Characterization of an entrained flow reactor for pyrolysis of coal and biomass at higher temperatures

Fuel 156 (2015) 254–266

Contents lists available at ScienceDirect

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

Characterization of an entrained flow reactor for pyrolysis of coal and biomass at higher temperatures Aime H. Tchapda a, Sarma V. Pisupati a,b,⇑ a b

John and Willie Leone Family Department of Energy and Mineral Engineering & EMS Energy Institute, The Pennsylvania State University, USA National Energy Technology Laboratory, US DOE, USA

h i g h l i g h t s  Use of CFD in the design of high pressure entrained flow reactor.  Biomass residence time was higher than coal particle.  Tar release was higher for biomass than from coal.  Specific surface areas of coal char lower due to pore coalescence.  Char showed less thermal annealing for biomass compared to coal.

a r t i c l e

i n f o

Article history: Received 16 September 2014 Received in revised form 8 April 2015 Accepted 8 April 2015 Available online 20 April 2015 Keywords: Coal Biomass Pyrolysis Pore coalescence Reactivity Thermal annealing

a b s t r a c t A laboratory-scale entrained flow reactor for gasification/pyrolysis of coal and biomass has been designed and constructed at the Pennsylvania State University. The pre-experimental numerical simulations have been used as an aid in the design of the reactor as well as understanding and explaining the experimental results. Post experimental modeling of the reactor has been carried out using the CFD package ANSYSFluent. Results from experiments conducted with the reactor are here presented. These initial characterization activities of the entrained flow reactor are carried out at atmospheric pressure. Modeling and experiments are conducted at three different temperatures: 1573 K, 1673 K and 1773 K. The CFD models show some particle and gas recirculation at the inlet of the reactor. The calculated residence time in the reactor is 0.5 s for biomass and 0.4 s for coal when the particles traveling distance is 0.65 m. Tar and CO are the dominant species at 1573 K in both coal and biomass conversions, however while tar reduces as the temperature increases, the CO formation increases. Fuel conversion varies significantly between coal and biomass. The minimum conversions observed during experiments were 86.7% for biomass and 56.8% for coal at 1573 K. Conversion rates as high as 90.5% were observed for biomass at 1773 K, while the maximum coal conversion observed was 64.0% at 1773 K. The BET surface area of coal chars obtained at 1573 K and 1673 K was similar and higher than that of the char obtained at 1773 K. This drop of surface area at 1773 K has been attributed to pore coalescence, following observation of the SEM images. The surface area of biomass chars does not vary significantly. The reactivity studies conducted on the chars reveal some thermal annealing at higher temperature for coal; this occurrence is observed to be less pronounced for biomass chars. Ó 2015 Elsevier Ltd. All rights reserved.

1. Introduction The continuous depletion of fossil fuel resources and the need to curb greenhouse gas (GHG) and pollution from fossil fuels utilization has stimulated the increasing use of biomass and ⇑ Corresponding author at: 126B Hosler Building, University Park, PA 16802, USA. Tel.: +1 814 865 0874. E-mail addresses: [email protected] (A.H. Tchapda), [email protected] (S.V. Pisupati). http://dx.doi.org/10.1016/j.fuel.2015.04.015 0016-2361/Ó 2015 Elsevier Ltd. All rights reserved.

opportunity fuels for power generation and fuel production. Co-firing coal and biomass has been assessed by various authors [1–3] and is suggested as a viable and readily applicable option for converting biomass into power, heat and chemicals [4–6]. Although coal and biomass naturally appear in solid form, their physical properties vary substantially, causing dissimilar behavior during their thermal conversion. High temperature gasification is well recognized for achieving high carbon conversion and producing a high quality syngas with

A.H. Tchapda, S.V. Pisupati / Fuel 156 (2015) 254–266

low methane and tar content [7,8]. Unlike fixed-bed and fluidizedbed, entrained flow gasifiers operate at higher temperature and higher heating rates, allowing better efficiency (higher carbon conversion) and cleaner syngas. However, the operating temperature of entrained flow reactors coincides with the fusion temperature of the major mineral matters found in biomass (Na, K, Ca). Moreover, the hydrodynamics of coal and biomass particles in an entrained flow reactor are likely to differ considerably, given the disparity of their size, shape and density. Coal particles are denser, smaller in size and are spherical or nearly spherical. Biomass particle on the other hand have a lower density, are 5–10 time order of magnitude bigger than coal particles and are prolate (cylinderlike) and oblate (disk-like) spheroids. These complexities necessitate a thorough perception of the conversion mechanism of the two fuels in an entrained flow reactor. Therefore, an entrained flow reactor aimed at studying the behavior of solid fuels (coal, coke, biomass, and waste) at high temperature, high pressure has been built at the Pennsylvania State University’s EMS Energy Institute. The present work outlines the development work involved in the construction of the entrained flow reactor following a simulated assisted design carried out in ANSYS Fluent and COMSOL as well as some results of experiments carried out at atmospheric pressure.

255

2. Reactor description and design motive 2.1. Basic layout and operation of the reactor The present reactor (Fig. 1) consists of five sub-systems: the feeding sub-system, the gas preheating and steam generator subsystem, the reaction sub-system, the species sampling and analysis sub-system and the reactor control and cooling sub-system. Fig. 2 shows a simplified 3D representation of the reactor created in order to facilitate modeling into Comsol Multiphisics. The raw coal and biomass used for these experiments were dried in an oven maintained at 60 °C for 24 h in order to get rid of any adsorbed moisture. The dried coal or biomass is loaded in the feeder which controls the feed rate of the particles into the reactor. The primary gas (transport gas), together with coal/biomass particles travel through the water cooled injection probe and are injected into the reactor with an initial velocity (Vp0 Þ. This initial particle velocity is controlled by the flow rate of the primary gas. Upon injection into the reactor, the particles travel the heated zone of 0.65 m length where reaction takes place. The secondary gas, including steam, passes through the preheater where it is heated and injected into the reaction section. The reaction section is electrically heated, using six Molybdenum disilicide (MoSi2 Þ heating

0.65 m

Fig. 1. Schematic of the reactor.

256

A.H. Tchapda, S.V. Pisupati / Fuel 156 (2015) 254–266

elements. The control system regulates the temperature of the reaction sub-system and the flow rate of gases. A chiller cooler maintains the temperature of the vessel below 60 °C. The distance travelled by the gas–particle mixture in the reactor is set by the adjustable water cooled collection probe. From the collection probe, char particles are separated from gases in the heated thimble filter. The gas mixture leaving the thimble filter passes through a tar collection train where tar and moisture are removed. The tar sampling method adopted during these tests is based on the ‘‘European Tar Protocol’’ [9–11]. The sampling unit consists of a heated bath, a cooled bath and six impinge bottles. Five of the impinge bottles are filled with isopropanol solution and the remaining one is left empty. Three of the bottles (bottles 1, 2 and 4) are placed in the heated bath and maintained at 35 °C while the other 3 (bottles 3, 5 and 6) are kept in the cooled bath at 20 °C. A better dispersion of the gas in the solvent solution is enhanced by adding glass frits around the inlet tube of bottles 2, 3 and 5. A vacuum pump pulls the gas mixture from the tar sampling train and sends it to a mass flow meter, then to the gas chromatograph where the various gas species present in the gas mixture are identified. 2.2. Modeling the temperature profile in the reactor The following assumptions have been made in modeling the reactor: – Fuel particles (coal and biomass) are assumed to be spherical. – Interaction or collision among particles in negligible, given the very low volume fraction of the particles in the reactor. – The temperature of any particle is uniform throughout at any given time.

– A swelling ratio of 1 has been adopted for both coal and biomass, owing to the very high heating rate encountered by the particles upon injection into the reactor. It is assumed that the particles will undergo a very fast, transient swelling (increased diameter) follow by fragmentation (reduced diameter). Because of the complexity involved in these mechanisms, the particle diameter has been assumed to remain unchanged. Moreover, It has been shown that swelling is not substantial at very high heating rate (>104 K/s) as in this reactor [12] and that the heating rate at which maximum particle swelling occurs for Pittsburgh # 8 coal is <104 K/s [13]. 2.2.1. Gas, refractory – insulation and vessel temperature One important design aspect of the entrained flow reactor was to estimate how effective the heat generated by the electrical heating elements could be retained within the reactor. In other words, how much heat is being lost from the reactor and how is the reactor shell temperature at maximum operating conditions. Solutions of this step of the design were carried out with the help of the software package Comsol, using a steady step conjugate heat transfer in the fluid (transport gas, reaction gas and purge gas) and solids (refractory-insulation, reactor tube) with surface to surface radiation between the heating elements and the inner refractory walls and outer reactor tube (Fig. 2). 2.2.2. Particle velocity As stated earlier, modeling the reaction section of the reactor has been carried out in Ansys – Fluent, considering a 2D axisymmetric geometry and steady state for the continuous phase (gas) while the discrete phase (particles) trajectory and velocity have been calculated in a time dependent frame. The particle velocity is dictated by the gas flow rate in the reactor, which requires the

Primary Stream inlet (particle & transport gas)

Water cooled injection probe

Secondary Stream inlet (773-973 K)

Purge gas Reactor tube

0. 7 m

Heating element RefractoryInsulation

GC Vacuum pump

Tar collection train

Water cooled collection probe

Fig. 2. A simplified 3D representation of the reactor for modeling purpose.

257

A.H. Tchapda, S.V. Pisupati / Fuel 156 (2015) 254–266

continuous phase velocity to be determined. In entrained flow systems, particle flow is said to be dilute because, due to the lower volume fraction of the discrete phase compared to the gas phase, particle motion is controlled by local aerodynamic forces [14]. The particle size varies between 75–250 lm for biomass and 45– 106 lm for coal. It is assumed that these size distributions follow aRosin–Rammler distribution with

"

  dp R dp ¼ 100  Exp  0 dp

!n #

The particle velocity in the reactor has been determined by solving the discrete phase momentum equation:

qg @v p ¼g 1 @t qp

! þ F D ðv g  v p Þ where;

ð3Þ

F D is the drag exerted by the gas phase on the discrete phase,   F D ¼ q18dl2  f Rep p p

:





f Rep ¼ C D

R(d) is the cumulative percent of material retained at mesh size 0 equal dp, dp is the particle size or mesh size, dp is the mean particle size, and n is a measure of the spread of particle sizes. Fig. 3 shows a plot of Log(Ln(R(d)) versus Log(dp). It is found that the spread parameter n is 2.7 and 4.9 for coal and biomass respectively while the average particle diameter is 109.1 lm and 198.2 lm for coal and biomass respectively. The steady state gas phase continuity and momentum equations that have been implemented in the 2D axisymmetric geometry are:

Continuity : r  ðqv Þ ¼ SP

ð1Þ

Momentum : r  ðqvv Þ ¼ rP þ r  s þ qg þ SP v

ð2Þ

q is the density of the gas; v is the gas phase velocity vector; Sp is the source term, due to mass addition from the discrete phase to the gas phase (devolatilization). P is the static pressure; qp is the particle density; g is the gravity; s is the stress (deviatoric) tensor. The source term Sp appearing n km

in Eqs. (1) and (2) is calculated as SP ¼ vpolumep where np is the number of particles in a computational cell after the equation has been discretized. k is the rate constant for the release of gaseous compound from the discrete phase (particle) to the gas phase. mp is the mass of a single particle and ‘‘volume’’ is the volume of a computational cell.

Rep is the Reynolds number, Rep ¼

qp dp jv p v g j l

and C D is the drag

coefficient.

Rep < 1;

CD ¼

1 < Rep < 1000; Rep > 1000;

Rep : 24

CD ¼

  24  1 þ 0:15  R0:687 ep Rep

:

C D ¼ 0:4:

The particle velocity is solved to be

Vp ¼ Vg þ



V p





ADt

 Vg e

þ



sv 1  qqgp g    f Rep

1  eADt



ð4Þ

f ðRep Þ q d2 where A ¼ sv and sv ¼ 18p lp V g is the gas velocity, V p is the particle velocity in the previous time step and Dt is the time step.

2.2.3. Particle temperature The following particle temperature equation has been solved and implemented in the CFD package Ansys-Fluent.

mp C p

p

  dm   @T P p ¼ hconv Ap T gT p þ ep Ap r h4R  T 4p þ DHdv ol @t dt

ð5Þ

where C p p is the particle specific heat; T g is the local gas temperature; T p is the particle temperature; hconv is the convection heat transfer coefficient from the gas phase to the discrete phase; Ap is 2

the area of the particle equals to p⁄dp , dp being the particle diameter; ep is the emissivity of the char particle; r is the Stefan Boltzmann constant (5.67 ⁄ 108 W=ðm2 K4 Þ; hR is the radiation  1=4 temperature 4Gr ; G is the incident radiation ðW=m2 Þ defined as R  I dX X¼2p ; I is the radiation intensity and X is the solid angle; DHdv ol is the heat of devolatilization. Solution of the particle temperature equation has been implemented in a user defined function (UDF) to be compiled and loaded into Ansys-Fluent during simulation. The particle temperature equation has been developed as follows:

mp C p

p

@T P dmp ¼ C 1 T þ C 2 h4R þ DHdv ol  C 1 T p  C 2 T 4p @t dt

C 1 ¼ hconv  Ap

where;

and C 2 ¼ ep  r  Ap :

  Tp. The term C 1 T p  C 2 T 4p is linearized as  C 1 þ C 2 T 3 p T p is the particle temperature in the previous time step.

TP ¼ Fig. 3. Plot of Log[Ln(R(d)] versus Log(dp) for coal (a) and biomass (b).

  C Dt C Dt C4 ð 3 Þ ð 3 Þ 1  e mp Cp p þ T p e mp Cp p C3

Dt is the time step

where;

ð6Þ

258

A.H. Tchapda, S.V. Pisupati / Fuel 156 (2015) 254–266

C 3 ¼ C 1 þ C 2 T 3 p

and C 4 ¼ C 1  T þ C 2 h4R þ

dmp DHdv ol : dt

Table 1 Proximate analysis of the fuels.

dm

The term dtp DHdv ol has been taken into account when calculating the particle temperature, but has been neglected when calculating the overall reactor temperature. This is explained by the low feed rate of particles (respectively 1.3 g/min and 1 g/min for coal and biomass) in the reactor. It is assumed that, given this low feed rate of particles, the heat absorbed by the particles during devolatilization is negligible compared to the heat generated by the heating elements.

Proximate analysis Switchgrass Coal

Moist %

Vol %

FC %

Ash %

9.3 2.9

71.6 36.2

12.2 52.9

7.0 8.9

Table 2 Ultimate analysis of the fuels. C%

2.2.4. Particle residence time The particle residence time is a very important factor in calculating the kinetic parameters in entrained flow systems. For a particle with residence time tk , a probability density function has been

H%

Ultimate analysis dry ash free basis Switchgrass 49.8 7.4 Coal 86.7 5.6

N%

S%

O%

1.4 1.4

0.5 2.0

40.9 4.4

mass flux with residence time t k defined as P ðt k Þ ¼ ParticleTotal . mass flux of particles

The mean particle residence time has been calculated by

tp ¼

nd X

t k  Pðt k Þ:

k

nd is the number diameters in theRosin–Rammler distribution. This takes into account the recirculation effect observed into the reactor. With this method, the particle residence time is higher than just applying the unsteady particle tracking equations. The average particle velocity is then calculated as V pav g ¼

Reactor Length Av erage residence time

giving

1.68 m/s and 1.23 m/s for coal and biomass respectively. 2.3. Testing procedure and reaction modeling

104 ðA1  A0 Þ A1 ð100  A0 Þ

2.3.2. Reaction modeling A global reaction modeling scheme has been adapted throughout this study where the devolatilization reaction is assumed to take place in a single reaction converting the solid fuels into gas products and char.

FuelðsÞ ! GasesðgÞ þ CharðsÞ

2.3.1. Testing procedure All experiments have been conducted under isothermal condition; this is achieved by programming the temperature controller to deliver a certain amount of power to the Super Kanthal heating element, therefore maintaining the reactor tube temperature to a set level. Experiments have been conducted at three set temperatures: 1573 K, 1673 K and 1773 K. The flow rate of the primary gas (transport gas) have been maintained at 10 l/min at standard temperature and pressure (STP) while that of the secondary gas has been maintained at 30 l/min at STP throughout the experiment. CO2 has been used as primary as well as secondary gas. A screw feeder has been used to accurately feed coal and biomass at rates of 1.3 g/min and 1 g/min, respectively. Tables 1 and 2 show proximate and ultimate analyses of coal and biomass used during these experiments. The char samples collected in the thimble filter have been subjected to various studies. The weight of tar dissolved in the solvent has been estimated using a gravimetric analysis. This analysis consists of pouring 50 ml of isopropanol tar mixture into a ceramic dish, letting it stay in a fume hood overnight, transferring it to a heating chamber at 105 °C for 60 min and record the weight of the remaining residue, mtar . The fuel conversion has been calculated using the ash tracer method. Assuming that the ash yield of the fuel remains constant during the whole reaction time, the fuel conversion is given by the following equation:

v¼

is retained in the solvent while the remaining light gases are measured in the flow meter: mlg . The mass balance conforms to mtot ¼ mlg þ mtar þ mH2O where the mass of water vapor is determined. The mass of each gas species identified in the gas chromatograph is calculated as mi ¼ xi  mlg where xi and mi are the mass fraction and the mass of gas specie i, respectively.

ð7Þ

where A0 is the ash yield of the original fuel; A1 is the ash yield of the char collected after reaction; v is the fuel conversion rate. Knowledge of the fuel conversion allows the estimation of the amount of gas produced during reaction mg ¼ X  mp0 ; where mp0 is the initial mass of fuel fed into the reactor and mg is the mass of gas produced. After the tar collection train, tar and water vapor

The motive for considering only devolatilization is based on the fact that the total volatile yield is highly dependent on the heating rate and temperature. The proximate volatile matter calculated at heating rates and temperatures much more lower than the operating conditions of our reactor cannot be a good predictor of the volatile yield in these conditions. Miller and Tillman [15] have proposed the following formula for estimating the maximum volatile yield from solid fuels:

MVY ð%Þ ¼ 0:697 

 0:18 VM  100 FC

ð8Þ

where MVY is the maximum volatile yield, VM and FC are the volatile matter and fixed carbon calculated by the standard proximate analysis method. Applying this formula, the maximum volatile yields for Switchgrass and coal are 95.8% and 65.1% respectively. These values are higher than the conversion obtained in the experiments, justifying the consideration of devolatilization as the main reaction taking place. The fuel particle (biomass or coal) is composed of volatiles, char and ash. The initial mass of the fuel particle can therefore be expressed as the sum of these individual components as:

mp0 ¼ ma þ mc1 þ mv 0

ð9Þ

where ma is the mass of ash, assumed constant during the whole conversion process. mc1 is the mass of char remaining in the particle at complete devolatilization. mv 0 is the initial mass of volatile in the particle. We can define the initial fraction of volatile in the particle as

Xv0 ¼

mv 0 mp0

or mv 0 ¼ X v 0  mp0

ð10Þ

At any time t during devolatilization, the particle mass can be represented as

259

A.H. Tchapda, S.V. Pisupati / Fuel 156 (2015) 254–266

mp ¼ ma þ mc1 þ mv

where;

ð11Þ

mv is the mass of volatile left in the particle at time t. The fraction of volatile remaining in the particle at any time t can also be expressed as:

Xv ¼

mv mp0

or mv ¼ X v  mp0

ð12Þ

mg ¼ mv 0  mv ¼ mp0  ðX v 0  X v Þ

ð13Þ

However, mg is also equal to the difference between the initial mass and the mass at time t, so that

mg ¼ mp0  mp

ð14Þ

The particle conversion, defined as the fraction of material that has been released from the particle during devolatilization is given by

mg mp0  mp mp ¼ ¼1 mp0 mp0 mp0

  mg ¼ mp0  mp ¼ ðm1 þ mv 0 Þ  mv 0  ekt þ m1 ¼ mv 0  mv 0  ekt

or

mp ¼ mp0  ð1  XÞ

ð15Þ ð16Þ

From Eqs. (11) and (16) ma þ mc1 þ mv ¼ mp0  ð1  XÞ. Expanding this equation gives the following relationship:

Xv ¼ Xv0  X

ð17Þ

This equation implies that during devolatilization, the maximum particle conversion is the fraction of the volatiles initially present in the particle. In other words, when particle conversion equals the fraction of the initial volatile, devolatilization is complete (X v ¼ 0).

Eqs. (13) and (21) show that

dmp ¼ k  mv ¼ kðmp  ma  mc1 Þ dt

ð18Þ

where k is the devolatilization reaction rate constant. Since mc1 is assumed constant during devolatilization, we can write m1 ¼ ma  mc1 . As the overall mass of particle remaining after complete devolatilization. Eq. (18) becomes

ð19Þ

Solving this equation gives

mp ¼ mv 0  ekt þ m1

ð20Þ

ð22Þ

which is the mass of volatiles remaining in the particle at time t. Eqs. (15) and (21) show that

X ¼ X v 0  ð1  ekt Þ

ð23Þ

This implies that at infinite time, the particle conversion equals the initial fraction of volatile in the particle during devolatilization. Eqs. (17) and (23) gives X v ¼ X v 0  ekt which is the fraction of volatile remaining in the particle at time t. Let’s define

X 0v ¼

mv ¼ ekt mv 0

ð24Þ

From Eq. (22) the rate constant is derived as:

ln



k¼

mv mv 0



t

  ln X 0v ¼ t

ð25Þ

Eqs. (23) and (24) give a relation between particle conversion (that is easily measured) and X 0v

X 0v ¼ 1 

dm

v . This expression can be The devolatilization rate r ¼ dtp ¼ dm dt verified by replacing mv by Eq. (12) and replacing mp by Eq. (16) and noting that X v and X vary inversely (Eq. (17)). The devolatilization reaction rate is proportional to the remaining volatile in the fuel particle, therefore,

dmp ¼ kðmp  m1 Þ dt

ð21Þ

mv ¼ mv 0  ekt

The mass of gas released during time t is given by

X¼

The equivalent mass of gas released (Eq. (14))

k¼

X Xv0

  ln 1  XXv 0 t

ð26Þ

¼ k0  eE=RT p

 3 2 ln 1  XXv 0 Ea 1 5 lnðkÞ ¼ lnðk0 Þ   ¼ ln4 t R Tp R is the universal gas constant and T p is the particle temperature. The particle density has been updated after accordingly, assumm

6m

ing no change in particle diameter qp ¼ Volpp ¼ pd3p . The initial partip cle densities used in the simulation are 1400 kg=m3 and 800 kg=m3 for coal and biomass respectively. R is the universal gas constant and T p is the particle temperature. The plot of ln(k) versus T1p is presented in Fig. 4. Calculations give activation energies of 79080.3 J/mol and 24206.2 J/mol for coal and biomass respectively while the pre-exponential factors are 1694:9 s1 for coal and 21:5 s1 for biomass.

Fig. 4. Plot of ln(k) version 1/T P .

260

A.H. Tchapda, S.V. Pisupati / Fuel 156 (2015) 254–266

The mass mg released from the particle to the gas phase consists of a mixture of light gases, tar and water vapor. Knowledge of the mole fraction of each gas specie from the GC and applying the analysis outlined in Section 2.3.1 the mass fraction of each gas specie is calculated. 2.4. Reaction products analysis 2.4.1. Gas analysis Real time gas analysis on the system is performed by the Agilent Micro GC 3000A. This gas chromatograph has two columns: a MolSieve 5A PLOT column and a Plot U column. Each column operates isothermally and separates the analytes from the sample into discrete peaks; these peaks are then detected with a thermal conductivity detector (TCD) and the GC software identifies individual compounds by retention times according to a preset list stored within its internal library. H2 ; O2 ; N2 ; CH4 , and CO are detected in the MolSieve column while CO2 , C2 H4 , C2 H6 , C2 H2 , H2 S and COS are detected in the Plot U column. 2.4.2. Char analysis The following analyses have been carried out on coal and biomass chars collected in the filter: Particle skeletal density (Helium density), BET surface area, SEM image analysis, Reactivity. The particle density has been measured using the Quantachrome Instruments’ Multipycnometer for skeletal (true) density analysis. The BET surface area has been carried out in the Micromeritics ASAP 2020 Surface Area and Porosity Analyzer. Liquid nitrogen at 77 K has been used as the adsorptive gas. Char reactivity has been carried out in a Perkin Elmer TGA7 Thermogravimetric Analyzer. Char samples were heated at a ramping rate of 20 °C/min to 800 °C with nitrogen as flowing at 50 ml/ min. The sample temperature was stabilized at 800 °C for 20 min and the gas was changed to CO2 . The Boudouard reaction involving char and CO2 was then allowed to proceed for 5 h. The SEM analysis was performed in a Helios NanoLab 660.

Fig. 5. Gas temperature.

2.5. Grid independence analysis In order to ensure that the solution to the model is independent to the mesh resolution, a mesh independence study has been carried out. The mesh used in the current simulation had a maximum element size of 0.001 mm which gave a total of 20,840 cell elements and 21,528 nodes. The mesh was then further refined to a maximum element size of 0.0008 mm and yielded a total of 32,580 cell elements and 33,440 nodes. Using the refined mesh, simulation of Switchgrass pyrolysis was run at 1773 K reaction tube temperature keeping all parameters the same. 3. Results and discussions 3.1. Reactor modeling The temperature distribution in the reactor is presented in Figs. 5 and 6. The primary gas (Fig. 2) is kept to ambient temperature while flowing in the water cooled injection probe. Upon injection in the reactor, the primary gas temperature rises to approximately 1800 K (Fig. 5) in a very short time. The secondary gas (Fig. 5) enters the reactor at ambient temperature and heat up to close the heating element temperature (2100 K). The double layer refractory-insulation of 20 cm thickness altogether provides a fairly good insulation, dropping the temperature from 1800 K to about 600 K. The shell temperature (Fig. 6) reaches 600 K indicating a requirement for a forced cooling system. Implementation of the forced cooling system has been achieved

Fig. 6. Refractory temperature.

by wrapping cooling coils (3/800 OD and 1/400 ID) on the side and top of the reactor shell in which cold water from a chiller cooler is flown, maintain the shell temperature below 333 K (60 °C). The particle velocity is presented in Fig. 7. On average, coal particles flow at a higher velocity than biomass particles even though their initial velocity (injection velocity) in the reactor is the same. It is estimated that at operating conditions (reaction gas flow rate,

A.H. Tchapda, S.V. Pisupati / Fuel 156 (2015) 254–266

261

occurring near the reaction tube wall where the gas temperature is higher (Fig. 8c) and possibly the adverse pressure gradient in the flow direction are large enough to overcome the momentum of the incoming secondary gas in the neighborhood of the wall surface. Therefore the incoming secondary gas shifts its trajectory toward the center of the reaction tube (Fig. 8a and b) and is followed by the recirculating particles. The particle flow in the reactor tube is limited to the Stokes and transition regimes as depicted by the particle Reynolds number (Fig. 9) with the Stokes regimes occurring predominantly near the wall of the reactor where viscous forces dominate. Fig. 10 shows the particle temperature in the reactor while Fig. 11 shows the heating rate and temperature of particles at the axis of the reactor. It can be observed that the particles reach their maximum temperature at about 200 mm traveling distance in the reactor. Heating rate as high as 25 ⁄ 104 K/s is observed at the initial heating of the particles. The heating rate sharply drops down to about 3000 K/s as the particle approaches its maximum temperature. Given the short time it takes to particles to reach their maximum temperature, the experiment can be accepted as isothermal, justifying the derivation of the rate parameters.

Pisburgh #8

Switchgrass

Fig. 7. Particle axial velocity (m/s).

collection probe position) the residence time of coal particles is approximately 0.5 s while that of biomass is about 0.7 s. Gas and particle recirculation is observed, according to the model, when feeding biomass (Fig. 8a and b). The particle velocity expressed in Eq. (4) gives an indication of why and how recirculation is taking place. In fact, for smaller size particles flowing in high temperature zones (high viscosity), term A in Eq. (4) becomes very big, therefore the second and third terms at the right hand side of Eq. (4) become very small and the particle velocity becomes close to the gas velocity. At the inlet of the reactor, the viscosity induced boundary layer

3.2. Global fuel conversion Fuel conversion varies from 57% to 64% for coal and 87% to 91% for biomass within the three temperatures (1573 K, 1673 K and 1773 K) investigated (Fig. 12). High fuel conversion is expected from biomass given the high content of oxygen (Table 2), which is almost ten times higher than coal’s. The proximate analysis of the remaining chars (Fig. 13) shows an increasing percentage of ash as the conversion increases. Up to 44% of ash yield is observed for Switchgrass pyrolyzed at 1773 K. The percentage of volatile in the remaining biomass char does not vary significantly. This is understandable, given the high level of conversion attained by these chars while the fixed carbon decreases as the conversion goes up. As for coal chars, the percentage of fixed carbon does

Fig. 8. Gas velocity (m/s) and temperature (K).

262

A.H. Tchapda, S.V. Pisupati / Fuel 156 (2015) 254–266

Parcle heang rate and temperature

30.3

2000 1800 1600 1400

20.3

HR_1773 HR_1673 HR_1573 Tp_1773 Tp_1673 Tp_1573

15.3

10.3

1200 1000 800

Temperature (K)

Heang Rate * 10000 (K/s)

25.3

600 400

5.3

200 0.3 0

100

200

300

400

500

600

0 700

Reactor axial posion (mm) Fig. 11. Particle temperature and heating rate (HR. Heating Rate, Tp: Particle temperature) calculated at the axis of the reactor.

Pisburgh #8

Switchgrass

and a decrease of tar, both for coal and biomass. This is explained by the secondary cracking reaction of high molecular weight compounds from volatile as they are released from the particles, to light gases. Although the amount of CO2 released decreases with increasing temperature, it does not vary much for biomass. For coal, CO2 release at 1673 K and 1773 K drops substantially compared to 1573 K. The amount of tar released during coal experiments in considerably lower than that of biomass experiments. Given the low feed rates of the fuel compared to the reaction gas

Fig. 9. Particle Reynolds number.

Fuel conversion 100 95

Conversion (%)

90 85 80 75

Coal

Switchgrass

70 65 60 55 50 1550

1600

1650

1700

1750

1800

T (K) Fig. 12. Fuel conversion as a function of temperature.

100%

80%

60%

Pisburgh #8

Switchgrass

Vol

40%

FC

Fig. 10. Particle temperature (K).

Ash

20%

not vary much whereas the percentage of volatile decreases as the level of conversion of the char increases. This observation can be justified by the fact that coal char conversion is not yet high therefore degradation of volatile-associated materials still dominates the conversion process. The mass fraction of gas species released during these experiments (Fig. 14) shows an increase of CO and H2 with temperature

0%

Fig. 13. Proximate analysis of coal and biomass chars.

263

A.H. Tchapda, S.V. Pisupati / Fuel 156 (2015) 254–266

Coal

Biomass

(a)

(b)

(c)

(d)

CO2

H2

tar

CO

CH4

H2O

C2H2

H2S

Fig. 14. Mass fraction of gas species.

flow rate, the fraction of CH4 , C2 H2 , H2 O and H2 S in the gas is small, despite the relatively high conversion of the fuels. The mass fraction of H2 O reduces with the increase of temperature, suggesting a possible endothermic gas phase reaction where steam is being consumed. It can also be noticed that the mass fraction of C2 H2 is substantial, especially for biomass. The cracking of tar may have contributed to the formation of C2 H2 . However, the lower concentration of CH4 is an indication that methane is being converted to C2 H2 in the temperature range of the experiments [16,17]. It is noticed that more CO is released from coal than biomass while the amounts of H2 released from coal and biomass are comparable. Hydrogen sulfide is not detected from biomass pyrolysis at all the temperatures tested.

3.3. Char properties As expected, the skeletal density of the char increases with conversion (Fig. 15). The change in density with respect to conversion still shows an exponential trend, this trend is expected to level off as the char conversion approaches 100%. At 1573 K, coal char with 56% conversion and biomass char with 87% conversion have similar density. The plot of specific surface area change versus temperature is reported in Fig. 16. Biomass chars reveal a slight increase in specific surface area from 1673 K to 1773 K. Unfortunately the data at 1573 K is not available (not enough sample was available) for a better appreciation of this increment. Coal chars at 1573 K and 1673 K have similar specific surface area, but at 1773 K the value

90

3.0

Density (g/cm3)

2.0 1.5

Coal Biomass

1.0 0.5 0.0 45

50

55

60

65

70

75

80

85

90

95

Conversion (%) Fig. 15. Particle density (skeletal) as a function of particle conversion.

Specific Surface Area (m2/g)

80

2.5

70 60 50

Coal

40

Biomass

30 20 10 0 1550

1600

1650

1700

1750

1800

T (K) Fig. 16. Specific surface area variation as a function of temperature.

264

A.H. Tchapda, S.V. Pisupati / Fuel 156 (2015) 254–266

Fig. 17. SEM images of coal chars.

100 90 80

X (%)

70 60 50 40 30

100SG-1573K

20

100SG_1673K

10

100SG_1773K

0 0

50

100

150

200

250

300

Time (min) Fig. 20. Biomass char conversion in the TGA, at 1073 K in CO2 . Fig. 18. Pore coalescence observed in the coal char generated at 1773 K.

40

100Coal_1573K

35

100Coal_1673K

30

X (%)

100Coal_1773K 25 20 15 10 5 0 0

50

100

150

200

250

300

Time (min) Fig. 19. Coal char conversion in the TGA, at 1073 K in CO2 .

drops to approximately 22 m2 /g. At first look, this drop in surface area seems implausible but a thorough look at the SEM images of the coal chars (Fig. 17 and 18) shows an advanced degree of degradation of the char structure at 1773 K. Pore coalescence is a wellknown phenomenon pertaining to entrained flow gasification and has been documented in the literature [18–20]. During the initial stage of conversion, the specific surface area increases as a result of pore growth. Continuous pore development leads to the merging of neighboring pore walls. As the conversion proceeds, pore wall collapse becomes important and the surface area begins

to decrease with further conversion. Mitchell et al. [18,20] have estimated that the maximum surface area occurs around 35% of coal conversion. If this estimation holds, then maximum surface area development has already been achieved and these chars have already embarked the declining side of surface area development, since their conversions are far beyond 35%. The char conversion graphs of the reactivity studies conducted in the TGA are presented in Fig. 19 and 20. Coal chars generated at 1673 K and 1773 K have similar reactivities while the char generated at 1573 K has a higher reactivity than the previous two (Fig. 19). It is interested to note that, despite the wide difference in specific surface area of the 1673 K and 1773 K chars, their reactivity is almost the same. This is an indication that coal char deactivation temperature lies between 1573 K and 1673 K. Although not very perceptible, some symptoms of char deactivation can also be identified from the graph of biomass char conversion in the TGA. The high conversion displayed by the biomass char generated at 1573 K at the beginning of the TGA test, compared to the 1673 K and 1773 K chars, is an indication that some mechanism has taken place. In fact, it is important to note the similarity in the shapes of the graphs of coal and biomass chars generated at 1573 K. Coal and biomass char deactivation at higher temperatures has been investigated by various authors [21–30]. The presence of innate recalcitrant carbon structures in coal or biomass may be responsible for the low carbon conversion sometimes observed at higher temperatures. However, there is also evidence that biomass char deactivation may also be responsible for the low carbon conversion. Various mechanisms by which char deactivation occurs have been identified. Thermal annealing also referred in the literature as thermal deactivation or graphitization

265

A.H. Tchapda, S.V. Pisupati / Fuel 156 (2015) 254–266

1770

0.924 0.922

Avg. Parcle Velocity (m/s) 0.918

1750

Max. Gas Temperature (K)

0.916

Max. Parcle Temperature (K)

1740

0.914

Temperature (K)

Velocity (m/s)

1760

Avg. Axial Gas Velocity (m/s)

0.920

1730 0.912 1720

0.910

20000

25000

30000

35000

Number of Cells Fig. 21. Velocities and temperatures obtained from actual and refined mesh.

3.4. Grid independence analysis results

1 0.9 0.8

% Error

0.7 0.6 0.5

Avg. Axial Gas velocity Avg. Parcle Velocity Max. Gas Temperature

Fig. 21 shows the graph of velocity (gas and particles) as well as temperature (gas and particles) for the actual mesh used in this simulation and a 1.5 mesh refinement. The error induced by this refinement is displayed in Fig. 22. The percentage error is within 1% of the variables investigated, suggesting that the results obtained in these simulations are not affected by the grid size.

0.4

Max. Parcle Temperature 0.3

4. Conclusions

0.2 0.1 0 Fig. 22. Percentage of error induced by mesh refinement.

is a phenomenon taking place at high temperatures where the char matrix is transformed into a highly ordered and unreactive carbon arrangement of aromatic structure. It has been found that the likelihood of char thermal annealing to occur is conditioned by the source of the char and its preparation. Guerrero et al. [22] identified thermal deactivation at 900 °C for eucalyptus char prepared at slow heating rate as a result of structural ordering as well as coalescence of the micro pores. In contrast, the same char prepared at rapid heating condition did not show any sign of graphitization. Womat et al. [28] analyzed the structural and compositional transformation occurring on coal and biomass chars. They concluded that, unlike coal char, the structural ordering of biomass chars is unlikely to happen because the abundant amount of oxygen present in biomass create a highly cross-linked carbon structure opposing ordering. Studies conducted by Hurt [23] on various chars also revealed a highly ordered carbon structure of the high rank coal chars while the low rank coal and biomass chars did not show any ordered structure. He concluded that solid fuels can be categorized into graphitizing and non-graphitizing carbon solids. Hurt then theorized that graphitizing carbons which comprise high rank coals undergo a fluid phase during pyrolysis while the fluid state is less prominent or absent in low rank coals and biomass. The ordering of the carbon structure then takes place during this fluid state forming a less reactive char as a result. The implication of this analysis is far-reaching in co-firing applications, especially for pulverized systems operating at higher temperatures. Such systems should consider the drop in coal char reactivity at higher temperatures and take appropriate measures to improve the subsequent heterogeneous char gas reaction following devolatilization. Blending coal with biomass could be beneficial in this situation since the reactivity of biomass char seems to be less affected by the heat treatment temperature.

The present paper has outlined the design, operation and testing of an entrained flow reactor. CFD approaches have aided to describe the hydrodynamics and heat transfer profile in the reactor. The average coal particle residence time is estimated at 0.4 s while that of biomass is estimated at 0.5 s. The CFD model showed some recirculation of particles and reaction gas at the inlet zone of the reactor. The calculated particle heating rate at the inlet of the reactor was of the order of 104 K/s. This causes the particle temperature to rise to maximum in a very short time. Carbon conversion for coal ranged from 57% to 64% while that for biomass from 87% to 91% in the temperature range investigated. As conversion increased, the ash content of the char increased for coal and biomass chars; meanwhile residual volatiles decreased and the fixed carbon almost remained constant in coal char whereas in biomass char, the fixed carbon decreased. The dominant species in the products at lower temperature (1573 K) were CO and tar whereas at higher temperature (1773 K), CO, tar and H2 were dominant species. More tar was released from biomass than from coal. The particle density (Helium density) increased with the level of conversion, for both coal and biomass. At 1773 K, the specific surface area of coal char decreased considerably due to pore coalescence. Coal chars produced at 1673 K and 1773 K had similar reactivity despite significant differences in their specific surface areas. Thermal annealing was observed above 1573 K in coal chars. This phenomenon was less noticeable in biomass chars, even though the char generated at 1573 K showed some difference in reactivity compared to the chars generated at 1673 and 1773 K. References [1] Van Loo S, Koppejan J. Handbook of biomass combustion and cofiring. London: Earthscan; 2008. [2] Tchapda AH, Pisupati SV. A review of thermal co-conversion of coal and biomass/waste. Energies 2014;7:1098–148. [3] Munir S. A review on biomass–coal co-combustion: current state of knowledge. Proc Pakistan Acad Sci 2010;47:265–87. [4] Sami KAM, Wooldridge M. Co-firing of coal and biomass fuel blends. Prog Energy Combust Sci 2001;27:171–214.

266

A.H. Tchapda, S.V. Pisupati / Fuel 156 (2015) 254–266

[5] Colechin M, Malmgren A. Best practice brochure: co- firing of biomass. COAL R287 DTI/Pub URN 05/1160, Department of Trade and Industry, UK, London; 2005. [6] Hoagland K. Why cofiring biomass with coal is hotter than ever. Biomass Mag 2013. [7] Henrich E, Weirich F. Pressurized entrained flow gasifiers for biomass. Environ Eng Sci 2004;21:53–64. [8] Higman C, Burgt Mvd. Gasification. New York (USA): Elsevier; 2008. [9] Neeft JPA. Rationale for setup of impinger train as used in the technical specification of sampling and analysis of tar and particles in the product gases of biomass gasification. CEN BT/TF 143, European Standardisation Institute (CEN); 2005. [10] Kamp Wvd, Wild Pd, Zielke U, Suomalainen M, Knoef H, Good J, et al. Tar measurement standard for sampling and analysis of tars and particles in biomass gasification product gas. In: 14th European biomass conference & exhibition, Paris, France; 2005. [11] Good J, Ventress L, Knoef H, Zielke U, Hansen PL, Kamp Wvd, et al. Sampling and analysis of tar and particles in biomass producer gases. CEN BT/TF 143, European Standardisation Institute (CEN); 2005. [12] Solomon PR, Hamblen DG. Pyrolysis. In: Schlosberg RH, editor. Chemistry of coal pyrolysis. New York: Plenum Press; 1985. [13] Gale TK, Bartholomew CH, Fletcher TH. Decreases in the swelling and porosity of bituminous coals during devolatilization at high heating rates. Combust Flame 1995;100:94–100. [14] Crowe CT. Review—numerical models for dilute gas–particle flows. J Fluids Eng 1982;104:297–303. [15] Miller BG, Tillman DA. Combustion engineering issues for solid fuel systems. Burlington: Academic Press; 2008. [16] Fincke JR, Anderson RP, Hyde T, Detering BA, Wright R, Bewley RL, et al. Plasma thermal conversion of methane to acetylene. Plasma Chem Plasma Process 2002;22:105–36. [17] Fincke JR, Anderson RP, Hyde TA, Detering BA. Plasma pyrolysis of methane to hydrogen and carbon black. Ind Eng Chem Res 2002;41:1425–35. [18] Mitchell RE, Ma L, Kim B. On the burning behavior of pulverized coal chars. Combust Flame 2007;151:426–36.

[19] Sørensen HS, Rosenberg P, Petersen HI, Sørensen LH. Char porosity characterisation by scanning electron microscopy and image analysis. Fuel 2000;79:1379–88. [20] Pusz S, Krzesin´ska M, Sme˛dowski Ł, Majewska J, Pilawa B, Kwiecin´ska B. Changes in a coke structure due to reaction with carbon dioxide. Int J Coal Geol 2010;81:287–92. [21] Beeley T, Crelling J, Gibbins J, Hurt R, Lunden M, Man C, et al. Transient hightemperature thermal deactivation of monomaceral-rich coal chars. In: Symposium (international) on combustion, vol. 26; 1996. p. 3103–10. [22] Guerrero M, Ruiz MP, Millera Á, Alzueta MU, Bilbao R. Characterization of biomass chars formed under different devolatilization conditions: differences between rice husk and eucalyptus. Energy Fuels 2008;22:1275–84. [23] Hurt RH. Char crystalline transformations during coal combustion and their implication for carbon burnout. DE-FG22-95PC95205, Office of Fossil Energy, DOE, Morgantown, WV (USA); 1997. [24] Keown DM, Hayashi J-i, Li C-Z. Effects of volatile–char interactions on the volatilisation of alkali and alkaline earth metallic species during the pyrolysis of biomass. Fuel 2008;87:1187–94. [25] Tremel A, Haselsteiner T, Nakonz M, Spliethoff H. Coal and char properties in high temperature entrained flow gasification. Energy 2012;45:176–82. [26] Tremel A, Spliethoff H. Gasification kinetics during entrained flow gasification – Part I; Devolatilisation and char deactivation. Fuel 2013;103:663–71. [27] Umeki K, Moilanen A, Gómez-Barea A, Konttinen J. A model of biomass char gasification describing the change in catalytic activity of ash. Chem Eng J 2012;207–208:616–24. [28] Wornat MJ, Hurt RH, Yang NYC, Headley TJ. Structural and compositional transformations of biomass chars during combustion. Combust Flame 1995;100:131–43. [29] Zolin A, Jensen A, Dam-Johansen K. Kinetic analysis of char thermal deactivation. Proc Combust Inst 2000;28:2181–8. [30] Zolin A, Jensen AD, Jensen PA, Dam-Johansen K. Experimental study of char thermal deactivation. Fuel 2002;81:1065–75.