MODELLING FINE PARTICLE FORMATION AND ALKALI
METAL DEPOSITION IN BFB COMBUSTION
JORMA JOKINIEMI1,2 AND OLLI SIPPULA1
1
Fine Particle and Aerosol Technology Laboratory, Department of Environmental Science, University
of Kuopio, FI-70211 Kuopio, Finland
2
VTT, Technical Research Centre of Finland, Fine Particles, FI-02044 VTT, Espoo, Finland
jorma.jokiniemi@uku.fi, tel. +358405050668
ABSTRACT
Fluidised bed combustion has been found to be a suitable process for producing energy
from biomass fuels. Behaviour of inorganic ash-forming species plays a major role when
operational problems such as bed agglomeration, fouling and corrosion of the heat
exchangers are considered. The size distribution and the morphology of the fly ash are
of interest when ash deposit formation is studied [1]. Coarse ash particles impact on the
heat exchanger surfaces with a high probability, but often do not adhere on the surface.
Particle retention on the deposit becomes much more favourable if the deposit surface
or the surface of the impacting particle is formed of sticky material. Enrichment of alkali
sulphates and chlorides has been observed on superheater deposits. Alkali compounds
and their solutions have low melting temperatures, usually close to heat exchanger
operating temperatures. Condensing on the pre-existing fly ash particles or on the heat
exchanger tubes they form a low viscosity sticky layer that highly contributes to the
deposition rates of impacting supermicron particles. Fine fly ash particles have also
gained increasing interest in relation to health effects. Once emitted from the stack fine
particles are carried along with flue gas and have a high probability of penetrating into
the alveolar regions of lungs. These particles have typically high specific surface area,
they can be enriched of toxic species and they penetrate more easily through the particle
separation devices as compared to supermicron particles.
The KCAR (Kuopio Center for Aerosol Research) model was used to simulate the
behaviour of fly ash particles during combustion process. In KCAR simulation program
formation and growth of aerosol particles, alkali and trace metal chemistry and
particle/vapour deposition have been modelled and connection of different phenomena is
taken into account in numerical models [2]. The program includes models for gas phase
chemical equilibrium and surface reaction kinetics, heterogeneous and homogenous
condensation, agglomeration and models for particle and vapour deposition. The main
flow is treated as one dimensional plug flow. Deposition of particles is represented with
boundary layer theories. Condensation of vapours is estimated using numerical fits of
vapours pressures and mass transfer equations.
The results indicate that the fine ash mode is composed mainly of condensed K, Cl, S
and Na. These species form molten solutions of chlorides, sulphates hydroxides and
carbonates at low temperatures, which enhance deposition rates due to the sticky layer
of salt solutions formed either on the surface of large particles or on the heat exchanger
tubes.
Keywords: fluidised bed boilers, fine ash particles, fouling, fume, biomass fuels
INTRODUCTION
In co-combustion of biomass and waste the volatilized species are both organically and
inorganically bounded. In Fluidized Bed Combustion (FBC) the fuel and small amounts
of the bed material transforms to vapours and fly ash particles of various sizes and
compositions. In biomass combustion the volatilized fly ash forming elements mainly are
alkali metals (K, Na), sulphur, chlorine and various metals (e.g. Zn). Prior to
volatilization alkali and trace elements may form non-volatile compounds by chemical
reactions inside the fuel particle and thus decreasing the volatility of these species.
Alkali salt deposition is a serious problem limiting and restricting successful operation of
FB boilers. Deposit layers formed on superheater platens reduce heat transfer
effectiveness, contribute to corrosion and, at the worst case, can even plug gas
passages. Effective removing of the deposit layer by soot blowing is a key parameter for
continuous operation of the boiler and is mainly determined by the deposit layer growth
rate and hardening. To understand these phenomena and to minimise the adverse effects
caused by particle and vapour deposition, one must understand aerosol dynamics in the
process. Vapour phase species release, gas phase chemistry, fume and residual ash
particle formation are essential to consider when deposition is modelled. Most important
parameters affecting the particle flux to the heat transfer surface are particle size and
concentration. Flue gas flow velocity, flue gas temperature and temperature difference
between gas and heat transfer surface, as well as vapour phase speciation are also
important parameters affecting the growth rate of the deposit layer.
A numerical aerosol model KCAR (Kuopio Center for Aerosol Research Aerosol
Model), has been developed to help the understanding of fine ash particle formation and
deposition. The KCAR model describes alkali species gas phase chemistry, fume particle
formation, growth and particle as well as vapour deposition in combustion processes. In
this paper we have concentrated on particle and vapour deposition modelling in the
superheater section of the FB boiler. Particle deposition mechanisms for surfaces
parallel to flue gas flow are turbulent impaction and thermophoresis, while
corresponding mechanisms on the front side of a cylinder in cross flow are inertial
impaction and thermophoresis. A detailed description of inertial impaction was added to
KCAR, compared with the earlier papers [3]. In addition, in the present work the
condensed liquid phase has been considered as a solution phase, which is especially
important when calculating vapour pressures of condensing salt vapours. To illustrate
the capabilities of the model, we have calculated deposit layer growth rates in the heat
transfer sections of a Bubling Fluidised Bed Boiler.
AEROSOL DYNAMICS
To be able to understand particle formation in boilers, one needs to know the basics of
aerosol dynamics first. Generally 100 m is considered the maximum size of an aerosol
particle. Aerosol dynamics cover such phenomena as gas phase reactions of condensable
species, homogeneous and heterogeneous condensation of these species to form and
grow aerosol particles, agglomeration of particles and deposition of particles and
vapour. In the following we briefly describe these phenomena.
Homogeneous Condensation (nucleation)
In the furnace volatilised alkali vapours can become supersaturated due to flue gas
cooling or chemical reactions. Saturation ratio of a certain vapour is defined as the ratio
of its partial pressure to its equilibrium pressure. If the saturation ratio is larger than
one, the vapour is supersaturated. For example, if KOH reacts with SO2, gaseous
sodium sulphate is formed. Because of the very low equilibrium vapour pressure of
K2SO4(g), this vapour becomes supersaturated immediately after formation. Thermodynamically, supersaturation is not a favoured state, and thus the vapour starts to
condense. Condensation decreases the partial vapour pressure toward its equilibrium
value. If there are no other particles in the flue gas, when the supersaturation starts to
increase, vapour molecules stick together and form new aerosol particles. This process
is called homogeneous nucleation and it requires a critical supersaturation, which is
much higher than one. When this critical supersaturation level has been reached, tiny
aerosol particles are formed at a rate, which can be predicted with the use of thermodynamics and kinetic considerations.
Heterogeneous Condensation
If there are other particles, such as metal oxide seeds around, the released alkali and
trace element vapours start to condense on the surface of these particles before any new
particles can be formed by homogeneous nucleation. This is due to the fact that
condensation on surfaces starts at lower supersaturation ratios than homogeneous
nucleation. The growth rate of the particles can be solved directly from the heat and
mass transfer equations for single particles with the use of numerical methods.
Agglomeration
Agglomeration is a process where particles collide with each other and stick together.
Collision rate is determined by Brownian and turbulent diffusion. When submicron
particles collide, they always stick together. Particle size, chemical composition and
process conditions determine the properties of these agglomerated particles. If the
colliding particles are liquid they form a new spherical liquid droplet and in the other
extreme the colliding solid particles stay together by Van der Waals attraction. In reality
agglomerates formed in combustion processes tend to sinter together and form particles
with a complex morphology.
Modelling Aerosol Dynamics in the KCAR Code
In the KCAR code we simulate the combustion process using elemental volatilisation
rates and possible initial seed particle size distribution as input data. The change in the
particle size and chemical composition spectrum is due to the mechanisms of chemical
reactions, homogeneous nucleation, vapour condensation, agglomeration and
deposition. The whole process is described by the General Dynamic Equation (GDE)
e.g [5]. In combustion processes steady-state conditions can be assumed. Thus we solve
the aerosol GDE in 1-dimensional stationary form along the flow direction. Mass size
distributions of different species can be calculated by solving the condensed phase
species GDEs, where the particle size spectrum is divided into a number of grid points:
dm jk
dm jk
dx
dx
dm jk
gtp
dx
agg
Vdk Ad
m
u V jk
(1)
Here mjk is the mass concentration of species j at the kth grid point corresponding to
radius rk, respectively. The first term at the right (gtp) corresponds to particle formation
due to homogeneous nucleation and growth by condensation and chemical reactions.
The second term (agg) describes agglomeration and the third term the rate of particle
removal due to deposition on boundary surfaces. Vdk is the particle deposition velocity,
Ad is the deposition area and V is the axial volume step. Particle velocity (u) is
calculated by taking into account gas velocity, Stokes drag and gravitation.
The rate of change in the molar concentration of a gas phase species is given by the gas
phase species conservation equation :
dc j
dx
dc j
dx
form
dc j
dx
gtp
vvA d
c
ug V j
(2)
where cj is the gas phase molar concentration of species j. The first term (form) at the
right represents the formation rate of these species, which can be calculated from local
gas phase chemical equilibrium or from reaction kinetics. The second term (gtp)
describes the vapour depletion rate by nucleation, condensation and chemical reactions
and the third term represents the depletion rate due to vapour deposition by
condensation and chemical reactions on structures. Vv is the vapour deposition velocity
and Ad is the deposition area. Here time is related to the axial position through the gas
velocity (ug). Particle and vapour deposition models are described in reference [3].
RESULTS
Fine particle measurements were carried out at a 66 MW Forssa biomass-fueled CHP
bubbling fluidized-bed plant. The measurements were carried out upstream of the twofield ESP at the flue gas temperature of 130-150 °C. The particle mass size distribution
and mass concentration was determined with Berner-type low-pressure impactor (BLPI)
and number size distribution was determined with an electrical low-pressure impactor
(ELPI). The values of process parameters and gas composition were collected
continuously. Two different combinations of biomass fuels were studied 1) wood waste
(75%) and forest residue (25%) and 2) wood waste (75%) and chipboard (25%). The
wood waste consisted of wood chips, saw dust and bark. Sand was used as bed
material. Even though the aim was to keep the power plant conditions as stable as
possible during the measurements with the two fuel combinations, the load of the power
plant varied slightly from day to day and even during the measurements [6]. In the
KCAR calculations following elements, gas phase chemical species and condensible
species were included:
Elements: C, Cl, S, H, O, N, Na, K
Gaseous species: C, CO2, CO, Cl, Cl2, HCl, S, SO2, SO3, H2S, H2SO4, H, H2,
OH, H2O, O, O2, N2, Na, NaCl, Na2Cl2, NaOH, Na2SO4, K, KCl, K2Cl2, KOH,
K2SO4
Condensible species: Na2CO3, NaCl, Na2SO4, NaOH, K2CO3, KCl, K2SO4,
KOH and Inert Ash Particles.
Input was taken from the measured particle and elemental size distributions [6]. It
was first assumed that the fine particle mode species were volatilised from the
furnace. As part of these species deposit during their transport from the furnace to
ESP we iteratively increased the released amounts so that finally the calculated
and measured results before ESP were equal. This method has shown to be very
useful, because we do not need to measure the released fractions in the hot area of
the furnace, but instead it is enough to measure release before ESP. In the
following work, it will be also necessary to compare releases calculated in this way
to actual measurements in the furnace, to validate the model approach.
In figure 1 there are the geometry of the modelled boiler and gas, surface and
steam temperatures measured by the boiler operators.
According to the calculations the amount of chlorine has a clear effect on the deposition
rates of alkali metal compounds. Measurements indicate that the concentration of fine
particles is approximately doubled when firing fuel 2, which is the same result as
obtained from Fact chemical equilibrium calculations. In the case when there is more
chloride compounds (mainly KCl) present in the gas phase there is also more
condensation on large particles and since deposition efficiencies of supermicron particles
are approximately tenfold compared to submicron particles, deposition rates of alkali
chlorides are also increased.
Chlorine concentration of the fuel seems to affect also the mass size distributions
calculated with KCAR. When firing fuel 2 the relative amount of Na 2SO4 and K2SO4 in
large particles was increased compared to firing fuel 1. Impaction is clearly the most
important deposition mechanism in all cases.
In the following the calculated mass and species size distributions together with
measured size distributions are presented (figures 2 and 3). In the calculations the gas
phase species reactions with large particles was not considered as there is not reliable
data to model this effect. The calculated results indicate that condensation on large
particles was insignificant. As an uncertainty the effect of particle shape on condensation
rate was checked, but it had only a minor effect on condensation rates. Thus we may
assume that the species present in large particle mode may not have vaporized
completely or that the vaporised species have reacted with large particles. Also the
effect of re-entrained particles should be considered here.
In the next figures (4 a,b,c) the alkali speciation is presented as a function of boiler
location. The calculated speciation is in good agreement with measurements of SO2 and
fine particle mode chemistry. The concentration of HCl is overestimated due to the lack
of data for KCl sulphation kinetics.
The deposition of condensible species is dependent of vapor phase speciation as well as
fine and coarse particle mode concentrations. The dependence is complicated and is not
analysed in detail here, but one can clearly see that for higher Cl concentration in fuel
also the deposition is higher. As the deposition layer grows thicker also the surface
temperature of the deposit is increased. This case was simulated by decreasing the
temperature difference between gas and deposit surface to half of its value for clean
surface. It can be seen now (figures 5 and 6) that in the first superheater (6.4 m)
deposition of the condensible species is decreased dramatically as most of the chlorides
are in the gas phase and do not condense on superheater surfaces.
In the last figure the deposition velocities of particles as a function of their size at
different locations is presented. Particles below 1 micron (submicron) deposit by
thermophoresis and larger particles (supermicron) by inertial impaction (on the frontal
side of the cylinder) and turbulent eddy impaction (inside the superheaters)
DISCUSSION AND CONCLUSIONS
Thermodynamic equilibrium models give fairly reasonable results at higher temperatures
but at lower temperatures formation of most species suggested by calculations is
retarded by reaction kinetics. It seems likely that in many cases alkali metals condense as
chlorides and do not react to sulphates and carbonates at lower temperatures, as stated
by equilibrium (e.g. Valmari et al, [7]. Thus, kinetic limitations should be taken into
account in the calculations of fine fly ash formation.
The chlorine concentration of fuel seems to have a significant role in the behaviour of
alkali metals. First, chlorine contributes to the volatilisation of alkali metals, thus
increasing the amount of alkali compounds in gas phase. Second, alkali chlorides formed
at high temperatures are quite stable and gas-to-particle conversion occurs by
homogenous or heterogeneous condensation rather than by chemical reaction. This can
be seen from the fact that, when burning fuels containing excessive amounts of chlorine,
more alkali metals are found in submicron particles [8,9].
When burning fuels with low chlorine concentrations alkali metals form initially alkali
hydroxides which later form sulphates and carbonates. Although the reaction kinetics of
chlorides and hydroxides with sulphur oxides is not entirely known, it seems that
hydroxides react more readily with sulphur dioxide and trioxide forming in this case
alkali sulphates, which may be found both in submicron and supermicron particles. In
addition, sulphur reacts with calcium and potassium forming CaSO4 and K2Ca2(SO4)3 .
Formation of particles and deposition of alkali metal compounds are complicated
phenomena in which thermodynamics play an important role. The calculated fly ash and
deposit properties can be further processed by equilibrium calculations to get indication
on the formation of molten solutions (containing potassium and sodium chlorides,
sulphates and carbonates). If these solutions are formed either on heat exchanger tubes
or on the surfaces of large particles, they significantly enhance the deposition rates. The
other possibility to reduce fouling and corrosion problems is to use fuels with low
chlorine concentration, high S/Cl, S/K ja Si/K ratios or use additives to increase the
melting point.
In the prediction of the composition of the deposited layer and its stickiness we need to
consider the amount of different alkali species. Alkali species in the superheater region
are deposited by vapour condensation and by thermophoresis and thus the deposit
composition can not be predicted from equilibrium calculations only.
ACKNOWLEDGEMENTS
The authors thank TEKES, Foster Wheeler, Metso and Forssan Energia Oy for funding
this study.
REFERENCES
1. Lind, T., Hokkinen, J., Jokiniemi, Aurela, M., Hillamo, R. (2003) Electrostatic
Precipitator Collection Efficiency and Trace Element Emissions from Co-Combustion of
Biomass and Recovered Fuel in Fluidized Bed Combustion, Environ. Sci. & Technol. 37
No. 12 (2842-2846).
2. Jokiniemi, J., Pyykönen, J., Mikkanen, P. and Kauppinen, E. (1996) Modeling fume
formation and deposition in kraft recovery boilers. Tappi Journal Vol. 79, No. 7. pp.
171 - 180.
3. Eskola, A., Jokiniemi, J., Vakkilainen, E. and Lehtinen, K. (1998) Modelling alkali
salt deposition on kraft recovery boiler heat exchangers in the superheater section. 1998
TAPPI Proceedings International Chemical Recovery Conference, Volume 3, Tampa,
USA 1 - 4 June 1998. pp. 469 - 486.
4. Neville, M. and Sarofim, A. F. (1982). The Stratified Composition of Inorganic
Submicron Particles Produced during Coal Combustion. 19th Symp. (Int´l) on
Combustion, pp. 1441-1449.
5. Friedlander, S. K. (1977). Smoke Dust and Haze. John Wiley & Sons, New
York.
6. Kurkela, J., Latva-Somppi, J., Tapper, U., Kauppinen, E. I. and Jokiniemi, J. (1998)
Ash formation and deposition onto heat exchanger tubes during fluidized bed
combustion of wood-based fuels. In Proceedings of International Conference on Ash
Behavior Control in Energy Conversion Systems, Yokohama, Japan. pp. 110-118.
7. Valmari T., Lind T. M., Kauppinen E. I., Am. Chem Soc. Energy & Fuels 1999, 13,
2, 390-395.
8. Lind, T., Kauppinen, E.I., Hokkinen, J., Jokiniemi, J.K., Orjala, M., Aurela, M.,
Hillamo, R. (2006) Effect of Chlorine and Sulfur on Fine Particle Formation in PilotScale CFBC of Biomass. Energy Fuels 20, pp. 61–68.
9. Sippula, O., Lind, T., Jokiniemi, J. (2008) Effects of chlorine and sulphur on particle
formation in wood combustion performed in a laboratory scale reactor. Fuel 87, pp.
2425-2436.
in
480 °C
[9] 700 °C; 9.2 m
out
[8] 750 °C; 8.5 m
511 °C
[7] 750 °C; 7.9 m
in
481 °C
in
[6] 870 °C; 6.4 m
403 °C
[10] 700 °C; 12.3 m
[11] 500 °C; 14.9 m
[12] 500 °C; 15.8 m
[13] 440 °C; 17.7 m
[5] 880 °C; 4
m
out
469
[16] 415 °C; 27.2 m out
[15] 420 °C; 25.2 m
276
°C
in
274
out
275
in
276
[14] 430 °C; 21.7 m
12
[4] 900 °C; 2
m
[3] 900 °C; 1.5 m
[2] 950 °C; 1
m
[1] 900 °C; 0.5 m
[0] 950 °C; 0
m
KCAR Nodalization of the
Forssa BFB Boiler
Fig. 1. KCAR nodalisation of the BFB boiler.
in
25 °C
Steam
[17] 125 °C; 38.2 m
Mass Size Distributions
1,20
1,00
dM/dlogDp [g/Nm**3]
dM/dlogDp
0,80
Fuel#2 exp.
0,60
0,40
0,20
0,00
0,01
0,10
1,00
10,00
100,00
1000,00
Dp [µm ]
Fig. 2. Comparison between simulated and measured particle size distributions before the ESP.
Fine particle Mass Size Distributions
0,06
NACL
dM/dlogDp [g/Nm**3]
0,05
KCL
K2SO4
0,04
K-exp
Cl-exp
0,03
0,02
0,01
0,00
0,01
0,10
1,00
10,00
Dp [µm ]
Fig. 3. Comparison between simulated and measured fine particle size mode before the ESP.
100,00
3
Chlorine concentration [g(Cl)/Nm]
5,0E-05
4,5E-05
4,0E-05
3,5E-05
3,0E-05
K2Cl2(g)
HCl(g)
Cl2(g)
2,5E-05
KCl(g)
2,0E-05
1,5E-05
KCl(p)
1,0E-05
NaCl(g)
5,0E-06
NaCl(p )
0,0E+00
0
4,4 7,19 8,7 12,7 13,6 14,5 17 27,8 30,2 32,6 35 37,4 38,8 39,7
X[m]
3
Potassium concentration [g(K)/Nm ]
5,0E-05
4,5E-05
4,0E-05
3,5E-05
3,0E-05
2,5E-05
KOH(g)
K2Cl2(g)
K2SO4(p)
2,0E-05
1,5E-05
KCl(g)
1,0E-05
5,0E-06
KCl(p)
0,0E+00
0
4,4 7,19 8,7 12,7 13,6 14,5 17 27,8 30,2 32,6 35 37,4 38,8 39,7
X[m]
3
Sodium concentration [g(Na)/Nm]
1,0E-05
9,0E-06
8,0E-06
7,0E-06
6,0E-06
NaOH(g)
Na2Cl2(g)
Na2CO3(p)
5,0E-06
4,0E-06
Na2SO4(p)
3,0E-06
NaCl(g)
2,0E-06
NaCl(p)
1,0E-06
0,0E+00
0
4,4 7,19 8,7 12,7 13,6 14,5 17 27,8 30,2 32,6 35
37,4 38,8 39,7
X[m]
Fig 4 a, b and c. Cl, K and Na particle and vapour phase speciation for fuel 2.
K
Deposition growth rates in the superheater section
condensible species only
NA
O
0,14
S
CL
0,12
C
Vd[mm/d]
0,1
0,08
0,06
0,04
0,02
19
17
15
13
11
9
7
5
0
X[m]
Fig. 5. Deposition growth rates for fuel 2.
Depositions growth rate in the super heater section
condensible species only
K
0,1
NA
Vd[mm/d]
0,09
O
0,08
S
0,07
CL
C
0,06
0,05
0,04
0,03
0,02
0,01
0
5,5
6,5
7,5
8,5
9,5
10,5
11,5
X[m]
Fig. 6. Deposition growth rates for fuel 2 dirty.
12,5
13,5
14,5
15,5
16,5
17,5
Particle Deposition velocities at different locations
1,00E+01
SH2-inlet
SH2-inside
Evaporator-inlet
Vd[m/s]
1,00E+00
Eko-inside
1,00E-01
1,00E-02
1,00E-03
0,001
0,01
0,1
1
dpa [µm]
Fig. 7. Particle deposition as a function of particle size.
10
100
1000