical packing. J. Eng.
Dev. 15 (4).
Ali, S.M., Chakraborty, A., 2016. Adsorption assisted double stage
cooling and desalination employing silica gel+ water and
AQSOA-Z02+ water systems. Energy Convers. Manag. 117, 193–
205.
Aristov, Y.I., Glaznev, I.S., Girnik, I.S., 2012. Optimization of
adsorption dynamics in adsorptive chillers: loose grains
configuration. Energy 46 (1), 484–492.
Çağlar, A., 2016. The effect of fin design parameters on the heat
transfer enhancement in the adsorbent bed of a thermal
wave cycle. Appl. Therm. Eng. 104, 386–393.
El Fadar, A., 2015. Thermal behavior and performance
assessment of a solar adsorption cooling system with finned
adsorber. Energy 83, 674–684.
Freni, A., Tokarev, M.M., Restuccia, G., Okunev, A.G., Aristov, Y.I.,
2002. Thermal conductivity of selective water sorbents under
the working conditions of a sorption chiller. Appl. Therm. Eng.
22 (14), 1631–1642.
Freni, A., Russo, F., Vasta, S., Tokarev, M., Aristov, Y.I., Restuccia,
G., 2007. An advanced solid sorption chiller using SWS-1L.
Appl. Therm. Eng. 27 (13), 2200–2204.
Freni, A., Maggio, G., Cipiti, F., Aristov, Y.I., 2012. Simulation of
water sorption dynamics in adsorption chillers: one, two and
four layers of loose silica grains. Appl. Therm. Eng. 44, 69–77.
Glaznev, I., Aristov, Y.I., 2010. Dynamic Aspects of Adsorptive
Heat Transformation. Advances in Adsorption Technologies.
Nova Science Publishers, ISBN, Hauppauge, NY, p. 107e163.
Guilleminot, J., Meunier, F., Pakleza, J., 1987. Heat and mass
transfer in a non-isothermal fixed bed solid adsorbent
reactor: a uniform pressure-non-uniform temperature case.
Int. J. Heat Mass Transf. 30 (8), 1595–1606.
Han, J., Cho, K.W., Lee, K.-H., Kim, H., 1998. Porous graphite
matrix for chemical heat pumps. Carbon 36 (12),
1801–1810.
Hong, S., Ahn, S.H., Kwon, O.K., Chung, J.D., 2015. Optimization of
a fin-tube type adsorption chiller by design of experiment.
Int. J. Refrigeration 49, 49–56.
Jemni, A., Nasrallah, S.B., 1995. Study of two-dimensional heat
and mass transfer during absorption in a metal-hydrogen
reactor. Int. J. Hydrogen Energy 20 (1), 43–52.
Jiang, Y., Ge, T.S., Wang, R.Z., Hu, L.M., 2015. Experimental
investigation and analysis of composite silica-gel coated fin-
tube heat exchangers. Int. J. Refrigeration 51, 169–179.
Lang, R., Roth, M., Stricker, M., Westerfeld, T., 1999. Development
of a modular zeolite-water heat pump. Heat Mass Transf. 35
(3), 229–234.
Le Bars, M., Worster, M.G., 2006. Interfacial conditions between a
pure fluid and a porous medium: implications for binary alloy
solidification. J. Fluid Mech. 550, 149–173.
Li, A., Thu, K., Ismail, A.B., Ng, K.C., 2016. A heat transfer
correlation for transient vapor uptake of powdered adsorbent
embedded onto the fins of heat exchangers. Appl. Therm.
Eng. 93, 668–677.
Liu, Y., Wang, R., Xia, Z., 2005. Experimental performance of a
silica gel–water adsorption chiller. Appl. Therm. Eng. 25 (2),
359–375.
Mauran, S., Prades, P., L’haridon, F., 1993. Heat and mass transfer
in consolidated reacting beds for thermochemical systems.
Heat Recov. Syst. CHP 13 (4), 315–319.
Mesalhy, O., Lafdi, K., Elgafy, A., Bowman, K., 2005. Numerical
study for enhancing the thermal conductivity of phase
change material (PCM) storage using high thermal
conductivity porous matrix. Energy Convers. Manag. 46 (6),
847–867.
Miltkau, T., Dawoud, B., 2002. Dynamic modeling of the combined
heat and mass transfer during the adsorption/desorption of
water vapor into/from a zeolite layer of an adsorption heat
pump. Int. J. Therm. Sci. 41 (8), 753–762.
Mitra, S., Aswin, N., Dutta, P., 2016. Scaling analysis and
numerical studies on water vapour adsorption in a columnar
porous silica gel bed. Int. J. Heat Mass Transf. 95, 853–864.
 international journal of refrigeration 80 (2017) 238–251
Nield, D.A., Bejan, A., 2006. Convection in Porous Media. Springer
Science & Business Media, Berlin.
Patankar, S., 1980. Numerical Heat Transfer and Fluid Flow. CRC
Press, Boca Raton, FL.
Pesaran, A.A., Mills, A.F., 1987. Moisture transport in silica gel
packed beds – II. Experimental study. Int. J. Heat Mass Transf.
30 (6), 1051–1060.
Poyelle, F., Guilleminot, J.-J., Meunier, F., 1999. Experimental tests
and predictive model of an adsorptive air conditioning unit.
Ind. Eng. Chem. Res. 38 (1), 298–309.
Sah, R.P., Choudhury, B., Das, R.K., 2016. A review on low grade
heat powered adsorption cooling systems for ice production.
Renew. Sustain. Energy Rev. 62, 109–120.
Saha, B.B., El-Sharkawy, I.I., Shahzad, M.W., Thu, K., Ang, L., Ng,
K.C., 2016. Fundamental and application aspects of
adsorption cooling and desalination. Appl. Therm. Eng. 97,
68–76.
San, J.-Y., Tsai, F.-K., 2014. Testing of a lab-scale four-bed
adsorption heat pump. Appl. Therm. Eng. 70 (1), 274–281.
Sapienza, A., Santamaria, S., Frazzica, A., Freni, A., Aristov, Y.I.,
2014. Dynamic study of adsorbers by a new gravimetric
version of the large temperature jump method. Appl. Energy
113, 1244–1251.
Sharafian, A., Bahrami, M., 2013. Adsorbate uptake and mass
diffusivity of working pairs in adsorption cooling systems. Int.
J. Heat Mass Transf. 59, 262–271.
Sharafian, A., McCague, C., Bahrami, M., 2015. Impact of fin
spacing on temperature distribution in adsorption cooling
system for vehicle A/C applications. Int. J. Refrigeration 51,
135–143.
Sladek, K.J., Gilliland, E.R., Baddour, R.F., 1974. Diffusion on
surfaces. II. Correlation of diffusivities of physically and
251
chemically adsorbed species. Ind. Eng. Chem. Fund. 13 (2),
100–105.
Solmuş, İ., Yamah C., Yildirim C., Bilen K., 2015. Transient
behavior of a cylindrical adsorbent bed during the adsorption
process. Appl. Energy 142, 115–124.
Stricker, M., 2002. Entwicklung einer mehrmodularen
Zeolith-Wasser-Adsorptionswärmepumpe. VDI-Verlag,
Dusseldorf.
Tamainot-Telto, Z., Critoph, R., 2001. Monolithic carbon for
sorption refrigeration and heat pump applications. Appl.
Therm. Eng. 21 (1), 37–52.
Wang, J., Wang, R., Wang, L., 2016. Water vapor sorption
performance of ACF-CaCl 2 and silica gel-CaCl 2 composite
adsorbents. Appl. Therm. Eng. 100, 893–901.
Wang, R., Wang, L., Wu, J., 2014. Adsorption Refrigeration
Technology: Theory and Application. John Wiley & Sons,
Hoboken, NJ.
Wang, X., Zimmermann, W., Ng, K.C., Chakraboty, A., Keller, J.U.,
2004. Investigation on the isotherm of silica gel+ water
systems. J. Therm. Anal. Calorim. 76 (2), 659–669.
Yong, L., Sumathy, K., 2004. Comparison between heat transfer
and heat mass transfer models for transportation process
in an adsorbent bed. Int. J. Heat Mass Transf. 47 (8),
1587–1598.
Youssef, P.G., Mahmoud, S.M., AL-Dadah, R.K., 2016. Numerical
simulation of combined adsorption desalination and cooling
cycles with integrated evaporator/condenser. Desalination
392, 14–24.
Zheng, X., Wang L.W., Wang R.Z., Ge T.S., Ishugan T.F., 2014.
Thermal conductivity, pore structure and adsorption
performance of compact composite silica gel. Int. J. Heat Mass
Transf. 68, 435–443.