- Email: [email protected]
Multi-scale simulation of chemical looping combustion in dual circulating fluidized bed
Applied Energy 155 (2015) 719–727
Contents lists available at ScienceDirect
Applied Energy journal homepage: www.elsevier.com/locate/apenergy
Multi-scale simulation of chemical looping combustion in dual circulating fluidized bed Wang Shuai a,⇑, Chen Juhui b,⇑, Lu Huilin a, Liu Guodong a, Sun Liyan a a b
School of Energy Science and Engineering, Harbin Institute of Technology, Harbin 150001, China School of Mechanical Engineering, Harbin University of Science and Technology, Harbin 150080, China
h i g h l i g h t s A modified multi-scale gas–solid flow-reaction coupled model is developed. Multi-scale characteristic of chemical looping combustion system is investigated. Predicted results show a good agreement with experimental data.
a r t i c l e
i n f o
Article history: Received 23 March 2015 Received in revised form 17 May 2015 Accepted 31 May 2015
Keywords: Computational fluid dynamics Chemical looping combustion Cluster Multi-scale Dual circulating fluidized bed
a b s t r a c t Chemical looping combustion (CLC) in an interconnected fluidized bed has attracted more and more attention owing to its novel technology with inherent separation of CO2. In recent years, some models have been developed to investigate the gas-particle flow and reactive characteristics during the CLC process. However, multi-scale structures in reactors make it complex to perform a simulation. In the current work, a multi-scale gas–solid flow-reaction coupled model is developed and applied to the simulation of the CLC process in a dual circulating fluidized bed (DCFB) system with consideration of the impact of multi-scale structures on chemical reactions, mass and heat transfer. By comparisons of gas pressure and gas components with experimental data, the present model shows a better prediction. The influence of clusters on the gas compositions and temperature field is analyzed. Ó 2015 Elsevier Ltd. All rights reserved.
1. Introduction With the global warming, carbon dioxide (CO2) emissions have attracted more and more attention. Chemical looping combustion is a promising CO2 capture technology, which isolates the fuel source from the oxygen source by oxygen carriers and achieves inherent separation of CO2 [1,2]. In the last two decades, diligent research efforts have been devoted to develop the CLC system [3–5]. Cuadrat et al. [6] investigated a coal fuelled CLC reactor using ilmenite as oxygen carriers. It was found that the optimization of the CLC with coal can result in energy production with high CO2 capture. García-Labiano et al. [7] exploited an energy potential of acid through the combustion of H2S using the CLC system. It was concluded that the H2S concentration of 20 vol.% was high enough to turn CLC into an auto-thermal process. Haonsen et al. [8] studied the CLC performance in a rotating bed reactor and discussed the effects of process parameters by means of experiments. Miller ⇑ Corresponding authors. E-mail addresses: [email protected] (S. Wang), [email protected] (J. Chen). http://dx.doi.org/10.1016/j.apenergy.2015.05.109 0306-2619/Ó 2015 Elsevier Ltd. All rights reserved.
et al. [9] emphasized that the Mg-promoted hematite enhanced oxygen transfer capacity of hematite and showed better performance for methane CLC. Zhang et al. [10] performed a comparison between fluidized bed and fixed bed for a pressurized coal-fuelled CLC process. It was demonstrated that the fluidized bed mode had the superiority for a long-term stable operation. Computational fluid dynamics (CFD) provides an effective insight into the hydrodynamics of the complicated multiphase flow and has been applied to the simulation of the CLC process [11–13]. Peng et al. [14] employed a CFD–DEM model to evaluate some factors that influenced the solid circulation rate (SCR) in chemical looping systems. The results indicated that the flow regime in the air reactor was the main mechanism to determine the fluctuation frequency and amplitude of SCR. Guan et al. [15] incorporated the kinetic theory of granular flow into a three-dimensional CFD model to predict the hydrodynamics characteristic in the interconnected fluidized bed for the CLC process. It was found that the drag model had a significant impact on the prediction of the gas–solid flow and the solids flow patterns in the fuel reactor (FR) depended strongly on the operating parameters in the air reactor (AR).
720
S. Wang et al. / Applied Energy 155 (2015) 719–727
Nomenclature a dc ds cp f F g h hdc Kdc M n Ndf
acceleration (m s2) cluster diameter (m) particle diameter (m) specific heat capacity (J kg1 K1) volume fraction of dense phase force acting on each particle or cluster (N) gravity (m s2) local heat transfer coefficient (J m2 s1 K1) inter-phase heat transfer coefficient (J m2 s1 K1) inter-phase mass exchange coefficient (m s1) Molar weight (g mol1) reaction order energy dissipation (W kg1)
r S T u U Xi
reaction rate (kmol m3 s1) mass source (kg m3 s1) temperature (K) velocity (m s1) superficial velocity (m s1) mass fraction of gas species
The dual circulating fluidized bed (DCFB) system including two interconnected circulating fluidized beds has the potential for the CLC process owing to its low particle attrition rate and excellent gas–solids contact, as shown in Fig. 1. Meanwhile, the staged fluidization of the air reactor controls the global circulation rate of the system [16]. Penthor et al. [17] presented the CLC performance in the DCFB system using a copper based oxygen carrier and revealed that the solids inventory and circulation were critical parameters for the fuel conversion. The impact of internal installation on the particles residence time was evaluated [18]. The results revealed that the effect of internal loop was enhanced when
Fig. 1. Sketch of a DCFB reactor system.
Xox
reduction or oxidization degree of oxygen carriers
Greek letters a specific surface area (m1) b drag coefficient (kg m3 s1) e volume fraction k thermal conductivity (W m1 K1) l viscosity (Pa s) q density (kg m3) Subscripts c cluster den dense phase dil dilute phase int interface g gas phase s particle phase
fluidization velocities were increased in reactors. Wang et al. [19,20] investigated the flow and reactive characteristic for the CLC process in the DCFB system by means of CFD modeling, where a cluster structure-dependent (CSD) drag model was used to consider the cluster impact. From the simulated results, we can find that there was an improvement of the predictions in some extent. However, the cluster effects on reactions and inter-phase heat transfer were neglected. Lu et al. [21] pointed out that an isolated particle had significantly higher mass and heat transfer rates than the particle in the cluster by an investigation of the gas-cluster mass and heat transfer. An increase of the cluster porosity can promote the gas-cluster mass and heat transfer rates. Hence, it is necessary to take into account the cluster impact on mass and heat transfer during the CLC process. In recent studies, Hou et al. [22,23] established the relationship between mass transfer and flow structures for fast fluidized beds and developed a multi-scale mass transfer (MSMT) model based on energy-minimization multi-scale (EMMS) method. In this model, the cluster impact on mass transfer was considered. The simulated results of the ozone decomposition process showed a better agreement with experimental data. The MSMT model was further validated for the catalytic oxidation of carbon monoxide in a circulating fluidized bed [24]. The results also demonstrated that the clusters played an important role in the gas–solid mass transfer. To describe the effect of clusters on chemical reaction, mass and heat transfer simultaneously, a multi-scale chemical reaction model coupling heat transfer was developed in our previous work [25], where the temperature discrepancy between cluster phase and dispersed phase was incorporated and the cluster structure-dependent (CSD) drag model was adopted to provide local structural parameters. By simulating the regeneration reaction of oxygen carriers in riser reactors, the multi-scale chemical reaction model promoted the accuracy of predicted results. This work focuses on the investigation of the multi-scale flow and reactive characteristics for the whole CLC process by means of CFD modeling. For the simulation of the DCFB system consisting of two circulating fluidized beds, employing a multi-scale model coupled with chemical reaction and heat transfer is more rational to reflect the influence of the temporal-spatial multi-scale structures on the chemical looping system, which is not available in recent literatures. The advantage of this approach lies in the more
721
S. Wang et al. / Applied Energy 155 (2015) 719–727
in-depth sight into the multi-scale structure performance during the reaction process. The formation of amorphous aggregates of oxygen carriers can be observed. Meanwhile, the simulation of the complete loop system including loop-seals and separators is carried out. The characteristics of multi-scale mass and heat transfer in reactors are analyzed. 2. Mathematical model In the current research, the two-fluid model is used as the basic framework of the multi-scale simulation with the kinetic theory of granular flow [26]. It is assumed that the solid phase is spherical with a uniform density and size. The governing equations mainly include continuity equations, momentum conservation equations, species transport equations and energy conservation equations. Detailed descriptions of the model can be found in Wang et al. [20]. To characterize the impact of cluster structures on the flow and reaction behaviors during the CLC process, the local flow in the control volume is separated into three sub-systems: the dense phase in the form of clusters, the dilute phase in the form of dispersed particles and the interface between the dense phase and
hgs ¼
t1 ¼
hdc
kg eden ¼ 2:0 þ dc
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 4qg cpg kg eden pt 1
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi Dg;i eden 4Dg;i eden ¼ 2:0 þ dc pt 1 dc jðU g;den =eden Þ ðU s;den =ð1 eden ÞÞj
ð1Þ
ð2Þ
ð3Þ
With local temperatures and heat transfer coefficients in the dilute phase and the dense phase, we can calculate the effective inter-phase heat transfer coefficient through the following equation:
ap ð1 f Þð1 edil Þhdil ðT g;dil T s;dil Þ þ ac f ð1 eden Þhint ðT g;dil T s;den Þ þ ðap ac Þf ð1 eden Þhden ðT g;den T s;den Þ ap ð1 eg ÞðT g T s Þ
the dilute phase. In each sub-system, gas–solid interaction, mass source and temperature owing to chemical reactions and heat transfer are solved respectively. Then these local parameters are integrated into the equations in the control volume to reflect the multi-scale effects. Here, the CSD drag model is employed to solve the multi-scale gas–solid interaction [27]. The expression of the inter-phase drag coefficient is given by Eq. (T1-1) in Table 1. In order to solve this multi-scale drag coefficient, the eight local structural parameters (Ug,dil, Ug,den, Us,dil, Us,den, edil, eden, f, dc) are required to be obtained by solving six equations (T1-2)–(T1-7) and a stability criterion of the minimum energy dissipation consumed by drag force (T1-8). To consider the cluster effects on reactions and heat transfer, the multi-scale chemical reaction model coupled with heat transfer was proposed and applied to simulate the riser reactors [25]. Here, the model is further modified with consideration of the discrepancy in oxygen carrier conversions between different phases. Detailed descriptions in the model can found in Table 2. The gas and solid species transport equations for different phases are expressed by Eqs. (T2-1), (T2-2), (T2-5) and (T2-6), respectively. The corresponding mass source terms are given by Eqs. (T2-3), (T2-4), (T2-7) and (T2-8). By integrating the obtained gas and solid species mass source terms in the dense phase and the dilute phase, we can derive the effective species mass source terms, which are incorporated into the two-fluid model to achieve the implement of the multi-scale chemical reaction model. In the model, the gas mass exchange between the dense phase and the dilute phase is considered according to La Nauze et al. [28]:
K dc;i
For the reaction rates at different sub-systems, the shrinking core methodology is selected and the related expressions are shown in Eqs. (T2-9)–(T2-11), where the reduction or oxidization of available surface area with reaction in progress is considered [29]. Following the previous work [25], the local gas and solid temperatures in the dense phase and the dilute phase are obtained by solving their respective energy balance equations, which are described by Eqs. (T2-12)–(T2-15), where the heat loss by radiation is assumed to be negligible. It can be found that except for the gas– solid heat transfer between phases, the gas heat transfer between the dilute phase and the dense phase is also taken into account, which is expressed as follows [22]:
ð4Þ
By means of the solution of the above multi-scale drag coefficient, the effective species mass source terms and the effective gas–solid heat transfer coefficient, the cluster-structure-dependent multi-scale model is established. 3. Model implementation description In this work, the DCFB system is selected as the objective of the simulation, which comprises fuel reactor and air reactor, as described in Kolbitsch et al. [16,30]. The lower loop seal is used to connect the two reactors and transport the reduced oxygen carriers. The internal loop seal and cyclone achieve internal recirculation of particles of the FR. The main system parameters are listed in Table 3 [16]. With respect to chemical reactions during the CLC process, the nickel oxide is chosen as oxygen carriers and the methane is used as the gas fuel. In the FR, the following three reactions are considered:
CH4 ðgÞ þ NiOðsÞ ! NiðsÞ þ 2H2 ðgÞ þ COðgÞ
ðR-1Þ
H2 ðgÞ þ NiOðsÞ ! NiðsÞ þ H2 OðgÞ
ðR-2Þ
COðgÞ þ NiOðsÞ ! NiðsÞ þ CO2 ðgÞ
ðR-3Þ
The carbon monoxide and hydrogen are generated as intermediates by Reaction (R-1) and simultaneously consumed by Reactions (R-2) and (R-3). In the AR, the regeneration reaction of oxygen carriers occurs and is written as follows:
O2 ðgÞ þ 2NiðsÞ ! 2NiOðsÞ
ðR-4Þ
Here, the kinetic parameters are chosen according to the Abad et al. [31]. Initially, the particles are filled in the FR and AR with the solid concentration of 0.5 and the static bed heights of 0.45 m and 0.3 m. The inlet velocities are specified at the bottom of the reactors and the pressure-outlets locate at the top of separators. At the wall,
722
S. Wang et al. / Applied Energy 155 (2015) 719–727
Table 1 Cluster structure-dependent drag model. 1. Equation for cluster structure-dependent drag coefficient e
(T1-1)
g ½nden F den þ ndil F dil þ nint F int bCSD ¼ jug u sj
2. Balance equations eg ¼ f eden þ ð1 f Þedil ug ¼ e1g ½fU g;den þ ð1 f ÞU g;dil
(T1-2) (T1-3)
us ¼ e1s ½fU s;den þ ð1 f ÞU s;dil
(T1-4)
3. Equation for momentum equation of the dense phase (T1-5)
nden F den þ nint F int ¼ f ð1 eden Þðqs qg Þðg þ as;den Þ þ f ð1 eden Þ @p @z 4. Equation for momentum equation of the dilute phase
(T1-6)
ndil F dil ¼ ð1 f Þð1 edil Þðqs qg Þðg þ as;dil Þ þ ð1 f Þð1 edil Þ @p @z 5. Equation for pressure drop balance nden F den f eden
ndil F dil nint F int ¼ ð1f þ ð1f þ qg ðag;dil ag;den Þ Þedil Þedil
(T1-7)
6. Stability criterion by minimization of the energy dissipation by drag force N df ¼ ð1e1g Þq ½nden F den U g;den þ ndil F dil U g;dil þ nint F int U g;dil ð1 f Þ ! minimum
(T1-8)
s
Table 2 Multi-scale chemical reaction and heat transfer model. 1. Gas species transport equations for the dense phase and the dilute phase @ðf eden qg X g;den;i Þ þ r ðf den g X g;den;i ug;den f den g Dg;i r X g;den;i Þ ¼ Sg;den;i @t @ðð1f Þedil qg X g;dil;i Þ þ r ðð1 f Þ dil g X g;dil;i ug;dil ð1 f Þ dil g Dg;i r X g;dil;i Þ @t P Sg;den;i ¼ m j¼1 r den;j M g;i K dc;i g c f den ðX g;den;i X g;dil;i Þ Pm Sg;dil;i ¼ j¼1 ðr dil;j þ r int;j ÞMg;i þ K dc;i g c f den ðX g;den;i X g;dil;i Þ
e
e
q
e q qa e
(T2-1)
q
e q
(T2-2)
¼ Sg;dil;i
(T2-3) (T2-4)
qa e
2. Solid species transport equations for the dense phase and the dilute phase @ðf ð1eden Þqs X s;den;i Þ þ r ðf ð1 den Þ s X s;den;i us;den Þ r ðf ð1 den Þ s Ds;i r ðX s;den;i ÞÞ ¼ Ss;den;i @t @ðð1f Þð1edil Þqs X s;dil;i Þ þ r ðð1 f Þð1 dil Þ s X s;dil;i us;dil Þ r ðð1 f Þð1 dil Þ s Ds;i r ðX s;dil;i ÞÞ @t P Ss;den;i ¼ m j¼1 ðr den;j þ r int;j ÞM s;i P Ss;dil;i ¼ m j¼1 r dil;j M s;i
e
e
q
e
q
(T2-5)
q
e
q
¼ Ss;dil;i
3. Reaction rates for the dense phase, the dilute phase, and the interface f eden X g;den;j qg n rden;j ¼ d60 s0 f ð1 eden Þð1 dds Þð1 X ox;den Þ2=3 k0 exp RTEden M g;j cl ð1f Þedil X g;dil;j qg n rdil;j ¼ d60 s0 ð1 f Þð1 edil Þð1 X ox;dil Þ2=3 k0 exp RTEdil M g;j ð1f Þe X q n dil g;int;j g rint;j ¼ d60 s0 f ð1 eden Þ dds ð1 X ox;den Þ2=3 k0 exp RTEint M g;j
(T2-6) (T2-7) (T2-8)
(T2-9) (T2-10) (T2-11)
cl
4. Energy balance equations of gas phase for the dense phase and the dilute phase lgt @ @t ðf eden qg c pg T g;den Þ þ r ðf eden qg ug;den c pg T g;den Þ ¼ r ½f eden kg þ c pg Prt r T g;den þ ac f eden hdc ðT g;dil T g;den Þ P þðap ac Þf ð1 eden Þhden ðT s;den T g;den Þ þ Sg;den;i cpg;i T g;den h i lgt @ @t ðð1 f Þedil qg c pg T g;dil Þ þ r ðð1 f Þedil qg ug;dil c pg T g;dil Þ ¼ r ð1 f Þedil kg þ c pg Pr t r T g;dil þ ap ð1 f Þð1 edil Þhdil ðT s;dil T g;dil Þ P þac f ð1 eden Þhint ðT s;den T g;dil Þ þ ac f eden hdc ðT g;den T g;dil Þ þ Sg;dil;i cpg;i T g;dil 5. Energy balance equations of solid phase for the dense phase and the dilute phase h i lst @ @t ðf ð1 eden Þqs c ps T s;den Þ þ r ðf ð1 eden Þqs us;den c ps T s;den Þ ¼ r f ð1 eden Þ ks þ c ps Prt r T s;den þ ac f ð1 eden Þhint ðT g;dil T s;den Þ P þðap ac Þf ð1 eden Þhden ðT g;den T s;den Þ þ Ss;den;i cps;i T s;den h i P lst @ Ss;dil;i cps;i T s;dil @t ðð1 f Þð1 edil Þqs c ps T s;dil Þ þ r ðð1 f Þð1 edil Þqs us;dil c ps T s;dil Þ ¼ r ð1 f Þð1 edil Þ ks þ c ps Prt r T s;dil þ
(T2-12) (T2-13)
(T2-14) (T2-15)
þap ð1 f Þð1 edil Þhdil ðT g;dil T s;dil Þ 6. Local heat transfer coefficients for the dense phase, the dilute phase, and the interface kg Nus;den eden ds kg Nus;dil edil ds kg Nus;int edil ð1f Þ dc
hden ¼
(T2-16)
hdil ¼
(T2-17)
hint ¼
no-slip and partial-slip conditions are adopted for gas and solid phases, respectively [32]. A constant temperature condition is used for the wall. Detailed operating parameters can be found in Table 3. Regarding the computational domain, a two-dimensional simulation is carried out. Following the previous investigation of the mesh independence [19,20], the grid sizes of 5 10 mm and 10 10 mm are used for the reactors and other components respectively. The model is implemented on the basis of the
(T2-18)
M-FIX program, which is an open-source CFD code written in FORTRAN to describe the multiphase flow with inter-phase exchanges. A solids volume fraction correction equation is employed instead of a solids pressure correction equation, as reviewed by Syamlal [33]. The above governing equations are conducted using finite differences equations with a higher order Total Variation Diminishing (TVD) scheme, which incorporates a modification into the higher order upwind scheme for hyperbolic
723
S. Wang et al. / Applied Energy 155 (2015) 719–727 Table 3 System properties and parameters for the simulations [16]. Unit
AR
FR
Reactor height Reactor diameter Diameter of cyclone Diameter of downcomer Height of downcomer Particle diameter Particle density Gas diffusivity
m m m m
4.1 0.15 0.15 0.05
3.0 0.159 0.15 0.05
m
0.5 135 3416 O2(0.178), N2(0.2)
Static bed height Initial concentration of particles Initial temperature Inlet gas flow Inlet gas temperature
m –
0.3 0.5
1.5 135 3416 CH4(0.2064), H2(0.634), CO(0.1921), CO2(0.1381), H2O(0.2178) 0.45 0.5
K N m3/h K
1223 138.0 300
1173 12.0 300
lm kg/m3 cm2 s1
Height from inlet of AR and FR (m)
Description
4
AR
FR TFR,0=1173K
Experimental data Present simulation Without multiscale simulation
3
2
air/fuel ratio=1.1 FR exit
TAR,0=1223K
Experimental data Present simulation Without multiscale simulation
air/fuel ratio=1.1 1
0 0
3
6
0
5
10
15
Pg (kPa)
(a)
4. Results and discussion Fig. 2(a) shows the simulated axial profiles of gas pressure along the height of the AR and FR in comparison with experimental results [16]. It can be seen that the pressure is higher at the bottom of the FR, and decreases rapidly with the height increased. At the upper section of the FR, the change of the FR pressure is not obvious, which reflects the distribution of solid concentration in the FR. The AR pressure spreads over the whole AR height compared to the FR pressure. Owing to a lower operating velocity and a higher solid concentration in the FR compare to those in the AR, there is a higher pressure in the FR. The predictions with no consideration of the multi-scale effects are also shown in Fig. 2(a). We can find that the present multi-scale model can give a better prediction. The gap between the model prediction and measured data results from the selectivity of uniform mean particle diameter in the simulation. In the experiments, the particles have a non-uniform distribution [16,30]. Additionally, the fluidization nozzles are arranged at the bottom of the reactors. The discrepancy of the inlet conditions between simulation and experiment is also a reason for this gap. Overall, the prediction by the model is consistent with measured data. To further validate the reaction model, the simulated gas compositions are compared to measured data as shown in Fig. 2(b). Meanwhile, the relative error between simulated results and experimental data is also given. It is found that the predictions by the multi-scale model are closer to experimental data [30]. The results without multi-scale effects show a lower fuel conversion and deviate from experimental data. Fig. 3 displays instantaneous distributions of the O2 molar fractions in the dense phase, dilute phase and grid cell at the quasi-steady state. A similar profile can be found that the O2 molar fraction at the bottom is high and decreases along the height. The difference between the dense phase and the dilute phase leads to a discrepancy of the profile of the O2 molar fraction. For the dense phase, more oxygen is consumed at the bottom of the reactor. As the height increases, there is a sharp
Dry gas composition (%)
systems. The scalar variables are set at the center of the cells while the vector variables are placed at the boundaries of the cells. The maximum residual for the species and energy balances is selected to be 104 and the maximum residual for other parameters is set to be 103. To run progresses at a high execution speed, the automatic time-step between 103 and 106 s is adopted. The simulation is performed over 30 s and the time-averaged variables are calculated from 10 to 30 s after reaching the quasi-steady state.
Relative error (%)
40
TFR= 1173K 30
Present simulation Without multiscale simulation
air/fuel ratio=1.1
20 10 1.0
Experimental data Present simulation Without multiscale simulation
0.9 0.8
air/fuel ratio=1.1
TFR= 1173K 0.1 0.0
CH4
CO
H2
CO2
(b) Fig. 2. Comparisons of simulated results and experimental data.
reduction of the O2 molar fraction. In contrast, the change of the O2 molar fraction in the dilute phase is not significant. A high solid concentration near the wall enhances the regeneration reaction of oxygen carriers and results in a low O2 molar fraction. The degree of the regeneration reaction influences the fuel conversion. Hence, it is necessary to increase the residence time of oxygen carriers. Fig. 4 shows instantaneous profiles of the CH4 molar fractions in the dense phase, dilute phase and grid cell at the quasi-steady state. The CH4 enters the bottom of FR and is continuously consumed with the height increased. The CH4 consumption rate in the dense phase is faster than that in the dilute phase, which reduces the CH4 molar fraction. However, there is still a higher CH4 concentration in the dilute phase. The non-uniformity caused by clusters brings out the heterogeneity of the CH4 concentration. Near the wall, the CH4 molar fraction approaches zero. It can be also found that the fuel partly flows into the loop seal, which weakens the fuel conversion. It is a key to reasonably control the operating parameters and arrange the position of loop seal to avoid the fuel leakage. Fig. 5 demonstrates instantaneous profiles of the CO2 molar fractions in the dense phase, dilute phase and grid cell at the quasi-steady state. The product CO2 is generated according to
724
S. Wang et al. / Applied Energy 155 (2015) 719–727
0.2 0.18 0.16 0.14 0.12 0.1 0.08 0.06 0.04 0.02
(a)
(b)
(c)
Fig. 3. Instantaneous profiles of O2 molar fraction at 10.0 s (a – dense phase; b – dilute phase; c – the grid).
0.2 0.19 0.18 0.17 0.16 0.15 0.14 0.13 0.12 0.11 0.1 0.09 0.08 0.07 0.06 0.05 0.04 0.03 0.02 0.01
(a)
(b)
(c)
Fig. 4. Instantaneous profiles of CH4 molar fraction at 10.0 s (a – dense phase; b – dilute phase; c – the grid).
(R-3). It can be observed that the CO2 molar fraction is improved along the axial direction. A higher reaction rate in the dense phase promotes the fuel conversion and produces more CO2 compared to that in the dilute phase.
Instantaneous profiles of gas temperatures at the quasi-steady state are presented in Fig. 6. In the AR, the regeneration reaction of oxygen carriers is exothermic reaction. Different reaction degrees between the dense phase and the dilute phase make a
725
S. Wang et al. / Applied Energy 155 (2015) 719–727
0.34 0.32 0.3 0.28 0.26 0.24 0.22 0.2 0.18 0.16 0.14 0.12 0.1 0.08 0.06 0.04 0.02
(a)
(b)
(c)
Fig. 5. Instantaneous profiles of CO2 molar fraction at 10.0 s (a – dense phase; b – dilute phase; c – the grid).
1232 1230 1228 1226 1224 1222 1220 1218 1216 1214 1212 1210 1208 1206 1204 1182 1181 1180 1179 1178 1177
(a)
(b)
(c)
Fig. 6. Instantaneous profiles of gas temperature at 10.0 s (a – dense phase; b – dilute phase; c – the grid).
726
S. Wang et al. / Applied Energy 155 (2015) 719–727
discrepancy of gas temperature. The temperature in the dense phase is slight higher than that in the dilute phase. Accordingly, in the FR, the temperature in the dense phase decreases more evidently than that in the dilute phase. As a result of inter-phase heat transfer between the dense phase and the dilute phase, the discrepancy between them is not obvious. Fig. 7 demonstrates the time-averaged distribution of gas compositions in the dense phase and the dilute phase along the axial and lateral directions. The similar profiles can be observed for the dense phase and the dilute phase. There is a significant discrepancy in gas species concentrations at the bottom of FR between the dense phase and the dilute phase. As the height increases, the difference becomes weak. The reason for this is that the solid volume fraction at the bottom of reactor is higher which promotes the degree of reactions in the dense phase more intensive and results in the change of gas species concentration more evident. For the CO2 and H2O as products, the gas species concentrations in the dense phase are higher that those in the dilute phase, which is reverse for the reactants. From the profile at the upper section of reactor, we can see that nearly full fuel conversion is achieved. For the lateral profile of gas compositions, the non-uniform distribution is recognized as shown in Fig. 7(b). The annulus-core structure in the reactor leads to a fast consumption of the reactants near the wall and the CH4 and CO concentrations approach zero. Accordingly, the CO2 concentration is enhanced. The local decrease of CO2 concentration at the left wall is attributed to the dilution of
gas from the AR, which is used to transport the regenerated oxygen carriers. The gas species concentration difference in the center between the dense phase and the dilute phase is more obvious than that near the wall. 5. Conclusion A multi-scale chemical reaction model coupling heat transfer is applied to investigate the CLC process in a dual fluidized bed system to consider the effect of clusters. The results indicate that the predictions by the multi-scale model are in better agreements with experimental data than those by the conventional model. The distribution of gas components and temperatures for the dense phase and the dilute phase is obtained. The change of gas species concentration in the dense phase is more significant along the axial direction compared to that in the dilute phase. However, the discrepancy of the temperatures in the dense phase and the dilute phase is not obvious. In the future work, the radiation model is required to be further considered and incorporated to the multi-scale chemical reaction model. Meanwhile, a three-dimensional simulation is expected to evaluate the impact of cyclone. Acknowledgments This research is conducted with financial support from the National Natural Science Foundation of China (51390494 and 51406045) and the Natural Science Foundation of Heilongjiang Province of China (Grant No. E201441). References
Fig. 7. Time-averaged profiles of gas compositions (a – axial direaction; b – lateral direaction).
[1] Lyngfelt A, Leckner B, Mattisson T. A fluidized-bed combustion process with inherent CO2 separation; application of chemical-looping combustion. Chem Eng Sci 2001;56:3101–13. [2] Ishida M, Jin H. A new advanced power-generation system using chemicallooping combustion. Energy 1994;19:415–22. [3] Fernández JR, Abanades JC. Conceptual design of a Ni-based chemical looping combustion process using fixed-beds. Appl Energy 2014;135:309–19. [4] Strohle J, Orth M, Epple B. Design and operation of a 1 MWth chemical looping plant. Appl Energy 2014;113:1490–5. [5] Lyngfelt A. Chemical-looping combustion of solid fuels-status of development. Appl Energy 2014;113:1869–73. [6] Cuadrat A, Abad A, García-Labiano F, Gayán P, de Diego LF, Adánez J. The use of ilmenite as oxygen-carrier in a 500 wth chemical-looping coal combustion unit. Int J Greenhouse Gas Contr 2011;5:1630–42. [7] García-Labiano F, de Diego LF, Gayán P, Abad A, Cabello A, Adánez J, et al. Energy exploitation of acid gas with high H2S content by means of a chemical looping combustion system. Appl Energy 2014;136:242–9. [8] Hakonsen SF, Blom R. Chemical looping combustion in a rotating bed reactor – finding optimal process conditions for prototype reactor. Environ Sci Technol 2011;45:9619–26. [9] Miller DD, Siriwardane R, Poston J. Fluidized-bed and fixed-bed reactor testing of methane chemical looping combustion with MgO-promoted hematite. Appl Energy 2015;146:111–21. [10] Zhang S, Xiao R, Zheng WG. Comparative study between fluidized-bed and fixed-bed operation modes in pressurized chemical looping combustion of coal. Appl Energy 2014;130:181–9. [11] Wang S, Liu GD, Lu HL, Chen JH, He YR, Wang JX. Fluid dynamic simulation in a chemical looping combustion with two interconnected fluidized beds. Fuel Process Technol 2011;92:385–93. [12] Wang XJ, Jin BS, Zhong WQ, Zhang Y, Song M. Three-dimensional simulation of a coal gas fueled chemical looping combustion process. Int J Greenhouse Gas Contr 2011;5:1498–506. [13] Deng ZY, Xiao R, Jin BS, Song QL. Numerical simulation of chemical looping combustion process with CaSO4 oxygen carrier. Int J Greenhouse Gas Contr 2009;3:368–75. [14] Peng ZB, Doroodchi E, Alghamdi YA, Shah K, Luo C, Moghtaderi B. CFD-DEM simulation of solid circulation rate in the cold flow model of chemical looping systems. Chem Eng Res Des 2015;95:262–80. [15] Guan YJ, Chang J, Zhang K, Wang BD, Sun Q. Three-dimensional CFD simulation of hydrodynamics in an interconnected fluidized bed for chemical looping combustion. Powder Technol 2014;268:316–28. [16] Pröll T, Kolbitsch P, Bolha‘r-Nordenkampf J, Hofbauer H. A novel dual circulating fluidized bed system for chemical looping processes. AIChE J 2009;55:3255–66.
S. Wang et al. / Applied Energy 155 (2015) 719–727 [17] Penthor S, Zerobin F, Mayer K, Proll T, Hofbauer H. Investigation of the performance of a copper based oxygen carrier for chemical looping combustion in a 120 kW pilot plant for gaseous fuels. Appl Energy 2015;145:52–9. [18] Guío-Péreza DC, Pröll T, Hofbauer H. Influence of ring-type internals on the solids residence time distribution in the fuel reactor of a dual circulating fluidized bed system for chemical looping combustion. Chem Eng Res Des 2014;92:1107–18. [19] Wang S, Yang YC, Lu HL, Xu PF, Sun LY. Computational fluid dynamic simulation based cluster structures-dependent drag coefficient model in dual circulating fluidized beds of chemical looping combustion. Ind Eng Chem Res 2012;51:1396–412. [20] Wang S, Lu HL, Zhao FX, Liu GD. CFD studies of dual circulating fluidized bed reactors for chemical looping combustion processes. Chem Eng J 2014;236:121–30. [21] Wang SY, Yin LJ, Lu HL, Yu Long, Jacques B, Hao ZH. Numerical analysis of interphase heat and mass transfer of cluster in a circulating fluidized bed. Powder Technol 2009;189:87–96. [22] Hou BL, Li HZ, Zhu QS. Relationship between flow structure and transfer coefficients in fast fluidized beds. Chem Eng J 2010;157:509–19. [23] Hou BL, Li HZ, Zhu QS. Relationship between flow structure and mass transfer in fast fluidized bed. Chem Eng J 2010;163:108–18. [24] Hou BL, Tang HL, Zhang HY, Shao GQ, Li HZ, Zhu QS. Experimental and theoretical investigation of mass transfer in a circulating fluidized bed. Chem Eng Sci 2013;102:354–64.
727
[25] Wang S, Zhang TY, Liu GD, Lu HL, Sun LY. Multi-scale heat transfer in fluidized bed reactors by Eulerian CFD modeling. Fuel 2015;139:646–51. [26] Gidaspow D. Multiphase Flow and Fluidization: Continuum and Kinetic Theory Description. Boston: Academic Press; 1994. [27] Wang S, Zhao GB, Liu GD, Lu HL, Zhao FX, Zhang TY. Hydrodynamics of gassolid risers using cluster structure-dependent drag model. Powder Technol 2014;254:214–27. [28] La Nauze RD, Jung K, Kastl J. Mass transfer to large particles in fluidised beds of smaller particles. Chem Eng Sci 1984;39:1623–33. [29] Cloete S, Johansen ST, Amini S. An assessment of the ability of computational fluid dynamic models to predict reactive gas-solid flows in a fluidized bed. Powder Technol 2012;215–216:15–25. [30] Kolbitsch P, Bolhàr-Nordenkampf J, Proll T, Hofbauer H. Comparison of two Nibased oxygen carriers for chemical looping combustion of natural gas in 140 kW continuous looping operation. Ind Eng Chem Res 2009;48:5542–7. [31] Abad A, Adánez J, García-Labiano F, de Diego LF. Mapping of the range of operational conditions for Cu-, Fe-, and Ni-based oxygen carriers in chemical looping combustion. Chem Eng Sci 2007;62:533–49. [32] Johnson PC, Jackson R. Frictional collisional constitutive relations for granular materials, with application to plane shearing. J Fluid Mech 1987;176:67–93. [33] Syamlal M, Rogers W, O’Brien TJ. MFIX documentation theory guide; Morgantown Energy Technology Center, Office of Fossil Energy, U.S. Department of Energy: Morgantown, WV, Technical Note, DOE/METC-94/ 1004, NTIS/DE94000087, 1993.
Contents lists available at ScienceDirect
Applied Energy journal homepage: www.elsevier.com/locate/apenergy
Multi-scale simulation of chemical looping combustion in dual circulating fluidized bed Wang Shuai a,⇑, Chen Juhui b,⇑, Lu Huilin a, Liu Guodong a, Sun Liyan a a b
School of Energy Science and Engineering, Harbin Institute of Technology, Harbin 150001, China School of Mechanical Engineering, Harbin University of Science and Technology, Harbin 150080, China
h i g h l i g h t s A modified multi-scale gas–solid flow-reaction coupled model is developed. Multi-scale characteristic of chemical looping combustion system is investigated. Predicted results show a good agreement with experimental data.
a r t i c l e
i n f o
Article history: Received 23 March 2015 Received in revised form 17 May 2015 Accepted 31 May 2015
Keywords: Computational fluid dynamics Chemical looping combustion Cluster Multi-scale Dual circulating fluidized bed
a b s t r a c t Chemical looping combustion (CLC) in an interconnected fluidized bed has attracted more and more attention owing to its novel technology with inherent separation of CO2. In recent years, some models have been developed to investigate the gas-particle flow and reactive characteristics during the CLC process. However, multi-scale structures in reactors make it complex to perform a simulation. In the current work, a multi-scale gas–solid flow-reaction coupled model is developed and applied to the simulation of the CLC process in a dual circulating fluidized bed (DCFB) system with consideration of the impact of multi-scale structures on chemical reactions, mass and heat transfer. By comparisons of gas pressure and gas components with experimental data, the present model shows a better prediction. The influence of clusters on the gas compositions and temperature field is analyzed. Ó 2015 Elsevier Ltd. All rights reserved.
1. Introduction With the global warming, carbon dioxide (CO2) emissions have attracted more and more attention. Chemical looping combustion is a promising CO2 capture technology, which isolates the fuel source from the oxygen source by oxygen carriers and achieves inherent separation of CO2 [1,2]. In the last two decades, diligent research efforts have been devoted to develop the CLC system [3–5]. Cuadrat et al. [6] investigated a coal fuelled CLC reactor using ilmenite as oxygen carriers. It was found that the optimization of the CLC with coal can result in energy production with high CO2 capture. García-Labiano et al. [7] exploited an energy potential of acid through the combustion of H2S using the CLC system. It was concluded that the H2S concentration of 20 vol.% was high enough to turn CLC into an auto-thermal process. Haonsen et al. [8] studied the CLC performance in a rotating bed reactor and discussed the effects of process parameters by means of experiments. Miller ⇑ Corresponding authors. E-mail addresses: [email protected] (S. Wang), [email protected] (J. Chen). http://dx.doi.org/10.1016/j.apenergy.2015.05.109 0306-2619/Ó 2015 Elsevier Ltd. All rights reserved.
et al. [9] emphasized that the Mg-promoted hematite enhanced oxygen transfer capacity of hematite and showed better performance for methane CLC. Zhang et al. [10] performed a comparison between fluidized bed and fixed bed for a pressurized coal-fuelled CLC process. It was demonstrated that the fluidized bed mode had the superiority for a long-term stable operation. Computational fluid dynamics (CFD) provides an effective insight into the hydrodynamics of the complicated multiphase flow and has been applied to the simulation of the CLC process [11–13]. Peng et al. [14] employed a CFD–DEM model to evaluate some factors that influenced the solid circulation rate (SCR) in chemical looping systems. The results indicated that the flow regime in the air reactor was the main mechanism to determine the fluctuation frequency and amplitude of SCR. Guan et al. [15] incorporated the kinetic theory of granular flow into a three-dimensional CFD model to predict the hydrodynamics characteristic in the interconnected fluidized bed for the CLC process. It was found that the drag model had a significant impact on the prediction of the gas–solid flow and the solids flow patterns in the fuel reactor (FR) depended strongly on the operating parameters in the air reactor (AR).
720
S. Wang et al. / Applied Energy 155 (2015) 719–727
Nomenclature a dc ds cp f F g h hdc Kdc M n Ndf
acceleration (m s2) cluster diameter (m) particle diameter (m) specific heat capacity (J kg1 K1) volume fraction of dense phase force acting on each particle or cluster (N) gravity (m s2) local heat transfer coefficient (J m2 s1 K1) inter-phase heat transfer coefficient (J m2 s1 K1) inter-phase mass exchange coefficient (m s1) Molar weight (g mol1) reaction order energy dissipation (W kg1)
r S T u U Xi
reaction rate (kmol m3 s1) mass source (kg m3 s1) temperature (K) velocity (m s1) superficial velocity (m s1) mass fraction of gas species
The dual circulating fluidized bed (DCFB) system including two interconnected circulating fluidized beds has the potential for the CLC process owing to its low particle attrition rate and excellent gas–solids contact, as shown in Fig. 1. Meanwhile, the staged fluidization of the air reactor controls the global circulation rate of the system [16]. Penthor et al. [17] presented the CLC performance in the DCFB system using a copper based oxygen carrier and revealed that the solids inventory and circulation were critical parameters for the fuel conversion. The impact of internal installation on the particles residence time was evaluated [18]. The results revealed that the effect of internal loop was enhanced when
Fig. 1. Sketch of a DCFB reactor system.
Xox
reduction or oxidization degree of oxygen carriers
Greek letters a specific surface area (m1) b drag coefficient (kg m3 s1) e volume fraction k thermal conductivity (W m1 K1) l viscosity (Pa s) q density (kg m3) Subscripts c cluster den dense phase dil dilute phase int interface g gas phase s particle phase
fluidization velocities were increased in reactors. Wang et al. [19,20] investigated the flow and reactive characteristic for the CLC process in the DCFB system by means of CFD modeling, where a cluster structure-dependent (CSD) drag model was used to consider the cluster impact. From the simulated results, we can find that there was an improvement of the predictions in some extent. However, the cluster effects on reactions and inter-phase heat transfer were neglected. Lu et al. [21] pointed out that an isolated particle had significantly higher mass and heat transfer rates than the particle in the cluster by an investigation of the gas-cluster mass and heat transfer. An increase of the cluster porosity can promote the gas-cluster mass and heat transfer rates. Hence, it is necessary to take into account the cluster impact on mass and heat transfer during the CLC process. In recent studies, Hou et al. [22,23] established the relationship between mass transfer and flow structures for fast fluidized beds and developed a multi-scale mass transfer (MSMT) model based on energy-minimization multi-scale (EMMS) method. In this model, the cluster impact on mass transfer was considered. The simulated results of the ozone decomposition process showed a better agreement with experimental data. The MSMT model was further validated for the catalytic oxidation of carbon monoxide in a circulating fluidized bed [24]. The results also demonstrated that the clusters played an important role in the gas–solid mass transfer. To describe the effect of clusters on chemical reaction, mass and heat transfer simultaneously, a multi-scale chemical reaction model coupling heat transfer was developed in our previous work [25], where the temperature discrepancy between cluster phase and dispersed phase was incorporated and the cluster structure-dependent (CSD) drag model was adopted to provide local structural parameters. By simulating the regeneration reaction of oxygen carriers in riser reactors, the multi-scale chemical reaction model promoted the accuracy of predicted results. This work focuses on the investigation of the multi-scale flow and reactive characteristics for the whole CLC process by means of CFD modeling. For the simulation of the DCFB system consisting of two circulating fluidized beds, employing a multi-scale model coupled with chemical reaction and heat transfer is more rational to reflect the influence of the temporal-spatial multi-scale structures on the chemical looping system, which is not available in recent literatures. The advantage of this approach lies in the more
721
S. Wang et al. / Applied Energy 155 (2015) 719–727
in-depth sight into the multi-scale structure performance during the reaction process. The formation of amorphous aggregates of oxygen carriers can be observed. Meanwhile, the simulation of the complete loop system including loop-seals and separators is carried out. The characteristics of multi-scale mass and heat transfer in reactors are analyzed. 2. Mathematical model In the current research, the two-fluid model is used as the basic framework of the multi-scale simulation with the kinetic theory of granular flow [26]. It is assumed that the solid phase is spherical with a uniform density and size. The governing equations mainly include continuity equations, momentum conservation equations, species transport equations and energy conservation equations. Detailed descriptions of the model can be found in Wang et al. [20]. To characterize the impact of cluster structures on the flow and reaction behaviors during the CLC process, the local flow in the control volume is separated into three sub-systems: the dense phase in the form of clusters, the dilute phase in the form of dispersed particles and the interface between the dense phase and
hgs ¼
t1 ¼
hdc
kg eden ¼ 2:0 þ dc
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 4qg cpg kg eden pt 1
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi Dg;i eden 4Dg;i eden ¼ 2:0 þ dc pt 1 dc jðU g;den =eden Þ ðU s;den =ð1 eden ÞÞj
ð1Þ
ð2Þ
ð3Þ
With local temperatures and heat transfer coefficients in the dilute phase and the dense phase, we can calculate the effective inter-phase heat transfer coefficient through the following equation:
ap ð1 f Þð1 edil Þhdil ðT g;dil T s;dil Þ þ ac f ð1 eden Þhint ðT g;dil T s;den Þ þ ðap ac Þf ð1 eden Þhden ðT g;den T s;den Þ ap ð1 eg ÞðT g T s Þ
the dilute phase. In each sub-system, gas–solid interaction, mass source and temperature owing to chemical reactions and heat transfer are solved respectively. Then these local parameters are integrated into the equations in the control volume to reflect the multi-scale effects. Here, the CSD drag model is employed to solve the multi-scale gas–solid interaction [27]. The expression of the inter-phase drag coefficient is given by Eq. (T1-1) in Table 1. In order to solve this multi-scale drag coefficient, the eight local structural parameters (Ug,dil, Ug,den, Us,dil, Us,den, edil, eden, f, dc) are required to be obtained by solving six equations (T1-2)–(T1-7) and a stability criterion of the minimum energy dissipation consumed by drag force (T1-8). To consider the cluster effects on reactions and heat transfer, the multi-scale chemical reaction model coupled with heat transfer was proposed and applied to simulate the riser reactors [25]. Here, the model is further modified with consideration of the discrepancy in oxygen carrier conversions between different phases. Detailed descriptions in the model can found in Table 2. The gas and solid species transport equations for different phases are expressed by Eqs. (T2-1), (T2-2), (T2-5) and (T2-6), respectively. The corresponding mass source terms are given by Eqs. (T2-3), (T2-4), (T2-7) and (T2-8). By integrating the obtained gas and solid species mass source terms in the dense phase and the dilute phase, we can derive the effective species mass source terms, which are incorporated into the two-fluid model to achieve the implement of the multi-scale chemical reaction model. In the model, the gas mass exchange between the dense phase and the dilute phase is considered according to La Nauze et al. [28]:
K dc;i
For the reaction rates at different sub-systems, the shrinking core methodology is selected and the related expressions are shown in Eqs. (T2-9)–(T2-11), where the reduction or oxidization of available surface area with reaction in progress is considered [29]. Following the previous work [25], the local gas and solid temperatures in the dense phase and the dilute phase are obtained by solving their respective energy balance equations, which are described by Eqs. (T2-12)–(T2-15), where the heat loss by radiation is assumed to be negligible. It can be found that except for the gas– solid heat transfer between phases, the gas heat transfer between the dilute phase and the dense phase is also taken into account, which is expressed as follows [22]:
ð4Þ
By means of the solution of the above multi-scale drag coefficient, the effective species mass source terms and the effective gas–solid heat transfer coefficient, the cluster-structure-dependent multi-scale model is established. 3. Model implementation description In this work, the DCFB system is selected as the objective of the simulation, which comprises fuel reactor and air reactor, as described in Kolbitsch et al. [16,30]. The lower loop seal is used to connect the two reactors and transport the reduced oxygen carriers. The internal loop seal and cyclone achieve internal recirculation of particles of the FR. The main system parameters are listed in Table 3 [16]. With respect to chemical reactions during the CLC process, the nickel oxide is chosen as oxygen carriers and the methane is used as the gas fuel. In the FR, the following three reactions are considered:
CH4 ðgÞ þ NiOðsÞ ! NiðsÞ þ 2H2 ðgÞ þ COðgÞ
ðR-1Þ
H2 ðgÞ þ NiOðsÞ ! NiðsÞ þ H2 OðgÞ
ðR-2Þ
COðgÞ þ NiOðsÞ ! NiðsÞ þ CO2 ðgÞ
ðR-3Þ
The carbon monoxide and hydrogen are generated as intermediates by Reaction (R-1) and simultaneously consumed by Reactions (R-2) and (R-3). In the AR, the regeneration reaction of oxygen carriers occurs and is written as follows:
O2 ðgÞ þ 2NiðsÞ ! 2NiOðsÞ
ðR-4Þ
Here, the kinetic parameters are chosen according to the Abad et al. [31]. Initially, the particles are filled in the FR and AR with the solid concentration of 0.5 and the static bed heights of 0.45 m and 0.3 m. The inlet velocities are specified at the bottom of the reactors and the pressure-outlets locate at the top of separators. At the wall,
722
S. Wang et al. / Applied Energy 155 (2015) 719–727
Table 1 Cluster structure-dependent drag model. 1. Equation for cluster structure-dependent drag coefficient e
(T1-1)
g ½nden F den þ ndil F dil þ nint F int bCSD ¼ jug u sj
2. Balance equations eg ¼ f eden þ ð1 f Þedil ug ¼ e1g ½fU g;den þ ð1 f ÞU g;dil
(T1-2) (T1-3)
us ¼ e1s ½fU s;den þ ð1 f ÞU s;dil
(T1-4)
3. Equation for momentum equation of the dense phase (T1-5)
nden F den þ nint F int ¼ f ð1 eden Þðqs qg Þðg þ as;den Þ þ f ð1 eden Þ @p @z 4. Equation for momentum equation of the dilute phase
(T1-6)
ndil F dil ¼ ð1 f Þð1 edil Þðqs qg Þðg þ as;dil Þ þ ð1 f Þð1 edil Þ @p @z 5. Equation for pressure drop balance nden F den f eden
ndil F dil nint F int ¼ ð1f þ ð1f þ qg ðag;dil ag;den Þ Þedil Þedil
(T1-7)
6. Stability criterion by minimization of the energy dissipation by drag force N df ¼ ð1e1g Þq ½nden F den U g;den þ ndil F dil U g;dil þ nint F int U g;dil ð1 f Þ ! minimum
(T1-8)
s
Table 2 Multi-scale chemical reaction and heat transfer model. 1. Gas species transport equations for the dense phase and the dilute phase @ðf eden qg X g;den;i Þ þ r ðf den g X g;den;i ug;den f den g Dg;i r X g;den;i Þ ¼ Sg;den;i @t @ðð1f Þedil qg X g;dil;i Þ þ r ðð1 f Þ dil g X g;dil;i ug;dil ð1 f Þ dil g Dg;i r X g;dil;i Þ @t P Sg;den;i ¼ m j¼1 r den;j M g;i K dc;i g c f den ðX g;den;i X g;dil;i Þ Pm Sg;dil;i ¼ j¼1 ðr dil;j þ r int;j ÞMg;i þ K dc;i g c f den ðX g;den;i X g;dil;i Þ
e
e
q
e q qa e
(T2-1)
q
e q
(T2-2)
¼ Sg;dil;i
(T2-3) (T2-4)
qa e
2. Solid species transport equations for the dense phase and the dilute phase @ðf ð1eden Þqs X s;den;i Þ þ r ðf ð1 den Þ s X s;den;i us;den Þ r ðf ð1 den Þ s Ds;i r ðX s;den;i ÞÞ ¼ Ss;den;i @t @ðð1f Þð1edil Þqs X s;dil;i Þ þ r ðð1 f Þð1 dil Þ s X s;dil;i us;dil Þ r ðð1 f Þð1 dil Þ s Ds;i r ðX s;dil;i ÞÞ @t P Ss;den;i ¼ m j¼1 ðr den;j þ r int;j ÞM s;i P Ss;dil;i ¼ m j¼1 r dil;j M s;i
e
e
q
e
q
(T2-5)
q
e
q
¼ Ss;dil;i
3. Reaction rates for the dense phase, the dilute phase, and the interface f eden X g;den;j qg n rden;j ¼ d60 s0 f ð1 eden Þð1 dds Þð1 X ox;den Þ2=3 k0 exp RTEden M g;j cl ð1f Þedil X g;dil;j qg n rdil;j ¼ d60 s0 ð1 f Þð1 edil Þð1 X ox;dil Þ2=3 k0 exp RTEdil M g;j ð1f Þe X q n dil g;int;j g rint;j ¼ d60 s0 f ð1 eden Þ dds ð1 X ox;den Þ2=3 k0 exp RTEint M g;j
(T2-6) (T2-7) (T2-8)
(T2-9) (T2-10) (T2-11)
cl
4. Energy balance equations of gas phase for the dense phase and the dilute phase lgt @ @t ðf eden qg c pg T g;den Þ þ r ðf eden qg ug;den c pg T g;den Þ ¼ r ½f eden kg þ c pg Prt r T g;den þ ac f eden hdc ðT g;dil T g;den Þ P þðap ac Þf ð1 eden Þhden ðT s;den T g;den Þ þ Sg;den;i cpg;i T g;den h i lgt @ @t ðð1 f Þedil qg c pg T g;dil Þ þ r ðð1 f Þedil qg ug;dil c pg T g;dil Þ ¼ r ð1 f Þedil kg þ c pg Pr t r T g;dil þ ap ð1 f Þð1 edil Þhdil ðT s;dil T g;dil Þ P þac f ð1 eden Þhint ðT s;den T g;dil Þ þ ac f eden hdc ðT g;den T g;dil Þ þ Sg;dil;i cpg;i T g;dil 5. Energy balance equations of solid phase for the dense phase and the dilute phase h i lst @ @t ðf ð1 eden Þqs c ps T s;den Þ þ r ðf ð1 eden Þqs us;den c ps T s;den Þ ¼ r f ð1 eden Þ ks þ c ps Prt r T s;den þ ac f ð1 eden Þhint ðT g;dil T s;den Þ P þðap ac Þf ð1 eden Þhden ðT g;den T s;den Þ þ Ss;den;i cps;i T s;den h i P lst @ Ss;dil;i cps;i T s;dil @t ðð1 f Þð1 edil Þqs c ps T s;dil Þ þ r ðð1 f Þð1 edil Þqs us;dil c ps T s;dil Þ ¼ r ð1 f Þð1 edil Þ ks þ c ps Prt r T s;dil þ
(T2-12) (T2-13)
(T2-14) (T2-15)
þap ð1 f Þð1 edil Þhdil ðT g;dil T s;dil Þ 6. Local heat transfer coefficients for the dense phase, the dilute phase, and the interface kg Nus;den eden ds kg Nus;dil edil ds kg Nus;int edil ð1f Þ dc
hden ¼
(T2-16)
hdil ¼
(T2-17)
hint ¼
no-slip and partial-slip conditions are adopted for gas and solid phases, respectively [32]. A constant temperature condition is used for the wall. Detailed operating parameters can be found in Table 3. Regarding the computational domain, a two-dimensional simulation is carried out. Following the previous investigation of the mesh independence [19,20], the grid sizes of 5 10 mm and 10 10 mm are used for the reactors and other components respectively. The model is implemented on the basis of the
(T2-18)
M-FIX program, which is an open-source CFD code written in FORTRAN to describe the multiphase flow with inter-phase exchanges. A solids volume fraction correction equation is employed instead of a solids pressure correction equation, as reviewed by Syamlal [33]. The above governing equations are conducted using finite differences equations with a higher order Total Variation Diminishing (TVD) scheme, which incorporates a modification into the higher order upwind scheme for hyperbolic
723
S. Wang et al. / Applied Energy 155 (2015) 719–727 Table 3 System properties and parameters for the simulations [16]. Unit
AR
FR
Reactor height Reactor diameter Diameter of cyclone Diameter of downcomer Height of downcomer Particle diameter Particle density Gas diffusivity
m m m m
4.1 0.15 0.15 0.05
3.0 0.159 0.15 0.05
m
0.5 135 3416 O2(0.178), N2(0.2)
Static bed height Initial concentration of particles Initial temperature Inlet gas flow Inlet gas temperature
m –
0.3 0.5
1.5 135 3416 CH4(0.2064), H2(0.634), CO(0.1921), CO2(0.1381), H2O(0.2178) 0.45 0.5
K N m3/h K
1223 138.0 300
1173 12.0 300
lm kg/m3 cm2 s1
Height from inlet of AR and FR (m)
Description
4
AR
FR TFR,0=1173K
Experimental data Present simulation Without multiscale simulation
3
2
air/fuel ratio=1.1 FR exit
TAR,0=1223K
Experimental data Present simulation Without multiscale simulation
air/fuel ratio=1.1 1
0 0
3
6
0
5
10
15
Pg (kPa)
(a)
4. Results and discussion Fig. 2(a) shows the simulated axial profiles of gas pressure along the height of the AR and FR in comparison with experimental results [16]. It can be seen that the pressure is higher at the bottom of the FR, and decreases rapidly with the height increased. At the upper section of the FR, the change of the FR pressure is not obvious, which reflects the distribution of solid concentration in the FR. The AR pressure spreads over the whole AR height compared to the FR pressure. Owing to a lower operating velocity and a higher solid concentration in the FR compare to those in the AR, there is a higher pressure in the FR. The predictions with no consideration of the multi-scale effects are also shown in Fig. 2(a). We can find that the present multi-scale model can give a better prediction. The gap between the model prediction and measured data results from the selectivity of uniform mean particle diameter in the simulation. In the experiments, the particles have a non-uniform distribution [16,30]. Additionally, the fluidization nozzles are arranged at the bottom of the reactors. The discrepancy of the inlet conditions between simulation and experiment is also a reason for this gap. Overall, the prediction by the model is consistent with measured data. To further validate the reaction model, the simulated gas compositions are compared to measured data as shown in Fig. 2(b). Meanwhile, the relative error between simulated results and experimental data is also given. It is found that the predictions by the multi-scale model are closer to experimental data [30]. The results without multi-scale effects show a lower fuel conversion and deviate from experimental data. Fig. 3 displays instantaneous distributions of the O2 molar fractions in the dense phase, dilute phase and grid cell at the quasi-steady state. A similar profile can be found that the O2 molar fraction at the bottom is high and decreases along the height. The difference between the dense phase and the dilute phase leads to a discrepancy of the profile of the O2 molar fraction. For the dense phase, more oxygen is consumed at the bottom of the reactor. As the height increases, there is a sharp
Dry gas composition (%)
systems. The scalar variables are set at the center of the cells while the vector variables are placed at the boundaries of the cells. The maximum residual for the species and energy balances is selected to be 104 and the maximum residual for other parameters is set to be 103. To run progresses at a high execution speed, the automatic time-step between 103 and 106 s is adopted. The simulation is performed over 30 s and the time-averaged variables are calculated from 10 to 30 s after reaching the quasi-steady state.
Relative error (%)
40
TFR= 1173K 30
Present simulation Without multiscale simulation
air/fuel ratio=1.1
20 10 1.0
Experimental data Present simulation Without multiscale simulation
0.9 0.8
air/fuel ratio=1.1
TFR= 1173K 0.1 0.0
CH4
CO
H2
CO2
(b) Fig. 2. Comparisons of simulated results and experimental data.
reduction of the O2 molar fraction. In contrast, the change of the O2 molar fraction in the dilute phase is not significant. A high solid concentration near the wall enhances the regeneration reaction of oxygen carriers and results in a low O2 molar fraction. The degree of the regeneration reaction influences the fuel conversion. Hence, it is necessary to increase the residence time of oxygen carriers. Fig. 4 shows instantaneous profiles of the CH4 molar fractions in the dense phase, dilute phase and grid cell at the quasi-steady state. The CH4 enters the bottom of FR and is continuously consumed with the height increased. The CH4 consumption rate in the dense phase is faster than that in the dilute phase, which reduces the CH4 molar fraction. However, there is still a higher CH4 concentration in the dilute phase. The non-uniformity caused by clusters brings out the heterogeneity of the CH4 concentration. Near the wall, the CH4 molar fraction approaches zero. It can be also found that the fuel partly flows into the loop seal, which weakens the fuel conversion. It is a key to reasonably control the operating parameters and arrange the position of loop seal to avoid the fuel leakage. Fig. 5 demonstrates instantaneous profiles of the CO2 molar fractions in the dense phase, dilute phase and grid cell at the quasi-steady state. The product CO2 is generated according to
724
S. Wang et al. / Applied Energy 155 (2015) 719–727
0.2 0.18 0.16 0.14 0.12 0.1 0.08 0.06 0.04 0.02
(a)
(b)
(c)
Fig. 3. Instantaneous profiles of O2 molar fraction at 10.0 s (a – dense phase; b – dilute phase; c – the grid).
0.2 0.19 0.18 0.17 0.16 0.15 0.14 0.13 0.12 0.11 0.1 0.09 0.08 0.07 0.06 0.05 0.04 0.03 0.02 0.01
(a)
(b)
(c)
Fig. 4. Instantaneous profiles of CH4 molar fraction at 10.0 s (a – dense phase; b – dilute phase; c – the grid).
(R-3). It can be observed that the CO2 molar fraction is improved along the axial direction. A higher reaction rate in the dense phase promotes the fuel conversion and produces more CO2 compared to that in the dilute phase.
Instantaneous profiles of gas temperatures at the quasi-steady state are presented in Fig. 6. In the AR, the regeneration reaction of oxygen carriers is exothermic reaction. Different reaction degrees between the dense phase and the dilute phase make a
725
S. Wang et al. / Applied Energy 155 (2015) 719–727
0.34 0.32 0.3 0.28 0.26 0.24 0.22 0.2 0.18 0.16 0.14 0.12 0.1 0.08 0.06 0.04 0.02
(a)
(b)
(c)
Fig. 5. Instantaneous profiles of CO2 molar fraction at 10.0 s (a – dense phase; b – dilute phase; c – the grid).
1232 1230 1228 1226 1224 1222 1220 1218 1216 1214 1212 1210 1208 1206 1204 1182 1181 1180 1179 1178 1177
(a)
(b)
(c)
Fig. 6. Instantaneous profiles of gas temperature at 10.0 s (a – dense phase; b – dilute phase; c – the grid).
726
S. Wang et al. / Applied Energy 155 (2015) 719–727
discrepancy of gas temperature. The temperature in the dense phase is slight higher than that in the dilute phase. Accordingly, in the FR, the temperature in the dense phase decreases more evidently than that in the dilute phase. As a result of inter-phase heat transfer between the dense phase and the dilute phase, the discrepancy between them is not obvious. Fig. 7 demonstrates the time-averaged distribution of gas compositions in the dense phase and the dilute phase along the axial and lateral directions. The similar profiles can be observed for the dense phase and the dilute phase. There is a significant discrepancy in gas species concentrations at the bottom of FR between the dense phase and the dilute phase. As the height increases, the difference becomes weak. The reason for this is that the solid volume fraction at the bottom of reactor is higher which promotes the degree of reactions in the dense phase more intensive and results in the change of gas species concentration more evident. For the CO2 and H2O as products, the gas species concentrations in the dense phase are higher that those in the dilute phase, which is reverse for the reactants. From the profile at the upper section of reactor, we can see that nearly full fuel conversion is achieved. For the lateral profile of gas compositions, the non-uniform distribution is recognized as shown in Fig. 7(b). The annulus-core structure in the reactor leads to a fast consumption of the reactants near the wall and the CH4 and CO concentrations approach zero. Accordingly, the CO2 concentration is enhanced. The local decrease of CO2 concentration at the left wall is attributed to the dilution of
gas from the AR, which is used to transport the regenerated oxygen carriers. The gas species concentration difference in the center between the dense phase and the dilute phase is more obvious than that near the wall. 5. Conclusion A multi-scale chemical reaction model coupling heat transfer is applied to investigate the CLC process in a dual fluidized bed system to consider the effect of clusters. The results indicate that the predictions by the multi-scale model are in better agreements with experimental data than those by the conventional model. The distribution of gas components and temperatures for the dense phase and the dilute phase is obtained. The change of gas species concentration in the dense phase is more significant along the axial direction compared to that in the dilute phase. However, the discrepancy of the temperatures in the dense phase and the dilute phase is not obvious. In the future work, the radiation model is required to be further considered and incorporated to the multi-scale chemical reaction model. Meanwhile, a three-dimensional simulation is expected to evaluate the impact of cyclone. Acknowledgments This research is conducted with financial support from the National Natural Science Foundation of China (51390494 and 51406045) and the Natural Science Foundation of Heilongjiang Province of China (Grant No. E201441). References
Fig. 7. Time-averaged profiles of gas compositions (a – axial direaction; b – lateral direaction).
[1] Lyngfelt A, Leckner B, Mattisson T. A fluidized-bed combustion process with inherent CO2 separation; application of chemical-looping combustion. Chem Eng Sci 2001;56:3101–13. [2] Ishida M, Jin H. A new advanced power-generation system using chemicallooping combustion. Energy 1994;19:415–22. [3] Fernández JR, Abanades JC. Conceptual design of a Ni-based chemical looping combustion process using fixed-beds. Appl Energy 2014;135:309–19. [4] Strohle J, Orth M, Epple B. Design and operation of a 1 MWth chemical looping plant. Appl Energy 2014;113:1490–5. [5] Lyngfelt A. Chemical-looping combustion of solid fuels-status of development. Appl Energy 2014;113:1869–73. [6] Cuadrat A, Abad A, García-Labiano F, Gayán P, de Diego LF, Adánez J. The use of ilmenite as oxygen-carrier in a 500 wth chemical-looping coal combustion unit. Int J Greenhouse Gas Contr 2011;5:1630–42. [7] García-Labiano F, de Diego LF, Gayán P, Abad A, Cabello A, Adánez J, et al. Energy exploitation of acid gas with high H2S content by means of a chemical looping combustion system. Appl Energy 2014;136:242–9. [8] Hakonsen SF, Blom R. Chemical looping combustion in a rotating bed reactor – finding optimal process conditions for prototype reactor. Environ Sci Technol 2011;45:9619–26. [9] Miller DD, Siriwardane R, Poston J. Fluidized-bed and fixed-bed reactor testing of methane chemical looping combustion with MgO-promoted hematite. Appl Energy 2015;146:111–21. [10] Zhang S, Xiao R, Zheng WG. Comparative study between fluidized-bed and fixed-bed operation modes in pressurized chemical looping combustion of coal. Appl Energy 2014;130:181–9. [11] Wang S, Liu GD, Lu HL, Chen JH, He YR, Wang JX. Fluid dynamic simulation in a chemical looping combustion with two interconnected fluidized beds. Fuel Process Technol 2011;92:385–93. [12] Wang XJ, Jin BS, Zhong WQ, Zhang Y, Song M. Three-dimensional simulation of a coal gas fueled chemical looping combustion process. Int J Greenhouse Gas Contr 2011;5:1498–506. [13] Deng ZY, Xiao R, Jin BS, Song QL. Numerical simulation of chemical looping combustion process with CaSO4 oxygen carrier. Int J Greenhouse Gas Contr 2009;3:368–75. [14] Peng ZB, Doroodchi E, Alghamdi YA, Shah K, Luo C, Moghtaderi B. CFD-DEM simulation of solid circulation rate in the cold flow model of chemical looping systems. Chem Eng Res Des 2015;95:262–80. [15] Guan YJ, Chang J, Zhang K, Wang BD, Sun Q. Three-dimensional CFD simulation of hydrodynamics in an interconnected fluidized bed for chemical looping combustion. Powder Technol 2014;268:316–28. [16] Pröll T, Kolbitsch P, Bolha‘r-Nordenkampf J, Hofbauer H. A novel dual circulating fluidized bed system for chemical looping processes. AIChE J 2009;55:3255–66.
S. Wang et al. / Applied Energy 155 (2015) 719–727 [17] Penthor S, Zerobin F, Mayer K, Proll T, Hofbauer H. Investigation of the performance of a copper based oxygen carrier for chemical looping combustion in a 120 kW pilot plant for gaseous fuels. Appl Energy 2015;145:52–9. [18] Guío-Péreza DC, Pröll T, Hofbauer H. Influence of ring-type internals on the solids residence time distribution in the fuel reactor of a dual circulating fluidized bed system for chemical looping combustion. Chem Eng Res Des 2014;92:1107–18. [19] Wang S, Yang YC, Lu HL, Xu PF, Sun LY. Computational fluid dynamic simulation based cluster structures-dependent drag coefficient model in dual circulating fluidized beds of chemical looping combustion. Ind Eng Chem Res 2012;51:1396–412. [20] Wang S, Lu HL, Zhao FX, Liu GD. CFD studies of dual circulating fluidized bed reactors for chemical looping combustion processes. Chem Eng J 2014;236:121–30. [21] Wang SY, Yin LJ, Lu HL, Yu Long, Jacques B, Hao ZH. Numerical analysis of interphase heat and mass transfer of cluster in a circulating fluidized bed. Powder Technol 2009;189:87–96. [22] Hou BL, Li HZ, Zhu QS. Relationship between flow structure and transfer coefficients in fast fluidized beds. Chem Eng J 2010;157:509–19. [23] Hou BL, Li HZ, Zhu QS. Relationship between flow structure and mass transfer in fast fluidized bed. Chem Eng J 2010;163:108–18. [24] Hou BL, Tang HL, Zhang HY, Shao GQ, Li HZ, Zhu QS. Experimental and theoretical investigation of mass transfer in a circulating fluidized bed. Chem Eng Sci 2013;102:354–64.
727
[25] Wang S, Zhang TY, Liu GD, Lu HL, Sun LY. Multi-scale heat transfer in fluidized bed reactors by Eulerian CFD modeling. Fuel 2015;139:646–51. [26] Gidaspow D. Multiphase Flow and Fluidization: Continuum and Kinetic Theory Description. Boston: Academic Press; 1994. [27] Wang S, Zhao GB, Liu GD, Lu HL, Zhao FX, Zhang TY. Hydrodynamics of gassolid risers using cluster structure-dependent drag model. Powder Technol 2014;254:214–27. [28] La Nauze RD, Jung K, Kastl J. Mass transfer to large particles in fluidised beds of smaller particles. Chem Eng Sci 1984;39:1623–33. [29] Cloete S, Johansen ST, Amini S. An assessment of the ability of computational fluid dynamic models to predict reactive gas-solid flows in a fluidized bed. Powder Technol 2012;215–216:15–25. [30] Kolbitsch P, Bolhàr-Nordenkampf J, Proll T, Hofbauer H. Comparison of two Nibased oxygen carriers for chemical looping combustion of natural gas in 140 kW continuous looping operation. Ind Eng Chem Res 2009;48:5542–7. [31] Abad A, Adánez J, García-Labiano F, de Diego LF. Mapping of the range of operational conditions for Cu-, Fe-, and Ni-based oxygen carriers in chemical looping combustion. Chem Eng Sci 2007;62:533–49. [32] Johnson PC, Jackson R. Frictional collisional constitutive relations for granular materials, with application to plane shearing. J Fluid Mech 1987;176:67–93. [33] Syamlal M, Rogers W, O’Brien TJ. MFIX documentation theory guide; Morgantown Energy Technology Center, Office of Fossil Energy, U.S. Department of Energy: Morgantown, WV, Technical Note, DOE/METC-94/ 1004, NTIS/DE94000087, 1993.