Experimental and Theoretical Study of Minimum Fluidization Velocity and Void Fraction of a Limestone Based CO2 Sorbent Chameera K

View metadata, citation and similar papers at core.ac.uk brought to you by CORE

provided by Elsevier - Publisher Connector

Available online at www.sciencedirect.com ScienceDirect

Energy Procedia 63 ( 2014 ) 1432 – 1445

GHGT-12 Experimental and theoretical study of minimum fluidization velocity and void fraction of a limestone based CO2 sorbent Chameera K. Jayarathnaa,b*, Britt M. Halvorsena, Lars-Andre Tokheima,b

aDepartment of Process, Energy and Environmental Technology, Telemark University College, Norway bTel-Tek,Research institute, Porsgrunn, Norway

Abstract

Calcium looping is a promising technology for CO2 capture as it may reduce considerably the energy penalty represented by the capture system. The CO2 capture efficiency will strongly depend on the reactor configuration, solids residence time and thermal operation. As these parameters are also interdependent, the impact on the process can be better understood by process modelling and flow modelling. Use of model results can be very valuable in the design of the process. Properties such as minimum fluidization velocity and void fraction related to limestone particles are important parameters for process simulations as well as for simulations of flow behaviour in the reactor. Limestone (calcite) with a high content of CaCO3 can be a good sorbent material as it is readily available at low cost. Limestone from a local cement plant was used in this study. The material which had been pre-crushed in a mill at the plant was classified into four different size classes: 120-150 μm, 150-180 μm, 180-300 μm and 300-500 μm. A lab-scale fluidized bed made in plexi-glass, with a diameter of 146 mm and a uniform air distribution, was used to determine the properties of the sorbent. Four separate experiments were run for each group of particles. The minimum fluidization velocity and the related void fraction were then determined. Multiphase flow simulations were then carried out, using the computational fluid dynamics (CFD) software BARRACUDA®. The simulation results compared reasonably well with the experimental results.

©© 20142013 The The Authors. Authors. Published Published by Elsevierby Elsevier Ltd. ThisLtd. is an open access article under the CC BY-NC-ND license (Selectionhttp://creativecommons.org/licenses/by-nc-nd/3.0/ and peer-review under responsibility). of GHGT. Peer-review under responsibility of the Organizing Committee of GHGT-12

Keywords: fluidized bed; limestone; minimum fluidization velocity; void fraction; solid sorbent; CO2 capture;, EMMS drag mode; Barracuda

* Corresponding author. Tel.: +47-35574000 E-mail address: [email protected]

1876-6102 © 2014 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/3.0/). Peer-review under responsibility of the Organizing Committee of GHGT-12 doi: 10.1016/j.egypro.2014.11.153 Chameera K. Jayarathna et al. / Energy Procedia 63 ( 2014 ) 1432 – 1445 1433

Nomenclature

݀௣ ൌ Mean particle diameter ܸ௦௢௟௜ௗ ൌ Volume of only the materials in the particle ݑ௚ ൌ Gas velocity bed without pores in the particles ܸ ൌ Volume of the voids in the particles bed ݑ௣ ൌ Particle velocity ௩௢௜ௗ௦ ߩ ൌ Gas density ߛൌ Restitution coefficient ௚ Radial distribution function ߩ ൌ Particle density ݃௢ ൌ ௣ ȣൌ Granular temperature ߩҧ ൌ Average particle density ௣ ܥൌ Instantaneous minus hydrodynamic velocity ߠ௚ ൌ Gas volume fraction (Initial) of the particle averaged over the velocity ߠ௣ ൌ Particle volume fraction (Initial) space. ߠ௠௙ ൌ Void fraction (at minimum fluidization) ܨ௣ ൌ Particle drag force Particle volume fraction at close packing ߠ௖௣ ൌ ݉௣ ൌ Particle mass ݌ൌ Gas pressure ݉௕௘ௗ ൌ Mass of the particle bed Gas stress tensor ߬௚ ൌ ߤ௚ ൌ Gas viscosity ݃ൌ Gravitational acceleration ݎ௣ ൌ Particle radius ܨൌ Momentum exchange rate per volume ܪൌ Depth of the bed between gas and particles ܪ௠௙ ൌ Depth of the bed (at minimum fluidization) ൌ Bed diameter ܦ ݔൌ Particle position ݐൌ Particle position ௕௘ௗ ߶௦ ൌ Sphericity of particles ܸ௣ ൌ Particle volume Ȱൌ Porosity of the particles ܸ ൌ Volume of the particle bed ௕௘ௗ ߩ஺௉ ൌ Apparent density of the particles ܸ ൌVolume of the particle bed (at minimum ௕௘ௗ೘೑ ߩ௠ ൌ Measured particle density fluidization)

1. Introduction

Earth is in great danger due to global warming. The average temperature of the earth’s atmosphere and ocean is rising continuously. During the last 100 years, the average surface temperature has increased about 0.8OC[1]. Scientists are working on the topic are certain that the major cause for the global warming is greenhouse gases emitted due to human activities. Deforestation and burning of fossil fuels are the two main reasons. According to Robinsson et al. [2], the total industrial CO2 production, primarily from burning coal, oil and natural gas and the production of cement, is currently about 8Gt carbon per year. During the past 10 years, many political efforts have been made to force worldwide agreement to the Kyoto treaty[3]. Removal of CO2 from gas streams has been a crucial unit operation for many decades to avoid corrosion and also to improve the calorific value of gas streams. More recently CO2 reduction has become an urgent need due to the greenhouse effect. The most common technology for capturing CO2 is by using amine-based CO2 solvents to absorb CO2 from the exhaust gas. However, other concepts may be more attractive from an energy penalty point of view. Using a solid sorbent at high temperature is a concept which is now being widely considered as an alternative. Development of more advanced solid sorbents is a continuous process, but a challenge is high sorbent production cost. CO2 capture from flue gas by carbonate looping may be an attractive alternative due to the cheap and readily available sorbent (limestone). The calcium looping (CaL) process, first brought up by Shimizu et al.[4], is regarded as one of the potential technologies.

In the CaL process, calcium oxide (CaO) is used as a regenerable solid sorbent to react with CO2. Due to the formation of calcium carbonate in the process, it is also called carbonate looping [5]. In calcium looping, calcium oxide (CaO) reacts with CO2 to form calcium carbonate (CaCO3) in a fluidized bed reactor (carbonator) at a temperature around 650 °C.

The calcination reaction of CaCO3(s) [6],

o –1 CaO(s) +CO2 (g) ĺCaCO3(s) ǻH = –178 kJ mol 1434 Chameera K. Jayarathna et al. / Energy Procedia 63 ( 2014 ) 1432 – 1445

The CaCO3 is separated from the cleaned exhaust gas by a gas/solid separator. In a second reactor, the calciner, CaCO3 decomposes into CaO and CO2 at a temperature close to 900 °C. The cleaned flue gas exiting from the carbonator can be released to the atmosphere. The regenerated CaO is separated from the CO2 in a gas/solid separator and recycled back to the carbonator. The basic idea is illustrated in the Figure . The fact that the heat transfer takes place at a temperature higher than the typical operational temperature of a coal fired power plant means that the energy penalty usually associated with CO2 capture processes can be greatly reduced.

Cleaned CO2 exhaust

gas Heat Heat

Carbonator CaCO3 Calciner Combustor o 650 C 900oC CaO

Fuel (CmHn) +Air Heat

Exhaust gas Figure : Calcium looping cycle

Many of the CaL concepts described in the literature [7-15] are based on fluidized bed (FB) technology. FB reactors are widely used in industry because of good mixing and a large contact area between the phases. It enhances the chemical reactions, heat transfer and mass transfer. Liquid like behaviour of the bed particles gives smooth, nearly isothermal operation conditions, and the operation can be easily controlled.

In FB reactors, the CO2 capture efficiency will depend on the reactor configuration, solids residence time and thermal operation. As these parameters are also interdependent, the impact on the process can be better understood by process modelling and flow modelling. Use of model results can be very valuable in the design of the process. Properties such as minimum fluidization velocity and void fraction related to limestone particles are important parameters for the process simulations as well as for Computational Fluid Dynamics (CFD) simulations of flow behaviour in the reactor. Both bubbling and fast fluidization regimes are relevant.

Limestone (calcite) with a high content of CaCO3 can be a good choice of sorbent material as it is readily available at relatively low cost. Precrushed limestone from a local cement plant was used in this study. The particles were classified into four different size classes. The mean particles size ሺ݀௣ሻ of each sample can be calculated as explained by Kunii and Levenspiel [16], equation no.().

§·Apertureupper mesh Aperturelower mesh d p ¨¸ ©¹2

A lab-scale fluidized bed was used to determine the minimum fluidization velocity and the related void fraction. Pressure drop measurements along the particles bed were used to pinpoint the minimum fluidization velocity. The experimental results were then compared with the results from CFD simulations carried out with a CFD software BARRACUDA® version 16.0.5. Chameera K. Jayarathna et al. / Energy Procedia 63 ( 2014 ) 1432 – 1445 1435

Numerous research articles are available on experimental work related to minimum fluidization velocity for different types of particles, but not many articles are available on limestone particles. The predictions of minimum fluidization velocity by using CFD are not that common and not many research articles are available. Pata and Hartman [17] studied the minimum fluidization velocity of lime and limestone particles in 1978. Their focus was to use limestone for commercial desulfurization of flue gas. By using air, set of experiments were carried out in an 85mm column with particles of average size ranging from 100 to 810μm. The minimum fluidization velocities of a bed in the same size region were determined from a plot of pressure drops vs. velocity of air at room temperature. Their experimental results were compared with the computed values from available fluidization mathematical models such as Ergun[16] and Kunii and Levenspiel[16]. Sau et al. [18] studied the minimum fluidization velocities in tapered fluidized beds, and several experiments were carried out with regular as well as irregular particles including lime stone with different sizes (500, 600, 800μm) and void fractions(0.245-0.305). The sphericity and density of the limestone used were 0.85 and 2785kgm-3, respectively. Subramani et al. [19] studied the variation of minimum fluidization in limestone particles at elevated temperatures, and their study was limited to Geldart group-B powders.

2. Computational model

With fast growing computational power, CFD simulations have attracted the attention in gas solids fluidization research due to its capabilities of providing far more information of the hydrodynamics and the reactions than the experimental approach. Both the Two Fluid Method (TFM) [20, 21] and the Discrete Element Model (DEM) [22, 23] are widely used in gas-solid fluidised bed CFD simulations. TFM treats gas and solid phases as interpenetrating continua and pseudo-fluid rheological properties of the solid phase are calculated based on the Kinetic Theory of Granular Flow (KTGF)[21]. Liang et al.[24] found, based on their literature review, that simulating coarse particle are more acceptable than simulating Geldart A or C particles by using the TFM due to the effect of higher inter- particle forces between smaller particles. The capability to deal with the effects of distribution of particle size and type together with TFM is not possible in some fluidized beds[24]. Deen et al. [22] mentioned as another drawback in TFM that it could overestimate the solids mixing in fluidized beds due to the omission of solids friction in the KTGF model.

The DEM model represents the Eulerian–Lagrangian CFD approach, in which each particle is tracked individually, and all the collisions are calculated without the need to consider the solid rheology as in TFM [24]. DEM is usually seen as a reliable method to study the hydrodynamics of gas-solid fluidized beds[22], but DEM is usually not affordable for large fluidized beds with over 2×105 particles due to the huge demand of computer power[22]. However William and Snider come up with a new Eulerian–Lagrangian multiphase flow model named as Computational Particle Fluid Dynamics (CPFD) [25] also its’ commercially developed platform known as BARRACUDA®, DEM could also be applied to larger systems. Through the parcel (or computational particle) concept and the multiphase particle in cell (MP-PIC) method[25], simulating large industrial fluidized-bed reactors is possible. There are different drag models available and the EMMS [26-28] model used in this study.

Qi et al. [27] studied the characteristics of gas-solid fluidized beds together with CFD and combined the Eulerian approach with EMMS theory to develop a new theoretical model for the drag between the gas and solid phases in dense fluidized systems. Jayarathna and Halvorsen [29] also did some studies regarding the minimum fluidization velocity in a gas-solid fluidized bed. Their experiments were performed in a cylindrical bed with spherical glass particles as the bed material. Minimum fluidization velocity and the expansion of the bed heights were observed for two different powders (based on particles diameter) and their mixtures. The commercial CFD code Fluent version 6.3 is used for the corresponding simulations using the Eulerian-Eulerian model.

The role of the drag coefficient model was studied by Wang et al. [30] in simulating dense gas-solid flow. They presented a new drag force coefficient model based on the EMMS model by analysing the heterogeneous characteristics of the dense gas–solid flow. In addition, an experimental study was performed to investigate the gas- solid two-phase flow behaviour in a dense circulating fluidized bed (CFB) reactor. The overall and local flow characteristics were determined by using the axial pressure profiles and solid concentration profiles. A research group from University of Athens and Centre for Research and Technology in Greece also did a detail study [31] on a 1436 Chameera K. Jayarathna et al. / Energy Procedia 63 ( 2014 ) 1432 – 1445

3D full-loop CFD isothermal simulation of a transparent plexi-glass small-scale CFB carbonator built by IFK, university of Stuttgart, Germany for the hydrodynamic investigation of the CFB reactor, utilized for the investigation of the calcium looping process. The work coupled the state of the art TFM approach with the advanced EMMS scheme for the calculation of drag exerted on the solid phase by the gas phase. Simulation results agreed quite well with experimental data, regarding the re-circulation flux and the pressure profile along the full-loop.

Here in this work computational studies have been performed on a three dimensional fluidized bed. The model is based on the Eulerian–Lagrangian multiphase flow model.

3. Computational set-up

The simulations are performed for each classified particle samples (Table 1). The whole particle size range in each sample class is used for by specifying minimum and maximum particle diameters for each sample. The simulations are run with the same conditions as in the cold flow experiments.

A three-dimensional Cartesian co-ordinate system is used to describe the cylindrical bed with a diameter of 146mm and a height of 1.2m, (0.2m shorter than the actual bed height in order to reduce the computation time). The computational grid is shown in Figure 2. The mesh size is about 15mm with 8700 control volumes. The simulations have been run for 100 seconds for each gas flow rate. The total pressure is monitored at 2.5cm and 12.5cm (see Figure 3 ) above the distributor plate, i.e. same as in the cold flow experiments.

10cm

Figure : Cylindrical mesh used in the simulations Figure : Transient data points for monitoring the total pressure 3.1. Model description

The continuity and momentum equations[20] for the gas phase without reactions and interface mass transfer in CPFD [25] [32]are,

wTUgg .0 TUgggu wt and

w()TUgggu .()() TUgggguu p TW gg TU ggg F wt The rate of momentum transfer between fluid and solid phases per unit volume is described by, Chameera K. Jayarathna et al. / Energy Procedia 63 ( 2014 ) 1432 – 1445 1437

ªº1 F ³³³ fVDuup UUpgp«» pdVddu ppp ¬¼«»U p where ݂ሺݔǡ ݑ௣ǡߩ௣ǡܸ௣ǡݐሻ is the particle probability function. The time evolution of ݂ is obtained by solving a Liouville equation for the particle distribution function [25]. wf fup fA 0 wt The particle acceleration balanceܣ[25] is given as, 111 ADuu gp pg W UUTpppU The particle normal stress߬ [25, 33] is, ªº2 (1 ) g WTUTUJ ¬¼pp pp oĬ and the granular temperature [25] is given by, 1 Ĭ C 2 3

The radial distribution function ݃௢ [20, 25] is given by, 1 ªº1/3 3 §·T p g «»1¨¸ o 5 «»¨¸T ¬¼©¹cp The EMMS drag model is used in the calculations[28], and The fluid drag on particles is

FDuup m pgp and the drag function ܦ [28] is,

ug Df 4.5 2 e U ppr

where ݂௘is a function of Reynolds number, ܴ݁ǡ and gas volume fraction, ߠ௣ǡ and adjustable coefficients ܿ଴ through ܿଵସ and ݊଴. Default coefficient values used in simulations, together with the EMMS model based on Yang et al. [28] are given in

Table

1 §·T p ° ¨¸ccRe T 0.74 18 ¨¸01 g ° TTgg©¹ ° no fccReeg ® 23 ZT t 0.74 andRe 1000 ° Re °candRe ZT tt 0.74 1000 ° 4 24 g ¯ c c d6 0.74T d 0.82 ° 5 4(T cc )2 g ° g 78

° c10 ZT ® c9 2 0.82 dg 0.97 ° 4(Tg cc11 ) 12 ° cc TT 0.97d 1 ° 13 14 gg ¯ 1438 Chameera K. Jayarathna et al. / Energy Procedia 63 ( 2014 ) 1432 – 1445

Table : Default model constant for EMMS model used in BARRACUDA®

Model Constants

co= 150 c5= -0.576 c10= 0.0038 c1= 1.75 c6= 0.0214 c11= 0.7789 c2= 1 c7= 0.7463 c12= 0.0040 c3= 0.15 c8= 0.0044 c13= -31.8295 c4= 0.44 c9= -0.0101 c14= 32.8295

4. Experimental set-up and procedure

A lab-scale fluidized bed with a uniform air distribution is used for the experiments (Figure 4) at ambient conditions. The bed is cylindrical and is made of Lexan plastic. The diameter and the height of the bed are 0.146m and 1.4m respectively. The gas flow rate is controlled by a pressure reduction valve and further with a Sierra mass flow controller. The setup is designed to record the pressure measurements online from 8 positions along the bed. The distance between two adjacent measuring points is 10cm, and signals are fed to a computer through a LabVIEW® programme.

Limestone from a local cement plant is used in this study. The material was extracted from the raw material grinding process with relatively wide particle size distribution (about 100-10000 μm). The particles were then classified into four different size classes (120-150 μm, 150-180 μm, 180- 300 μm and 300-500 μm) by screening. The screening method reported by Wong is followed [34]. These size classes are within a size range which may be relevant in calcium looping when both reaction time and separation are taken into account.

The particle density is measured separately for each particle class by a Pycnometer. According the Pycnometer density analysis method [35], the true volume of the solid materials in the particle is measured based on the Figure cold flow fluidized bed experimental Archimedes’ principle of fluid displacement and gas expansion(Boyle’s setup law). Ideally, a gas is used as the displacing fluid since it penetrates the finest porous assuring maximum accuracy. For this reason helium is used, since its small atomic dimension enables entry into crevices and porous approaching 0.2nm. In other words, the density measured here is not exactly the apparent particle density; it is the density of the materials in the particles excluding the porous.

Measured particle densities by the gas Pycnometry method are shown in Table . Assumed all the pores of the particles are excluded from the measurements, which mean these values are the particle density at 0% porosity. Based on this assumption it is possible to illustrate the connection between the apparent particle density and the porosity of the particle as shown in equation () [By solving equation () to equation ()].

VVbed solid V pores V voids

mp U AP VVsolid pores Chameera K. Jayarathna et al. / Energy Procedia 63 ( 2014 ) 1432 – 1445 1439

V ĭ pores VVsolid pores

mp Um Vsolid

? UUAP m > 1)@

Four separate experiments were run for each group of particle. A pouring bed height is 157mm. Initial bed weight is measured prior to each experiment. Pressure drop measurements along the particles bed were used to determine the minimum fluidization velocity. The initial void fraction and the void fraction at minimum fluidization are calculated based on weight, measured density and bed volume.

Table : Experimental data

Bed design Height 1.4 m Diameter 0.146 m Particle size range (diameter) Particle range 120-150 μm 150-180 μm 180-300 μm 300-500 μm Mean particle size 135 μm 165 μm 240 μm 400 μm Measured particle density (kg/m3) 2801 2795 2814 2834 Fluidization air ( at ambient conditions)[34] Density (kg/m3) 1.1707 Viscosity (N.s/m2)*107 183.6

5. Results and Discussion

The minimum fluidization velocity was measured for the four different groups of particles. The results are presented in Figure and compared to experimental data from Pata and Hartman[17].Simulations were carried out with the same groups of particles and the particle properties used in the simulations are presented in Table . The apparent densities are calculated based on typical limestone porosities reported in the literature. According to a report on particle porosity [37], the porosity of limestone particles can vary from 7 to 56%, hence the apparent density of each particle sample can vary as shown in Figure . Even though there is a considerable variation of the apparent particle density of the reference particles within the above mentioned porosity range, those are still within the Geldart B particle group as indicated in Figure . By averaging the reported limestone porosity data [37], the simulations for all the four samples are run with 30% particle porosity. Minimum fluidization velocities from simulations and experiments are in quite good agreement with each other, and they also comply well with literature data reported by Pata and Hartman [17]. A slight difference between the experimental data from this work and the literature data could be due to the higher uncertainty of the measurements they had with the technology in 1978. They used glass inserts in the particle bed for the pressure measurement and those probes could disturb the hydrodynamics of the bed. A slight deviation of predicted minimum fluidization velocity by the simulations, compared to the experimental data, could be due to the error related to the selected porosity of the particles.

Table : Basic particle properties used in the simulations apart from the default settings

Particle size range (μm) 120-150 150-180 180-300 300-500 Calculated apparent 1961 1956 1970 1984 particle density (kg/m3) [at Ȱ ൌ ͵ͲΨ ] Initial void fraction 0.35 0.34 0.32 0.31 1440 Chameera K. Jayarathna et al. / Energy Procedia 63 ( 2014 ) 1432 – 1445

0.160

0.140 m/s

), 0.120 o (u 0.100

velocity 0.080 gas 0.060

0.040

Superficial 0.020

0.000 100 150 200 250 300 350 400 450 500

Meanparticlediameterforeachsample(dp),μm

PataandHartman(1978) Thiswork(Coldflowexperiments) Thiswork(Barracuda)

Figure : Minimum fluidization velocity versus mean particle diameter from the cold flow bed experiments, simulated values and literature data.

3000

3) 2800 Ͳ

2600 (kgm 2400

particles 2200 of 2000

1800 Density 1600

1400 Apparent

1200 0 102030405060 Porosityofparticles(%)

Sampleno.1(120Ͳ150)um Sampleno.2(150Ͳ180)um Sampleno.3(180Ͳ300)um Sampleno.4(300Ͳ500)um

Figure : Particle apparent density variation based on the porosity of particles Chameera K. Jayarathna et al. / Energy Procedia 63 ( 2014 ) 1432 – 1445 1441

Figure : Particles samples used in the experiments in Geldart’s classification of powders chart

At a relatively low gas flow, the regime is at fixed bed conditions, and the pressure drop is approximately proportional to the superficial gas velocity. With a gradual increase in velocity the maximum pressure dropሺο݌௠௔௫ሻ of the fixed bed stage is reached, a value which is slightly higher than the static pressure of the bed. With a further gas velocity increase, the fixed bed unlocks, and the voidage increases from ߠ௚ to ߠ௠௙ at the minimum fluidization velocityሺݑ௠௙ሻ. As a result, the pressure drop of the bed drops to its static pressure [16]. Based on this behaviour it is possible to pinpoint the value of ݑ௠௙ in a pressure drop ሺο݌ሻ versus superficial gas velocityሺݑ௢ሻ plot as explained by Kunii and Levenspiel [16]. Experimental and computational, Pressure drop versus superficial gas velocity for 120-150μm sized limestone particle are shown in Figure

The predicted pressure profiles are quite accurate for the fixed bed, and slightly higher than the experimental values for the fluidized or bubbling bed. However, with increased velocities, above minimum fluidization, the predictions of pressure drop improve. An explanation to this may be that the EMMS drag model gives more accurate results with higher gas velocities, when particle to particle interactions are less pronounced.

16.000

14.000

12.000

10.000 (mbar) 8.000

6.000 p/10cm (ݑ݂݉)fromexperiments ȴ 4.000 ݑ݂݉)fromBarracuda) 2.000

0.000 0.000 0.050 0.100 0.150 0.200 Superficialgasvelocity(m/s)

Experimentaldp,120Ͳ150um Barracudadp(120Ͳ150)um

Figure : Pressure drop versus superficial gas velocity for 120-150μm sized limestone particle 1442 Chameera K. Jayarathna et al. / Energy Procedia 63 ( 2014 ) 1432 – 1445

Particle Volume Fraction

0.625

0.562

0.500

0.438

0.375

0.313

0.251

0.188

0.126

0.063

0.001

Figure : Pictures from the BARRACUDA® simulations (120-150μm particles), changes of the particle bed with increased gas flow rate Chameera K. Jayarathna et al. / Energy Procedia 63 ( 2014 ) 1432 – 1445 1443

In Figure , snaps from the 3D simulations with increased gas flow rates are presented The change of the bed height at the minimum fluidization stage is moderate, as visualized in the CFD simulations result plots, and it is difficult to measure the bed height for calculating the bed voidage at minimum fluidization ሺߠ௠௙ሻstage. Leva [38] has reported that for most solids a close approximation can be obtained by substituting the value ߠ௠௙ for the value ߠ௚ obtained by pouring the solids carefully from one container in to another. Pata and Hartmen [17] followed that procedure to determine the ߠ௠௙ in their experiments with limestone since they found it very difficult to measure the height. But Leva’s method is highly depending on the person that runs the experiments and the vibration from the surrounding. In this study the bed height measurements are taken and bed voidage is calculated [equation ()].

݉௕௘ௗ Ͷ݉௕௘ௗ ߠ௠௙ ൌͳെቆ ቇൌͳെቆ ଶ ቇ ߩ஺௉ܸ௕௘ௗ೘೑ ߨܦ௕௘ௗߩ஺௉ܪ௠௙

Figure shows the increment of the bed voidage in-between the initial and the minimum fluidization conditions for each particle sample. Increments of the bed voidage in between these 2 stages are smaller in the small particle samples compared to bigger particles. It is about 2.4% increase of the void fraction in 120-150μm particle sample and 8.2% increase in 300-500μm particles. It could be because of the inter particles forces are much stronger in smaller particle bed compared the bed with the larger particle.

0.360 0.355 0.350 0.345 0.340 0.335 fraction 0.330

Void 0.325 0.320 0.315 0.310 0.305 0 100 200 300 400 500 Meanparticlediameter(μm)

Initialvoidfraction VoidfractionatMinimumfluidization

Figure : Difference between the void fraction at the initial and the minimum fluidization stages in the cold flow experiments (Mean particle diameter is used as the x-axis instead of the sample number)

1444 Chameera K. Jayarathna et al. / Energy Procedia 63 ( 2014 ) 1432 – 1445

6. Conclusion

The apparent density of each particle sample can vary considerable based on the porosity of particles. Minimum fluidization velocities from simulations and experiments are in quite good agreement with each other, and they also comply well with literature data. The predicted pressure profiles are quite accurate for the fixed bed, and slightly higher than the experimental values for the fluidized or bubbling bed. The EMMS drag model gives more accurate results with higher gas velocities, when particle to particle interactions are less pronounced. Increments of the bed voidage in between these two stages are smaller in the small particle samples compared to bigger particles. It could be because of the inter particles forces are much stronger in smaller particle bed compared the bed with the larger particle. Measuring the particle porosity for the specific particle sample is recommended as the future work.

Acknowledgements This work is financially supported by ALSTOM and GASSNOVA through FIRCC project. NORCEM (Brevik, Norway) Cement Company is highly acknowledged for providing the limestone particles. Franz Hafenbrädl from Tel-Tek has supported with the density measurements. Chandana Rathnayaka, Anette Mathisen and Eksath de Silva from Tel-Tek also acknowledge for their time spent on valuable discussions.

Reference 1. Choices, A.s.C., Environmental Effects of Increased Atmospheric Carbon Dioxide. 2011. The National Academies Press.(ISBN 978-0-309-14585-5): p. 15. 2. Robinson, A.B., N.E. Robinson, and W. Soon, Environmental Effects of Increased Atmospheric Carbon Dioxide. Journal of American Physicians and Surgeons, 2007. 12(3): p. 79-90. 3. Shimizu, T., et al., A Twin Fluid-Bed Reactor for Removal of CO2 from Combustion Processes. Chemical Engineering Research and Design, 1999. 77(1): p. 62-68. 4. Lasheras, A., et al., Carbonate looping process simulation using a 1D fluidized bed model for the carbonator. International Journal of Greenhouse Gas Control, 2011. 5(4): p. 686-693. 5. Chang, M.H., et al., Design and Experimental Investigation of Calcium Looping Process for 3-kWth and 1.9-MWth Facilities. Chemical Engineering & Technology, 2013. 36(9): p. 1525-1532. 6. Bennaceur, K.G.D.K.T.T.C. CO2 capture and storage a key carbon abatement option. 2008; Available from: http://www.sourceoecd.org/9789264041400. 7. Kremer, J., et al., Continuous CO2 Capture in a 1-MWth Carbonate Looping Pilot Plant. Chemical Engineering & Technology, 2013. 36(9): p. 1518-1524. 8. Ströhle, J., et al., Carbonate looping experiments in a 1 MWth pilot plant and model validation. Fuel, 2014. 127(0): p. 13-22. 9. Dieter, H., et al., Development of the calcium looping CO2 capture technology from lab to pilot scale at IFK, University of Stuttgart. Fuel, 2014. 127(0): p. 23-37. 10. Alstom, Fully Integrated Regenerative Carbonate Cycle (FIRCC), Pre-study description. 2012, Alstom. 11. Hatzilyberis, K.S., Design of an indirect heat rotary kiln gasifier. Fuel Processing Technology, 2011. 92(12): p. 2429-2454. 12. Junk, M., et al., Thermodynamic Evaluation and Cold Flow Model Testing of an Indirectly Heated Carbonate Looping Process. Chemical Engineering & Technology, 2013. 36(9): p. 1479-1487. 13. Junk, M., et al., Thermodynamic evaluation and cold flow model testing of an indirectly heated carbonate looping process, in 2nd International Conference on Chemical Looping, Darmstadt, Germany. 2012. 14. Hoeftberger, D. and J. Karl, Self-Fluidization in an Indirectly Heated Calciner. Chemical Engineering & Technology, 2013. 36(9): p. 1533-1538. 15. Kunii, D.L.O., Fluidization engineering. 1991, Boston,Mass.: Butterworths. 16. Jiradilok, V., et al., Kinetic theory based CFD simulation of turbulent fluidization of FCC particles in a riser. Chemical Engineering Science, 2006. 61(17): p. 5544-5559. 17. Yang, N., et al., Simulation of Heterogeneous Structure in a Circulating Fluidized-Bed Riser by Combining the Two-Fluid Model with the EMMS Approach. Industrial & Engineering Chemistry Research, 2004. 43(18): p. 5548-5561. Chameera K. Jayarathna et al. / Energy Procedia 63 ( 2014 ) 1432 – 1445 1445

18. Pata, J. and M. Hartman, Minimum Fluidization Velocities of Lime and Limestone Particles. Industrial & Engineering Chemistry Process Design and Development, 1978. 17(3): p. 231-236. 19. Sau, D.C., S. Mohanty, and K.C. Biswal, Minimum fluidization velocities and maximum bed pressure drops for gas–solid tapered fluidized beds. Chemical Engineering Journal, 2007. 132(1–3): p. 151-157. 20. Subramani, H.J., M.B. Mothivel Balaiyya, and L.R. Miranda, Minimum fluidization velocity at elevated temperatures for Geldart’s group-B powders. Experimental Thermal and Fluid Science, 2007. 32(1): p. 166-173. 21. Gidaspow, D., Multiphase Flow and Fluidization: Continuum and kinetic theory description. 1993, 24-28 Oval Road, London: Academic Press, Inc. 22. Gidaspow, D., J. Jung, and R.K. Singh, Hydrodynamics of fluidization using kinetic theory: an emerging paradigm: 2002 Flour-Daniel lecture. Powder Technology, 2004. 148(2–3): p. 123-141. 23. Deen, N.G., et al., Review of discrete particle modeling of fluidized beds. Chemical Engineering Science, 2007. 62(1–2): p. 28-44. 24. Zhu, H.P., et al., Discrete particle simulation of particulate systems: Theoretical developments. Chemical Engineering Science, 2007. 62(13): p. 3378-3396. 25. Liang, Y., et al., A critical validation study on CPFD model in simulating gas–solid bubbling fluidized beds. Powder Technology, 2014. 263(0): p. 121-134. 26. Snider, D.M., An Incompressible Three-Dimensional Multiphase Particle-in-Cell Model for Dense Particle Flows. Journal of Computational Physics, 2001. 170(2): p. 523-549. 27. Qi, H., et al., Modeling of drag with the Eulerian approach and EMMS theory for heterogeneous dense gas–solid two-phase flow. Chemical Engineering Science, 2007. 62(6): p. 1670-1681. 28. Jayarathna, C.K. and B. Halvorsen, Experimental and computational study of particle minimum fluidization velocity and bed expansion in a bubbling fluidized bed, in SIMS 50. 2009: Fredericia, Denmark. p. 5. 29. Wang, X.Y., et al., Experiment and CFD simulation of gas–solid flow in the riser of dense fluidized bed at high gas velocity. Powder Technology, 2010. 199(3): p. 203-212. 30. Nikolopoulos, A., et al., High-resolution 3-D full-loop simulation of a CFB carbonator cold model. Chemical Engineering Science, 2013. 90(0): p. 137-150. 31. Lun, C.K.K., et al., Kinetic theories for granular flow: inelastic particles in Couette flow and slightly inelastic particles in a general flowfield. Journal of Fluid Mechanics, 1984. 140: p. 223-256. 32. Wong, A.C.-Y., Use of angle of repose and bulk densities for powder characterization and the prediction of minimum fluidization and minimum bubbling velocities. Chemical Engineering Science, 2002. 57(14): p. 2635-2640. 33. Instruments, Q. Automatic Pycnometers. 2013 [cited 2014 29/08/2014]; 07120 Rev.C:[Available from: http://www.quantachrome.com/pdf_brochures/07120.pdf. 34. Incropera, F.P., Fundamentals of Heat and Mass Transfer. 2006: John Wiley \& Sons. 35. Bottrell, S.H., et al., Controls on bacterial sulphate reduction in a dual porosity aquifer system: the Lincolnshire Limestone aquifer, England. Chemical Geology, 2000. 169(3–4): p. 461-470. 36. Laboratory, A.N. Total Porosity. 2014 [cited 2014 02/09/2014]; Available from: http://web.ead.anl.gov/resrad/datacoll/porosity.htm. 37. Leva, M., Fluidization. McGraw-Hill series in chemical engineering. 1959, New York: McGraw-Hill.