You are on page 1of 11

Subscriber access provided by UNIV DE BURGOS

Article

Bed Agglomeration Characteristics during


Fluidized Bed Combustion of Biomass Fuels
Marcus hman, Anders Nordin, Bengt-Johan Skrifvars, Rainer Backman, and Mikko Hupa
Energy Fuels, 2000, 14 (1), 169-178 DOI: 10.1021/ef990107b Publication Date (Web): 17 January 2000
Downloaded from http://pubs.acs.org on February 26, 2009

More About This Article


Additional resources and features associated with this article are available within the HTML version:

Supporting Information
Links to the 8 articles that cite this article, as of the time of this article download
Access to high resolution figures
Links to articles and content related to this article
Copyright permission to reproduce figures and/or text from this article

Energy & Fuels is published by the American Chemical Society. 1155 Sixteenth Street
N.W., Washington, DC 20036

Energy & Fuels 2000, 14, 169-178

169

Bed Agglomeration Characteristics during Fluidized Bed


Combustion of Biomass Fuels
Marcus O
hman* and Anders Nordin
Energy Technology Centre, Department of Chemistry, Inorganic Chemistry, Ume University,
P.O. Box 726, S-941 28 Pite, Sweden

Bengt-Johan Skrifvars, Rainer Backman, and Mikko Hupa


Process Chemistry Group, University, FIN-20500 Turku, Finland
Received June 2, 1999

The in-bed behavior of ash-forming elements in fluidized bed combustion (FBC) of different
biomass fuels was examined by SEM/EDS analysis of samples collected during controlled
agglomeration test runs. Eight fuels were chosen for the test. To cover the variations in biomass
characteristics and to represent as many combinations of ash-forming elements in biomass fuels
as possible, the selection was based on a principal-component analysis of some 300 biomass fuels,
with respect to ash-forming elements. The fuels were then combusted in a bench-scale fluidized
bed reactor (5 kW), and their specific agglomeration temperatures were determined. Bed samples
were collected throughout the tests, and coatings and necks formed were characterized by SEM/
EDS analyses. On the basis of their compositions, the corresponding melting behaviors were
determined, using data extracted from phase diagrams. The bench-scale reactor bed samples
were finally compared with bed samples collected from biomass-fired full-scale fluidized bed
boilers. In all the analyzed samples, the bed particles were coated with a relatively homogeneous
ash layer. The compositions of these coatings were most commonly constricted to the ternary
system K2O-CaO-SiO2. Sulfur and chlorine were further found not to participate in the
agglomeration mechanism. The estimated melting behavior of the bed coating generally correlated
well with the measured agglomeration temperature, determined in the 5 kW bench-scale fluidized
bed reactor. Thus, the results indicate that partial melting of the coating of the bed particles
would be directly responsible for the agglomeration.

Introduction
The fate of ash-forming elements during combustion
of biomass fuels is important for the efficiency and
availability of biomass-fired boilers. Ash-related operating problems, such as slagging and fouling, have been
reported extensively in the literature for most conventional combustion technologies. Due to the inherent
advantages of low process temperatures, isothermal
operating conditions, and fuel flexibility, fluidized bed
technology has been found to be the most suitable
approach to converting a wide range of biomass fuels
into energy. With increasing operational experience of
fluidized bed boilers, however, bed agglomeration has,
more than occasionally, been reported as a major
problem.1-3 Bed agglomeration may, in the worst cases,
result in total defluidization of the bed, resulting in an
unscheduled plant shut down. Recent studies4-7 have
further indicated that certain new biomass fuels, such
* Corresponding author.
(1) Skrifvars, B.-J.; Hupa, M.; Hiltunen, M. Ind. Eng. Chem. Res.
1992, 31, 1026-1030.
(2) Dawson, M. R.; Brown, R. C. Fuel 1992, 71, 585-592.
(3) Salour, D.; Jenkins, B. M.; Vafei, M.; Kayhaian, M. Biomass
Bioenergy 1993, 4, 117-133.
(4) Nordin, A. Fuel 1995, 74, 615-622.
(5) Viktoren, A. Thermal Engineering Research Foundation, Report
no. 416, 1991.

Figure 1. Scores (fuel abbreviations) and loading elements


(within rectangles) for the two significant principal components
for 300 solid fuels and the studied fuel samples (within circles).
Plot marks (abbreviations) for the fuel types are as follows:
wood (W); bark (B); wood residue/logging debris (Wr); barley
(b); reed canary grass (c1); reed canary grass - delayed harvest
(c2); Lucerne (l); rape (r); wheat straw (we); peat (P); coal (C);
bagasse (ba); cane trash (le); timothy (t); municipal solid waste
(M); refuse-derived fuels (RDF); salix (s).

as different types of energy crop, would be especially


problematic. Several authors8-11 have identified differ-

10.1021/ef990107b CCC: $19.00 2000 American Chemical Society


Published on Web 01/17/2000

170

O
hman et al.

Energy & Fuels, Vol. 14, No. 1, 2000


Table 1. Fuel and Bed Material Characteristics

dry substance
asha
Nab
Kb
Cab
Mgb
Alb
Feb
Sib
Sb
Clb
Pb
a

wheat
straw

wood

peat

90.3
5.9
0.596
20.7
6.90
1.73
0.385
0.361
18.2
2.88
4.92
2.52

92.0
0.45
2.80
3.91
15.3
3.03
1.82
0.964
6.63
2.22
2.22
1.65

46.8
5.4
0.629
1.41
14.8
0.435
2.58
17.1
16.4
7.41
0.370
0.946

cane
trash

wood
residue

reed canary
grass

bark

RDF

93.5
5.5
0.222
13.3
4.79
2.17
2.70
1.61
20.6
1.8

53.3
3.2
0.603
5.89
21.7
1.79
1.77
1.54
10.3
1.25
0.313
1.35

90.5
5.7
0.907
2.96
3.92
0.763
0.761
0.600
36.0
1.75
0.526
0.911

90.6
3.0
1.16
6.36
27.6
2.35
1.18
2.73
6.84
1
0.333
1.24

61.3
16.4
3.04
2.53
12.4
1.10
10.4
1.24
17.3
0.976
2.32

1.00

bed
material

0.0297
0.0497
0.0879
0.0778
0.0957
0.0860
46.3
<0.00523

Dry basis, wt %. b wt % of ash.

Figure 2. A typical agglomeration test run.

ent physical appearances of ash and bed-material in the


bed, either a coating of the bed material by another
material, or ash particles mixed with bed particles.
Previous attempts to explain the mechanisms of bed
agglomeration have included viscous flow (silicate melt)
sintering,8,12,13 reactive liquid (salt melt) sintering,8,14,15
and chemical reaction sintering.16 Although they all may
occur, a precise and quantitative description of bed
(6) Baxter, L. L.; Miles, T. R.; Miles, T. R.; Jenkins, B. M.; Milne,
T. A.; Dayton, D. C.; Bryers, R. W.; Oden, L. L. Sandia report SAND968587*UC-1301, 1996.
(7) Thambimuthu, K. V.; Clift, R. Proc. Fluidization VI, 547-554.
(8) Nordin, A.; O
hman, M.; Skrifvars, B.-J.; Hupa, M. In Applications
of Advanced Technology to Ash-related Problems in Boilers. Baxter,
L., DeSollar, R., Eds.; Plenum Press: New York, 1996, 353-366.
(9) Skrifvars, B.-J.; Sfiris, G.; Backman, R.; Widegren-Dafgrd, K.;
Hupa, M. Energy Fuel 1997, 11, 843-848.
(10) Manzoori, A. R. Thesis, University of Adelaide, Australia, 1990.
(11) Latva-Somppi, J.; Kurkela, J.; Tapper, U.; Kauppinen, E. I.;
Jokiniemi, J. K.; Johanson, B. Proceedings of the International Conference on Ash Behavior Control in Energy Conversion Systems, Pacifico
Yokohama, Japan, 1998, pp 119-126.
(12) Lin, W.; Krusholm, G.; Johansen, K. D.; Musahl, E.; Bank, L.
Proc. Fluidized Bed Combust. 1997, 2, 831-837.
(13) Ergudenler, A.; Ghaly, E. Biomass Bioenergy 1993, 4, 135-147.
(14) Manzoori, A. R.; Agerwal, P. K. Fuel 1993, 72, 1069-1075.
(15) Manzoori, A. R.; Agerwal, P. K. Fuel 1992, 71, 513-522.
(16) Skrifvars, B. J.; Anthony, E. J.; Hupa, M. Proc. Fluidized Bed
Combust. 1997, 2, 819-821.

agglomeration processes, during fluidized bed combustion of biomass fuels, has not yet been presented. This
inspired us to more closely study the characteristics of
the bed particles during actual agglomeration processes.
Our main interest in the present work was to elucidate the mechanism of bed agglomeration during combustion of different biomass fuels, particularly with
respect to the formation of a possible sticky coating on
the bed particles that could be responsible for this
process. Information was gathered by repeated sampling
during controlled agglomeration processing of several
representative biomass fuels, as well as analysis of
corresponding bed samples from full-scale trials.
Experimental Section
Fuels and Bed Material Used. Eight different types of
biomass fuels were used in the study. The fuels were chosen
based on a principal component analysis 17 (PCA) of a compilation of about 300 different samples of Nordic biomass fuels,18
classified with respect to elements important for ash formation.
The PCA resulted in two significant components, where the
(17) Wold, S. Technometrics 1978, 20, 397.
(18) Nordin, A. Biomass Bioenergy 1994, 6, 339-347.

Bed Agglomeration during Fluidized Bed Combustion

Energy & Fuels, Vol. 14, No. 1, 2000 171

Figure 3. Back scattered electron image of a polished cross section of a bed sample (wood) from the 18 MWth bed boiler (to the
right) and from the 5 kW bench-scale rigg (to the left).
Table 2. Bulk Bed Characteristics of Bed Samples Prior to the External Heating Phase
wheat straw
Na
K
Ca
Mg
Al
Fe
Si
S
Cl
P

0.074a
1.7
0.50
0.15
0.16
0.087
45
0.16
0.06
0.18

45b
36
30
31
90
53
96
24
5.3
30

wood
0.17a
1.6
3.1
0.56
0.27
0.14
43
0.15
0.01
0.096

25b
160
85
71
50
46
94
28
1.9
24

peat
0.082a
0.21
0.43
0.04
0.28
0.60
46
0.16
<0.01d
0.039

46b
54
12
23
39
14
96
9.1
<11d
17

cane trash
0.067a
0.80
0.62
0.13
0.37
0.37
45
0.06
0.02
0.061

82b
25
51
22
51
79
93
14
c
25

wood residue
0.037a
0.42
0.52
0.088
0.17
0.37
45
0.01
<0.01d
0.052

46b
79
28
38
71
170
94
9.7
<39d
45

reed canary
grass
0.21a
0.54
0.52
0.11
0.26
0.13
46
0.06
0.04
0.12

86b
72
52
44
96
60
87
14
32
55

bark
0.096a
1.1
1.8
0.22
0.24
0.22
44
0.08
<0.01d
0.14

RDF
32b
69
28
45
65
30
99
34
<13d
47

0.51a
0.34
0.98
0.16
0.65
0.27
45
0.05
0.04
0.092

72b
55
34
49
27
76
98
23
7.7
c

a wt % of bed sample. b % retained in bed of total input (total input ) fuel-ash + bed mtrl). c Uncertain values because of low concn in
fuel. d Uncertain values because of low concn in bulk bed.

statistical significance (32+22%) was determined by cross validation.17 From the PCA, eight reference fuels were chosen to
represent most of the variation in the contents of ash-forming
elements in normal biomass fuels. The results from the global
PCA of the 300 samples, including the eight reference fuels
(circled) used in the present study, are illustrated in Figure
1. This plot displays the directions of the maximum variation
and co-variation in the multivariate space. As a result of the
large number of samples, each point in the plot may represent
several samples. The corresponding loading plot, i.e., variable
plot (denoted by squares), is superimposed on the score plot.
The Figure shows that the fuel samples often separate into
different groups depending on the variation in the relative
amounts of different ash-forming elements. Fuels which contain relatively large amounts of alkali metals, chlorine, and
phosphorus are found in the upper right quadrant, whereas
fuels enriched in silicon are located at the upper left. Bark,
wood, and salix, with relatively large alkaline earth metal
contents, are found in the lower right quadrant, whereas peat
samples, with relatively large amounts of aluminum, iron, and
sulfur, are grouped in the lower left part of the Figure. The
characteristics of the eight reference fuels are also given in
Table 1.
The fuels were further pelletized to a diameter of 6-8 mm
and a length of 5-15 mm, and then used for the controlled
bed agglomeration tests.19 The tendency to cause bed agglomeration in a normal sand bed was tested by determination
of the characteristic agglomeration temperature. The bed
material used during the experiments was normal quartz sand
(see Table 1), initially sieved to a size between 200 and 250
m. The sand contained more than 98% SiO2 and only small
amounts of mineral impurities. The same type of quartz sand
is used by a large number of FBC operators in Sweden.
Controlled Fluidized Bed Agglomeration Tests. The
controlled fluidized bed agglomeration method has previously

been more fully described by O


hman and Nordin (1998),19 and
only a brief description is given here. The bench-scale reactor
(5 kW), is constructed from stainless steel, being 2 m high,
100 mm and 200 mm in bed and freeboard diameters,
respectively. To obtain isothermal conditions in the bed, and
to minimize the significant influence of cold walls in such a
small-scale unit, the reactor is equipped with electrical wall
heating elements, equalizing the wall and bed temperatures.
The agglomeration tests were initiated by loading of the bed
with a certain ash-to-bed material ratio, under normal FBC
conditions. The excess oxygen concentration was controlled to
6%dry. A fluidization velocity of four times the minimum
fluidization velocity was used, and the bed temperature was
maintained at 760 C for all fuels except wheat straw. To avoid
agglomeration during the ashing procedure for this fuel, a bed
temperature of 650 C was used. At an ash amount corresponding to 6 wt % ash in the bed, the fuel feeding was stopped
and the operation was switched to external heating. Several
initial runs have shown that only 1.5% of ash in the bed is
sufficient for agglomeration to occur. The bed was then heated,
at a rate of 3 C/min, to the point where it agglomerated. To
maintain a combustion atmosphere in the bed during the
external heating phase, propane was mixed with the primary
air in a chamber prior to the air distributor.
The onset of bed agglomeration was determined by monitoring differential pressures and temperatures in the bed. The
detection of initial bed particle cohesion was facilitated by PCA
by considering all bed-related variables (three temperatures
and four differential pressures) simultaneously. A typical
illustration of fluctuations in temperatures and differential bed
pressures versus time in a controlled bed agglomeration test
run is shown in Figure 2. The final defluidization can clearly
be seen here as a drop in the bed pressure.
Bed Samples and Bed Analyses. Throughout the experimental runs, representative samples of the bed material were

172
Energy & Fuels, Vol. 14, No. 1, 2000
O
hman et al.

Figure 4. Example of SEM/EDS elemental mapping for one bed sample (wood) from the bench-scale tests.

Bed Agglomeration during Fluidized Bed Combustion

Energy & Fuels, Vol. 14, No. 1, 2000 173

Figure 5. Mole percentage (%) of elements, except oxygen, of different ash components in the bulk bed.
collected by an air-cooled suction probe, equipped with a
cyclone separator. These samples, as well as the final agglomerated beds, were subjected to elemental analysis by
inductively coupled plasma atomic emission spectrometry
(ICP-AES) and wet chemical methods. Both bed samples and
agglomerates were also characterized with scanning electron
microscopy (SEM) combined with energy-dispersive X-ray
analysis (EDS). The samples for the SEM/EDS analysis were
mounted in epoxy, cut by a diamond saw, and polished, and
the resulting cross-sections were carefully examined by SEM/
EDS spot and elemental mapping methods.
(19) O
hman, M.; Nordin, A. Energy Fuels 1998, 12, 90-94.

In addition, bed material samples from two full-scale


biomass fired boilers were characterized with SEM/EDS. Both
plants, an 18 MWth hot water boiler and a 125 MWth combined
heat and power plant, fired wood residues at bed temperatures
of approximately 850 C. Details about these boilers, as well
as firing conditions and other tests performed during the bed
samplings, can be found in the literature.20,21
To avoid contributions from the bed material, special care
was taken not to collect EDS data from spots close to the
interface between coatings and particles. On the basis of
semiquantitative SEM/EDS analyses of coatings on the bed
particles, the melting behavior of the coating material was
finally estimated by extracting corresponding data from phase

174

Energy & Fuels, Vol. 14, No. 1, 2000

O
hman et al.

Figure 6. Manually performed SEM/EDS spot analyses of cross-sectioned bed samples. The analyses were always made on the
coating of the bed particle or the neck formed between two particles.
diagrams.22 The resulting melting behavior was then compared
with the determined fuel specific agglomeration temperature.

Results and Discussion


The SEM examinations of the bed material samples
showed that a coating with a thickness of 10-50 m
was formed around all bed particles. An example of this
coating can clearly be viewed from the SEM backscatter
micrographs shown in Figure 3. To the right, a micrograph of a bed sample from the 18 MWth CFB boiler is
shown; to the left, a similar image is shown for a bed

sample collected from the 5 kW bench-scale reactor


during the controlled bed agglomeration test of wood.
The coatings can be distinguished on the bed particles
as a lighter periphery surrounding the darker SiO2
cores. The black backgrounds seen in the micrographs
are the epoxy in which the bed samples were mounted.
Figure 4 is an example of the SEM/EDS element
mapping analysis for the same bench-scale bed sample
as in Figure 3. A light area indicates a higher concentration of the element in question than a dark area.
Corresponding element maps for all the other fuels

Bed Agglomeration during Fluidized Bed Combustion

Energy & Fuels, Vol. 14, No. 1, 2000 175

Figure 7. Comparison between fuel ash and bed particle coating characteristics.

showed that all coatings consisted of different ratios of


Si, K, Ca, Fe, Al, and Na, and thus different silicates.
The results of the elemental analyses of the bulk bed
samples are shown in Table 2 (column 1 for each fuel).
As expected, the element silicon dominates the samples
composition. Although a major fraction of the ashforming elements introduced by the fuel was retained

in the bed, it was depleted by some elements, especially


sulfur and chlorine (see Table 2, column 2 for each fuel).
The total concentrations of these elements in the bulk
bed were low (see Figure 5), which also seems to be the
case for the full-scale boilers.9,20,21 This was further
supported by a careful study of ash formation during
waste sludge and bark incineration.23 The significant

176

Energy & Fuels, Vol. 14, No. 1, 2000

O
hman et al.

Figure 8. K2O-CaO-SiO2 ternary diagram with some solidus temperatures (extracted from Morey et al.22) and the compositions
of the different bed particle coatings. Plot marks (abbreviations) for the different bed samples are the following: wood (W); bark
(B); wood residue (R); wheat straw (w); cane trash (l). Circles indicate component compositions of intermediate phase. The
components are indicated as follows: K2O (100), CaO (010), SiO2 (001).

vaporization and transport of K, Na, Cl, and S from the


bed was determined by impactor sampling and elemental analysis for both full-scale FBC and corresponding
runs in the bench-scale reactor. The vaporization during
combustion in the bench-scale unit was also found to
increase linearly with increased bed temperature. In
addition, no vaporization was found during the external
heating phase, even when the temperature was increased to 960 C. Thus, the coating characteristics seem
to be preserved during the external heating period and
the procedure is only an effective and reliable means to
determine the coating specific agglomeration temperature.
To quantify the elemental amounts found in the
coatings, a large number of EDS spot analyses were
performed on each bed sample and the results are
(20) Skrifvars, B.-J.; Backman, R.; Hupa, M.; Sfiris, G.; byhammar,
T.; Lyngfelt, A. Fuel 1998, 77, 65-70.
(21) Makdessi, R.; Johnson, J.; Hupa, M.; Skrifvars, B.-J.; Lauren,
T.; Backman, R. Thermal Engineering Research Foundation, Report
no. 621, 1997.
(22) Morey, G. W.; Kracek, F. C.; Bowen, N. L. J. Soc. Glass Technol.
1930, 14, 158.
(23) Latva-Somppi, J.; Kauppinen, E. I.; Kurkela, J.; Tapper, U.;
O
hman, M.; Nordin, A.; Johanson, B. Combust. Sci. Technol. 1999,
134, 433-456.

summarized in Figure 6. Each bar represents the result


of one spot analysis taken from the coating of a bed
particle or the neck formed between two bed particles.
In all coatings, silicon proved to be the dominating
element. Other elements most frequently detected were
potassium and calcium, but in significantly varying
proportions. For the peat bed coatings, iron was also
found as a major element and for the RDF and reed
canary grass, the bed coatings further contained a
significant fraction of aluminum. For the RDF bed
sample, sodium was also found in moderate amounts.
Further, by comparing the results from the analyses
of the coatings with corresponding fuel ash elemental
data (see Figure 7), it can be seen that the distribution
of the elements differs significantly. The implication of
this comparison is that a standard ash analysis of the
fuel may not provide a very reliable prediction of the
bed agglomeration tendency since the elemental content
of the actual coating may differ considerably.
The results of the SEM/EDS analyses indicate that
the layers covering the bed particles are homogeneous,
but the elemental distributions in the coatings vary
significantly between bed samples from different fuels.

Bed Agglomeration during Fluidized Bed Combustion

Energy & Fuels, Vol. 14, No. 1, 2000 177

Figure 9. Melting behavior of different spot characteristics of bed coatings from different bed samples (solid lines). The initial
bed agglomeration temperatures, determined with the controlled bed agglomeration test, are represented by the broken lines.

However, the outermost layer of the coating may be


composed of a heterogeneous mixture of different particles. With time, however, homogenization seems to
take place. This seems to be in agreement with the
recent SEM studies of agglomerates from burning forest
residue in a 35 MWth CFB boiler.24 Samples collected
when firing wood, bark, wood residues, cane trash, and
wheat straw show that the overall compositional distributions of the major fraction of the bed particle
coatings are mainly limited (>90% of the material) to
the ternary system K2O-CaO-SiO2.22 Figure 8 shows
the K2O-CaO-SiO2 ternary diagram with some solidus
(initial melting) temperatures and the compositions of
the bed material coatings from using different fuels. The
chemical compositions of these samples are mainly
restricted to the SiO2 rich corner in Figure 8, and
silicates with these compositions have a first melting
temperature of 720 C, while a small addition of calcium
will shift this value to roughly 1080 C.
Previous results8,19,25 have indicated that the chemical
characteristics, and thereby the melting behaviors of the
coatings, are very important for the bed agglomeration
process. If the coating has a high enough fraction of
molten phase, it will cause bed agglomeration, and, in
the most severe cases, defluidization. For silicate melts,
the viscosity of the melt has to be taken into consideration to determine the fraction needed for stickiness.26
It was therefore of interest to determine the melting
(24) Kauppinen, E. I., et al. LIEKKI 2 Tech. Rev. 1993-1998 1998,
639-690.
(25) Skrifvars, B.-J.; O
hman, M.; Nordin, A.; Hupa, M. Energy Fuels
1999, 13, 359-363.
(26) Skrifvars, B.-J.; Hupa, M.; Backman, R.; Hiltunen, M. Fuel
1994, 73 (2), 171-176.

behaviors of the different bed coatings, and compare the


melting temperatures with the corresponding specific
agglomeration temperatures.
Owing to a lack of thermodynamic data for several
intermediate phases in the system K2O-CaO-SiO2,
thermodynamic multicomponent, multiphase equilibrium calculations could not be used to accurately determine the melting behaviors. Instead, the evaluation was
performed by extracting melting behavior data from the
phase diagram K2O-CaO-SiO2.22 Only coatings that
had elemental compositions (>90%) within the composition triangles in Figure 8 were included in the comparison. The resulting fractions of melt (solid lines) versus
temperature are shown in Figure 9 together with the
determined initial agglomeration temperature (broken
lines), determined by the controlled fluidized bed agglomeration method.
The melting behavior is very sensitive to the relative
amounts of potassium and calcium in the sample.
Coatings with a relatively high fraction of potassium
and a lower fraction of calcium (wheat straw and wood)
will contain a very large amount of melt at temperatures
below 900 C. On the other hand, coatings with a
relatively high portion of calcium and a lesser portion
of potassium, i.e., samples from bark, cane trash, and
wood residue, will not contain a large amount of melt
until temperatures well above 900 C. The specific
agglomeration temperature agrees fairly well with the
temperatures where the melting behavior evaluation
indicates a high fraction of melt in the coating. The
results therefore seem to confirm the hypothesis that
melt formation is responsible for agglomeration. For
wood residues, however, a significantly higher (100 C)

178

O
hman et al.

Energy & Fuels, Vol. 14, No. 1, 2000

Figure 10. Illustration of the important chemical subprocesses of the bed agglomeration mechanism.

initial melting temperature than agglomeration temperature was obtained, possibly because of the external
minerals introduced during processing of the fuel.
Thus, the following important chemical subprocesses
of bed agglomeration are suggested (Figure 10):
(1) Ash deposition on the bed material is probably
dominated by a combination of (i) attachment of small
particles to the bed particle surfaces; (ii) condensation
of gaseous alkali species (KCl, KOH, K2SO4, K) on bed
particles; and (iii) chemical reaction of the gaseous alkali
on the surfaces.
(2) As the continuous deposition on the bed particles
proceeds, the inner layer of the coating is probably
homogenized and strengthened via sintering. In the
present study, alkali sulfates and chlorides are not
found in the homogeneous layer of the coating and thus
do not participate in the agglomeration process. The
characteristics and thickness of the coatings produced
during the 8-30 h run in the present work seem to
agree with the experiences from several previous fullscale tests.9,20,21,27
(3) The melting behavior of the homogeneous silicate
layer seems to control the adhesive forces, which are
(27) Latva-Somppi, J. Thesis, Technical Research Centre of Finland,
1998.

responsible for the final temperature-controlled agglomeration process.


Conclusions
The following conclusions could be drawn from the
present work:
The elements found in the coatings were mainly fuel
related. However, further work is needed to elucidate
any potential interactions between coatings and bed
materials.
Bed particles were found to be coated with a
relatively homogeneous ash layer.
The melting of the coating material was identified
to be the reason for formation of a sticky layer on the
bed particles, causing bed agglomeration and defluidization.
The elemental characteristics and the corresponding
melting behavior estimations of the bed particle coatings
suggest that silicate melts are responsible for the
agglomeration.
Acknowledgment. Financial support from the Swedish National Board for Industrial and Technical Development (NUTEK) and the LIEKKI Combustion Research Program in Finland is gratefully acknowledged.
EF990107B

You might also like