Rosario Porrazzo, Graeme White, Raffaella Ocone: Sciencedirect

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 11

Fuel 136 (2014) 4656

Contents lists available at ScienceDirect

Fuel
journal homepage: www.elsevier.com/locate/fuel

Aspen Plus simulations of uidised beds for chemical looping


combustion
Rosario Porrazzo, Graeme White, Raffaella Ocone
Chemical Engineering, Heriot-Watt University, Edinburgh EH14 4AS, UK

h i g h l i g h t s
 Aspen Plus is applied to chemical looping combustion.
 The bubbling bed and the circulating bed are modelled using appropriate combinations of ideal reactors.
 The effect of the shrinking-core model for the non-catalytic heterogeneous reactions is investigated.
 The methodology is checked against experimental data from the literature.

a r t i c l e

i n f o

Article history:
Received 1 October 2013
Received in revised form 19 June 2014
Accepted 24 June 2014
Available online 16 July 2014
Keywords:
Chemical looping combustion
Aspen Plus
Fluidised beds
Carbon capture

a b s t r a c t
Chemical Looping Combustion (CLC) is a technology able to generate energy whilst managing CO2
emissions. A system composed by two interconnected uidised beds is often used in CLC: the two uidised beds are employed for carrying out the oxidation and reduction reactions of the metal oxide
employed as oxygen carrier. In this work, a model to implement uidised bed systems in Aspen Plus is
presented. Depending on the hydrodynamic regimes, two different models are considered: one of the
two uidised beds, called the fuel reactor, is modelled according to the two-phase theory (i.e. emulsion
and bubble phase) whilst the other bed, called the air reactor, is assumed to operate in the fast uidisation regime. Kinetic equations for heterogeneous gas/solid reactions are also considered in the model.
Simulation tests for each uidised bed are carried out, and comparisons are made with experimental data
from the literature. A comparison with the largely used Gibbs reactor model is carried out showing the
advantages of using the models developed here. In addition, the net heat duty of the whole process is
calculated and the role of the main variables that affect the process is investigated.
2014 Elsevier Ltd. All rights reserved.

1. Introduction
Carbon capture from power plants is one of the remits of
governments worldwide as part of their duty to mitigate climate
change. The increase in anthropogenic greenhouse gases is
believed to be related to the planet temperature rise. CO2 accounts
for up to 64% of emissions of enhanced greenhouse effect [1] constituting about 15% of the ue gas stream from coal combustion
power plants, where the total amount of CO2 produced from coal
based power plants accounts for more than 40% of all anthropogenic CO2 emissions [2]. The target for the future is an energy
mix which involves the development of technologies based on
nuclear and/or renewable energies with lower emissions of CO2.
Meanwhile, a viable energy solution for the reduction of CO2
emissions could consist in coupling fossil fuel energy conversion
Corresponding author.
E-mail address: r.ocone@hw.ac.uk (R. Ocone).
http://dx.doi.org/10.1016/j.fuel.2014.06.053
0016-2361/ 2014 Elsevier Ltd. All rights reserved.

technologies with economical capture, transportation, and safe


storage solutions.
A number of carbon capture techniques can be implemented in
the operation of fossil fuel based power plants. They can be divided
into post-combustion capture, pre-combustion capture, and
oxy-combustion. Post-combustion capture removes CO2 from the
combustion ue gases whereas in the pre-combustion capture fuel
reacts with oxygen obtained from the air separation unit (ASU), in
the presence of steam, to produce syngas. The CO in the syngas is
converted to CO2 through the watergas shift reaction (WGS). The
products of the WGS reaction contain high CO2 and H2 amounts:
CO2 is captured whilst H2 is used to generate electricity through
a combined cycle power generation system. In the oxy-combustion
capture, combustion takes place using oxygen instead of air, to
generate a concentrated CO2 and H2O vapour streams.
Fossil fuel based power plants, using traditional post-combustion
separation units, involve large energy losses and additional
costs for the electricity generated. The existing techniques for

47

R. Porrazzo et al. / Fuel 136 (2014) 4656

Nomenclature
a
Ab
Ae
At
CA
DBB
db
dp
DR
F
Fs
g
HR
Hti
i
k
kCSTR
kPFR
Kbc
Kbe
Kce
ks
Lf
Lm
P
Qb
Qe
r
R
r0
rc

decay index
cross sectional area of the bubble phase (m2)
cross sectional area of the emulsion phase (m2)
cross sectional area of the whole bed (m2)
molar concentration of A gas reactant (kmol/m3)
diameter of the bubbling bed (m)
bubble diameter (m)
particle diameter (m)
diameter of the riser (m)
component molar owrate (kmol/s)
inlet solid mass owrate (kg/s)
gravity (m/s2)
height of the riser (m)
height of i stage (m)
number of stages
kinetic pre-exponential factor (s1)
kinetic pre-exponential factor in CSTR reactors (s1)
kinetic pre-exponential factor in PFR reactors (s1)
mass transfer coefcient between bubble and cloud
(s1)
overall mass transfer coefcient between bubble and
emulsion (s1)
mass transfer coefcient between cloud and emulsion
(s1)
kinetic pre-exponential factor (m/s)
height of the uidised bed modelled (m)
height of the packed solid loading (m)
operative pressure (atm)
total volumetric owrate of the bubble phase (m3/s)
total volumetric owrate of the emulsion phase (m3/s)
reaction rate (kmol/(m3 s))
external radius of the particle (m)
reaction rate (kmol/(m2 s))
particle radius (m)

CO2 capture include monoethanolamine (MEA) scrubbing technology. In this process, the cooled ue gases enter into the absorber
unit where fresh amine solvent is used to remove CO2 from the
gas stream. The spent amine solvent is regenerated in the stripper
unit where the temperature is higher than in the absorber unit;
CO2 is then recovered at lower pressure. MEA and other solvents
present drawback such as equipment corrosion in presence of O2
and energy intensive solvent regeneration [35]. In addition, the
presence of ue gas contaminants, such as SOx and NOx, has a negative impact on solvent based process performance.
Materials such as zeolites, alumina molecular sieves and activated carbon are often employed for selectively adsorb CO2 in
applications for the production of H2 from syngas and in natural
gas sweetening; however, intensive adsorbent regeneration often
implies a high energy penalty [6]. Porous membranes are also used
to separate gases of different molecular sizes. However, since the
amount of gas treated is low, various operational set-ups need to
be implemented in practice; those comprise multistage operation
or stream recycling, for example [6].
Compared to the methods described above, Chemical Looping
Combustion (CLC) is considered to be both pre-combustion capture
and oxy-combustion and it is potentially the technology best suited for the efcient, low cost and low energy capture of CO2 from
ue gases [711]. CLC may operate, in principle, with a variety of
fuel types, including carbonaceous fuel such as coal-derived syngas
and natural gas. The process uses a transitional metal oxide as oxygen carrier to transfer the oxygen from the air to the fuel reactor.
The oxygen carrier is circulated between the fuel reactor, where

T
Ub
Ubr
Ue
Umf
Uo
Ut
Vb
VCSTR
Ve
VPFR
V
W
XB
z = z4
z1
zi
d

l
eb
ee
ef
emf
es(z)
es
esd
esi
qg
qs
r

operative temperature (C)


bubble supercial gas velocity (m/s)
bubble supercial gas velocity at minimum uidisation
(m/s)
emulsion supercial gas velocity (m/s)
supercial gas velocity at minimum uidisation (m/s)
inlet supercial gas velocity (m/s)
particle terminal velocity (m/s)
bubble volume (m3)
CSTR volume (m3)
emulsion volume (m3)
PFR volume (m3)
total reaction volume (m3)
solid inventory in the bed (kg)
average solid conversion
height of the riser (m)
height of dense phase (m)
height of i stage (m)
diffusion coefcient (m2/s)
gas viscosity (kg/(m s))
gas voidage in the bubble phase
gas voidage in the emulsion phase
gas voidage in the whole uidised bed
gas voidage at minimum uidisation conditions
solid voidage at generic height z
solid voidage at the exit of the uidised bed
solid voidage in the dense phase
solid voidage in the i stage
gas density (kg/m3)
solid density (kg/m3)
volumetric fraction of the bubble phase

it is reduced through combustion, and the air reactor, where it is


oxidised in contact with air (Fig. 1).
A generalised description of the overall reaction in the fuel reactor can be written as follows:

2n mMy Ox Cn H2m ! 2n mM y Ox1 mH2 O nCO2

The reduced metal oxide MyOx1 is then transported to the air


reactor where it is re-oxidised to MyOx:

My Ox1 1=2O2 air ! M y Ox air : N2 unreacted O2

Thus, the air is never mixed with the fuel avoiding NOx emissions [12] and producing a stream of CO2 and H2O vapour; the latter can be easily separated from the CO2 through condensation. The
gas stream leaving the air reactor contains nitrogen and un-reacted
oxygen. These gases can be released to the atmosphere with minimal negative environmental impact.
Transition metals are good oxygen carrier. In particular, NiO/Ni,
CuO/Cu, Fe2O3/Fe3O4, Mn2O3/Mn3O4 have been investigated with
different inert support materials such as: Al2O3, TiO2, SiO2, ZrO2,
and bentonite, to increase their reactivity, durability and uidising
properties [13]. Hossain et al. [13] discussed the different properties obtained when different metal oxides and inert support materials are mixed. The ideal oxygen carrier particles should show
good oxygen-carrying capacity, high oxidation and reduction reactivity, good mechanical strength, suitable particle size, density and
pore structure to achieve high uidisability and reaction rate.
The hot air leaving the combustor is used to drive a steam turbine/gas turbine combined cycle system for electricity generation.

48

R. Porrazzo et al. / Fuel 136 (2014) 4656

Fig. 1. Circulating uidised bed system for CLC.

Whereas the reduction reaction of the metal oxide is often endothermic, the oxidation reaction of the metal oxide is exothermic.
The heats of reaction depend on the fuel type and on the metal
oxide used as oxygen carrier. The heat integration between the
two reactors can reduce the energy loss by recovering the lowgrade heat while producing a larger amount of high grade heat
as shown in Fig. 2 [14].
The overall generation of heat equals the heat of combustion.
Depending on the metal oxide utilised, the thermal energy released
in the oxidation reactor is usually larger than the energy required
by direct combustion of the fuel [14,15]. Also, the heat absorbed in
the reduction reactor is at low temperature and heat is released at
high temperature in the oxidation reactor. These features imply
that the combustion system can be highly efcient [15]. The freeof-water CO2, obtained after condensation of the water vapour,
can be captured or/and used for other applications.
When information on the CLC process is needed, process simulation such as Aspen Plus, Aspen Hysys, PRO/MAX, PRO/II are
implemented. Those packages are largely used for whole plant
modelling given their ability to simulate a variety of steady-state
processes ranging from single unit operation to complex processes
involving many units. Such codes work with different standard
blocks that represent the main unit operations in the simulated
process (e.g. PFR and CSTR reactors, absorbers, distillation columns,
etc.). Mass and heat balances are within the blocks. Inlet and outlet
mass and heat streams link the blocks to each other. When an
attempt is made to simulate CLC processes with one of the process
packages, uidised beds would need to be represented, and those
are not employed specically in of the packages aforementioned.
Nevertheless, an accurate representation of the process cannot
neglect the complex hydrodynamics and kinetics happening in
the reactors. A way to solve this issue has been proposed by Jafari
et al. [16] who employed a number of basic blocks (e.g. ideal

reactors, usually PFRs and CSTRs) combined in a fashion that could


simulate the real hydrodynamics and/or kinetics. Following this
strategy, uidised beds have been implemented in Aspen Plus for
a number of process set-ups (e.g. Sarvar-Amini et al. [17] modelled
a uidised bed membrane reactor; Sotudeh-Gharebaagh et al. [18]
and Liu et al. [19] modelled a uidised bed for coal combustion;
Sohi et al. [20] have modelled a uidised bed for gas natural
combustion).
To the best of our knowledge, uidised beds for CLC purpose
have been modelled exclusively by using a Gibbs reactor which
is based on the minimisation of the Gibbs free energy [14,21]. Such
minimisation leads to an overestimation of the conversions of the
species since effects such as gas by-pass and mass transfer and/or
kinetic limitations, for instance, are neglected. Those effects are of
importance in determining the real conversion. Additionally, if
ideal conditions are assumed, i.e. the Gibbs reactor is employed,
the real reactor size cannot be estimate.
The aim of this work is to attempt to model uidised bed reactors which mimic the real uidised bed reactors employed in CLC.
Aspen Plus is adopted by using different combinations of standard
blocks of reaction such as CSTRs and PFRs to take into account
different hydrodynamic regimes. Additionally, heterogeneous,
non-catalytic reactions are incorporated in the model for the rst
time, to take into account the kinetics of reaction (shrinking-core
model). An estimation of the sizes and of the optimal operating
conditions for the CLC reactors is obtained and comparisons are
made with the Gibbs reactor, normally implemented when modelling uidised beds in Aspen Plus [14,21].
2. System description
In this work, the most common set-up of the CLC process is
assumed, and therefore two interconnected uidised beds working
in different hydrodynamic regimes are considered. Specically, the
reduction of the metal oxide is carried out in a bubbling bed while
the air reactor, where the metal oxides are oxidised, is assumed to
work in the fast uidisation regime.
2.1. The fuel reactor

Fig. 2. Heat integration scheme for CLC.

The bubbling regime is characterised by a value of the supercial gas velocity, Uo, higher than the uidisation velocity, Umf, and
lower than the terminal velocity of the isolated particles, Ut [22].
The bubbling bed can be modelled according to the two phase theory [22] where the bubble phase, with low content of solids, and
the emulsion phase, characterised by perfect mixing of gas and
solids, are considered as co-existing within the bed. According to
Davidson and Harrison [23], the emulsion phase is at minimum
uidisation conditions and the gas in excess with respect to the

R. Porrazzo et al. / Fuel 136 (2014) 4656

49

minimum uidisation velocity (Uo > Umf) is transferred to the bubble phase. Additional assumptions are [23]:
 The bubble diameter, db, is constant along the bed height.
 The reactor operates at isothermal conditions.
 The radial mass solid gradient within the bed is neglected.
Table 1 reports the correlations and equations adopted in this
work for the bubbling bed.
2.2. The air reactor
Fast uidisation is characterised by the supercial velocity of
the inlet gas, Uo, greater than the terminal velocity of an isolated
particle, Ut [22]. In fast uidisation, perfect mixing of the gas and
the solid is assumed. The solid volume fraction is assumed to
remain constant in the bed radial direction, while two zones along
the bed height are identied: the dense and the lean phases; the
latter is divided into lower acceleration region, upper acceleration
region and completely uidised region [18,19,22] (Fig. 3). The relative height of those regions varies with the inlet supercial gas
velocity. The solid voidage of the dense phase is assumed constant
whereas the solid voidage of the lean phase decreases along the
bed height. The lean phase is usually referred to as the transport
disengagement height (TDH) and an exponential change in the solids loading is assumed to describe the variation of voidage, e, along
the TDH [22]:

es  es z az
e
es  esd

where es is the solid voidage at the exit of the uidised bed, esd is the
solid voidage of the dense phase, a is the decay index. esd is a function of inlet supercial gas velocity.
2.3. Reduction and oxidation kinetics
In this work, pure methane and NiO/Ni oxygen carrier supported by bentonite (80 lm diameter) are chosen as fuel and solid
reactant, respectively. Pure air is used in the air reactor to oxidise
Ni metal. This reaction has been chosen because of the large
amount of kinetic data available from the literature useful to
implement and validate the model proposed.
The heterogeneous non-catalytic reaction occurring in both the
riser and fuel reactor is:

Agas bBsolid ! cC products :

Various studies have been carried out to characterise the reduction and oxidation behaviour of metal oxides and diverse gases

Table 1
List of correlations and equations applied for the bubbling bed model.
U mf

Supercial gas velocity in the emulsion


phase
Rise bubbles velocity at Uo = Umf
Rise bubbles velocity at Uo Umf
Volumetric fraction of bubble phase

U e  emf

Fluidised bed voidage


Height of the packed solid loading

ef = r * eb + (1  r) * ee
W
Lm q At 1
em

Height of uidised bed

1em
Lf Lm1
ef

Ubr = 0.711 * (g * db)0.5


Ub = Uo  Ue + Ubr
e
r UUob U
U e

 

Mass transfer coefcient between


bubble and cloud

K bc 4:5 

Mass transfer coefcient between cloud


and emulsion

K ce 6:77 

Overall mass transfer coefcient


between bubble and emulsion

1
K be

K1ce K1bc

Ue
db

5:85 

demf U b
3
db

0:5

d0:5 g 0:25
5
db4

Fig. 3. Trend of the solid fraction (1  e) in the riser.

have been considered including CO, H2, CH4 [12,24,25]. Many


authors have used kinetic models, based on shrinking core and
changing grain size, to represent the chemical kinetics of the metal
oxides [13]. The best t with experimental data is achieved with
one of these two models depending on the gas reactant and the
metal oxides considered.
To describe the reactions on the solid particles the un-reacted
core model is considered in this work; this is a good approximation
of the real behaviour, as conrmed by a number of studies
[13,24,2629]. The model assumes that the reaction happens on
the surface separating the un-reacted solid from the reacted shell.
The initial reaction surface corresponds to the initial external surface of the solid. The thickness of the reacted shell increases with
time, determining the shrinking core of un-reacted solid [24,30].
The heterogeneous reaction proceeds via three steps: external
mass transfer, internal mass transfer (internal diffusion within
the particle), chemical reaction.
Within the fuel reactor the reduction reaction is:

CH4 4NiO ! CO2 2H2 O 4Ni

While within the riser the oxidation reaction is:

O2 2Ni ! 2NiO

The following assumptions are made [31]:


 The particles are spherical.
 The external mass transfer step is fast compared to the internal
diffusion and reaction steps.
 The reaction is rst order with respect to the concentration of
the reactant gas.
 The particle volume remains constant.
 The reaction is isothermal.
Ruy et al. [24] demonstrate that the reduction rate for NiO/Ni
particles supported by bentonite is controlled by the chemical
reaction while the oxidation is controlled by the internal diffusion.
Garcia-Labiano et al. [26,27] studied the CLC kinetics and the variations in the structure of the oxygen carrier were considered
together with various geometries; the changing grain size model
was utilised. Small particles (3070 lm) were selected to minimise
mass transfer limitations. The shrinking core model with the reaction being the controlling step describes well the experimental
data [13]. Indeed, the oxygen carrier particles used in CLC have
small diameter and high internal porosity, consequently the
assumption is shown to be a reasonable one [28].
Data from the literature report that oxidation and reduction
times of the oxides selected in this study are of the same order
of magnitude. Sung and Sang [32] found a reduction conversion

50

R. Porrazzo et al. / Fuel 136 (2014) 4656

rate of 4%/min for NiO supported by bentonite and an oxidation


conversion rate of 12%/min in experiments with circulating uidised beds at constant temperature. Lyngfelt et al. [33] reported a
reduction conversion rate of 4%/min for pure NiO and 7%/min for
NiO supported by bentonite (60/40%) and an oxidation conversion
rate of 13%/min for pure Ni and 21%/min for Ni supported by
bentonite (60/40%) in uidised beds. Garcia-Labiano et al. [26]
reported high reactivity of nichel oxygen carriers with full conversion for both reduction and oxidation reactions in the range of
2060 s that means 300100%/min of conversion respectively.
Based on these results, the same kinetic pre-exponential factor ks
([=] m/s) for the oxidation and reduction is assumed; the reaction
is the controlling step:

3. Aspen Plus implementation

as plug ow whilst completely mixed conditions are assumed in


the emulsion phase. The whole reactor can be divided along the
axial direction into several stages, with each stage considered as
made up of two parallel ideal sub-reactors: a PFR to represent
the gas ow through the bubbles and a CSTR to represent the gas
ow through the emulsion. A rst order reaction with respect to
the concentration of the gas is assumed in each sub-reactor. Mass
transfer between the two sub-reactors in each stage (i.e. the PFR
and CSTR) occurs at their respective exit streams [16], before entering the next stage, as shown in Fig. 4.
Aspen Plus provides calculator blocks where user dened calculations can be inserted; those can be written in FORTRAN or Excel
can be used. Such calculators are used to link the sub-reactors in
each stage as well as each stage to the next. Specically, the calculator block MTr Ci (in Excel) is used to modify the outlet molar gas
ow-rate from each sub-reactor by solving the mass transfer term
between bubble and emulsion phase (Eqs. (12)(15)). Transfer
block functions (i.e. TrBi and TrEmi see Fig. 4), insure that the
streams between stages verify mass continuity for each component. In this way, the gas molar ow-rate along the bed is always
redistributed between bubble and emulsion phase. Another calculator block (Feed C), written in Excel, is used to dene the feed
conditions. For a xed value of solid inventory, W, as the inlet
supercial velocity of the methane, Uo, varies, the volume of each
sub-reactor varies (the volumetric fraction of the bubble phase,
r, the gas velocity of the bubbles, Ub, the volumetric owrate of
the bubble phase, Qb, are xed once the supercial velocity is
xed). However, a change in the solid inventory determines a variation of the total height of the uidised bed, as reported in Table 1,
which results in a change in the volume of each sub-reactor.
Aspen Plus provides only a limited number of kinetic rate
expressions; therefore, a FORTRAN code is written (i.e. Kin Ci
see Fig. 4) to implement the un-reacted core model in each subreactor. Wegstein convergence solver [34] is used for solving the
mass balance on the streams from and to the sub-reactors in each
stage. Mass balance equations for each component in each stage
are reported in Table 2.
Considering the mass transfer between the emulsion and the
bubble phase, the values of methane concentrations at the beginning of the i + 1 stage are dened as:

3.1. Bubbling bed

C CH4bi1 C CH4bi  K be  C CH4bi  C CH4ei 

Hti
Ub

12

The hydrodynamics of bubbling beds is rather complex (e.g.,


[22,23]). In this work, the gas ow through the bubbles is treated

C CH4ei1 C CH4ei K be  C CH4bi  C CH4ei 

Hti  r 

Ue
1r

13

area particles
 r0
reaction v olume

r 0 ks  C A

8
3

where CA is the molar concentration of the gas ([=] kmol/m ).


For spherical particles:

area particles
4  p  r 2c  N part

reaction v olume reaction v olume


4  p  r 2c  V tot  1  e

4
 p  R3  V tot  e
3

where rc is the average particle radius at the reaction surface, R is


the external radius of the particle, Vtot is the whole volume of the
system and e is the voidage of the system. According to the shrinking core model [30]:
1

r C R  1  X B 3

10

where XB is the average conversion of the solid reactant. Finally the


kinetic rate expression is given by:
2

2
6  1  e  ks  C A  1  X B 3
k  C A  1  X B 3
dp  e

11

where k is the kinetic pre-exponential factor ([=] s1).

Table 2
Mass balance equations for each component in each stage of the bubbling bed model.
Bubble phase
C CH4bi1  U b  Ab  C CH4bi  U b  Ab  K be  C CH4bi  C CH4ei  V bi  Ab  eb 
R zi
r CH4i dz 0
F NiObi1  F NiObi  4  Ab  eb  zi1
R zi
F CO2bi1  F CO2bi Ab  eb  zi1 r CH4i dz 0
R zi
r CH4i dz 0
F H2Obi1  F H2Obi 2  Ab  eb  zi1
R zi
F Nibi1  F Nibi 4  Ab  eb  zi1 r CH4i dz 0

R zi

zi1

r CH4i dz 0

Emulsion phase
 d 
 r CH4  V CSTRi 0
C CH4ei1  U e  Ae  C CH4ei  U e  Ae K be  C CH4bi  C CH4ei  V ei  1d
FNiOe(i1)  FNiOei  4 * rCH4 * VCSTRi = 0
FCO2e(i1)  FCO2ei + rCH4 * VCSTRi = 0
FH2Oe(i1)  FH2Oei + 2 * rCH4 * VCSTRi = 0
FNie(i1)  FNiei + 4 * rCH4 * VCSTRi = 0
Bubble volume Vb
Emulsion volume Ve
PRF volume VPFR
CSTR volume VCSTR

Vb = V * r
Ve = V * (1  r)
VPFR = Vb * eb
VCSTR = Ve * ee

51

R. Porrazzo et al. / Fuel 136 (2014) 4656

Fig. 4. Aspen Plus bubbling bed scheme.

where CCH4bi and CCH4ei are the methane concentrations at the exit
of the i stage for the bubble and the emulsion phase, respectively;
Kbe is the overall mass transfer coefcient between the bubble
and the emulsion phase; Ub and Ue are the bubble and the emulsion
supercial gas velocity, respectively; Hti is the height of i stage.
A different expression is applied if the concentrations of the
component at i + 1 stage become negative or if the mass driving
force principle is not satised:

C CH4bi1



1
Q
 C CH4bi C CH4ei  e
2
Qb

14



1
Q
 C CH4ei C CH4bi  b
2
Qe

15

C CH4ei1

where Qb and Qe are the total volumetric owrates of the bubble


and the emulsion phases, respectively. The distribution of the reactant gas between the two phases affects largely the reaction; whilst
the methane mass transfer is explicitly considered (see Eqs. (12)
(15)), the mass transfer of the products (CO2 and H2O) between
the two phases is assumed to be negligible since the kinetics is
assumed to be unaffected by the products.
An increase in the number of stages results in an increase in the
amount of gas transferred from the bubble phase to the emulsion
phase and thus available for the reaction. The effect of the number
of stages on the nal gas conversion varies depending on the
hydrodynamic and kinetic variables, i.e. on the ratio between the
inlet supercial gas velocity Uo and the inlet supercial gas velocity
at minimum uidisation condition Umf, and the ratio between the
kinetic coefcient k and the mass transfer coefcient Kbe. When
Uo  Umf, since it is assumed that the bubble phase takes the
excess of gas, with respect to the emulsion phase, the volumetric
fraction of the bubble phase, r, is high; additionally, if the number
of stages increases, the mass transfer between the two phases also
increases and so the conversion (more gas available for the reaction). When Uo  Umf, the volumetric bubble fraction r is low compared to the emulsion volumetric fraction, and the amount of gas
that by-passes the emulsion phase is small; in this case, the presence of the bubbles is less relevant and the mass transfer between
the two phases is low. In this situation, the number of stages
affects slightly the gas conversion. When k  Kbe, the mass transfer
between the two phases is the controlling step and most of the
unreacted gas remains in the bubble phase while the gas in the
emulsion phase is quickly consumed; an increase in the number
of stages allows a for large amount of fresh gas coming from the
bubbles to react with the solid particles. When k  Kbe the kinetic
term becomes the controlling step and a large amount of unreacted

Table 3
Parameters used for the simulation of the bubbling bed.
Parameter

Value

Units

Source

T
P
dp

750
1
8.00E05
0.191
2489
0.000027
1
0.78
0.0096
0.019
0.03
0.26
0.38
0.46
0.1
0.277
98
2
0.5
0.9
1
0.45
0.18
0.57
1.28
6.50E05
9.57
5
3.31
4.41E04
33.06
3.67

C
atm
m
kg/m3
kg/m3
kg/(m s)
m
m2
m/s
m/s
m
m/s
m/s
m/s
m/s
kg/s
% w/w
% w/w

m
m2/s
s1
s1
s1
m/s
s1
s1

Ref. [24]
Assumed
Ref. [24]
Aspen Plus
Aspen Plus
Aspen Plus
Assumed
Calculated
Calculated
Calculated
Ref. [25]
Calculated
Calculated
Calculated
Assumed
Assumed
Assumed
Calculated
Assumed
Assumed
Assumed
Assumed
Calculated
Calculated
Calculated
Aspen Plus
Calculated
Calculated
Calculated
Ref. [24]
Calculated
Calculated

qg
qs
l
DBB
At
Umf
Ue
db
Ut
Ubr
Ub
Uo
Fs
NiO inlet
Ni inlet

emf
eb
Lm

em
r
ef
Lf
d
Kbc
Kce
Kbe
ks
kCSTR
kPFR

database
database
database

database

gas is present in the emulsion phase; in this case the gas conversion is slightly affected by the mass transfer and by the number
of stages: the effect of the by-pass of gas in the bubble phase is
not very relevant since the gas in the emulsion phase is not largely
consumed.
The right number of stages to model the system depends on
its kinetics and hydrodynamics and the details of the choice made
will be discussed below. Experimental data taken from the literature are used to validate the multi-stages model [13,14]. In
Table 3 the values of the parameters used for the simulation of
the bubbling bed are reported. The solid fed into the system is
in stoichiometric quantity; in the air reactor, the solid reacts up
to the given conversion (almost unity); the oxidised solid is then
sent into the fuel reactor where all the methane reacts and the
nal conversion is dictated by the oxidised solid circulating in
the system.

52

R. Porrazzo et al. / Fuel 136 (2014) 4656

3.2. Riser implementation


The different amount of solid in the different regions of the riser
affects the reaction. Since the kinetic rate changes depending on
the solid voidage (Eq. (11)), the riser is split into a number of CSTRs
in series [18,19]. In the case under exam, the system is split into
four CSTRs: one CSTR is assumed to represent the dense phase
and three CSTRs are assumed to mimic the lean phase characterised by three different mean voidages calculated in the following
way:

esi es 

esd  es

a  zi  zi1

 eazi  eazi1

16

with i = 2, 3, 4. z1 is the height of the dense phase which, summed of


all to other zones, up to z4, gives the height of the whole riser. The
relationship between a and Uo plotted by Kunii and Levenspiel [22]
shows that the product of a times Uo is a constant and its value is
between 4 and 12.
The previous system of three equations (Eq. (16)) is solved with
the following constrains:

z2  z1 z3  z2 z4  z3

17

z 4 z 1 3  Dz

18

And the constrain of the solid mass balance:

Lm  1  em z1  esd z 

4
X

esi

19

i2

where Lm * (1  em) is the height of the packed solid loading equal to


W/(qs * At). W is the inventory of solid in the system, qs is the solid
density and At is the area of the column.
Most of the reaction occurs in the dense zone because of the
large quantity of reactant solid present. A calculator block (Feed
C) written in Excel is used to dene the operating condition of
the feed and the volume of each sub-reactor. The calculator block,
by solving Eqs. (16)(19), determines the height of the column and
the volume of each CSTR for an assumed value of the solid inventory, diameter of the column, and inlet supercial gas velocity Uo. A
FORTRAN code is written into the calculator blocks (Kin Ci) to modify in each CSTR the kinetic rate expression, as reported in the case
of the bubbling bed (Section 3.1). Broyden convergence solver [34]
is used for solving the mass balance in each CSTR. In Fig. 5 the
Aspen Plus scheme for the fast uidisation regime is shown.
Mass balance equations for each component in each CSTR reactor are reported in Table 4. In Table 5 the values of parameters used
for the simulation of riser are reported.

Table 4
Mass balance equations for each component in each CSTR of the fast uidisation
model.
FO2(i1)  FO2i  rO2i * VCSTRi = 0
FNi(i1)  FNii  2 * rO2i * VCSTRi = 0
FNiO(i1)  FNiOi + 2 * rO2i * VCSTRi = 0

Table 5
List of the values of parameters used for the simulation of the riser.
Parameter

Value

Units

Source

T
P
dp

750
1
8.00E05
0.23889
6033
4.7E05
0.51
3.54.5

0.013
0.35

0.22
5
95
0.251

4.41E04

C
atm
m
kg/m3
kg/m3
kg/(m s)
m
m
m2
m/s
m/s
m/s
kg/s
% w/w
% w/w
m

m/s

Assumed
Assumed
Assumed
Aspen Plus database
Aspen Plus database
Aspen Plus database
Assumed
Assumed
Calculated
Calculated
Calculated
Calculated
Exit fuel reactor
Exit fuel reactor
Exit fuel reactor
Assumed
Ref. [22]
Calculated
Calculated
Calculated
Assumed

qg
qs
l
DR
HR
At
Umf
Ut
Uo
Fs
NiO
Ni
Lm

esd1
esd2
esd3
esd4
ks

4. Results and discussion


4.1. Bubbling bed
In Fig. 6 the conversion of methane is reported as a function of
the number of stages. The inlet gas supercial velocity, Uo, and the

Fig. 6. Conversion of gas vs number of stages n at different k keeping Uo and Fs


constants.

Fig. 5. Aspen Plus fast uidisation scheme.

R. Porrazzo et al. / Fuel 136 (2014) 4656

Fig. 7. Conversions of both gas and solid vs Uo at n = 5 keeping Fs constant.

53

Fig. 9. Conversion of gas vs solid inventory W at n = 5.

Fig. 8. Conversions of both gas and solid vs Fs at n = 5 keeping Uo constant.

solid owrate, Fs, are kept constant. The ratio between the mass
transfer coefcient Kbe and the pre-exponential factor k is varied
with Kbe kept constant.
As mentioned previously, a k value lower than Kbe implies that
the kinetics is the controlling step and therefore the contribution of
gas by-pass to the overall gas conversion is lower. In this case, the
overall gas conversion is not affected greatly by the number of
stages, therefore the system can be modelled with a fewer stages.
Conversely, a higher value of k implies the mass transfer to be the
controlling step: consequently, a good gas redistribution between
the bubble and the emulsion phase is relevant. In this case, the variation of the overall gas conversion with the number of stages is
larger than the conversion obtained in the case of kinetic control.
After 5 stages, the system reaches a plateau and therefore the
multi-stage model assumes 5 stages as the maximum number of
stages. In general, the increase in the number of stages determines
the increase in the conversion for both gas and solid and the conversion values could overpredict the experimental data since the
hydrodynamics of the emulsion phase moves from well mixed to
plug ow. In the case analysed here, k is taken equal to 33 s1.
Experimental data taken from the literature are considered to justify the choice made [13,14]. A number of authors have found that
the methane conversion, using NiO/Ni supported by bentonite or
Al2O3 in a range of temperature of 10001400 K, is more than
90% with gas residence time in the range of 15 s [13,14]. The
gas residence time in the present system is about 13 s. This condition leads to two considerations:

Fig. 10. Conversions of both gas and solid vs riser diameter DR.

 The gas residence time is long enough to expect the methane


conversion higher than 90%; this condition is achieved with 5
stages and the conversion of both gas and solid reaches a
plateau.
 The gas residence time from the literatures [13,14], for methane
conversion higher than 90%, is found to be lower than the one
calculated; consequently, to reduce the gas residence time,
the height of the uidised bed is reduced (and thus the solid
inventory see Table 1). (As it can be seen in Fig. 9, the same
conversion will be obtained, since for W > 350 kg, the conversion is practically constant.)
In Fig. 7 the conversions of both methane and solid, when Uo
varies and the inlet solid mass ow-rate is kept constant are
reported. A comparison with the Gibbs reactor model is presented
in Fig. 7.
The increase in Uo determines a higher by-pass of the gas in the
bubble phase with a consequent decrease in the gas conversion.
When Uo decreases the reactant gas becomes the limiting reactant
and thus the solid conversion decreases. For a value of the inlet gas
supercial velocity higher than 0.1 m/s the solid becomes the

54

R. Porrazzo et al. / Fuel 136 (2014) 4656

Fig. 11. Conversions of both gas and solid vs riser height HR.

limiting reactant and the calculation is stopped. The sharply


increase of solid conversion depends on the high kinetic rate (k is
equal to 33 s1) of NiO reduction. Runs with lower kinetic rate
are carried out and they show a slow increase of the solid conversion when Uo increases; this is a consequence of the fact that the
decrease of both gas residence time and gassolid contact time
becomes more relevant. Finally, the slope of the solid conversion
vs the inlet supercial gas velocity Uo is a function of the kinetic
rate.
Considering the comparison between the model proposed here
and the largely used Gibbs reactor model, the Gibbs reactor
assumes that the free energy of the reaction is minimum. Thus,
the hydrodynamics is not considered and no information is available on the size of the reactors. The comparison between the two
models shows that the results are close when the inlet supercial
gas velocity is close to Umf (equal to 0.01 m/s): in this case, the volumetric fraction in the bubble phase is small and thus its inuence
on the overall gas conversion can be neglected. Most of the gas is
consumed in the CSTR reactors: the assumption that the reaction
kinetics is fast, allows for a nearly complete methane conversion.
When Uo increases, the volumetric fraction of the bubble phase
increases and thus gas bypass is observed; consequently, the difference between the present model and the Gibbs reactor becomes
more pronounced. In this case, a Gibbs reactor gives higher conversions than the multistage system implemented here.
In Fig. 8 the conversions of both methane and solid vs Fs, at constant Uo and n equal to 5, are shown. It can be seen that when Fs
increases the solid conversion decreases for a constant value of
Uo. This is a consequence of the fact that for a value of Fs higher
than 0.277 kg/s the gas becomes the limiting reactant. The solid
residence time for the reduction reaction of the NiO expressed as
ratio between the whole mass of solid charged into the system
W and the stoichiometric solid owrate [33] is 64 min: that corresponds to a solid conversion rate of 1.55%/min. This reaction time
is quite large. A sensitivity analysys is necessary to know the right
loading of solid to reach more than 90% in the gas conversion. In
Fig. 9 the relationship between the solid inventory and the gas conversion is shown.
Fig. 9 shows that in the interval 3601060 kg of solid inventory
the gas conversion is unchanged. Thus, the same degree of gas
conversion can be achieved by choosing a lower solid inventory.

Fig. 12. Conversions of both gas and solid vs height of the packed solid loading Lm.

Fig. 13. Conversions of both gas and solid vs molar stoichiometric ratio Fair/FCH4.

Consequently, the new value of 360 kg is chosen and that corresponds to a value of Lm of 33 cm. This corresponds to a reaction
time of 21 min and a solid conversion rate of 4.6%/min: such value
is very close to the experimental data reported by Lyngfelt et al.
[33] (7%/min of reduction reaction rate with NiO supported 60/
40 and 4%/min of reduction reaction rate with pure NiO) for NiO/
Ni supported reacting with methane. Based on measured kinetics
and assuming complete gassolid contact, a very small solid
inventory of around 1020 kg/MW would be sufcient to reach full
conversion in the fuel reactor, or equivalently, to reach the equilibrium conversion [28]. However, in bubbling uidised beds, the
gassolids contact is not complete, since gas by-pass is induced
by the presence of the bubbles. Consequently, predictions based
on the kinetics and assuming complete gassolids contact would
underpredict the solid inventory, and much larger solid inventory
would be needed, around 500 kg/MW [29,35]. In this case, the
plant generates 0.75 MW, thus the solid inventory from the

R. Porrazzo et al. / Fuel 136 (2014) 4656


Table 6
Optimised values of the variables that affect the riser performance.
Uo (m/s)

DR (m)

HR (m)

Lm (m)

ks (m/s)

Fair/FCH4

1.79

0.8

3.5

0.25

4.41E04

1.25

Table 7
Heat duty calculation of the CLC system.

Bubbling bed (T = 750 C)


BB
BB
BB
BB
BB

the same temperature of the fuel reactor are assumed. Higher temperature and higher values of Ks will lead to higher conversion as
founded by Lyngfelt et al. [33] (21%/min) and Sung and Sang [32]
(12%/min). The value of the inlet supercial gas velocity, Uo,
obtained is in good agreement with data from the literature (about
100 Umf) [14]. Volumetric gas ow in the air reactor is approximately 10 times larger than that of CH4 (11 times in this case) as
mentioned by Abad et al. [28] and Johansson et al. [36].
4.3. Net heat duty

Heat duty

Cstr1
Cstr2
Cstr3
Cstr4
Cstr5

55

24.06
58.87
30.69
16.49
8.97

kW
kW
kW
kW
kW

Riser (T = 750 C)
pfr1
pfr2
pfr3
pfr4
pfr5

BB
BB
BB
BB
BB

Tot BB

2.69
0
0
0
0

kW
kW
kW
kW
kW

Cstr1
Cstr2
Cstr3
Cstr4

415.58
296.65
78.07
45.28

kW
kW
kW
kW

141.76

kW

Tot Riser

835.57

kW

literature should be around 375 kg: this value is close to the operating value of 360 kg.
4.2. Riser
The inlet solid owrate in the riser (air reactor) must match the
rate circulating in the bubbling bed fuel reactor. Usually a cyclone
between the two beds allows for the separation between gas and
solid products from the air reactor. The solid particles from the
bubbling bed return to the air reactor by gravity.
In the air reactor the variables that can be changed, apart from
operative temperature and pressure, are: diameter DR, height HR,
solid inventory and molar stoichiometric ratio between air and fuel
reactants. A sensitivity analysis is carried out to understand how
the variables taken into account inuence the conversions of both
gas and solid. The sensitivity analysis gives information on the values of the variables that optimise the process. In Fig. 10 the conversions of both gas and solid, as a function of the riser diameter, are
shown. The increase in the diameter of the riser, DR determines a
decrease in Uo (at constant inlet molar air owrate), which implies
a higher gas residence time, higher solidgas contact time and
higher conversions. In Fig. 11 the conversions of both gas and solid
as function of the height of the riser HR are shown:
An increase in the height of the riser, HR, corresponds to a higher
residence time for the gas and higher conversions. In Fig. 12 the conversions of both gas and solid, as a function of the solid inventory,
expressed as height of packed solid loading Lm (Table 1), are shown.
The increase in the height of packed solid loading determines a
higher kinetic rate reaction as a function of solid voidage (Eq. (11))
and therefore higher conversions. In Fig. 13 the conversions of both
gas and solid, as a function of the molar stoichiometric ratio
between air and methane, are reported. It can be noted that the
increase of the molar stoichiometric ratio between air and methane determines an increase of the solid conversion and a decrease
of gas conversion because of the excess of gas with respect to the
stoichiometric amount.
Analysis of Figs. 1013 leads to the optimal choice of the values
of the variables that affect the riser performance and those are
reported in Table 6.
The oxygen molar conversion is about 73% and the nickel molar
conversion is about 96%. The data reported in Table 6 allow for the
solid owrates in the two reactors to match. In the riser the solid
inventory is about 417 kg, the solid oxidation reaction time is
about 31 min and the oxidation solid conversion rate is about
4%/min. The solid oxidation conversion rate is close to the reduction one because the same kinetic pre-exponential factor Ks and

Finally, a check of the net heat duty of the whole process is carried out. It is expected that the net heat duty of the process
matches the net heat duty produced by the direct combustion of
methane with air [14,15]. The direct combustion of methane is
implemented in Aspen Plus using a stoichiometric reactor at the
same temperature of the two uidised beds (750 C), with 92.5%
of methane molar conversion as obtained in the fuel reactor. The
simulated results show that the net heat duty of the process is
693.81 kWatts that is equal to the net heat duty produced by
methane direct combustion (Table 7). In the riser, the heterogeneous exothermic reaction occurs mainly in the dense phase and
the heat produced decreases along the bed height as a consequence
of the decreasing of the reaction due to a decrease of solid particles.
In the bubbling bed reactor, the heat consumed by the endothermic reaction decreases along the bed height; this is a consequence
of the decrease in methane concentration resulting in the reduction of the reaction; this trend is observed from CSRT2 up to CSRT5.
CSTR1 takes care of the split of the feed between bubble and emulsion phase and it is introduced to initialise the sequence of the
stages. Finally, the CLC power plant is assumed to work at atmospheric pressure, with the fuel reactor usually working in adiabatic
conditions and with the circulating solid ow-rate from the air
reactor sustaining the endothermic reaction [14,15]; the riser usually works at temperatures higher than those in the fuel reactor
and the difference in temperature depends on the amount of solid
ow-rate circulating between the two reactors. The extra energy
produced by the exothermic reaction in the riser is employed in
a steam turbine cycle for electricity production.
5. Conclusions
In this work, a circulating chemical looping combustion process
is studied. The system is characterised by a fuel reactor operating
in the bubbling regime, and a circulating bed air reactor operating
in the fast uidisation regime. The reactions taking place are
non-catalytic heterogeneous reactions and the shrinking core model
is applied to describe the kinetic mechanism. The Davidson and
Harrison model [23] is applied to describe the hydrodynamics of
the bubbling bed and bubble phase and emulsion phase are considered. The riser is described considering only one phase with solid
particles concentration varying along the bed height. Air and methane are considered as inlet gas reactants for the riser and the fuel
reactor respectively. NiO/Ni is used as oxygen carrier between
the two uidised beds which undergo red-ox reactions. The whole
system is modelled in Aspen Plus where modications are made to
consider the hydrodynamics and the kinetics of the system in a
novel way. The results obtained show that:
 The model allows for including hydrodynamic and kinetic
mechanisms although its complexity is lower than CFD models
implementing more rigorous description of the system at the
particle level (e.g. two-uid models).
 The model gives a good estimation of the main variables
involved, namely diameter and height of the two beds, solid
inventory, molar stoichiometric ratio between air and fuel.

56

R. Porrazzo et al. / Fuel 136 (2014) 4656

 A comparison with a uidised bed modelled using the Gibbs


reactor is carried out showing differences in terms of gas/solid
conversion since the gas by-pass through the bubble phase is
neglected in a Gibbs reactor model; consequently, the model
proposed here is more accurate (realistic) than the Gibbs one.
 The model makes it possible to estimate the thermal efciency
of the process and to undertake an economic analysis.
As a concluding remark, more complicated reactor set-ups,
could in principle be modelled by implementing the appropriate
combination of sub-reactors.
Acknowledgement
The Engineering and Physical Sciences Research Council
(EPSRC) support is acknowledged under the Grant No. EP/
F034482/1.
References
[1] Bryant EA. Climate process and change. Cambridge (UK): Cambridge University
Press; 1997.
[2] National Council for Science and the Environmental. In: Blockstein DE,
Shockley MA, editors. Energy for a sustainable and secure future: a report of
the sixth national conference on science, policy and the environment,
Washington, DC; 2006.
[3] Abu-Zhara MRM, Schneiders LHJ, Niederer JPM, Feron PHM, Versteeg GF. CO2
capture from power plants: Part II. A parametric study of the technical
performance based on mono-ethanolamine. Int J Greenhouse Gas Control
2007;1(1):13542.
[4] Abu-Zhara MRM, Schneiders LHJ, Niederer JPM, Feron PHM, Versteeg GF. CO2
capture from power plants: Part I. A parametric study of the technical
performance based on mono-ethanolamine. Int J Greenhouse Gas Control
2007;1(1):3746.
[5] Singh D, Croiset E, Douglas PL, Douglas MA. Techno-economic study of CO2
capture from an existing coal-red power plant: MEA scrubbing vs. O2/CO2
recycle combustion. Energy Convers Manage 2003;44(19):307391.
[6] Herzog HJ. The economics of CO2 capture. In: Proceedings of the fourth
international conference of greenhouse gas control technologies. London:
Pergamon Press; 1999. p. 1016.
[7] Ritcher H, Knoche K. Reversibility of combustion process. ACS Symp Ser
1983;235:7185.
[8] Ishida M, Zheng D, Akehata T. Evaluation of a chemical-looping combustion
power-generation system by graphic exergy analysis. Energy 1987;12:14754.
[9] Wolf J, Anheden M, Yan J. Comparison of nickel- and iron-based oxygen
carriers in chemical looping combustion for CO2 capture in power generation.
Fuel 2005;84:9931006.
[10] Kerr HR. Capture and separation technologies gaps and priority research
needs. In: Thomas D, Benson S, editors. Carbon dioxide capture for storage in
deep geologic formationsresults from the CO2 capture project, vol. 1. Oxford
(UK): Elsevier Ltd.; 2005 [chapter 38].
[11] IPCC special report on carbon dioxide capture and storage; 2005.
[12] Ishida M, Jin H. Novel chemical-looping combustor without NOx formation. Ind
Eng Chem Res 1996;35:246972.

[13] Hossain MM, de Lasa HI. Chemical-looping combustion (CLC) for inherent CO2
separations a review. Chem Eng Sci 2008;63:443351.
[14] Fan Liang-Shih, John Easton C. Chemical looping systems for fossil energy
conversions. Wiley; 2010.
[15] Hatanaka T, Matsuda S, Hatano H. A new-concept gassolid combustion
system MERIT for high combustion efciency and low emissions. In:
Proceedings of intersociety energy conversion engineering conference, vol.
30; 1997. p. 9448.
[16] Jafari R, Sotudeh-Gharebagh R, Mostou N. Modular simulation of uidized
reactors. Chem Eng Technol 2004;27:2.
[17] Sarvar-Amini A, Sotudeh-Gharebagh R, Bashiri H, Mostou N, Haghtalab A.
Sequential simulation of a uidized bed membrane reactor for the steam
methane reforming using ASPEN PLUS. Energy Fuels 2007;21:35938.
[18] Sotudeh-Gharebaagh R, Legros R, Chaouki J, Paris J. Simulation of circulating
uidized bed reactors using Aspen Plus. Fuel 1998;77(4):32737.
[19] Liu B, Yang X, Song W, Lin W. Process simulation of formation and emission of
NO and N2O during coal decoupling combustion in a circulating uidized bed
combustor using Aspen Plus. Chem Eng Sci 2012;71:37591.
[20] Sohi AH, Eslami A, Sheikhi A, Sotudeh-Gharebagh R. Sequential-based process
modeling of natural gas combustion in a uidized bed reactor. Energy Fuels
2012;26:205867.
[21] Li F, Kim HR, Sridhar D, Wang F, Zeng L, Chen J, et al. Syngas chemical looping
gasication process: oxygen carrier particle selection and performance. Energy
Fuels 2009;23:41829.
[22] Kunii D, Levenspiel O. Fluidization engineering. 2nd ed. Boston: ButterworthHeinmen; 1991.
[23] Davidson JF, Harrison D. Fluidized particles. Cambridge University Press; 1963.
[24] Ryu HJ, Bae DH, Han KH, Lee SY, Jin GT, Choi JH. Oxidation and reduction
characteristics of oxygen carrier particles and reaction kinetics by unreacted
core model. Korean J Chem Eng 2001;18:8317.
[25] Mattisson T, Lyngfelt A, Cho P. The use of iron oxide as an oxygen carrier in
chemical-looping combustion of methane with inherent separation of CO2.
Fuel 2001;80:195362.
[26] Garcia-Labiano F, de Diego LF, Adanez J, Abad A, Gayan P. Temperature
variation in the oxygen carrier particles during their reduction and oxidation
in a chemical-looping combustion system. Chem Eng Sci 2005;60:85162.
[27] Garcia-Labiano F, Adanez J, de Diego LF, Gayan P, Abad A. Effect of pressure on
the behavior of copper-, iron-, and nickel-based oxygen carrier for chemicallooping combustion. Energy Fuel 2006;20:2633.
[28] Abad A, Adnez J, Garca-Labiano F, De Diego LF, Gayn P, Celaya J. Mapping of
the range of operational conditions for Cu-, Fe-, and Ni-based oxygen carriers
in chemical looping combustion. Chem Eng Sci 2007;62:53349.
[29] Abad A, Adnez J, Dueso C, Garca-Labiano F, De Diego LF, Gayn P. Modeling of
the chemical looping combustion of methane using a Cu-based oxygen-carrier.
Combust Flame 2010;157:60215.
[30] Levenspiel O. Chemical reaction engineering, vol. 83. New York: Wiley; 1965.
[31] Marban G, Garcia-Calzada M, Fuertes AB. Kinetics of oxidation of CaS particles
in the regime of low SO2 release. Chem Eng Sci 1999;54:77.
[32] Sung RS, Sang DK. Chemical-looping combustion with NiO and Fe2O3 in
thermobalance and circulating uidized bed reactor with double loops. Ind
Eng Chem Res 2006;45:268996.
[33] Lyngfelt A, Leckner B, Mattisson T. A uidized-bed combustion process with
inherent CO2 separation; application of chemical-looping combustion. Chem
Eng Sci 2001;56:310113.
[34] Gupta SK. Numerical method for engineers. New Age International; 1995.
[35] Lyngfelt A. Oxygen carriers for chemical looping combustion 4000 h of
operational experience. Oil Gas Sci Technol 2011;66(2):16172.
[36] Johansson E, Mattisson T, Lyngfelt A, Thunman H. A 300 W laboratory reactor
system for chemical looping combustion with particle circulation. Fuel
2006;85:142838.

You might also like