A model of wood flash pyrolysis in fluidized bed reactor

Document Sample
A model of wood flash pyrolysis in fluidized bed reactor Powered By Docstoc
					                             Renewable Energy 30 (2005) 377–392

      A model of wood flash pyrolysis in fluidized
                    bed reactor
                 Zhongyang Luo Ã, Shurong Wang, Kefa Cen
     Institute for Thermal Power Engineering, Zhejiang University, Hangzhou, Zhejiang 310027, China
                          Received 13 December 2002; accepted 23 March 2004


   With a view of exploiting renewable biomass energy as a highly efficient and clean energy,
liquid fuel from biomass pyrolysis, called bio-oil, is expected to play a major role in future
energy supply. At present, fluidized bed technology appears to have maximum potential in
producing high-quality bio-oil. A model of wood pyrolysis in a fluidized bed reactor has
been developed. The effect of main operation parameters on wood pyrolysis product distri-
bution was well simulated. The model shows that reaction temperature plays a major impor-
tant role in wood pyrolysis. And a good agreement between experimental and theoretical
results was obtained. It was shown that particles less than 500 lm could achieve a high heat-
ing-up rate to meet flash pyrolysis demand.
# 2004 Published by Elsevier Ltd.

Keywords: Wood; Pyrolysis; Model; Fluidized bed reactor

1. Introduction

   With concerns over energy shortage and CO2 emission, biomass is now being
considered as an inexhaustible and clean energy resource all over the world. Being
the world’s second largest country of CO2 emission after the USA, China’s
recoverable fossil fuel reserves have a CO2 emission potential of some 225 Gt,
which would all be released to the atmosphere by 2040 [1]. This emission may
cause a disruption of the climate and ecological balance. At the same time, biomass
is abundant in China, and accounts for about 17% of the nation’s primary
energy structure. Of late, thermochemical conversion is considered to be the most

     Corresponding author. Tel.: +86-571-8795-2440; fax: +86-571-8795-1616.
     E-mail address: zyluo@mail.hz.zj.cn (Z. Luo).

0960-1481/$ - see front matter # 2004 Published by Elsevier Ltd.
378                  Z. Luo et al. / Renewable Energy 30 (2005) 377–392


  A       pre-exponential factor, sÀ1
  B       viscous effusion degree, m2
  Bi      Biot number
  C       heat capacity, J/(kg K)
  d       particle size, m
  D       diffusivity, m2/s
  E       activation energy, J/mol
  Fo      Fourier number
  g       acceleration of gravity, m/s2
  H       volume enthalpy, J/m3
  J       mass transfer rate, kg/(m2 s)
  k       reaction rate coefficient, sÀ1
  DL      particle distance between two steps, m
  m       mass, kg
  M       molecular weight
  Nu      Nusselt number
  p       pressure, Pa
  Pr      Prandtl number
  q       reaction heat, J/kg
  Q       reaction heat rate per volume, J/(m3 s)
  r       particle radius, m
  R       gas constant, 8.314 J/(mol K)
  Re      Reynolds number
  S       cross-sectional area, m2
  T       reaction temperature, K
  u       flow velocity, m/s
  v       volumetric flow rate, m3/s
  V       volume, m3
  z       reactor height, m
  a       heat transfer coefficient, W/(m2 K)
  q       density, kg/m3
  e       voidage of inner particle
  s       time, s
  k       thermal conductivity, W/(m K)
  r       5:67 Â 10À8 W/(m2 K4)
  n       emissivity of particle surface
  w       drag coefficient
  l       kinetic viscosity, kg/(m s)

  0       initial time
                     Z. Luo et al. / Renewable Energy 30 (2005) 377–392            379

  c         char
  d         heat convection
  e         reaction end
  f         molecular diffusion
  g         gas
  G         overall gas out particle
  i         corresponding pyrolysis product
  l         fluidizing gas
  m         dense bed
  n         viscous flow
  o         bio-oil
  r         radiation heat transfer
  R         particle boundary
  s         solid particle
  v         volatile
  w         wood
  x         suspension bed

common and convenient method for the biomass being utilized as an energy
resource. This includes combustion, gasification, liquefaction and carbonization. In
all these processes, liquefaction is expected to play a major role in future biomass
utilization. Liquefaction can increase volumetric heat content, decrease transpor-
tation costs, and increase the use of liquid fuels in more equipment. There are sev-
eral ways available for biomass thermochemical liquefaction. However, flash
pyrolysis appears to be the most promising method of producing bio-oil. Fluidized
bed technology is one of the most efficient and economic methods of actualizing
flash pyrolysis, as it offers high heating rate, rapid devolatilization, convenient char
collection, and re-utilization. A fluidized bed flash pyrolysis reactor, operating
at atmospheric and nitrogen atmosphere, has been successfully developed at the
Zhejiang University in China. Sawdust, wood, and rice straw were fed continu-
ously at a maximum rate of about 3 kg/h directly into the reactor by way of a
unique solid feeder. At optimum condition, a maximum bio-oil yield of up to 60%
and 50%, respectively, was obtained [2].
   Besides experimental research, numerical simulation is another extremely impor-
tant tool for reactor design and experimental data interpretation. Therefore, a
model was developed in this paper to simulate biomass pyrolysis behavior in our
fluidized bed reactor.

2. Theory

  In the model, wood particle is considered to be spherical in shape. When the
wood particle is exposed to medium temperature when being fed into the dense bed
380                   Z. Luo et al. / Renewable Energy 30 (2005) 377–392

in a fluidized bed reactor, heating occurs immediately by conduction, radiation,
and convection. As the temperature of the particle rises rapidly from the outer sur-
face to core, pyrolysis decomposition reaction is likely to take place at the front at
first, and then progress inwards until it reaches the core. The generated vapors dif-
fuse through pore and out of particle to the bulk of the gas. Meanwhile, homo-
geneous secondary cracking occurs inside the porous particle and in the bulk of
gaseous phase. Along with pyrolysis, particle shrinkage takes place, which may
influence the transport phenomenon. Generally, to maximize bio-oil production,
the temperature of the dense bed is to be set at an optimized and medium value.
The temperature of the suspension bed is to be lower than that of the dense bed so
as to restrain homogeneous secondary cracking. It is reasonable to develop corre-
sponding equations for wood pyrolysis both is dense bed and suspension bed.
  On the basis of the above discussion, the model is developed with the following

1. The wood particle maintains the shape of a sphere during pyrolysis.
2. Fresh wood particle has uniform distributions in porosity, density, conductivity,
   and specific heat capacity.
3. Reaction temperature variation of dense bed and suspension bed is neglected, since
   the temperature is accurately adjusted to a predetermined value in pyrolysis progress.
4. Bulk parameters are only functions of an axial location, but are uniform along
   the radial direction in a fluidized bed reactor.
5. Wood or char particles have far larger specific heat capacity than volatile tem-
   peratures released during particle thermo-decomposition. Therefore, the volatile
   temperature leaving the particle is considered to be readily heated to the tem-
   perature of wood or char that it flows. At the same time, the solid temperature
   is also constant, while the volatile temperature flows through the wood particle.
6. Nitrogen can be heated up to reaction temperature in a very short distance less
   than 5 mm from air-distributor. Therefore, the wood particle is exposed to hot
   nitrogen atmosphere immediately, as soon as it is fed into the dense bed, with
   the feeding entrance to the reactor being 10 mm apart from the gas-distributor.
7. Wood particle has no free water as a pre-drying procedure is adopted before the

3. Model equations

3.1. Chemical reaction kinetics

  The raw material of wood is considered homogeneous, while the reaction pro-
ducts are grouped into three main components: gas, bio-oil and char. Therefore, a
two-stage and semi-global model is adopted to describe the wood thermal
decomposition. Wood undergoes thermal degradation to gas, bio-oil, and char by
way of primary reactions (k1, k2, k3); then the bio-oil may be changed either to
                    Z. Luo et al. / Renewable Energy 30 (2005) 377–392          381

gas by cracking (k4), or to char by repolymerization (k5) in the following second-
ary reactions. The Primary reactions are represented as first order in wood mass
and having an Arrhenius type. The Secondary reactions are assumed to occur in
gas phase within the pores of solid material and bulk gas phase in reactor, which
are mainly related to bio-oil concentration.
  For gas and bio-oil, the corresponding kinetic expressions are associated with
mass transfer phenomenon, which would be described better in mass diffusion
equation. the kinetic equations for wood and char can be written as:

         ¼ Àðk1 þ k2 þ k3 Þmw                                                   ð1Þ

         ¼ kl mw þ k5 mo                                                        ð2Þ

     kl ¼ Al expðÀEl =RTÞ                                                       ð3Þ

     e ¼ Vv =V ¼ ðVg þ Vo Þ=V                                                   ð4Þ

where mw ¼ qw V , mc ¼ qc V , mo ¼ eqo V and mg ¼ eqg V . qw and qc are the appar-
ent solid-phase densities that are correlated with particle volume, while qo and qg
are determined on the basis of gaseous volume. Thus, Eqs. (1) and (2) could be
expressed in other ways:

     @ðqw V Þ
              ¼ Àðk1 þ k2 þ k3 Þqw V                                            ð5Þ

     @ðqc V Þ
              ¼ Àk1 qw V þ k5 eqo V                                             ð6Þ

3.2. Enthalpy balance

  The Wood particle is heated up to rapidly to the reaction temperature when
exposed to hot atmosphere in the dense bed by way of intensive convection, con-
duction, and radiation. Along with wood pyrolysis progress, endothermic and exo-
thermic reactions may take place. The solid particle undergoes shrinkage with
devolatilization Simultaneously, which also influences enthalpy balance. All these
can be described as:

     DH                   X          H @V
        ¼ divðk grad TÞ þ       Qi À                                            ð7Þ
     Ds                   i¼1;5
                                     V @s
382                               Z. Luo et al. / Renewable Energy 30 (2005) 377–392

  It is obvious that the enthalpy flow of solid substances is zero; thus, the enthalpy
balance equation for particle pyrolysis in dense bed can be written in detail as:
       @                                                                            @
           ½ðqw Cw þ qc Cc þ eqo Co þ eqg Cg ÞTŠ þ u ½ðqo Co þ qg Cg ÞTŠ
      @s                                                                           @r
      |fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl} |fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}
                                   ð1Þ                                                          ð2Þ
            1 @                @T
          ¼ 2             r2 k            þ ðk1 q1 þ k2 q2 þ k3 q3 Þqw þ ðk4 q4 þ k5 q5 Þeqo
           r @r                 @r          |fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}
           |fflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflffl}                                           ð4Þ
                ðqw Cw þ qc Cc þ eqo Co þ eqg Cg ÞT @V
              À                                                                                                                   ð8Þ
                                                V                                     @T

where (1) represents enthalpy change rate, (2) describes enthalpy flow, (3) figures
thermal conduction flux, (4) refers to heat effect during particle thermal-degradati-
on, and (5) reflects the effect of shrinkage on enthalpy.
  The related initial and boundary conditions are listed as:
      Ts¼0 ¼ T0 ;
                                                                      0       r       rR
      @r r¼0
      Àk         ¼ ðad þ ar ÞðTm À TR Þ
         @r       r¼rR

where ad and ar correspond to heat transfer coefficients of convection and radi-
ation, respectively. So far, no generally accepted equation is available for convec-
tion coefficient of the bulk gaseous phase to the fine particle in the dense
bed. Agarwal et al. [3] provided an equation (Eq. (10)) to calculate the convection
coefficient in the dense bed. However, if the Reynolds number is very small, the
Nusselt number will be found below the lower limit of 2 when using Eq. (10),
which means that at this time, convection in the dense bed would be weaker than
that in packed bed. At this time, Scott and Piskorz [4] recommended that Nu value
could be chosen as 2 or 4.

      Nu ¼ 0:03Re1:3                                       Re < 100
      Nu ¼ 2:0 þ 0:6Re1=2 Pr1=3                            Re > 100
  The Radiation coefficient of ar in Eq. (9) can be calculated by the following
            À 2    2
     ar ¼ nr Tm þ TR ðTm À TR Þ                                          ð11Þ
  When the wood particle is fed into the dense bed, the char layer will immediately
appear to surround particles due to thermo-degradation. The radiation of char is
very close to that of black-body and an emissivity of 1 could be selected [5].
  The Above-mentioned enthalpy balance equation could also be applied in the
suspension bed. However, heat transfer in the suspension bed is really weaker than
                            Z. Luo et al. / Renewable Energy 30 (2005) 377–392                       383

that in the dense bed, where the Biot number is only about 10À3–10À2. Thus, a par-
ticle can be considered isothermal, whose temperature can be written as:
     T À Tx
             ¼ eðÀBiFoÞ                                                                             ð12Þ
     Tm À Tx

3.3. Component diffusion

      @                                                                              eq @V
          ðeqo Þ þ divðqo  þ J fo Þ ¼ k2 qw À ðk4 þ k5 Þeqo À o
                               u                                                                   ð13Þ
     |fflfflfflffl{zfflfflfflffl}  |fflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflffl} |fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl} |fflfflfflffl{zfflfflfflffl}
                                                                                      V ffl @s   ffl
                               ð2Þ                            ð3Þ
         ð1Þ                                                                             ð4Þ

     @                                                eqg @V
        ðeqg Þ þ divðqg  þ J fg Þ ¼ k3 qw þ k4 eqo À                                               ð14Þ
     @s                                                V @s
where (2) represents mass convection and diffusion, (3) refers to chemical reaction,
while (4) materializes the effect of volume shrinkage on component diffusion.
  The parameter u in above equations can be expressed as:
                       Bo pMv
      Jnv ¼ qv u ¼ À            rp                                             ð15Þ
where B is a coefficient relevant to porosity [6]:

     B ¼ 4 Â 10À11 Â e3 =ð1 À eÞ2                                                                   ð16Þ
  Gaseous products are assumed to accord with the ideal gas law:
           qv RT
     p¼                                                                                             ð17Þ
  Thus, u can be deduced as:
                B @ðr2 pÞ
     u¼À                                                                                            ð18Þ
               lv r2 @r
  Volatile diffusion flux is expressed in the Fick law:
                 Dv @ðr2 qfv Þ
     Jfv ¼ À                                                                                        ð19Þ
                 r2    @r
  The related initial and boundary conditions are listed as:
      qo ¼ qg ¼ 0;      0                       r     R       s¼0
                    ¼0                                                                             ð20Þ
       @r r¼0 @r r¼0

3.4. Particle volume shrinkage

  Particle volume shrinkage will occur during thermo-degradation. Chan et al. [5]
and Lee et al. [7] pointed out that marked shrinkage would take place during
pyrolysis, especially in the case of larger particles. Here, particle voidage is
384                     Z. Luo et al. / Renewable Energy 30 (2005) 377–392

assumed to vary with the apparent densities of char and wood, which is written as:
                 qs À qw0
      e ¼ e0 þ             ðee À e0 Þ                                            ð21Þ
                 qce À qw0
  In the same way, particle volume is calculated as follows:
                   ms À mw0
      V ¼ V0 þ               ðVe À V0 Þ                                          ð22Þ
                   mce À mw0

3.5. Residence time of gaseous products and solid particle

  Along with reactor height from inlet to outlet during wood pyrolysis, the particle
undergoes thermo-degradation, and massive volatile temperature is released from
the inner particle to bulk gaseous phase through the solid surface continuously.
Therefore, the bulk gas flow will become larger and larger due to additive volatile
temperature released from the wood particle, which will reduce gas residence time.
Scott and Piskorz [4] pointed out that this phenomenon would lead to a decrease
of gaseous product residence time by at least 15%. Here, a plug-flow model is suit-
able for describing the phenomenon:
              e  S  DL
      DsG ¼                                                                      ð23Þ
                ðvl þ vv Þ
  Up to now, no generally accepted equation for particle residence time is avail-
able. Conti et al. [8] indicated that the solid biomass was unlikely to participate in
aggregation or back-mixing but was directly elutriated. Thus, the wood particle is
assumed to move in the direction of gas flow, with only a slip between gas flow [9]:
      dus   wqg                 q s À qg    us dqs
          ¼       ðuG À us Þ2 À          gÀ                                      ð24Þ
      ds    ds qs                   qs      qs ds
         < 9:689ReÀ0:78 ð1 þ 0:147Re0:82 Þ 0:1 Re 5
      w ¼ 9:689ReÀ0:78 ð1 þ 0:227Re0:55 Þ 5 Re 40                                ð25Þ
           9:689ReÀ0:78 ð1 þ 0:838Re0:82 Þ 40 Re 400

3.6. Secondary cracking in bulk gaseous phase

   When the volatile temperatere is released from particle into bulk gaseous phase,
it still undergoes secondary cracking due to the hot atmosphere in the reactor.
However, the repolymerization of volatile to char could be ignored because of the
not so high reaction temperature in the reactor. Boroson et al. [10] drew a con-
clusion that reforming probability in the gaseous phase was too small to be con-
sidered even at high temperatures about 500–800 C. Meanwhile, the average
freedom course between volatile molecules was so large that mutual collision
hardly took place, thus reducing reforming probability. Therefore, secondary
                         Z. Luo et al. / Renewable Energy 30 (2005) 377–392                    385

cracking of the volatile molecule could be described as follows:

       @ðqo Þ      @ 2 qo uG @qo
              ¼ Do       À       À k6 qo                                                     ð26Þ
        @s          @z2    e @z

       @ðqg Þ      @ 2 qg uG @qg
              ¼ Dg       À       þ k6 q o                                                    ð27Þ
        @s         @z2     e @z

4. Parameter selection

   Many researchers have obtained various chemical reaction kinetic parameters of
wood pyrolysis. The cited kinetic parameters applied in our model are listed in
Table 1.
   In the following sections, the wood pyrolysis behavior in our developed fluidized
bed reactor (Fig. 1) is simulated in detail. The fluidized bed reactor has a diameter
of 80 mm and a height of 1200 mm, which is operated within temperatures
between 450 and 700 C. Nitrogen flow at about 3.5 N m3/h is used to fluidize

sand in the bed, and sweep the volatile temperature quickly off the hot zone. The
Wood particle has 660 kg/m3 density, 2.4 kJ/kg K specific heat capacity, and
0.3 W/m K conduction coefficient [11]. Meanwhile, the char has 0.84 kJ/kg K in
specific heat capacity, and 0.1 W/m K in conduction coefficient [7]. Drummond
and Drummond [12] indicated that the primary reaction was slightly endothermic
when the temperature was about 500–550 C, while the secondary reaction tended
to be exothermic. Generally, the endothermic stage mainly takes place during
levoglucosan generation, while that of exothermic stage in the secondary reaction.
The wood devolatilization reaction heat might range from À418 to 418 kJ/kg,
while a reaction heat about 42 kJ/kg would be liberated in the following secondary
cracking [5,13].

Table 1
Kinetic parameters selection
Pre-exponential         Value (sÀ1)          Activation          Value        References
factor                                       energy              (kJ/mol)
A1                      1:08 Â 107           E1                  121          Chan et al. [5]
A2 (sÀ1)                 2:0 Â 108           E2                  133          Chan et al. [5]
A3 (sÀ1)                 1:3 Â 108           E3                  140          Chan et al. [5]
A4, A6 (sÀ1)            3:09 Â 106           E4, E6              108          Boroson et al. [10]
A5 (sÀ1)                1:48 Â 106           E5                  144          Chan et al. [5]
386                  Z. Luo et al. / Renewable Energy 30 (2005) 377–392

               Fig. 1. Fluidized bed reactor system for flash pyrolysis of biomass.

5. Results and discussion

5.1. Effect of dense bed temperature

   There is no doubt that reaction temperature plays an active role in wood pyrol-
ysis, as it does in a fluidized bed reactor. Fig. 2 shows the effect of dense bed
temperature on pyrolysis product distribution along with reactor height. With the
dense bed temperature increase, it is wood that will mostly undergo thermo-
degradation in the dense bed. At a dense bed temperature of 500 C, about 80% of
wood was decomposed to gas, char, and bio-oil, While it was almost completed
when the temperature rose up to 600 C. Therefore, higher temperatures will pro-
mote wood thermo-decomposition. However, the reaction of bio-oil secondary
cracking to gas will become more violent when the temperature increases beyond
some value. Though wood pyrolysis takes place more rapidly and completely at a
dense bed temperature of 700 C, all bio-oils were decomposed to gas for second-
ary cracking. Thus, a temperature of 500 C is more suitable for maximizing bio-
oil production, which will not only efficiently promote wood pyrolysis rate but also
restrain secondary cracking to a certain extent. it is also obvious that gas pro-
duction will increase with temperature all along. This is the reason why gasification
is often operated at higher temperatures.
                         Z. Luo et al. / Renewable Energy 30 (2005) 377–392                       387

Fig. 2. Wood pyrolysis product distribution vs. dense bed temperature (suspension bed temperature:
450 C; feed rate: 1 kg/h; particle size: 200 lm; black dot: dividing point). (a) dense bed temperature
   v                                  v                                 v
450 C; (b) dense bed temperature 500 C; (c) dense bed temperature 600 C; (d) dense bed temperature
700 C.

5.2. Effect of suspension bed temperature

  When compared with the dense bed temperature, the suspension bed tempera-
ture has minor effects on wood pyrolysis behavior. At an optimum dense bed
temperature of about 500 C, only a small fraction of wood undergoes thermal
degradation within suspension bed. However, bio-oil secondary cracking may take
place violently within the suspension bed if the temperature is higher. Therefore,
with a view of restraining bio-oil secondary cracking, it is important to control the
suspension bed temperature at a reasonable value.
  As shown in Fig. 3, secondary cracking of bio-oil becomes more acute with an
increase of suspension bed temperature, also accompanied by an increase of gas
production. In their detailed research on secondary cracking, Stiles and Kandiyoti
[14] indicated that more than 30% bio-oil would take part in secondary cracking
when volatile temperatures stayed at 600 C for 3.5 s than for Q.25s. Fig. 3(b) has
shown that nearly half of the bio-oil was transformed to gas caused by secondary
cracking at 600 C. However, secondary cracking could be neglected when the
388                      Z. Luo et al. / Renewable Energy 30 (2005) 377–392

Fig. 3. Wood pyrolysis product distribution vs. suspension bed temperature (dense bed temperature:
500 C; feed rate: 1 kg/h; particle size: 200 lm; black dot: dividing point). (a) suspension bed tempera-
         v                                       v
ture: 550 C; (b) suspension bed temperature: 600 C.

temperature is below 450 C according to Fig. 2(b). In accordance with the result
proposed by Samolada and Vasalos [15]. However, it is not reasonable to set the
temperature of the suspension bed to a value lower than the setting point of bio-
oil, in order to avoid the bio-oil being unexpectedly pre-cooled before entering into
the quencher. Generally, according to the experimental results in our fluidized bed
reactor, it is more suitable to set suspension bed temperature within the range of
about 400–450 C.
5.3. Effect of particle size

   Generally, particle size has mutual effects on pyrolysis behavior. On the one
hand, the fine particle will be carried by gas flow more easily, which will lead to a
short residence time and incomplete decomposition. On the other hand, bio-oil
production will be lower if adopting excessive large wood particles because of slow
heating up. Therefore, a wood particle of a moderate size should be more reason-
able for bio-oil production. As shown in Fig. 4, wood particle size is often set in
the order of microns so as to meet flash pyrolysis and pneumatic transportation.
Within this range, particle size has little influence on product distribution. After
all, the non-sensitivity of product distribution on particle size will lower pretreat-
ment complexity of the raw material.
5.4. Effect of feed rate

   Until now, the effect of feed rate on pyrolysis behavior had not been reported.
The adjustment of feed rate will change gas flow and the volatile temperature in
the reactors which will influence wood decomposition and bio-oil secondary crack-
ing. Fig. 5 indicates that bio-oil production increases with feed rate. If wood could
be completely decomposed, more volatile temperature will be yielded at the same
interval when the feed rate becomes higher, which will enhance gas flow. Therefore,
volatile residence at the hot atmosphere will be shortened and secondary cracking
                        Z. Luo et al. / Renewable Energy 30 (2005) 377–392                      389

Fig. 4. Wood pyrolysis product distribution vs. wood particle size (dense bed temperature; 550 C;
suspension bed temperature: 450 C; feed rate: 1 kg/h; black dot: dividing point). (a) particle size:
100 lm; (b) particle size: 300 lm.

Fig. 5. Wood pyrolysis product distribution vs. feed rate (dense bed temperature: 600 C; suspension
bed temperature: 450 C; particle size: 200 lm; black dot: dividing point). (a) feed rate: 0.5 Kg/h;
(b) feed rate: 5 Kg/h.

will be restrained to some extent. Unlike reactor height which influences wood
pyrolysis behavior mostly in suspension bed, feed rate plays an active role both in
dense bed and in suspension bed. Because the temperature of the dense bed is gen-
erally higher than that of the suspension bed, feed rate will restrain the bio-oil sec-
ondary cracking more efficiently. However, it is not advantageous to increase the
feed rate when incomplete wood decomposition occurs at lower temperatures.
5.5. Verification of model

   Based on the above analysis, it is clear that the reaction temperature plays the
most important role in pyrolysis behavior. Fig. 6 shows the experimental result car-
ried out on our fluidized bed reactor, which is compared with model calculation.
As shown in Fig. 6, the modeling result is in good consonance with the
390                       Z. Luo et al. / Renewable Energy 30 (2005) 377–392

Fig. 6. Experimental and theoretical results of ultimate product distribution (particle size: 200 lm; feed
rate: 1 kg/h; black dots: experimental data). (a) bio-oil production; (b) gas production.

experimental results, while a dense bed temperature of 500 C, and lower suspen-
sion bed temperature are optimum for bio-oil production.

6. Study of heating-up rate

   During pyrolysis process, no matter what the reaction temperature is, the wood
particle should go through proper temperature areas. Rapid heating-up will make
the particle pass low temperature area more quickly in order to reduce charring
and maximize bio-oil production. However, the heating-up rate is affected by many
operational parameters, such as particle characteristics and heat flux. Therefore, if
the reactor type is pre-determined, temperature and particle size will be the two
most important factors influencing heating-up rate.
   Kothari and Antal [16] suggested that flash pyrolysis could be divided into two
steps in succession: heating-up stage and volatile release stage. At heating-up stage,
the weight loss of particle could be neglected and the particle is rapidly heated. As
the particle temperature reaches up to a predetermined value, the volatile tempera-
ture begins to be released out of the particle, which is called the volatile release
stage. In fact, the above two stages could not be completely separated. However,
when compared with the reaction period and the residence time of particle, the
time for the particle being heated up to 95% of reaction temperature is too short to
consider pyrolysis process, especially in the case of high flux. Therefore, wood
decomposition could be ignored during the calculation of heating-up rate in flui-
dized bed reactor. Only in the situation higher reaction temperature or larger par-
ticle is it necessary to the above two stages consider simultaneously. In this paper,
particle core, where heating-up rate is the slowest, is selected as the basis to study
the effect of temperature and particle size on the heating-up rate.
   As shown in Table 2, the heating-up rate of a fine particle could achieve up to
an order of 104 C/s, which is also verified in other research. In some cases, fine

particles might attain a very high heating-up rate, up to the order of 105 C/s

[8,17]. However, the actual heating-up rate depends not only on heat transfer but
                           Z. Luo et al. / Renewable Energy 30 (2005) 377–392                    391

Table 2
Heating-up rate vs. particle size and temperature
                   v                                v                           v
Size           450 C                         550 C                       700 C
               UT95%          UT350          UT95%         UT350         UT95%       UT350
                v              v              v             v             v           v
               ( C/s)         ( C/s)         ( C/s)        ( C/s)        ( C/s)      ( C/s)
100            2:2 Â 104      3:4 Â 104      2:9 Â 104     5:3 Â 104     4:1 Â 104   8:1 Â 104
300            6:3 Â 102      1:0 Â 103      8:4 Â 102     1:6 Â 103     1:2 Â 103   2:5 Â 103
500            2:1 Â 102      3:4 Â 102      2:8 Â 102     5:5 Â 102     4:0 Â 102   8:6 Â 102

also on pyrolysis progress, which implies that the actual heating-up rate is smaller
than the values listed in Table 2.
   Generally, the heating-up is often calculated on the basis of 95% of reaction tem-
perature. However, the essential effect of heating-up rate on flash pyrolysis is to
shorten particle residence time at low temperatures. This will decrease the charring
opportunity and increase the bio-oil production. Thus, it is more reasonable to sel-
ect a moderate temperature value but not a reaction temperature as the baseline
for calculating the heating-up rate. Here, the temperature of 350 C is selected as
the calculation base, at which mostly volatile temperature is released out of the
particle. The related result is also listed in Table 2, which shows that all particles
less than 500 lm are located in flash pyrolysis range. When one Compares the
results in Table 2 with those shown in the above figures, it is clear that the pyrol-
ysis rate is far slower than the heating-up rate. Therefore, it is more reasonable to
modify the name ‘‘flash pyrolysis’’ as ‘‘pyrolysis at flash heating-up’’.

7. Conclusions

   An integrated model, combined reaction kinetics with mass and heat transfer,
was proposed to predict pyrolysis behavior in a fluidized bed reactor. The Model-
ing results showed that the reaction temperature played the most important role in
wood pyrolysis behavior. A moderate dense bed temperature of 500 C, and lower
suspension bed temperature were more suitable for optimizing bio-oil production,
at which the wood particle could undergo a complete decomposition and bio-oil
secondary cracking be well restrained.
   Besides this, particle size and feed rate have mutual effects on pyrolysis behavior.
The Fine particle will enhance rapid heating-up, but may lead to incomplete par-
ticle decomposition at lower temperatures. In the case of higher dense bed tem-
perature, volatile secondary cracking will be efficiently restrained when the feed
rate is increased. However, an excessive large feed rate might cause incomplete
wood particle decomposition.
   Rapid heating-up will make particle passing low temperature area more quickly
so as to reduce charring and maximize bio-oil production. The study of heating-up
rate showed that it is more suitable to choose a moderate temperature, and not
392                      Z. Luo et al. / Renewable Energy 30 (2005) 377–392

normal reaction temperature as the evaluation criterion. It was proved that parti-
cles less than 500 lm could meet the flash pyrolysis requirement.


  This work was supported by National Natural Science Foundation of China,
under Project Nos. 50025618, 29976039 and 50176046.

 [1] Bach W, Fiebig S. China’s key role in climate protection. Energy 1998;23(4):253–70.
 [2] Luo ZY, Wang SR, Dong LJ, Fang MX, Hong J, Gu YL, et al. Research on flash pyrolysis of
     biomass for bio-oil in a fluidized bed reactor. Reno: Proceedings of the 16th International Confer-
     ence on Fluidized Bed Combustion. 2001, p. 5.
 [3] Agarwal PK, Genetti WE, Lee YY. Models for devolatilization of coal particles in fluidized beds.
     Fuel 1984;63:1157–65.
 [4] Scott DS, Piskorz J. The continuous flash pyrolysis of biomass. The Canadian Journal of Chemical
     Engineering 1984;62:404–12.
 [5] Chan WR, Kelbon M, Krieger BB. Modelling and experimental verification of physical and
     chemical process during pyrolysis of a large biomass particle. Fuel 1985;64:1505–15.
 [6] Cen KF, Ni MJ, Luo ZY, Yan JH, Chi Y, Fang MX, et al. Theory, design and operation on
     circulating fluidized bed boilers. Beijing: Chinese Electric Power Press; 1998 [in Chinese].
 [7] Lee CK, Chaiken RF, Singer JM. Charring pyrolysis of wood in fires by laser simulation. 16th
     International Symposium on Combustion. 1976, p. 1459–70.
 [8] Conti L, Scano G, Boufala J. Bio-oils from arid land plants: flash pyrolysis of EUPHORBIA
     CHARACIAS bagasse. Biomass and Energy 1994;7(1–6):291–6.
 [9] Hastaoglu MA, Hassam MS. Application of a general gas-solid reaction model to flash pyrolysis of
     wood in a circulating fluidized bed. Fuel 1995;74(5):697–703.
[10] Boroson ML, Howard JB, Longwell JP, Peters WA. Product yields and kinetics from the vapor
     phase cracking of wood pyrolysis tars. AICHE Journal 1989;35(1):120–8.
[11] Dubianzhiren. Wood application. Shanghai: Shanghai Science and Technology Press; 1986 [Zhang
     QL, Zhang QS, Zhang SY, Trans.; in Chinese].
[12] Drummond AF, Drummond IW. Pyrolysis of sugar cane bagasse in a wire-mesh reactor. Industrial
     and Engineering Chemistry 1996;35:1263–8.
[13] Koufopanos CA, Papayannakos N, Maschio G, Lucchesi A. Modelling of the pyrolysis of biomass
     particles. Studies on kinetics, thermal and heat transfer effects. The Canadian Journal of Chemical
     Engineering 1991;69:907–15.
[14] Stiles HN, Kandiyoti R. Secondary reactions of flash pyrolysis tars measured in a fluidized bed
     pyrolysis reactor with some novel design features. Fuel 1989;68:275–82.
[15] Samolada MC, Vasalos IA. A kinetic approach to the flash pyrolysis of biomass in a fluidized bed
     reactor. Fuel 1991;70:883–9.
[16] Kothari V, Antal MJJ. Numerical studies of the flash pyrolysis of cellulose. Fuel 1985;64:1487–93.
[17] Scott DS, Piskorz J. The flash pyrolysis of aspen-poplar wood. The Canadian Journal of Chemical
     Engineering 1982;60:666–74.

Shared By:
Description: A model of wood flash pyrolysis in fluidized bed reactor