1-s2.0-s0016236113008387-main (1)

8/10/2019 1-s2.0-S0016236113008387-main (1)

http://slidepdf.com/reader/full/1-s20-s0016236113008387-main-1 1/12

Computational fluid dynamics modeling of biomass fast pyrolysis

in a fluidized bed reactor, using a comprehensive chemistry scheme

Pelle Mellin ⇑, Efthymios Kantarelis, Weihong Yang

KTH Royal Institute of Technology, Division of Energy and Furnace Technology, Brinellvägen 23, SE 100 44 Stockholm, Sweden

h i g h l i g h t s

We have used an advanced kinetic model for pyrolysis coupled with CFD in 3D. The yields are compared to experimental results and shows a not too far-off prediction.

The simulations are very time consuming but makes it possible to explore secondary reactions.

For example, a number of thermal cracking reactions are applied to the tar components to see the effect.

a r t i c l e i n f o

Article history:

Received 18 June 2013

Received in revised form 6 August 2013

Accepted 5 September 2013

Available online 17 September 2013

Keywords:

CFD

Fluidized bedFast pyrolysis

Pyrolysis oil

Pyrolysis

a b s t r a c t

The CFD modeling for fast pyrolysis has previously focused on the major pyrolysis products; liquid, char

and gas. This paper introduces a new approach to biomass pyrolysis; integrating a complex scheme of

reactions including formation of such components as levoglucosan. The 3-D simulation takes into account

the complex breakdown of each biomass subcomponent, the fluid dynamics of the process as well as the

heat and momentum transfer of three Eulerian phases.

The pyrolysis products include reference species that reflects the composition of the bio oil, gas fraction

and char fraction. A number of reactions are in addition applied to account for the thermal cracking of tar

compounds and the final compositions are compared to experimental yields. The results show that the

predicted pyrolysis products reflect the experimental yields satisfactorily, apart from the water content

which is under predicted. Most importantly though, the approach is computationally feasible and it

should be useful for future work.

2013 Elsevier Ltd. All rights reserved.

1. Introduction

Pyrolysis of biomass is considered as a potential source of fuel

for various applications such as transportation, combined heat

and power production as well as reduction agents. From a practical

standpoint, fast pyrolysis means processing with optimized

conditions for liquid production which implies fast heating rate

and a temperature not exceeding 500 C.Modeling such processes is challenging in many ways and even

more so in fluidized bed reactors. Those challenges are linked to

the complex thermo-physical environment of fluidized bed

reactors and the complex structure of biomass and the decomposi-

tion with an immense number species as products.

Much of the earlier CFD simulation work assumed three

superficial components as representatives for the liquid, gaseous

and solid components. Some studies elaborated on the non-unifor-

mity of a reacting biomass particle and some studies focused more

on the kinetics and the composition of the samples related to

pyrolysis products [1]. No one has yet been able to device a

complete model which takes into account all phenomena of fast

pyrolysis in fluidized beds but both approaches improved the

outcome and lead to increased understanding.

From the CFD direction, Papadikis [2] argues that: ‘‘The most

crucial of the assumptions is that the particle is assumed to

maintain uniform temperature along its radius’’. In contrast,Dupont et al. [1] claims that ‘‘the accurate knowledge of the

reaction kinetics appears to be a crucial parameter for a reliable

modeling of the pyrolysis process’’. This work can be seen as a

unifying effort, where the in-depth chemistry is included in a

comprehensive CFD model which accounts for heat and mass

transfer as well as entrainment of particles.

Previously, a CFD model was developed (described in Mellin

et al. [3]) with a two phase framework for time-effectively

studying the behavior of the gas. Now this model has been

extended in order to report the specific components of pyrolysis

and thus obtain a composition of the liquid, gas and char. Therefore

a new kinetic model has been implemented as well as a third

0016-2361/$ - see front matter 2013 Elsevier Ltd. All rights reserved.http://dx.doi.org/10.1016/j.fuel.2013.09.009

⇑ Corresponding author. Tel.: +46 87909022; fax: +46 8207681.

E-mail address: [email protected] (P. Mellin).

Fuel 117 (2014) 704–715

Contents lists available at ScienceDirect

Fuel

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

http://dx.doi.org/10.1016/j.fuel.2013.09.009

mailto:[email protected]


http://www.sciencedirect.com/science/journal/00162361

http://www.elsevier.com/locate/fuel

http://www.elsevier.com/locate/fuel

http://www.sciencedirect.com/science/journal/00162361


mailto:[email protected]


http://crossmark.crossref.org/dialog/?doi=10.1016/j.fuel.2013.09.009&domain=pdf

8/10/2019 1-s2.0-S0016236113008387-main (1)


Eulerian phase which in greater detail renders the biomass flow.

Experiments are in addition made alongside the simulation to

compare with the results. More information on the basis of the

numerical model can be found in the previous publication [3], as

well as more details on tuning of the drag law. The far-reaching

purpose of the work is up-scaling of the technology.

In this paper, the pilot reactor is firstly described with analysis

results of the pyrolysis products. Then the numerical model is

described with focus on the additions to the previous work; finally

the results, comparison with experiment, discussion and conclu-

sion is presented. Throughout the work we made extensive use

of User Defined Functions (UDF) and solution methods included

in the commercial software package ANSYS Fluent 14.5.

2. Pilot reactor

A pilot fast pyrolysis setup with has been assembled at KTH,Sweden. The setup includes a preheater for the fluidizing-gas, the

fluidized bed reactor, a cyclone and a scrubber. A schematic

overview of the plant is shown in Fig. 1, with the domain of the

CFD model indicated.

The biomass is introduced by a screw feeder directly into the

fluidized bed at height of 5.75 cm above the distributor plate.

The biomass is fed at rate of 2 kg/h, which is the designed capacity

of the rig. The biomass is a mix of pine and spruce, see proximate

and ultimate analysis in Table 1.

The gas species and the main groups of bio oil components are

given in Table 2 which is reported in Kantarelis et al. [4], therein

referred to as case S/B: 0. All specific components of the bio oil

can be found in the same paper as well as the measurements

methods and more detailed descriptions of the plant.

Nomenclature

Alphabetic letters A pre-exponential factor (s1)a interfacial area concentration (m2 m3)C heat capacity (J kg-1 K-1)c coefficient (–)d diameter (m)E activation energy (J mol1)e coefficient of restitution (–) g gravitational acceleration (m s2)h heat transfer coefficient (W m2 K1)K momentum exchange (kg m3 s1)k heat conductivity (W m1 K1)m mass (kg)Q heat transfer (W)q heat transfer per surface area (W m2)R universal gas constant (J mol1 K1)S source term, due to e.g. reactions ([kg, J. . .]m3 s1)T temperature (K)t time (s)u intrinsic velocity (m s1)v velocity (m s1)Y mass fraction (–)

Greek lettersa volume fraction (–)q density (kg m3)l viscosity (Pa s)p pi (–)s stress (Pa)u sphericity (–)

Dimensionless numbersRe Reynolds number (-)Pr Prandtl number (-)Nu Nusselt number (-)

Common subscripts g gass sandb biomassi any specie

mf minimum fluidizationD drag

Biomass

Windbox

Bubblingfluidized

bedreactor

Char

Fluidizing gaspreheater

Circulation

with cooling

Cyclone

Liquid

Scrubber

Gas

CFDmodel

Fig. 1. Cross section of the bubbling fluidized bed reactor system with boundariesindicated for the CFD model.

Table 1

Biomass composition [4].

Analysis Parameter

Moisture

(wt%)

Volatile

matter

(wt%)db

Ash

(wt%)dbFixed

carbon

(wt%)db

HHV

(MJ/kg)

Proximate

analysis

9.8 83 0.31 16.6 20.46

C, wt%db H, wt%db Oa,

wt%db

N, wt%db S, ppmdb

Ultimate

analysis

50.7 6.1 42.71 0.18 <120

a

By difference.db Dry basis.

P. Mellin et al. / Fuel 117 (2014) 704–715 705

http://-/?-

http://-/?-

http://-/?-

http://-/?-

http://-/?-

http://-/?-

http://-/?-

http://-/?-

http://-/?-

http://-/?-

http://-/?-

http://-/?-

http://-/?-

http://-/?-

http://-/?-

http://-/?-

8/10/2019 1-s2.0-S0016236113008387-main (1)


3. Modeling approach

The Euler–Euler multiphase framework with three phases is ap-

plied in the computation; one gas phase and two granular phases:

sand and biomass. Several CFD studies treated the biomass parti-

cles as dilute in the sand and vapor phase, which is a reasonable

assumption. The biomass particles are then tracked in a lagrangian

manner and treated as an additional layer to the two-phase

framework.

When such an arrangement is made the phase coupling be-

tween the two Eulerian phases and the lagrangian one are often

simplified. Usually the drag of the two Eulerian phases correctly

influences the momentum of the lagrangian phase; while not the

other way around. Often the conservation equations for the Euleri-an phases do not contain a term for drag caused by the lagrangian

phase.

In this case where the amount of biomass particles can have

non-negligible effect on the other phases in the system, the

three-Eulerian system was chosen. In addition the Eulerian defini-

tion of a phase in Fluent, allow the use of a stiff chemistry solver

and chain reactions for solid species.

The model with phases and interactions are shown in Fig. 2. The

mass exchange includes drying and reactions. The defined pyroly-

sis reactions correspond to the scheme developed by Ranzi et al. [5]

with cracking reactions proposed by Blondeau and Jeanmart [6].

The mass exchange is two-ways since the cracking of some com-

pounds produce solid char.

A transient second order formulation is used with a fixed timestep. QUICK is used for the volume fraction coupling, which corre-

sponds to third order accuracy. No turbulence is accounted for

since the flow is assumed laminar. The residuals are allowed to de-

cline below 1 103 before progressing to the next time step.

Interested readers can find further information on the computa-

tional solution procedure in the ANSYS Fluent theory guide [7];

see section on The Pressure-Based Segregated Algorithm. In the

same guide the full set of conservation equations and correlations

used in this publication can be found.

3.1. Species conservation and reactions

The multi-fluid CFD calculation is based on continuity equations

that are solved for all iterations in each time step, in each cell and

phase. Continuity equations account for mass, momentum, energy

and species; for the fluid phase the following momentum Eq. (1) is

solved, where a denotes volume fraction, q the density, v the veloc-

ity, _m g the mass exchange between phases and S the source term

defined by the user.

@

@ t ða g q g Þ þ r ða g q g ~v g Þ ¼ _m g þ S g ð1Þ

For all the species in a phase, such as the compounds in the

fluid, the following analogous Eq. (2) is solved, where Y denotesthe mass fraction of any specie i.

@

@ t ða g q g Y iÞ þ r ða g q g Y i~v g Þ ¼ _m g þ S g ð2Þ

Heterogeneous reactions between phases is included in the

sources term S and for this simulation the primary pyrolysis reac-

tions (1–15, in Fig. 3) corresponding to the scheme developed by

Ranzi et al. [5] is implemented. All reactions are considered first or-

der kinetically controlled, however the heat supply rate to the par-

ticle and the concept of competing reactions makes the picture

more complex. Heat of the reactions is collected from Calonaci

et al. [8]. The primary pyrolysis reactions are in detail described, to-

gether with kinetics and reaction heat, in Appendix A (see Table A).

All the tar species are assumed rapidly evaporating from thebiomass particle, primarily since the fluidization provides swift

flushing of the pyrolysis vapors but also because the general

knowledge is lacking on this particular step (see Dofour et al. [9]

for more discussion on the evaporative step). Hence the gaseous

species trapped in the metaplast, as described in Ranzi et al. [5]

are also directly released into the gas phase.

The thermal cracking of tar is usually effective above 500 C

according to Fagbemi et al. [10] and is commonly known to be fa-

vored by contact with char particles and long residence times.

Hoekstra et al. [11] on the other hand performed an experimental

study indicating that low-mineral char actually is not catalytically

active in a fluidized bed, while in the same study observing that tar

cracking can occur above 400 C. In any case a need to model the

thermal cracking is obvious as the reactor temperature is almost500 C and the gas residence time quite long, around 2 s. In the

Table 2

Pyrolysis product composition, with the rest being mostly water [4].

Product Composition

Gas Specie H2 CH4 CO CO2 C2 C3 C4 C6 C2–C3 ole fins Tota l

wt%wb 0.088 1.581 9.249 8.620 0.461 0.272 0.039 0.005 0.356 20.672

Oil Group Acids Ketones Aldehydes Furans Sugars Phenols Methoxy-phenols Catechols Benzenes Total

wt%wb 9.218 14.044 0.406 4.600 1.741 1.560 6.305 2.039 0.271 34.574

Char Element C H O N Ash Total

wt%wb 14.432 0.541 2.598 0.162 0.280 18.022

wb Wet basis.

H e a t

M a s s

M o m

e n t u m

Heat

Momentum

H

e a t M o m e n t u m

FeedstockSolid 1

(biomass, b)

Solid 2

(sand, s)

Fluid

(gas, g)

E n t r a i n e d s o l i d

Gas and vaporsGas (fluidizing)

Fig. 2. The model in principal with phases and exchange of heat, mass and

momentum.

706 P. Mellin et al./ Fuel 117 (2014) 704–715

8/10/2019 1-s2.0-S0016236113008387-main (1)


model the cracking is considered as extra-particle reactions ap-

plied to the macromolecules with products according to the in-

tra-particle reactions from Blondeau and Jeanmart [6] with

kinetics from Park et al. [12]. These are hypothetical reactions that

have not been proven experimentally but are still helpful to ex-

plore this aspect; note that some solid Char forms due to reaction

3, 4 and 9. The thermal cracking reactions shown in Fig. 3 and are

in detail described in Appendix A, see Table B.

3.2. Energy conservation and phase exchange

The conservation equation for energy in the gas phase is givenby (3), where h g is the internal heat transfer coefficient, h gs is the

heat transfer coefficient between gas-sand and h gb is the heat

transfer coefficient between gas–biomass. As the conservation

equations are based volumetric flow variables, the heat transfer

coefficient is multiplied by the interfacial area concentration a gs

and a gb respectively.

@

@ t ða g q g h g Þ þ r ða g q g ~v g h g Þ ¼ a g

@ p

@ t þ s g : r~v g

r~q g þ S þ h gba gbðT g T bÞ þ h gsa gsðT g T sÞ ð3Þ

h gs is estimated based on the Nusselt number correlation (4) from

Gunn [13] which is applied for the heat transfer between the sandand the gas phase (5).

Biomass C6H8.46O3.9

tw%80.7tw%16.1tw%01.81tw%39.03tw%72.24

lom%87.2lom%56.0lom%99.11lom%20.04lom%75.44

2 1 9 11 12

15 H2O 1 CellA C2H4 1 Acetone 11 CO2

16 Char 2 7 6 H2O 1 LignOH 1

3 CH4

CO2 CO2

4 1 Xylan CH4 H2 H2O

1 LVG 8 C2H4 Char 2 Methanol

CO2 10.1 p-Coumaryl2 CH4

H2 10.08 Phenol2 C2H4

CH2O 0.35 LignCC CO2

CO2 Methanol2 H2

CO2 H2 Ethanol2 Char

H2 OCO 2 H2O H2O 1 Lign

CH4 CO2 Char 2 CH4 14

Acetaldehyde2 Formaldehyde2 C2H4

Char 2 Methanol2 H2 11 Lumped-phenol2 11 H2O

HAA2 Ethanol2 CO2 12 CO2

Glyoxal2 H2 r ahCO 2 Formaldehyde2

HMFU2 CH4 10.3 p-Coumaryl2 Methanol2

Acetone2 C2H4 10.2 Phenol2 Acetaldehyde2

Char 2 10.35 Acrylic-acid2 CH4

C2H4

H2

End product Char 2

Solid intermediete product Acetone2

Tar subject to thermal cracking

Thermal cracking reactions (1-10)

CO2 CO2 CO2 CO2 CO2 12.5 CO2 CO CO2 12 CO2 CO2

C2H4 H2 C2H4 C2H4 H2 11.5 H2 H2 H2 C2H4 C2H4

C2H4 Char 2 Char C2H4 11.75 C2H4 Char 2

81.05%

Cell

C6H10O5

LignOH

18.95%

100%

11.14%

88.86%

60.21%39.79%

41.68%

HCell1 HCell2

100%

2.08.0

0.41

1

0.495

1.32

15.0

1

58.32%

0.7

7.052.061.0

Primary pyrolysis reactions (1-15)

OngiLHngiLCngiLlleCH

0.25 5.735

54.08.0

40% 60%99.97% 100% 100%

C5H8O4 C15H14O4 C16H10O6(OCH3)4 C17H13O4(OCH3)5

5

0.03%

13

51.4521.057.08.0

2

0.2

52.016.0

521.059.0

521.052.0

521.08.09.0

4.11.0

5.02.0

1

0.7

0.65

0.6

1

1.8

6.4

15

Acrylic-acid

3 0.5

HMFU Acetone p-Coumaryl Phenol Xylan LVG

1

1 2 3 4 5 6 7 8 9 10

1.5 0.5

1.25

1

1

2

1

2

2

HAA Glyoxal Lumped-phenol

0.675 0.6

0.65

0.5

5.5

0.2

0.2

526.052.0 0.4

52.02.0

1

3

3

3

2.5

1 0.5

1.5

2.5 1.5

1

2

10 100%

Fig. 3. Shows the reactions pathways implemented in the model, with greyed-out percentages of the predicted results from the model, at 10.75 s.


http://-/?-

http://-/?-

8/10/2019 1-s2.0-S0016236113008387-main (1)


h gs ¼ 6k g a g asNus

d2

s

ð4Þ

Nus ¼ ð7 10a g þ 5a2 g Þð1 þ 0:7Re0:2s Pr

1=3Þ

þ ð1:33 2:4a g þ 1:2a2 g ÞRe0:7s Pr

1=3 ð5Þ

For the heat transfer to a particle immersed in a fluidized bed,

several expressions exist and a commonly used one (6) is describedby Mickley and Fairbanks [14] (with adaption for CFD by Papadikis

et al. [15]).

h ¼

ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffikmðqC ÞmS

r ð6Þ

For the most important heat transfer to the biomass phase; the

Eulerian framework requires a separate contribution from each

phase (sand and gas). As result the expression (6) has been modi-

fied to account only for the heat conduction from the solid phase,

see (7) with km now is e sks and (qC p)m being esqs(C p)s.

hsb ¼

ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffia2s ksqsC p;spt

r ð7Þ

The exchange coefficient h gb (8) between the gas and the bio-mass is instead calculated according to the well-known Ranz–Mar-

shall equation [16,17], with correction for the local gas volume

fraction. An implicit assumption is then made; that heat transfer

from gas inside the bed occurs through convection and not by con-

duction as implied in the theory by Mickley and Fairbanks [14]. By

implementing the heat transfer in parallel no discontinuity at an

interface of dense region is encountered.

h gb ¼ k g Nubdb

ð8Þ

Nub ¼ 2 þ 0:6Re1=2b Pr1=3 g : ð9Þ

3.3. Momentum conservation and phase exchange

The momentum equation for the gas is expressed in (10), where

p is pressure, s g equals the stress tensor, g is the gravitational force,

K gs is the phase momentum exchange coefficient for gas-sand, K gb

the exchange coefficient for gas-biomass, _m gb~v g is the momentum

transfer associated to mass transfer between gas and biomass.

Equivalent equations are solved for the two granular phase.

@

@ t ða g q g ~v g Þ þ r ða g q g Y g ~v g ~v g Þ ¼ a g rp þ rs g

þ a g q g ~ g þ K gsð~v g ~v sÞ þ K gbð~v g ~v bÞ þ _m gb~v g ð10Þ

K gs is estimated by (10), according to the work by Syamlal and

O’Brien [18]. We have in addition tuned the drag law according pre-

dict the right minimum fluidization velocity umf (c 1 and c 2 in (11)are approximated to 9.19 and 0.28). See Ref. [3] for how for the

gas-sand drag law tuning was carried out. The drag coefficient c Dis given by (12) and the terminal velocity v r ,s for the sand particles

are defined in (13) with coefficients in (14).

K gs ¼3asalq f 4v

2r ;sdsc DResv r ;s

j~v s ~v g j ð11Þ

c D ¼ 0:63 þ 4:8 ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiRes=v r ;s

p !2

ð12Þ

v r ;s ¼ 0:5 A0:06Res þ ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

ð0:06ResÞ2

þ0:12Resð2B AÞ þ A2q

ð13Þ

A ¼ a4:14 g B ¼

ac 1 g ;a g > 0:85

c 2a1:28 g ;a g 6 0:85

(

K gb is described by (15) which is the drag law from Morsi and

Alexander [19] with u added as an additional input to the drag coef-

ficient (u is set to 0.6). c D is the drag coefficient given by (16) and c 1,

c 2, c 3, c 4 are coefficients defined in (17)–(20) respectively.

K gb ¼18l g

qbd2

b

c DReb24

j~v b ~v g j ð15Þ

c D ¼ 24

Rebð1 þ c 1Re

b2b

Þ þ c 3Rebc 4 þ Reb

ð16Þ

c 1 ¼ expð2:3288 6:4581u þ 2:4486u2Þ ð17Þ

c 2 ¼ 0:0964 þ 0:5565u ð18Þ

c 3 ¼ expð4:905 13:8944uþ 18:4222u2 10:2599u3Þ ð19Þ

c 4 ¼ expð1:4681 þ 12:2584u 20:7322u2 þ 15:8855u3Þ ð20Þ

K sb is the exchange coefficient for sand and biomass, the parti-

cle–particle drag term developed by Syamlal [20] given by (21) is

used (with esb is set to 0.9). g 0,sb is the radial distribution function,

calculated according to Lun et al. [21].

K sb ¼ 3ð1 þ esbÞðp=2 þ c fr ;sbp

2=8Þasqsabqbðds þ dbÞ2 g 0;sb

2p qsd3s þ qbd

3b

j~v s ~v bj

ð21Þ

Bulk viscosity for the sand phase is calculated according to Lun

et al. [21]. The solid shear viscosity is generally assumed to be a

sum of three components, the collisional viscosity calculated

according to Ref. [22], the kinetic viscosity according to Ref. [18]

and the frictional viscosity (assumed negligible due solids volumefraction being far from the maximum packing limit). The granular

temperature is a component of both the collisional viscosity and

the kinetic viscosity and is estimated based on kinetic theory, from

Ref. [18].

3.4. Solution strategy and time step size

The rate of various reactions sometimes differs by several or-

ders of magnitude. As the products of some reactions are reactants

in others, the large difference in rate can cause stability problems.

The system can in that sense be called stiff and for this work we

have used a Stiff Chemistry Solver; which is applied to each

time-step and first solves the flow field with all reaction rates set

to zero. This gives an initial solution which aids in convergenceof the next step which is computing the complete reacting flow

field. The used Stiff Chemistry Solver is integrated in ANSYS Fluent

and is based on the Double precision Variable-coefficient Ordinary

Differential Equation solver (DVODE); see for example Brown et al.

[23]. This solver introduces an error and as a result the Stiff Chem-

istry Solver was only used until the calculation stabilized.

When performing the calculation, a first stage without any bio-

mass feeding was used to obtain a continuously converging simu-

lation with only fluidizing gas and sand. After 1.4 s, the biomass

feeding was switched on and a short time step of 1 105 s was

used. After a short elapse of time, a step size of 1 104 s was

used. Much later in the calculation, the step size could be increased

to a final value of 5 104 s with an average of 25 iterations per

time step. The physical time computed in total is 10.75 s whichcorresponds to over two months of nonstop computation.


8/10/2019 1-s2.0-S0016236113008387-main (1)


8/10/2019 1-s2.0-S0016236113008387-main (1)


most of the water formation occurred due to secondary tar reac-

tions (STR) and only a small amount during primary biomass

decomposition.

A considerably less amount of work has been done on the

homogenous reactions that do not (or only slightly) alter theamount of liquid. In the context of this model, those would

be the reactions that follow the formation of e.g. Levoglucosan

and contribute to the vast variety of molecules found in pyro-

lysis oil. These are likely the bulk source of water; as seen in

the work by Zhang et al. [27], most of the typical pathways

of continued levoglucosan decomposition involve sheddingwater.

(a) (b)

(c) (d) 0 . 9

7 9

0 . 0

0 0

0 . 0

0 0

0 . 0

2 1

0 . 9

7 9

0 . 0

0 0

0 . 0

0 0

0 . 0

2 1

0 . 8

7 7

0 . 0

1 3 0 . 0 9 3

0 . 0

1 7

0 . 8

0 8

0 . 0 3 0

0 . 1

4 6

0 . 0

1 6

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

C H O Ash

E l e m

e n t a l w t f r a c t i o n o f c h a r

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

Gas Liquid Char

W

t f r a c t i o n o f p y r o l y s i s

p r o d u c t s

Without tar cracking With tar cracking With tar cracking incl. unreacted biomass Experiment

0 . 4

6 3

0 . 0

6 7

0 . 4

7 0

0 . 4

5 8

0 . 0

6 5

0 . 4

7 7 0

. 5 7 6

0 . 0

6 7

0 . 3

5 7

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

C H O

E l e m e n t a l w t f r a c t i o n o f

o r g a n i c l i q u i d

0 . 1 1 7

0 . 1

9 4

0 . 1

3 8

0 . 2

4 9

0 . 0 6 3

0 . 1

3 4

0 . 2

4 6

0 . 1

9 8

0 . 3

0 3

0 . 1

1 9

0 . 1

4 3

0 . 0 6 4

0 . 2

8 3

0 . 4

7 7

0 . 0

4 6

0.0

0.1

0.2

0.3

0.4

0.5

0.6

CH4 H2 CO2 CO C2H4

M o l f r a

c t i o n o f p e r m e n e n t

g a s e s

Without tar cracking With tar cracking Experiment

H2O

fraction

0 . 1 9

1

0 . 5

9 3

0 . 1

0 1 0 . 2

5 9

0 . 4 3 8

0 . 1

4 5 0 . 2

4 8 0

. 4 2 1

0 . 1 7 8

0 . 2 0

7 0 . 3

4 6

0 . 1 8

0

0 . 1 5 9

0 . 1

1 5

0 . 2

5 4

0 . 1

5 2

fraction

Organic

Fig. 4. The pyrolysis products, at 10.75 s. Where (a) is the elemental composition of the biomass, (b) the volume fraction of gas species (c) the elemental composition incl. ash

of the char fraction, (d) the overall pyrolysis products summed up on wet basis.

Fig. 5. Shows the velocity profile of the different phases along the volume fraction, at 10.75 s. Note that volume fraction of biomass is displayed with a logarithmic scale.


8/10/2019 1-s2.0-S0016236113008387-main (1)


In fact several groups of primary pyrolysis products are known

to produce water when they react (dehydrate) homogenously in

the vapor phase [11]. A study by Faix et al. [28] concerning lignin

clearly shows water formation by TGMS, at temperatures usually

above primary biomass decomposition. A common product of lig-

nin is methoxyphenols, such as isoeugenol, and forms by dehydra-

tion according to [29]. The formation mechanism of sugars(monosaccharaides) is dehydration according to [30] and a com-

mon product is furans which will also form by dehydration [31].

In the model all furans are represented by hydroxymethylfurfural

(HMFU) with a yield of about 1% for ‘‘Without tar cracking’’ and

0.66% ‘‘With tar cracking’’. This is too low as Table 2 states about

4.6% furans on wet basis (5.1% on dry basis) in the experiment.

As a result of further dehydration of organic tar, the oxygen con-

tent of the liquid will decrease as well. As of now it is too high as

visible in Fig. 4(a).

4.2. Instantaneous results: Volume fraction and velocity

Fig. 5 shows the velocity magnitude and volume fraction of gasbiomass and sand along a surface in the reactor. The biomass is

quite well distributed in the fluidized bed and as visible some

floats on top of the bed, in the splash zone.

The volume fraction along the centerline of the reactor is shown

in Fig. 6. The velocity in y-direction for the three phases is shown in

Fig. 7; the acceleration at the outlet is very rapid (for the gas phase

velocity is at most 21.82 m/s). The biomass phase velocity strongly

correlates with the sand phase in the dense region and correlates

with the gas phase in the less dense region. This means that themomentum is easily transferred to the biomass particles and more

easily so by the sand phase. The pressure in the reactor is also

shown in Fig. 7. A decline in pressure clearly correlates with an

acceleration of the gases. The pressure drop over the bed in the

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

0 0.3 0.6 0.9

V o l u m e f r a c t i o n

V

u

m

a

o

Reactor height, meactor height m

Gas

Sand

Biomass

Fig. 6. Shows the volume fraction of gas, sand and biomass along the centerline, at

10.75 s.

Atmosphericpressure

101

103

105

107

109

-1

-0.5

0

0.5

1

1.5

2

2.5

3

0 0.3 0.6 0.9

P r e s s u r e ,

k P a

V e l o c i t y , m / s

Reactor height, m

Gas y-velocity

Sand y-velocity

Biomass y-velocity

Predicted pressure

Fig. 7. The velocity for gas, sand and biomass as well as gas pressure, as function of

reactor height, at 10.75 s.

Fig. 8. Shows the temperature of all phases, at 10.75 s. A surface along the Z = 0

plane shows a color map in degrees centigrade.

0

50

100

150

200

250

300

350

400

450

500

0 0.3 0.6 0.9

H e a t t r a n s f e r c o e f f i c i e n

t , W / m 2 / K

Reactor height, m

Gas-biomass

Sand-biomass

Total

Fig. 9. The heat transfer coefficient at the interface of gas-biomass and sand-biomass, at 10.75 s.


http://-/?-

http://-/?-

http://-/?-

8/10/2019 1-s2.0-S0016236113008387-main (1)


experiment is 1212.1 ± 128.9 Pa while the predicted value is

1246 Pa.

4.3. Instantaneous results: temperature and heat transfer

The temperature of the phases is shown in Fig. 8. The tempera-

ture is quite uniform in the lower part of the reactor and the gas

and sand temperature is close. The low temperature zone caused

by biomass feeding is small but clearly visible in the biomass

phase. As the volume fraction of biomass is low, no significant

amount of heat is transferred from the gas and sand which meansthe temperature does not drop noticeably.

The heat transfer coefficient is presented in Fig. 9, as the fluctu-

ations indicate the local volume fraction of gas and sand is factored

in. In the lower part (dense region) of the bed the heat transfer

from the sand dominates while higher up, the heat transfer from

gas obviously dominates. At the outlet of the reactor, the heat

transfer coefficient for the gas phase increases as the Reynolds

number rises at higher velocities.

Since the temperature of the sand and gas is very close, the heattransfer coefficients of each may be added together which is shown

as a black line in Fig. 9. The total heat transfer coefficient varies be-

tween 200 and 320 W/m2/K which is close to the range found by

Papadikis et al. [32] for a slightly smaller but otherwise similar

reactor. The total heat transfer coefficient also shows that high

heat transfer is possible in the splash zone. Here the high velocity

of gas results in high convective heat transfer, due to rupture of

bubbles.

Fig. 10 shows the heat transfer with the temperature difference

factored in. Some biomass goes directly upwards from the feeding

line and enters the splash zone. A low temperature contributes to a

very fast actual heat transfer. Biomass entering this region is thus

not necessarily negative and Fig. 9 together with Fig. 10 shows that

overall heat transfer actually is highest here.

4.4. Biomass conversion, residence time and Species

The total product formation rate was, in the case of ‘‘Without

tar cracking’’ 6.63 104 kg/s which means some accumulation

of biomass occurs during the beginning of the simulation. The li-

quid, gas and water that escape the outlet of the reactor amounts

to 3.99 104 kg/s (see Table C in Appendix A for specific compo-

nents). This is the basis for ‘‘With tar cracking’’ added with the char

formation; from both primary and secondary pyrolysis reactions

integrated over the whole domain. The total product formation

rate for the case ‘‘With tar cracking’’ is 4.66 104 kg/s. With the

unreacted biomass factored in the total output rate is

4.86 104

kg/s with specific solid components shown inTable D. This means some time still remains for the output rate

to become equal to the input rate 5.56 104 kg/s and as a result

of the cases ‘‘With tar cracking’’ and ‘‘With tar cracking incl. unre-

acted biomass’’ should be treated as indicative.

The unreacted biomass is shown in Table D, in Appendix A,

amounts to a small percentage of the fed biomass. The percentages

of Cellulose, Hemicellulose and Lignin in the unreacted biomass are

48.7%, 23.2% and 28.0% respectively while in the biomass it is

0

5000

10000

15000

20000

25000

30000

0 0.3 0.6 0.9

H e a t t r a n s f e

r , W / m

H

a

e

W

/

m

Reactor eactor

height, meight m

Gas-biomass

Sand-biomass

Total

Fig. 10. Shows the heat transfer coefficient multiplied with the temperature

difference between gas-biomass and sand-biomass, at 10.75 s.

Fig. 11. Path lines along the velocity field of gas, at 10.75 s.

1 6

1 0 9

1 0 8

7 3

5 1

2 8

1 7

1 2

1 8

7 7 4

2 2 1 5

2 6 ( t > 5 s )

0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0

0

0.05

0.1

0.15

0.2

0.25

0.3

0

20

40

60

80

100

120

R e s i d e n c e t i m e d i s t r i b u t i o n ( R T D )

N u m b e r o f p a t h l i n e s

Time, s

Fig. 12. The number of path lines, following the velocity field of the gas phase, in

each time span represented by bars. The fitted gas residence time distribution isrepresented by a curve with axis on the right.


8/10/2019 1-s2.0-S0016236113008387-main (1)


42.2%, 31.0% and 26.8%. While this could change, it is interestingthat Hemicellulose reacts to a higher degree compared to Cellulose

and Lignin. This could be due to Hemicellulose beginning to react

at lower temperatures (according to Ref. [5]) and thus getting a

head start.

In order to get a sense of the residence time, mass-less particle

tracking along the velocity field of the gas phase was made. An

example is shown in Fig. 11, where 21 mass-less particles were

released from the inlet. In total 480 were released at different

points in time from the biomass inlet. 468 reached the outlet and

the distribution is shown in Fig. 12, with the required time for

reaching the outlet registered on the horizontal axis. The average

gas residence time is 1.8 s with around 4% (16 path lines) residing

in the reactor less than 1 s. The velocity field for the biomass has

not yet stabilized enough and actual particle tracking may be nec-essary to evaluate the precise particle residence time distribution.

We can however safely assume that the average particle residencetime is higher than the average gas residence time, 1.8 s, since the

biomass particles collides with the sand particles and backflow is

possible along the flight upwards. This gives the particles time to

react but a 100% conversion within the framework of the model

cannot be expected or is not even reasonable. Within the model

framework a 100% conversion would mean that the char fraction

entirely consists of carbon and ash.

The reaction pathways for the first Case is given in percent

in Fig. 3. LVG is highly favored in this case as almost all the cel-

lulose decomposes along this pathway. LignO decomposes very

slowly compared to LignH as most of the LignOH are formed

from LignH while more LignO is fed to the system. Conse-

quently, some unreacted LignO is still inside the reactor and

at 10.75 s, it amounts to 2.07

10

4

kg of LignO, while only0.04 104 kg of LignH.

Fig. 13. Shows some selected species in profile along the Z = 0 plane, at 10.75 s.

Table A

Primary pyrolysis reactions (Reaction 1-15) from [5] with reaction heat from [8] and drying (Reaction 16) from [6].

Reaction A (s1) E

(kJ mol1)

Dh

(kJ kg1)

1 Cell? CellA 8 1013a 192.5a 447.7

2 Cell? 5H2O + 6Char 8 107 125.5 1087.8

3 CellA? LVG 4Ta 41.8a 732.2

4 Ce llA? 0.95HAA + 0.25Glyoxal + 0.2Acetaldehyd + 0.25HMFU + 0.2Acetone + 0.16CO2 + 0.23CO + 0.9H2O + 0.1CH4 + 0.61Char 1 109a 133.9 899.6

5 H Cell? 0.4HCell1 + 0.6HCell2 1 1010a 129.7a 548.1

6 HCell1? 0.75H2 + 0.8CO2 + 1.4CO + 0.5Formaldehyde 3 109a 113.0a 447.7

7 HCell1? Xylan 3Ta 46.0a 707.1

8 HCell2? CO2 + 0.5CH4 + 0.25C2H4 + 0.8CO + 0.8H2 + 0.7Formaldehyde + 0.25Methanol + 0.125Ethanol + 0.125H 2O + Char 1 1010 138.1 259.4

9 L ig nC? 0.35LignCC + 0.1 pCoumaryl + 0.08Phenol + 0.41C2H4 + H2O + 0.495CH4 + 0.32CO + CO + H2 + 5.735Char 4 1015 202.9 602.5

10 LignH? LignOH + Acetone 2 1013 156.9 523.0

11 L ig nO? LignOH + CO2 1 109 106.7 510.4

12 LignCC? 0.3pCoumaryl + 0.2Phenol + 0.35Acrylic-acid + 0.7H2O + 0.65CH4 + 0.6C2H4 + 1.8CO + H2 + 6.4Char 5 106 131.8 288.7

13 LignOH? Lign + H2O + Methanol + 0.45CH4 + 0.2C2H4 + 2CO + 0.7H2 + 4.15Char 3 108 125.5 100.4

14 L ig n? Lumped-phenol 8Ta 50.2a 577.4

15 Lign? H2O + 2CO + 0.2Formaldehyde + 0.4Methanol + 0.2Acetaldehyd + 0.2Acetone + 0.6CH 4 + 0.65C2H4 + 0.5H2 + 5.5Char 1.2 109a 125.5a209.2

16 H2O(l)? H2O(g) 5.3 1010 88 2260.0

a Modified for high temperature according to Ref. [6].


http://-/?-

http://-/?-

8/10/2019 1-s2.0-S0016236113008387-main (1)


Table B

Secondary pyrolysis reactions [6].

Reaction A (s1) E (kJ mol1) Dha (kJ kg1)

1 HMFU? 3CO + 1.5C2H4 4.28 106 108.0 642.7

2 Acetone? 0.5CO2 + 0.5H2 + 1.25C2H4 4.28 106 108.0 1878.2

3 pCoumaryl? CO2 + 2.5C2H4 + 3Char 4.28 106 108.0 359.6

4 Phenol? 0.5CO2 + 1.5C2H4 + 2.5Char 4.28 106 108.0 143.1

5 Xylan? 2CO2 + H2 + 1.5C2H4 4.28 106 108.0 563.0

6 LVG? 2.5CO2 + 1.5H2 + 1.75C2H4 4.28 106 108.0 1701.67 HAA? 2CO + 2H2 4.28 106 108.0 3562.7

8 Glyoxal? 2CO + H2 4.28 106 108.0 156.6

9 Lumped-phenol? 2CO2 + 3C2H4 + 3Char 4.28 106 108.0 693.8

10 Acrylic-acid? CO2 + C2H4 4.28 106 108.0 912.9

a By balance.

Table C

Product yields for specific components, with formation rate from primary pyrolysis and outflow rate for the case ‘‘With tar cracking’’.

Specie Formula Included in Formation rate of primary pyrolysisa Outflow rateb

(kg/s) (wt%) (kg/s) (wt%)

Methane CH4 Gas yield 1.38E-05 2.08 1.12E-05 2.40

Carbon-monoxide CO Gas yield 5.18E-05 7.81 4.42E-05 9.49

Carbon-dioxide CO2 Gas yield 4.52E-05 6.82 4.53E-05 9.72Hydrogen H2 Gas yield 2.87E-06 0.43 2.57E-06 0.55

Water, gas H2O Liquid yield 7.60E-05 11.46 7.40E-05 15.88

Nitrogen N2 – – – 6.39E-04 -

Formaldehyde CH2O Liquid yield 2.01E-05 3.03 1.46E-05 3.13

Acetaldehyde CH3HCO Liquid yield 1.82E-06 0.27 9.37E-07 0.20

Methanol CH3OH Liquid yield 9.59E-06 1.45 7.07E-06 1.52

Glyoxal C2H2O2 Liquid yield 2.97E-06 0.45 1.39E-06 0.30

Ethylene C2H4 Gas yield 1.31E-05 1.98 1.73E-05 3.71

Hydroxyacetaldehyde (HAA) C2H4O2 Liquid yield 1.17E-05 1.76 5.46E-06 1.17

Ethanol C2H5OH Liquid yield 5.85E-06 0.88 4.57E-06 0.98

Acrylic-acid C3H4O2 Liquid yield 1.17E-05 1.76 1.87E-09 0.00

Acetone C3H6O Liquid yield 5.80E-06 0.87 6.41E-06 1.38

Xylan C5H8O4 Liquid yield 3.97E-05 5.99 4.75E-05 10.19

Levoglucosan (LVG) C6H10O5 Liquid yield 2.64E-04 39.82 1.03E-04 22.10

Phenol C6H5OH Liquid yield 4.38E-06 0.66 3.27E-06 0.70

Hydroxymethylfurfural (HMFU) C6H6O3 Liquid yield 6.44E-06 0.97 3.01E-06 0.65

pCoumaryl C9H10O2 Liquid yield 8.74E-06 1.32 6.52E-06 1.40

Lumped-phenol C11H12O4 Liquid yield 2.89E-07 0.04 6.41E-08 0.01

Char C Char yield 6.72E-05 10.13 6.76E-05c 14.51

a Used for the case ‘‘Without tar cracking’’.b Used for the case ‘‘With tar cracking’’.c Value used for the case ‘‘With tar cracking’’, not the actual outflow rate.

Table D

All solid species with feeding and outflow rate.

Specie Formula Feeding rate Outflow ratea

kg/s wt% kg/s wt%

Cellulose (Cell) C6H10O5 2.11E-04 38.01 2.05E-06 10.76

Activated cellulose (CellA) C6H10O5 - - 7.22E-06 37.95Hemicellulose (HCell) C5H8O4 1.55E-04 27.81 1.81E-10 0.00

Hemicellulose 1 (HCell1) C5H8O4 – - 5.80E-07 3.05

Hemicellulose 2 (HCell2) C5H8O4 – - 3.84E-06 20.18

C-rich lignin (LignC) C15H14O4 9.04E-05 16.28 2.57E-06 13.52

H-rich lignin (LignH) C16H10O6(OCH3)4 8.06E-06 1.45 1.34E-08 0.07

O-rich lignin (LignO) C17H13O4(OCH3)5 3.54E-05 6.37 1.81E-08 0.10

C-rich lignin (LignCC) C15H14O4 – – 7.99E-07 4.20

OH-rich lignin (LignOH) C19H22O8 – – 1.85E-06 9.75

Lignin (Lign) C11H12O4 – – 8.10E-08 0.43

Water, liquid H2O 5.44E-05 9.8 8.85E-12 0.00

Ash SiO2 1.56E-06 0.28 1.08E-07b –

Char C – – 1.02E-06b –

a Used for the case ‘‘With tar cracking incl. unreacted biomass’’.b Values not used.


8/10/2019 1-s2.0-S0016236113008387-main (1)


Fig. 13 shows the molar concentration of some selected species

in the domain. The species propagates throughout the domain, a

higher concentration of species in the midsection indicates that

the results have not stabilized fully yet. An accumulation of bio-

mass is seen on the left hand side, opposite to the feeding line,

which results in concentrated gas release. This accumulation is also

visible in the volume fraction of the biomass, see Fig. 5. Other accu-

mulations are also visible in Fig. 5 but high heat transfer in combi-nation with an almost complete evaporation of water cause the

high pyrolysis reaction rate shown Fig. 13.

5. Conclusion

The main conclusions of this study can be summarized by the

following paragraphs.

The primary pyrolysis scheme used in this work is possible to

implement in a comprehensive CFD model. The thermal cracking

reactions are helpful but should be replaced or put in parallel with

other secondary reactions that can explain the formation of more

water.

The even distribution of biomass in the bed suggests good oper-

ational conditions. Pressure drop as compared with experiment is

accurate. Total heat transfer rate was found to be highest in the

splash zone.

The momentum of the sand is easily transferred to the biomass

particles and more so compared to the gas phase. The residence

time for the gas is 1.8 s on average and the biomass particles

should reside at least as long in the reactor, which is sufficient time

for conversion.

Evaluating the percentages of different pathways in the reaction

scheme, shows that almost all the cellulose is converted into levo-

glucosan. As indicated by the model results, the conversion of

hemicellulose is more complete than cellulose and lignin.

Acknowledgments

This work was supported by the Swedish National Infrastructurefor Computing (SNIC 001-11-26) via PDC. KIC Innoenergy and The

Swedish Energy Agency (Contract No. 33284-1) are acknowledged

for funding of the project. Preem, Boson Energy, Sveaskog and The

Division of Chemical Technology at KTH are acknowledged for

support in other ways. We are grateful to all above named

organization for the opportunity to carry out this research.

Appendix A

See Tables A–D.

References

[1] Dupont C, Chen L, Cances J, Commandre J, Cuoci A, Pierucci S, et al. Biomasspyrolysis: Kinetic modelling and experimental validation under hightemperature and flash heating rate conditions. J Anal Appl Pyrol2009;85:260–7.

[2] Papadikis K, Gerhauser H, Bridgwater AV, Gu S. CFD modelling of the fastpyrolysis of an in-flight cellulosic particle subjected to convective heattransfer. Biomass Bioenergy 2009;33:97–107.

[3] Mellin P, Zhang Q, Kantarelis E, Yang W. An Euler–Euler approach to modelingbiomass fast pyrolysis in fluidized-bed reactors – focusing on the gas phase.Appl Therm Eng 2013;58:344–53.

[4] Kantarelis E, Yang W, Blasiak W. Production of liquid feedstock from biomassvia steam pyrolysis in a fluidised bed reactor. Energ Fuel 2013;27:4748–59.

[5] Ranzi E, Faravelli T, Frassoldati A, Migliavacca G, Pierucci S, Sommariva S.Chemical kinetics of biomass pyrolysis. Energy Fuel 2008;22:4292–300 .

[6] Blondeau J, Jeanmart H. Biomass pyrolysis at high temperatures: prediction of gaseous species yields from an anisotropic particle. Biomass Bioenergy2012;41:107–21.

[7] ANSYS FLUENT 14.5, Theory Guide; ANSYS, Inc., 2012.[8] Calonaci M, Grana G, Hemings EB, Bozzano G, Dente M, Ranzi E.

Comprehensive kinetic modeling study of bio-oil formation from fastpyrolysis of biomass. Energy Fuel 2010;24:5727–34.

[9] Dufour A, Ouartassi B, Bounaceur R, Zoulalian A. Modelling intra-particlephenomena of biomass pyrolysis. Chem Eng Res Des 2011;89:2136–46.

[10] Fagbemi L, Khezami L, Capart R. Pyrolysis products from different biomasses:application to the thermal cracking of tar. Appl Energy 2001;69:293–306 .

[11] Hoekstra E, Westerhof RJM, Brilman W, et al. Heterogenous and homogenousreactions of pyrolysis vapors from pine wood. AIChE J 2012;58:2830–42 .

[12] Park WC, Atreya A, Baum HR. Experimental and theoretical investigation of heat and mass transfer processes during wood pyrolysis. Combust Flame2010;157:481–94.

[13] Gunn DJ. Transfer of heat or mass to particles in fixed and fluidized beds. Int JHeat Mass Transfer 1978;21:467–76.

[14] Mickley HS, Fairbanks DF. Mechanism of heat transfer to fluidised beds. AIChE J 1955;1:374–84.

[15] Papadikis K, Gu S, Bridgwater AV. CFD modelling of the fast pyrolysis of biomass in fluidised bed reactors. Part B: heat, momentum and mass transportin bubbling fluidised beds. Chem Eng Sci 2008;64:1036–45.

[16] Ranz WE, Marshall WR. Evaporation from drops: Part I. Chem Eng Prog1952;48:141–6.

[17] Ranz WE, Marshall WR. Evaporation from drops: Part II. Chem Eng Prog1952;48:173–80.

[18] Syamlal M, Rogers W, O’Brien TJ. MFIX documentation: theoryguide. Morgantown, WV: U.S. Department of Energy; 1993 .

[19] Morsi SA, Alexander AJ. An investigation of particle trajectories in two-phaseflow systems. J Fluid Mech 1972;55:193–208.

[20] Syamlal M. The particle–particle drag term in a multiparticle model of fluidization. VA: Springfield; 1987.

[21] Lun CKK, Savage SB, Jeffrey DJ, Chepurnity N. Kinetic theories for granularflow: inelastic particles in couette flow and slightly inelastic particles in ageneral flow field. J Fluid Mech 1984;140:223–56.

[22] Gidaspow D, Bezburuah R, Ding J. Hydrodynamics of circulating fluidized beds,kinetic theory approach. In: Paper presented at: proceedings of the 7thengineering foundation conference on fluidization, 1992.

[23] Brown PN, Byrne GD, Hindmarsh AC. VODE: a variable-coefficient ODE solver.

Siam J Sci Stat Comp 1989;10:1038–51.[24] Alexander T. Water reuse in paper mills: measurements and control problems

in biological treatment, licentiate thesis. Lund Institute of Technology; 2003 .[25] Boroson ML, Howard JB, Longwell JP, Peters WA. Product yields and kinetics

from the vapor phase cracking of wood pyrolysis tars. AIChE J 1989;35:120–8 .[26] Morf P, Hasler P, Nussbaumer T. Mechanisms and kinetics of homogeneous

secondary reactions of tar from continuous pyrolysis of wood chips. Fuel2002;81:843–53.

[27] Zhang X, Yang W, Blasiak W. Thermal decomposition mechanism of levoglucosan during cellulose pyrolysis. J Anal Appl Pyrol 2012;96:110–9 .

[28] Faix O, Jakab E, Till F, Székely T. Study on low mass thermal degradationproducts of milled wood lignins by thermogravimetry-mass-spectrometry.Wood Sci Technol 1988;22:323–34.

[29] Patwardhan PR, Satrio JA, Brown RC, Shanks BH. Product distribution from fastpyrolysis of glucose-based carbohydrates. J Anal Appl Pyrol 2009;86:323–30.

[30] Lin YC, Cho J, Tompsett GA, Westmoreland PR, Huber GW. Pyrolysis kineticsand mechanism of cellulose. J Phys Chem C 2009;113:20097–107.

[31] Shen DK, Gu S, Bridgwater AV. Study on the pyrolytic behaviour of xylan-basedhemicellulose using TG–FTIR and Py–GC–FTIR. J Anal Appl Pyrol

2010;87:199–206.[32] Papadikis K, Gu S, Bridgwater AV. CFD modelling of the fast pyrolysis of

biomass in fluidised bed reactors: modelling the impact of biomass shrinkage.Chem Eng J 2009;149:417–27.


http://refhub.elsevier.com/S0016-2361(13)00838-7/h0005
























































































































































http://-/?-

http://-/?-

http://-/?-

http://-/?-

1-s2.0-s0016236113008387-main (1)

Documents