Ash Fusion Temperatures For Biomass

November 1999
ECN-C-99-090
CONVERSION OF BIOMASS,{PRIVATE }
PREDICTION AND SOLUTION METHODS
FOR ASH AGGLOMERATION AND RELATED
PROBLEMS
Contract JOR3-95-0079
Final Report
1 March 1996 to 1 March 1999
A. van der Drift1)
A. Olsen2)
1)
ECN, Holland; 2) Ris, Denmark
Revisions
A
November 1999
B
Final version
Made by:
Approved:
A. van der Drift
J. Prij
Checked by:
Issued:
J. Beesteheerde
C.A.M. van der Klein
ECN Fuels Conversion & Environment
Preface
This report describes the results obtained in a project financed by the European Commission,
(contract number JOR3-95-0079 in the framework of the JOULE III programme), NOVEM
(Netherlands agency for energy and the environment, project 355196/1030), ECN and Ris.
Originally it has been a project of three partners: ECN (NL), Ris (DK) and Aalborg Boilers.
After the amendment of 31 August 1998, only ECN and Ris were left to meet the objectives. In
the same amendment, the duration of the project has been extended by 6 months.
The present report has been edited by A. van der Drift and A. Olsen, however many people have
been involved. Within ECN, Herman Bodenstaff, Joep van Doorn, Frank Ligthart, Rian Visser,
John Neeft and many more have contributed. Snnik Clausen, Torben D. Petersen, Jrgen
Bilde-Srensen, Jesper Vejl Carstensen have been working on the project for Ris. Also
Marcus hman and Christoffer Boman from ETC (S), Peder Stoholm from Danish Fluid Bed
Technology ApS (DK), Karin Poulsen from ReaTech (DK), B. J. Gerrits from TNO (NL),
Susanne Westborg and Henning Munk Jensen from dk-TEKNIK (DK), Jaana Latikainen
Luntama and Antero Moilanen from VTT (FIN) and Johnny Helbo Hansen, Peter Binderup
Hansen and Kate Wieck-Hansen from I/S MIDTKRAFT have contributed to the result of the
project (all, except MIDTKRAFT, acting as subcontractor from Ris).
Partners and addresses:
ECN (Energy Research Foundation)
contact: A. van der Drift
Email: vanderdrift@ecn.nl
tel: (31) 224-564515
Ris
contact: A. Olsen
Email: aksel.olsen@risoe.dk
tel: (45) 4677 4677
PO Box 1
NL-1755 ZG Petten
The Netherlands
PO Box 49
DK-4000 Roskilde
Denmark
Keywords
ash, biomass, agglomeration, sintering, defluidisation
ECN-C-99-090
CONTENTS
EXECUTIVE SUMMARY
1.
OBJECTIVES
2.
ACTIVITIES
3.
RESULTS 9
3.1 TASK 1: Information and material collection
3.2 TASK 2: Selection of biomass fuels, bed materials and additives
3.3 TASK 3: Description of the lab-scale apparatus and test procedures
3.3.1 Chemical laboratories at ECN and Ris (DK-TEKNIK) (a)
3.3.2 BFBG: bubbling fluidized bed gasification facility (5 kWth ) at ECN (b)
3.3.3 BFBC: bubbling fluidized bed combustion facility (5 kWth )
by Ris (ETC) (c)
3.3.4 CFBG: circulating fluidized bed gasification facility (500 kWth )
at ECN (d)
3.3.5 DTA/TGA: differential thermal analysis and thermo-gravimetric
analysis at ECN and Ris (e)
3.3.6 Compression strength measurement facility by Ris (TNO) (f)
3.3.7 Standard facilities for measuring standard ash melting temperatures
by ECN (NMi) and Ris (DK-TEKNIK) (g)
3.3.8 SEM and HTM (high temperature microscope) at Ris (h)
3.3.9 BFBC: bubbling fluidized bed combustion facility by Ris (VTT) (i)
3.3.10 Other facilities
3.4 TASK 4: Construction and modification of lab-scale fluidised bed
3.5 TASK 5: Development and description of an agglomeration model
3.6 TASK 6: Performance of lab-scale and bench-scale tests
3.6.1 Chemical laboratories at ECN and Ris (DK-TEKNIK) (a)
3.6.2 BFBG: bubbling fluidized bed gasification facility (5 kWth ) at ECN (b)
3.6.3 BFBC: bubbling fluidized bed combustion facility (5 kWth) by
Ris (ETC) (c)
3.6.4 CFBG: circulating fluidized bed gasification facility (500 kWth )
at ECN (d)
3.6.5 DTA/TGA: differential thermal analysis and thermo-gravimetric
analysis at ECN and Ris (e)
3.6.6 Compression strength measurement facility by Ris (TNO) (f)
3.6.7 Standard facilities for measuring standard ash melting temperatures
by ECN (NMi) and Ris ( DK-TEKNIK) (g)
3.6.8 SEM and HTM: high temperature microscope at Ris (h)
3.6.9 BFBC: bubbling fluidized bed combustion facility by Ris (VTT) (i)
3.7 TASK 7: Description of the total methodology for prediction of ash
agglomeration and related problems, and identification of potential solutions
3.7.1 Growing and harvesting
3.7.2 Fuel pre-treatment
3.7.3 Blending
3.7.4 Fluid dynamic control
3.7.5 Bed particle management
3.7.6 Temperature control
3.7.7 Other measures
3.8 TASK 8: Conclusive evaluation
9
11
12
12
12
15
16
17
18
19
19
20
21
22
22
24
24
28
32
36
39
43
44
45
46
46
47
48
48
49
50
52
53
55
4.
COMPARISON OF PLANNED ACTIVITIES AND WORK ACCOMPLISHED
58
5.
CONCLUSIONS
59
ECN-C-99-090
6.
REFERENCES
60
ANNEX A. A literature survey on agglomeration, mechanism, prediction and prevention

ANNEX B. Overview of prediction methods reported in literature
ANNEX C. Publications, conference presentations and patents
ECN-C-99-090
EXECUTIVE SUMMARY
When biomass is used as fuel for thermal conversion plants, minerals from the fuel can be
responsible for major problems. Generally, these problems are associated with the existence and
development of low melting compounds or eutectics, which form sticky layers. In a fluidised
bed, this can result in bed-agglomeration and defluidisation. This causes local high temperature,
which often accelerates the process. It ultimately can lead to a completely sintered bed content
with a glassy phase gluing the bed particles together and shut down of the plant.
The main objective of this project is to develop a methodology to predict ash/bed agglomeration
and sintering problems, to indicate related problems and, furthermore, to identify solution
methods to make different types of biomass streams more viable for energy production.
Within the present study, selected fuels are subjected to different existing methods together with
some new ones, in order to determine the agglomeration temperature. The selected fuels are
verge grass, Danish wheat straw (both stored dry and partly leached due to rainfall), sewage
sludge, cacao shells and willow as a reference. The methods used within the study are chemical
analysis of fuel and ashes, determination of standard ash melting temperatures, compression
strength measurements of the ash, DTA/TG analysis of the ash, SEM and ESEM (high
temperature environmental scanning microscopy), two different lab-scale bubbling fluidised bed
combustion facilities, a lab-scale bubbling fluidised bed gasifier and a circulating fluidised bed
gasifier. The lab-scale facilities have been used to test potential measures to reduce the problem
of agglomeration. These measures are the use of additives (kaolin, magnesite, dolomite, gibbsite
and sewage sludge) and non-quartz bed materials (alumina and mullite).
The work performed within the project has lead to the following results. Chemical analysis of
the fuel can give a first indication of whether there might be an agglomeration problem during
thermal conversion. In general a high K-content means an increased risk for agglomeration.
However, the K-content alone is not a good indicator. Also chlorine proved to be very
important.
From the methods used in the project, laboratory fluid bed agglomeration experiments seem to
give the most reliable information about conditions and temperatures where agglomeration takes
place. Contrary to methods like DTA, compression strength and ash melting temperatures, all
processes that might be relevant for agglomeration actually can occur during fluid bed
experiments: fuel-bed material interactions, volatilization and condensation, shear forces,
temperature homogeneity and accumulation. Standard tests have been developed where process
temperature is gradually increasing until agglomeration. These tests have been applied in the
project at three different laboratories: ECN (NL), ETC (S) and VTT (FIN). They have proved to
be accurate and reproducible.
It has been shown that the addition of kaolin, magnesite, dolomite and sewage sludge
significantly reduce the risk of agglomeration. The agglomeration temperatures increased with
at least 60C in these cases.
During combustion experiments, measured particle temperatures appeared to be up to 100C
higher than the bed temperature. This might have a large influence on agglomeration. Because
gasification, contrary to combustion, is a process where peak temperatures are lower or even
absent, one might expect that agglomeration during gasification will occur at higher
temperatures than during combustion. This however can not be concluded from the experiments,
ECN-C-99-090
illustrating the importance of other factors like the design of fluid beds (the gas distribution, the
type of nozzles, etc.).
It can be concluded that not only the type of fuel and other chemical input is determining the
agglomeration temperature, also other factors like gas distribution, size of bed material, type of
nozzles, cyclone efficiency in CFBs, etc. can have an important role. This means that results
from lab-scale facilities can be interpreted in a relative way (comparing fuels and evaluate
possible solution) but should always be used with great care when trying to draw conclusions
for full-scale plants. Nevertheless, standardised lab-scale bubbling fluidized bed experiments, as
developed and used in this project, seem to be the most reliable tools for the prediction of
agglomeration.
ECN-C-99-090
1.
OBJECTIVES
The main objective of this project is to develop a methodology to predict ash agglomeration and
sintering problems, to indicate related problems and, furthermore, to identify solution methods
to make different types of biomass streams more viable for energy production.
ECN-C-99-090
2.
ACTIVITIES
The activities within the project are divided into a number of tasks:
Task 1: Information and material collection
Task 2: Selection of biomass fuels and other materials
Task 3: Description of the lab-scale apparatus and test procedures
Task 4: Construction and modification of lab-scale fluidised bed
Task 5: Development and description of an agglomeration model
Task 6: Performance of lab-scale and bench-scale tests
Task 7: Description of the total methodology for prediction of ash agglomeration and related
problems, and identification of potential solutions
Task 8: Conclusive evaluation
In the following chapters, results obtained within each task will be presented subsequently.
ECN-C-99-090
3.
RESULTS
In this chapter, the results within each task will be described as separate sub-chapters. Within
task number 8 (conclusive evaluation, chapter 3.8) the information obtained within the different
tasks will be combined.
3.1
TASK 1: Information and material collection
Evil elements
In the literature there is good agreement as to what is seen as the source of ash agglomeration
and related problems. Generally alkali metals (mainly potassium in case of biomass) are
considered as the main component causing the agglomeration problems. In biomass, alkali
metals concentrate in biomass-parts of most vigorous growth. Likely problematic fuels are
therefore residues of agricultural (annual) crops, young energy crops and all other fuels
containing high amounts of young organic material. Problems are more severe when a high
alkali content is combined with high contents of sulphur (in combustion), chlorine, silica and
phosphorus. In annex 1 a review of relevant literature is given.
Apart from chemistry, system design and operational conditions are very important factors
determining the time-scale of the process. Temperature is the most important factor. The higher
the temperature the more severe the problems are. Its important to realise that local peak
temperatures may initiate the problem in cases where only the average temperature is measured.
For high-sulphur fuels there appears to be a temperature window where in-bed sulphur retention
is less and agglomeration problems are reduced. The use of silica-sand bed material promotes
agglomeration due to reaction of alkali compounds with the silica to form low-melting surfaces
on the particles. The time-scale of the agglomeration process and maximum allowable
temperatures in the bed are furthermore strongly influenced by fluidisation conditions (related to
both temperature homogeneity and friction between particles), gas atmosphere, bed and fuel
particle size, and ash re-circulation from cyclones (in CFBs).
Mechanism
Mentioned elements (K, Na, Cl, S, Si, P) form low-melting compounds or eutectics which
become deposited on the bed particles, coating them with a sticky layer. Another possible
mechanism can be distinguished where particles are chemically bonded together by gas-solid
reactions.
Prediction
Three possible prediction methods can be distinguished:
Temperature comparison method: This method is a comparison between the process
temperature (during gasification or combustion) and some typical temperature measured in
a laboratory. The so-called softening temperature of an ash sample, considered as the
temperature where agglomeration starts, can be determined using standardised methods.
This temperature can also be estimated using the ash composition. In general this method is
considered as very inaccurate. It suffers from the limitation that gas-phase reactions,
reactions between gas components and the ash, condensation of alkali vapours, the
influence of particle-to-particle collisions in fluidised-beds and partitioning of elements can
not be adequately simulated in the laboratory. Furthermore, thermal ashing to prepare a
laboratory ash can create artefacts in the ash and, more important, gives a substantial alkaliloss, artificially increasing the ash-fusion temperatures. Also the accumulation of certain
components in a real process is not included in laboratory tests. These methods can
ECN-C-99-090
therefore not give absolute agglomeration temperatures, but some seem useful for gaining a
more fundamental understanding of ash behaviour in thermal conversion
Agglomeration lab test: This method is a laboratory test with ash and bed material. In this
test the temperature is determined at which agglomerates are formed. This method should
result in more realistic predictions since volatilisation/condensation and accumulation of
components and gas-gas and gas-solid reactions can take place, depending on the specific
experimental facility and test procedure.
Mathematical prediction method: The mathematical prediction method concerns the
calculation of possible low-melting mixtures of components availa ble in the system. The
chemical composition of the materials used have to be measured. These methods usually are
equilibrium calculations and therefore do not include kinetics.
Annex 1 contains a literature review on agglomeration. In annex 2, the prediction tools reported
in literature are summarised.
Prevention
Adding inert high-melting minerals (CaO, MgO), sulphur-sorbents (limestone, dolomite) and
alkali-sorbents (clay minerals) to the bed is reported to reduce agglomeration problems in
fluidised-beds, but mechanisms are not fully understood. In some reported cases, sulphursorbents even decreased ash deformation temperatures and increased fouling. Blending of straw
with wood and coal is reported to be successful to some extent. In full-scale practice, the bedmaterial is refreshed regularly before a critical ash-accumulation (alkalis) or agglomeration level
is reached.
A commercial combustion facility at Grenaa, Denmark
The Grenaa CFB1 CHP2 plant was commissioned January 1992. The plant, which is based on a
conventional steam cycle, is producing electricity, district heat and process steam for nearby
industry. It can be fired with coal and coal/straw mixtures. Typical data are shown below.
Owner
Location
Year of commissioning
Max. coal / straw share
Annual straw consumption
Annual coal consumption
Boiler capacity
Steam flow rate
Steam pressure
Feed water temperature
Superheating temperature
Electric output (net at max. heat duty)
Process steam export
Hot water for district heating (max.)
I/S Midtkraft
Grenaa, DK
1992
%/%
100 / 60
ton
70,000
ton
38,000
MW
78
kg/s
29
MPa
9.2
C
170
C
505
MW
17.8
MW
25
MW
32
by energy
at 50 % straw
at 50 % straw
(no re-heating)
8.3 bar, 210C
A typical content of bed material in the system is 24 t. By adding a similar amount of coal fly
ash from a nearby grate fired boiler (20 t) and some fresh sand (3 t) every 24 hour, it is possible
to keep the concentration of water soluble potassium in the re-circulating bed particles below
2000 ppm. This concentration is believed to be a safe limit for avoiding problems due to
agglomeration and deposition in the boiler. However, 4000 ppm has also been achieved without
any problems.
Severe bed sintering/agglomeration seems not to be a major part of the history of the plant. The
reason is probably the frequent change of bed particles, improved temperature control and other
measures (including design modifications), that has been realised in order to deal with other
problems. However, bed sintering/agglomeration is occasional seen as a consequence of
1
2
CFB: circulating fluidised bed

CHP: combined heat and power (electricity)
ECN-C-99-090
10
improper plant control. In particular the operator has to be careful when performing a sudden
switch from a volatile to a less volatile fuel such as from a high share of straw to coal only.
Hence, the control system tends to overshoot in this situation, which may lead to a severe
temperature excursion within the first half-hour or so.
3.2
TASK 2: Selection of biomass fuels, bed materials and additives
Criteria
Biomass materials tested within the project were chosen according to several criteria. Firstly the
fuel should have potential to cause agglomeration problems during thermal treatment. This
potential is related to the chemical composition, especially concerning the list of elements
mentioned in the previous chapter known to cause agglomeration problems. Secondly it should
be a material that is relevant in a European background. In other words: it should be available
for energy conversion in certain quantities either now or in the near future. As a reference
willow has been chosen which is a possible energy-crop in Northern Europe and is considered
to be non-problematic for agglomeration. Bed materials and bed additives are chosen because of
reported potential for reducing agglomeration problems.
Biomass fuels
The fuels chosen are:
1. verge grass,
2. Dutch wheat straw,
3. Danish wheat straw,
4. weathered Danish wheat straw (subjected to rainfall)
5. sewage sludge and straw/sludge-mixtures,
6. cacao shells,
7. willow (as a reference fuel)
Bed materials
The following bed materials were chosen for the lab-scale fluidized bed tests.
1. silica: SiO 2 ,
2. alumina: Al2 O3 ,
3. mullite: Al2 O3 .SiO 2
Silica is the most common bed material. Since silica is considered to play a role in
agglomeration, alumina has been chosen as a more inert bed material. Finally mullite is chosen
because its cheaper than pure alumina and might show the inert character of alumina, see also
annex 1.
Bed additives
The additives chosen for tests in some fluidized bed experiments are based on literature
findings, see annex 1. Chosen additives all represent different (suspected) mechanisms for
improving agglomeration:
1. kaolin: Al2 SiO 5 (OH)4 ,
2. magnesite: MgO,
3. gibbsite: Al(OH)3 ,
4. dolomite: MgCO3 .CaCO3
Kaolin is considered as a chemical getter of alkali's. It binds the alkali metals into a harmless
component. Magnesite probably influences the viscosity of the melt. Battelle reports positive
results obtained with MgO. Gibbsite on the other hand decomposes to very porous Al2 O3
capable of physically gettering alkali's. Dolomite, known for it suitability as tar cracker and
NH3 -reduction agent, is proven to reduce agglomeration in straw combustion when used as bed
material by VTT. Besides this, dolomite is cheap.
ECN-C-99-090
11
Not all combinations of fuels, additives and bed materials as mentioned above were tested. A
selection has been made for each type of experiment. In table 3.1 the combinations of selected
materials and experiments are given.
Table 3.1
Experimental programme with selected biomass materials, marked grey are

experiments actually performed within the project. Also indicted are combinations
of tests and materials reported in literature [1] but worthwhile for comparison.
BFBG: bubbling fluidized bed gasification, BFBC: bubbling fluidized bed
combustion, CFBG: circulating fluidized bed gasification, DTA/TGA: differential
thermal analysis/thermo-gravimetric analysis, HTM: high temperature microscope
facility
experiment/analysis
1.
2.
3.
4.
5.
6.
7.
verge Dutch Danish Danish sewage cacao willow

grass straw straw weathered sludge shells
straw
a. chemical labs
chemical analysis
b. BFBG
gasification (5 kWth)
c. BFBC
combustion and controlled aggl. (5 kWth)
d. CFBG
gasification (500 kWth)
e. DTA/TGA
melting temp. of ash and ash/bed
(*)
[1]
[1]
[1]
f. compression strength compr. strength of sintered ash pellets

g. standard facilities
standard ash melting temperature
h. HTM
heating while observing ash/sand
i. BFBC
combustion and controlled aggl. (5 kWth)
* two different sewage sludges are tested, also in combination with Danish straw (fuel #3)
3.3
TASK 3: Description of the lab-scale apparatus and test procedures
3.3.1
Chemical laboratories at ECN and Ris (DK-TEKNIK) (a)
The selected biomass fuels, see chapter 3.2., were analysed in chemical laboratories of ECN and
DK-TEKNIK as a subcontractor of Ris. Fuels number 3, 4 and 7 were ashed at three different
temperatures (450, 550 and 650C) and the ash was subsequently analysed. Some samples were
analysed by both laboratories to be able to compare results. For the analyses, standard
procedures were used as indicated by [2] for ECN and several, mostly ISO, standards for DKTEKNIK.
3.3.2
BFBG: bubbling fluidized bed gasification facility (5 kWth ) at ECN (b)
The ECN lab-scale bubbling fluidized bed gasification facility (BFBG), called WOB, has been
designed and constructed specially for the project. The design fuel rate is about 1 kg/h being
about 5 kWth . Figure 3.1 shows a picture and figure 3.2. shows a schematic representation of the
facility. It consists of the following parts.
Feeding system
The feeding system consists of a couple of closed vessels of 20 litre equipped with a mechanism
to prevent bridging of the fuel. A dosing screw conveyer moves the fuel towards another screw,
which takes care of quick transport into the hot gasification reactor. Fuel feeding rate is adjusted
according to the calibration of feeding rate with number of revolutions per minute of screw
conveyor. The above described test facility has been used to test different bed materials and
additives in order to detect possible positive effects. The additives have been added into the
system by mixing it with the biomass fuel beforehand.
ECN-C-99-090
12
Figure 3.1 BFB gasifier at ECN

flare
cyclone
fuel gas
sampling
gasification
vessel
feeding bunkers
ash
feeding screw
air
Figure 3.2 Schematic drawing of the BFB (bubbling fluidized bed) gasification set-up (called
WOB) at ECN used to do agglomeration tests
Gasification vessel
The vessel is an electrically heated metal construction with a internal diameter of 74 mm. The
diameter increases to 108 mm at a height of 500 mm. Total length is 1100 mm from air-inlet to
fuel gas outlet. The air is supplied via a sintered metal distribution plate. Temperatures are
measured at 50, 100, 150, 225 mm height above the distribution plate using 2 mm
thermocouples centred in the vessel. At 30 mm height a thermocouples measures the
temperature close to the wall opposite to the fuel inlet, which is at the same height. 65 mm
underneath the distribution plate, the temperature of the incoming air is measured in the centre
of the pipe. The bed material used is either 0.5 mm or 0.27 mm. (the minimum fluidisation
velocity is 10 resp. 3 cm/s in case silica sand is used). The air flow used is 16 nl/min (velocity at
ECN-C-99-090
13
800C is 24 cm/s) for the 0.5 mm sand and 10 nl/min min (velocity at 800C is 15 cm/s) for the
0.27 mm bed material. The amount of bed material in all cases is 1 kg.
Fuel gas treatment
The fuel gas passes a cyclone to separate particles from the gas. The resulting ash passes a valve
system to be able to remove the ashes during operation. The fuel gas is sampled for analysis just
after the cyclone. The hot fuel gas passes an electrically heated pipe and is flared.
Test procedure
Two different kind of tests have been defined for the WOB:
Test X: In a so called X-test the biomass material is converted by air-blown gasification at
750 during 4 hours while feeding the material with about 1 kg/h. Fresh bed material is used
for every test. The amount of air is fixed to obtain an equivalence ratio between 0.25 and
0.30, being a realistic value for full-scale plants. Subsequently the temperature of the bed is
raised stepwise (by adjusting the electrical heating of the gasification vessel) with 25 every
30 minutes until agglomeration occurs, see figure 3.3. Both the fuel feeding rate and air
flow remain unchanged. The 4 hours of stationary operation makes up the biggest part of
the total duration of the experiment. This ensures a certain ash accumulation in the system
which is independent on the agglomeration behaviour of the fuel tested. The result of a Xtest is a temperature at which agglomeration is starting to occur. The point where
agglomeration starts is characterised by a sudden drop in pressure usually together with a
dramatic decrease of temperature homogeneity in the bed.
Test Y: In a Y-test, a temperature is chosen, lower than the temperature found in the X-test,
at which the conversion takes place until agglomeration. The result of a Y-test is a time-toagglomeration at a certain temperature. Fresh bed material is used for every test.
temperature [C]
850
Y-test
825
800
775
X-test
750
725
0
time [hours]
Figure 3.3 Schematic representation of the two types of standard agglomeration tests in labscale fluidized bed gasification facility at ECN
Measurements
Standard measurements comprise the concentration of gaseous components CO, H2 , CO2 and
CH4 in the fuel gas, the solid fuel mass flow, the ash mass flow and the incoming air and
nitrogen volumetric flow. The concentration of nitrogen in the fuel gas is calculated and is used
for determining the total volumetric flow of the fuel gas.
During some tests, bed material was collected, both during the gasification test (after four hours
at 750C) and after actual agglomeration, for further chemical analysis.
ECN-C-99-090
14
3.3.3
BFBC: bubbling fluidized bed combustion facility (5 kWth ) by Ris (ETC) (c)
Lab-scale bubbling fluidized bed combustion experiments took place at ETC (Energiteknisk
Center in Lule/Pite in Sweden), as a subcontractor of Ris. The experiments were performed
by Ris staff in co-operation with the local scientists. In figure 3.4 a sketch of the fluidized bed
combustion system at ETC is given. Quartz sand (542 gram) has been used with a diameter of
200-250m, unless stated otherwise. The velocities used were about 4 times the minimum
fluidisation velocity, which is 7 cm/s. The fuel was fed as pellets of 5 mm diameter and approx.
7 mm long.
Figure 3.4 Sketch of fluidised bed combustion facility at ETC in Sweden

The bed is normally operated at 5 kWth , but the power may be higher. The inner diameter of the
bed section is 100 mm and the reactor height is 1.8 m. The fuel is fed above the bed (upper bed
feeding) which requires the fuel to be pelletized. The fuels have been pelletized without the
addition of any kind of binder material.
The test procedure is to ash a certain amount of fuel at fixed temperature ("ashing temperature")
by combustion and subsequently stop the biomass fuel feeding and raise the bed temperature
externally (by supplying propane flue gas with the same [O 2 ] concentration as the biomass flue
gas) with 2-3C/min until agglomeration occurs. This point can be detected by looking at the
temperatures in and pressure drop over the bed. The point where agglomeration starts is
characterised by either a sudden drop in pressure or a sudden change in temperature.
During several tests the particle temperature (both sand and biomass/char) is measured using a
FTIR pyrometer. An infrared-probe was therefore developed for measuring temperatures of
burning particles inside the ETC fluidized-bed reactor. Optical access to the bed is obtained
ECN-C-99-090
15
through a hot sapphire window at the end of a ceramic probe-tip (not cooled). Radiation spectra
(1.4 5.7 m) from the interior of the fluidized bed reactor are measured at three positions with
a Fourier transform infrared spectrometer connected to the probe with a mid-infrared optical
fiber. At the relatively long wavelengths, temperature measurements are hardly disturbed by
soot radiation. Figure 3.5 shows a drwaing of the probe for temperature measurements.
f e
Figure 3.5 The infrared probe used for measurements of particle temperatures in a fluidized
bed. Via an infrared fiber, a FTIR spectrometer is coupled. Letters a: sapphire
window; b: ceramic tube; c: water-cooled mount; d: lens (CaF2 ); e: connector for
the optical fiber and f: purge gas
3.3.4
CFBG: circulating fluidized bed gasification facility (500 kWth ) at ECN (d)
The ECN circulating fluidized bed gasification facility, called BIVKIN, is designed to gasify
biomass and waste streams and to do research on the conversion itself and the fuel gas
characteristics. It is an atmospheric air blown facility of about 500 kWth . Figure 3.6 shows a
schematic of the facility, see also [3].
2nd cyclone
flare
st
1 cyclone
sampling for
analysis
riser: gasifier
feeding bunkers
ash
Sec. air
20%
downcomer
rotary valves
20 cm
h=6 m
feeding screw
50-100 kg/h
primary air
sealpot
N2
10-20 m3/h
Figure 3.6 Schematic drawing of CFB (circulating fluidized bed) facility at ECN, called
BIVKIN
Feeding system
It is equipped with various feeding systems of which two can be used simultaneously. The straw
used as fuel within this project has been fed using a simple closed bunker and a (non-air tight)
rotary valve for the dosing of the fuel. Since the straw was in the form of pellets, no problem
ECN-C-99-090
16
with bridging occurred. The valve dumps the fuel on a screw transporting the material into the
bottom of the reactor.
Gasification vessel
The gasification vessel (or reactor or riser) is a 20 cm diameter and 6 meter high refractory lined
pipe. Thermocouples and pressure measuring points are present at different location along the
length of the reactor.
Circulation system
The gas flow, containing large quantities of sand (and char) passes the first cyclone, which is
part of the circulation system. The particles separated from the gas fall down into the
downcomer and passes a bubbling fluidized bed seal operated on nitrogen. The total nitrogen
flow (feeding system and seal) is 15 Nm3 /h in average. The particles are flowing back into the
bottom of the reactor.
Fuel gas treatment
The fuel gas passes a second cyclone separating the ash from the gas. The ash passes a valve
system for continuous removal. The fuel gas is sampled for analysis just after the second
cyclone. The fuel gas is flared.
Measurements
Standard measurements comprise the concentration of components CO, H2 , CO2 , CH4 , ethene,
ethane, NH3 , HCl, water, benzene, toluene and xylene in the fuel gas, particles and light tars
(molecular weight up to 300 g/mol) in the gas, the solid fuel mass flow, the ash mass flow and
the incoming air and nitrogen volumetric flow. The concentration of nitrogen in the fuel gas is
calculated and is used for determining the total volumetric flow of the fuel gas. The particles in
the fuel gas and the ash from the second cyclone have been analysed for chlorine and sulphur as
well as the ash-content (550C). Samples were taken from the sealpot every hour and
subsequently analysed on potassium.
Test procedure
The process temperature has been chosen to be low (750C) in order to avoid agglomeration.
The test took 10 hours. The question was how great the penalty was concerning carbon
conversion and tar emission. Since agglomeration wasnt expected to occur within 10 hours at
750C, the concentration of K in the bed has been monitored during the test in order to measure
accumulation. The bed material used is 0.5 mm diameter riversand containing 96.5 wt% SiO 2 ,
1.6 wt% Al2 O3 , 0.4 wt% Fe2 O3 , 0.7 wt% K2O and 0.2 wt% Na2 O. The amount of air was
adjusted, with constant fuel feeding rate, until a stationary situation was reached at desired
temperature. Total heat losses are about 40 kW due to active cooling of feeding screw and
convection heat losses of the whole system.
3.3.5
DTA/TGA: differential thermal analysis and thermo-gravimetric analysis at

ECN and Ris (e)
In the DTA/TGA apparatus the temperature difference between the ash sample and an inert
material is measured during heating, to reveal exothermic and endothermic processes in the ash.
Endothermic peaks are associated with decomposition, vaporisation or melting. This
information allows location of the initial melting temperatures. The weight loss of the sample is
simultaneously measured to reveal (in combination with chemical analysis) e.g. decomposition
of carbonates (CO2 release), vaporisation of alkalis and chemical reactions with the gas phase.
Both ECN and Ris did perform measurements with their DTA/TGA facility. The test apparatus
and conditions are seen in table 3.2.
ECN-C-99-090
17
Table 3.2
Ris and ECN test apparatus and conditions for DTA/TGA-analysis

unit
Ris
apparatus
TA Instruments SDT 2960
sample holder
Al2 O3 (>99.5 %)
sample mass
mg
6-16
gas flow rate
ml/min 500 or 80
temperature range C
20-1000/1200
heating rate
C/min 10
atmosphere
N2 or N 2 /O2 (80% / 20%)
sample drying
sample kept dry
ECN
Seiko TG/DTA 320U
Platinum
11
80
100-1000/1200
10
N2
In situ, 6 hours at 100 C
In all of the Ris and ECN tests the temperature was increased to at least 1000C (1200 C in
some of the tests). In the Ris tests it was emphasised to minimise the uptake of moisture in the
time period between ashing and the DTA/TGA test and hence a period for drying at constant
temperature in the DTA/TGA was not included. ECN included an approx. 6 hour in situ drying
period, and therefore the plots based on ECN data start at approx. 100 C.
ECN performed in situ cooling and reheating in all of the tests. However, since no thermal
reactions could be detected during an extra heat up at ECN, such tests were not performed at
Ris.
The tests were carried out using the ash of different fuels, made at different temperatures. Ris
also did tests with a mixture of silica sand (0.1 mm) and ash (with different ash/sand ratios) in
order to measure the influence of the sand. In table 3.3 an overview is given of the DTA/TGA
tests performed with the different biomass fuels.
Table 3.3
Experimental programme performed with the DTA/TGA facilities at ECN and Ris.
All experiments were carried out both in pure nitrogen and nitrogen/oxygen
(80%/20%) atmosphere, except the sand/ash samples
sample
ashing
temperature
pure ash
450C
550C
650C
ash and sand
3.3.6
550C
fuel #3:
unweathered
wheat straw
Ris: N2 and N2 /O2
ECN: N2
Ris: N2 and N2 /O2
ECN: N2
Ris: N2 and N2 /O2
ECN: N2
Ris: N2
(0, 50, 75, 90
and 100% sand)
fuel #4:
weathered
wheat straw
Ris: N2 and N2 /O2
ECN: N2
Ris: N2 and N2 /O2
fuel #7:
willow
Ris: N2 and N2 /O2
ECN: N2
Ris: N2 and N2 /O2
Ris: N2 and N2 /O2
Ris: N2 and N2 /O2
Compression strength measurement facility by Ris (TNO) (f)
Compression strength measurements were performed at TNO in the Netherlands as a

subcontractor of Ris. TNO is a research institute with some experience with this kind of tests
and especially with the preparation of samples for the compression tests.
When a compressive force is applied in an axial direction on a cylindrical pellet, the pellets tend
to swell. Depending on the cylinders elasticity and plasticity, cracks will form in the pellet as
the applied force increases. At a certain point, the force reaches a maximum and causing the
pellet to break. The measured compression strength will be a function of both the specific
ECN-C-99-090
18
strength of the pellet, the pellet size and the testing device. Both the pellet size and the device is
kept constant. The procedure used is according to a standard procedure based on methods used
in Abo Academy in Finland:
Low temperature ashing
milling if necessary to size under 10 mm
drying at 105 C till constant weight
pyrolysis in inert nitrogen atmosphere below 500C
ashing of char in nitrogen/air mixture below 500C to obtain a C content < 1 % (total
C according to standard ASTM E 1019)
Sample preparation
grinding of the ash below < 0.25 mm
sieving below < 0.125 mm
Pelletising
compression to pellets in a mould with a pressure of 10 kg/cm2
pellet dimensions are: length = 1 cm and diameter = 1 cm
Heat treatment
four hours in air at 600, 700, 750, 800, 850 and 900C
Compression strength measurement
in standard equipment: three pellets per temperature
3.3.7
Standard facilities for measuring standard ash melting temperatures by ECN

(NMi) and Ris (DK-TEKNIK) (g)
Standard ash melting temperatures are determined both by DK-TEKNIK and NMi in the
Netherlands. NMi did the tests in a reducing atmosphere (20% CO in N2 ), whereas Ris used
dry air as a medium. Using standard procedure ASTM D1857),four typical temperatures were
determined:
IDT = initial deformation temperature
ST = softening temperature
HT = hemispherical temperature,
FT = fluid temperature (total fusion)
3.3.8
SEM and HTM (high temperature microscope) at Ris (h)
The SEM spectroscopy measurements were performed with a JSM-840 scanning electron
microscope equipped with a NORAN Voyager version 3.5 energy dispersion x-ray
spectrometer. In order to obtain a reasonable statistics at least 3 measurements of each type of
particles present in the sample were always taken. The measurements included - if possible also the agglomerated transition bridge between particles. The measurements were performed
on bed ash samples before and after agglomeration in the fluidised bed.
ESEM (HTM) investigations were performed using ElectroScan ESEM model E-3 instrument
equipped with a high temperature stage and a Kevex energy dispersive x-ray spectrometer. In
figure 3.7 the main part of the apparatus is shown. With this facility its possible to heat a
sample in a hot stage while observing it. In order to prevent blowing the powder out from the
crucible at a certain temperature (above 400-500C), probably because of electrostatic forces, it
appeared to be necessary to pelletize the sample first. Because this was not desirable, the design
of the hot stage was changed and the problem was solved.
ECN-C-99-090
19
The high temperature stage is a specially designed electrically heated unit. As mentioned, this
unit was improved during the project. The improved unit was also capable of reaching higher
temperatures (1500C instead of 900C).
Figure 3.7 Electroscan ESEM model 20 used to perform the high temperature microscopy
(HTM) measurements at Ris
3.3.9
BFBC: bubbling fluidized bed combustion facility by Ris (VTT) (i)
VTT performed the experiments as a subcontractor of Ris. Both weathered and fresh wheat
straw (fuel #3 and 4) was combusted in a small fluidized-bed reactor. The suitable feedstock
fraction for the fuel feeding equipment was above 0.56 mm. The fuel is fed with a screw feeder
using a small amount of nitrogen (0.5 l/min) as a purging gas. The feeding tube is located 10 cm
above the air distributor. A schematic diagram of the lab-scale atmospheric fluidized-bed
reactor is shown in Figure 3.8. The diameter of the bed and freeboard is 3.5 cm, and the height
from the air distributor to the gas outlet pipe is 53.5 cm. The bed thermocouple is located in the
fluid-bed region of the reactor, the height from the air distributor being 6 cm, and the freeboard
thermocouple is located 21 cm higher. The bed material used is silica sand (350-560 m). The
fluidization velocity was 0.4 m/s.
3.3.9.1
Test procedure
At a given temperature, a fixed amount of fuel (50 gram) is combusted in the bed containing 65
gram of bed material. The fuel feeding rate is used to control the bed temperature, sometimes
together with electric heaters. After the fuel batch is fed, the reactor is emptied and a new
experiment starts at a higher temperature (steps of 25C) until signs of ash sintering are
observed. The test series started at the temperature of 700C.
ECN-C-99-090
20
Figure 3.8 Sketch of bubbling fluidized bed combustion facility at VTT in Finland
3.3.10
Other facilities
During the project several facilities have been used with no successful outcome. In this chapter
these will be shortly mentioned together with the reason why it was decided to stop using them.
BFBG: bubbling fluidized bed conversion facility at ECN (200 W)
At ECN a 200 W bubbling fluidized bed was constructed. It is designed to operate at a load of
about 50 gram/hour and can operate from pyrolysis- to combustion mode. The biomass was fed
by a calibrated screw feeder through a cooled tube from the top of the reactor. The quartz
reactor had an internal diameter of 40 mm. The whole reactor was placed in a furnace in order to
minimise heat loss. During the first tests it appeared that the feeding system failed. The
problems were associated with the fact that the feeding-pipe passed through the hot zone which
resulted in unwanted hold-up of material and premature pyrolysis, unless the measures taken to
prevent early warming up of the material. This meant unstable conditions in the gasification
vessel. Fixing the problem seemed hopeless given the small size. It was therefore decided to not
continue the experiments with this facility and instead use an existing gasification set-up at the
Technical University of Delft in the Netherlands.
BFBG: bubbling fluidized bed gasification facility at Delft Technical University (5 kWth )
The gasification facility at TUD is a 1 kg/h bubbling fluidized bed (about 5 kW) with pneumatic
feeding system. During the test runs with straw and road side grass, it turned out that the TUDfacility has some major disadvantages. The feeding system often failed feeding the material with
constant flow. Because of the feeding system being pneumatic, the air flow is relatively fixed
and high. Because of this, and the absence of external heating/cooling, reactor-temperature and
air/fuel-ratio are not independently controllable. Another major draw back is the fact that there
are no pressure measurements in the bed which can be used to detect agglomeration in an early
stage.
ECN-C-99-090
21
All the problems and disadvantages of existing test units have shown us that the specific
demands that exist for a test facility to use in the present project are hard to meet. It was decided
to build a new facility at ECN. This facility has been used in the project and is described in
more detail in chapter 3.3.2.
Sinterdisk
The sinterdisk is a metal plate where ash/sand mixtures can be pressed into pellets and
subsequently heat treated up to 900C. After this it should be possible to press the (sintered)
pellet out for further analysis like SEM and compression strength measurement. The test is
based on similar tests done in the US to characterise coal. It turned out that the pellets didnt
come out in one piece, making further analysis impossible. It was decided to stop the sinterdiskexperiments since also similar tests were performed by TNO (compression strength
measurement of ash pellets, see chapter 3.3.6) and Ris (high temperature microscope, see
chapter 3.3.8).
3.4
TASK 4: Construction and modification of lab-scale fluidised bed

The facility has been designed and constructed for the project. It has been equipped with
sampling points, temperature and pressure points etc. in such a way that agglomeration could be
determined in an early stage. The design combines the best features of several other facilities
around the world.
BFBC: bubbling fluidized bed combustion facility (5 kWth) at ETC (c)
The bubbling fluidized bed combustion unit has been adjusted for the purposes of the project.
Since agglomeration is very dependent on temperature, Ris built a fibre-optic IR-probe which
can be inserted at various places in the bed or in the free board. Insertion of the probe in the bed
required some modification of the fluid bed system. The probe is connected to an Fourier
transform infrared (FTIR) spectrometer. This probe measures the temperature of the different
particles during the formation of the ash and during the agglomeration process. Furthermore the
gas temperature might be extracted from absorption bands of CO2 .
CFBG: circulating fluidized bed gasification facility (500 kWth) at ECN (d)
The circulating fluidized bed gasification facility has been equipped with a facility to take bed
material samples for further analyses. This has been done by creating a nitrogen-blown L-valve
in the sealpot.
3.5
TASK 5: Development and description of an agglomeration model
Bed samples collected when firing DK straw (fuel #3) and DK weathered straw (fuel #4) show
that the overall compositional distributions of the major fraction of the bed particle coatings are
mainly limited (>85 % of the material) to the ternary system K2O-CaO-SiO 2 [4] . Figure 3.9
shows the K2 O-CaO-SiO 2 ternary diagram with some solidus (initial melting) temperatures and
the compositions of the bed material coatings from using the two different straws (average value
of approximately 20 EDS spot analysis). The chemical compositions of these samples are
mainly restricted to the SiO 2 rich corner in figure 3.9, and silicates with these compositions have
a first melting temperature of 720C.
ECN-C-99-090
22
010
0
10
CaO
25
75
50
50
011
930C
75
1080C
25
1020C
136
830C
111
800C
126
129
825C
213
101
100
KO
4110
895C
770C
25
102
50
720C
720C
740C
10
0
890C 216
106
104
001
75
100
SiO
Figure 3.9 K2 O-CaO-SiO2 ternary diagram with some solidus temperatures (extracted from
[4]) and the composition of the different bed particle coatings. Plot marks
(abbreviations) for the different bed samples are: 1: wheat straw (fuel #3) and 2:
weathered wheat straw (fuel #4)
Previous results [5,6] have indicated that the chemical characteristics, and thereby the melting
behaviours 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[6] . It was therefore of interest to
determine the melting behaviours of bed coatings when burning DK straw (fuel #3) and DK
weathered straw (fuel #4), and compare the melting temperatures with the corresponding
agglomeration temperatures.
Owing to a lack of thermodynamic data for several intermediate phases in the system K 2 O-CaOSiO2 , thermodynamic multi-component, multiphase equilibrium calculations could not be used
to accurately determine the melting behaviour of the bed particle coatings. Instead, extracting
melting behaviour data from the phase diagram K2O-CaO-SiO 2 performed the evaluation. The
average composition (see figure 3.9) of approximately twenty EDS spot analysis from bed
sample coatings after agglomeration were included in the comparison.
The resulting fractions of melt (solid lines) versus temperature are shown in figure 3.10 together
with the determined initial agglomeration temperature (vertical lines), determined by the
controlled fluidized bed agglomeration method.
ECN-C-99-090
23
wt% molten material .
100
80
60
fuel #3
fuel #4
40
20
0
700
800
900
1000
1100
1200
temperature [C]
Figure 3.10 Melting behaviour of different spot characteristics (average values) of bed coatings
from bed samples when burning wheat straw (fuel #3) and weathered wheat straw
(fuel #4). The initial bed agglomeration temperatures, determined with the
controlled bed agglomeration test, are represented by the vertical lines
Apparently around 20% molten material seems to be enough to initiate agglomeration in the
fluidised bed system.
3.6
TASK 6: Performance of lab-scale and bench-scale tests
3.6.1
Chemical laboratories at ECN and Ris (DK-TEKNIK) (a)
Analyses of biomass fuels

The selected biomass fuels, see chapter 3.2, have been analysed for their chemical composition
and heating value. Table 3.4 shows the results. The results are stored in a public database, called
Phyllis. It is available on the Internet (http://www.ecn.nl/phyllis).
ECN-C-99-090
24
Table 3.4
Chemical composition of the seven selected biomass/waste fuels as used in the

project. In the right columns, the average composition is given of several fuels
according to the Phyllis database on chemical composition of biomass [7]. HHV:
higher heating value, dry: on dry basis, daf: on dry and ash-free basis.
1
ash mixed
ash 450
ash 550
ash 650
ash 815
volatiles
HHV
C
H
O
N
S
Cl
F
Al
As
B
Ba
Ca
Cd
Co
Cr
Cu
Fe
Hg
K
Mg
Mn
Mo
Na
Ni
P
Pb
Sb
Se
Si
Sn
Sr
Te
Ti
V
Zn
wt% dry
wt% dry
wt% dry
wt% dry
wt% dry
wt% daf
kJ/kg daf
wt% daf
wt% daf
wt% daf
wt% daf
wt% daf
wt% daf
wt% daf
mg/kg dry
mg/kg dry
mg/kg dry
mg/kg dry
mg/kg dry
mg/kg dry
mg/kg dry
mg/kg dry
mg/kg dry
mg/kg dry
mg/kg dry
mg/kg dry
mg/kg dry
mg/kg dry
mg/kg dry
mg/kg dry
mg/kg dry
mg/kg dry
mg/kg dry
mg/kg dry
mg/kg dry
mg/kg dry
mg/kg dry
mg/kg dry
mg/kg dry
mg/kg dry
mg/kg dry
mg/kg dry
verge
grass
------ wheat straw -----Dutch

Danish
Danish
(dry)
(wet)
15.3
79.0
19880
49.9
5.68
40.9
2.31
0.17
0.74
0.0034
7.2
7.1
6.7
4.6
80.7
19920
49.4
6.42
43.1
0.48
0.17
0.43
0.0011
11.9
10.1
1.0
13000
1.0
1.2
11
10
0.04
21000
1900
150
1500
1.6
2400
14
1.0
1.0
38000
10
78.3
18730
49.7
5.76
43.0
0.71
0.17
0.70
0.0014
940
2.5
5.3
23
3700
0.6
1.5
3.1
2.6
580
13200
690
17
2.3
320
1.3
760
2.9
3.0
2.0
34000
2.0
25
1.0
6.1
44.0
3.3
3.3
2.9
sewage
sludge
cacao
shells
willow
0.13
12400
6100
6000
100
60
1030
700
600
11100
20070
10.2
75.1
20540
52.5
5.70
38.2
3.36
0.24
0.02
0.0011
900
1.5
47
26
2330
0.60
1.2
6.1
39
840
0.05
26550
5185
74
0.5
59
11.4
3725
19
1.0
1.0
7050
2.8
1.0
37
1.7
12
wheat
straw
sewage
sludge
willow
6.6
39
1.9
81.0
19470
49.1
5.87
43.6
0.78
0.15
0.45
0.0011
940
2.5
5.3
23
2625
0.2
1.5
3.1
3.3
290
0.02
9133
863
8.5
2.3
831
1.3
465
0.89
3.0
2.0
20250
2.0
25
87.6
21890
52.7
7.51
31.0
6.71
2.09
0.23
0.025
20300
5.6
62
400
39000
1.4
6.6
64
450
50800
2.3
1675
3700
405
6.4
895
170
21900
180
0.3
0.4
32600
24
350
37
1.7
12
420
14
930
83.1
19800
49.7
6.06
43.6
0.63
0.06
0.01
0.0030
59.5
1.3
9.8
4.0
5720
2.4
0.6
14
13
68
0.05
2894
524
9.7
0.5
210
26
708
238
2.5
1.0
618
1.7
14
1.0
4.0
0.28
97
4.0
2.9
1.6
36.4
82.1
24100
53.0
8.20
29.0
8.46
1.11
0.22
average values from Phyllis
2.0
91
81.2
19250
49.2
5.99
44.2
0.51
0.036
0.02
0.010
100
1.2
9.0
4.0
4650
2.35
0.6
27
26
110
0.05
2550
530
11
0.5
205
41.5
650
238
2.5
1.0
1280
1.7
14
1.0
9.0
0.4
64
From the table, some observations can be made. The K- and Cl-content of straw decreases
significantly by allowing rain to wash the material. At the same time, the ash content also is
reduced by washing. This may be explained by the leaching out of inorganic material. The Kcontent, generally thought to have a main role in agglomeration, is very high for verge grass and
cacao shells (more than 2 wt% of the fuel). Although less in concentration, also the wheat straw
(not washed by rain) contains large amounts of K. Sewage sludge however contains much less
K but has relatively high amounts of sulphur, calcium and iron. Furthermore, verge grass and
Dutch wheat straw contain very much silicon compared to the other fuels, probably due to sand.
ECN-C-99-090
25
In table 3.5 some indices used for predicting problems concerning agglomeration are given. The
alkali-index is the ratio of the sum of K2 O and N2O and the higher heating value. Values higher
than 0.17 kg/GJ are considered to cause possibly problems, values exceeding 0.34 kg/GJ almost
certainly cause problems concerning slagging and fouling. Another index is the molar ratio
between potassium and chlorine. For combustion purposes also sulphur has to be considered.
The idea about this ratio is that it gives an indication of how volatile the K is. Chlorine and (for
combustion) sulphur are important elements forming volatile components with potassium.
Table 3.5
Some indices for the fuels used in the project

1
verge
grass
alkali-index
K/(Cl)-ratio
K/(Cl+2S)-ratio
kg/GJ
mol/mol
mol/mol
1.6
2.6
1.7
wheat straw
Dutch
Danish
Danish
(dry)
(wet)
0.98
1.7
1.1
0.81
2.6
1.4
0.38
4.1
average values from Phyllis
sewage
sludge
cacao
shells
willow
wheat
straw
sewage
sludge
willow
0.56
2.5
0.2
1.7
98
4.4
0.18
12
2.4
0.7
1.9
1.1
0.24
0.6
0.03
0.19
19
1.8
From the table it can be seen that only willow and sometimes sewage sludge show a low alkaliindex. For cacao shells it turns out that it contains high amounts of K but hardly any chlorine to
make the K volatile in gasification atmosphere. This might reduce the agglomeration problems
for cacao shells considerably.
Analyses of ashes from biomass fuels
The composition of the ashes produced by three fuels, ashed at different temperatures, is given
in tables 3.6 to 3.8. Note that the elements are reported as oxides, according to international
agreements. This does not mean that the elements are actually present as oxides.
Table 3.6
Ash composition [wt%] of willow, ashed at different temperatures
willow
ashing temp. [C]:
analysis by:
P2 O5
SO3
Na2 O
MgO
Al2 O3
SiO2
K2 O
CaO
TiO2
Fe2 O3
CO2
Cl
sum
ECN-C-99-090
450
ECN
7.3
2.6
2.2
4.1
2.6
27
10.1
25
0.15
1.4
0.9
0.41
84
550
ECN
10.2
3.2
2.5
4.8
0.3
2
12.1
37
0.03
0.3
2.2
0.75
76
650
ECN
10.8
3.6
2.57
5.1
0.3
2
12.8
38
0.02
0.2
1.8
0.67
77
450
Ris
10.0
2.5
2.6
4.6
1.7
17
12.0
30
550
Ris
11.0
2.4
2.7
4.8
1.8
17
13.0
32
650
Ris
13.0
4.1
3.1
5.0
2.0
20
13.0
36
1.4
16
0.75
99
1.4
15
0.49
102
1.6
7.5
0.16
105
26
Table 3.7
Ash composition [wt%] of wheat straw (dry), ashed at different temperatures
wheat straw (dry)

ashing temp. [C]:
analysis by:
P2 O5
SO3
Na2 O
MgO
Al2 O3
SiO2
K2 O
CaO
TiO2
Fe2 O3
CO2
Cl
sum
Table 3.8
450
ECN
2.7
4.1
0.22
2.1
0.2
59
17.4
6
0.02
0.1
0.3
1.7
94
550
ECN
2.1
4.0
0.25
1.9
0.1
56
21.5
6
0
0.1
0.2
3.3
96
650
ECN
2.4
5.2
0.49
2.1
0.2
56
22.0
6
0.02
0.1
0.1
1.3
96
450
Ris
2.5
3.7
0.16
2.0
<0.2
51
21.0
7
0.09
0.85
4
92
550
Ris
2.5
3.1
0.16
2.0
51
21.0
7
650
Ris
2.3
4.1
0.19
1.8
<0.2
51
22.0
7
0.69
3.7
91
0.1
0.22
3.8
93
Ash composition [wt%] of weathered wheat straw (wet), ashed at different

temperatures
weathered wheat straw (wet)

ashing temp. [C]:
analysis by:
P2 O5
SO3
Na2 O
MgO
Al2 O3
SiO2
K2 O
CaO
TiO2
Fe2 O3
CO2
Cl
sum
450
ECN
3.8
5.8
0.2
2.6
0.3
39
21
12
0.02
0.2
0.8
0.8
87
550
ECN
4.2
5.6
0.3
2.8
0.4
45
19
13
0.02
0.2
0.7
0.6
92
650
ECN
4.3
6.4
0.4
3.1
0.4
47
22
14
0.02
0.2
0.5
0.9
99
450
Ris
4.0
4.5
0.2
2.6
0.4
37
23
15
550
Ris
4.1
4.4
0.3
2.4
0.4
39
24
15
650
Ris
3.6
5.9
0.3
2.8
0.5
40
25
16
0.33
7.37
1.0
95
0.3
2.79
0.7
93
0.3
0.99
0.4
96
The following observations can be made. In general ECN and Ris report similar results. The
chemical analysis results show some variance for the same fuel, probably due to the
heterogeneous nature of biomass and the small size of the analysed samples. The high silicon
content in willow ashed at 450 C is likely to be due to contamination with sand or earth. The
sum of all oxides is 70-80% for willow and about 85% for straw. This probably means that
certain elements are measured in too low concentrations. This might well be the case for SiO 2
which is difficult to dissolve and is present in relatively large quantities.
The differences between the ashes made by combustion at different temperatures are rather
small. Ashing at temperatures going from 450 to 650C does neither result in a significant loss
of alkalis expressed in wt% of the ash nor does it result in a significant decrease of the ash
content (see tables 3.6 to 3.8). There is a decrease in carbonate content in some of the ashes with
increasing temperature. This however is probably not so important as carbonates decompose
before they melt and a loss of carbonate would therefore not influence the melting behaviour
determined in DTA/TG or other techniques.
ECN-C-99-090
27
3.6.2
Tests have been performed with different fuels. Using Dutch straw, different bed materials and
additives are used as well as bed material sizes. Figure 3.11 shows a typical result of a so-called
X-test, see chapter 3.3.2. for description of procedure. In figure 3.12 a detail of the last 20
minutes is given.
Standard agglomeration test
wheat straw (bed material:silica sand), 23 December 1998
200
850
25
intervals
bed temperatures
150
550
start of
standard test
agglomeration
(end of test)
50
450
pressure drop
350
temperature under air distributor
-50
250
-400
100
30' intervals
4 hours at 750C
650
pressure drop over bed [mbar]
temperature [C]
750
-100
-350
-300
-250
-200
-150
-100
-50
50
time to agglomeration [minutes]
Figure 3.11 Result of X-test with Dutch wheat straw in standard agglomeration test in BFBG at
ECN
From the figure its clear that the temperatures within the bed start to deviate as soon as
agglomeration is starting. The differences between the different thermocouples keep increasing
when continuing the conversion process. At a certain time also the pressure drop across the bed
starts to change suddenly. This usually is considered as the moment agglomeration starts. From
the figure however its clear that about 7 minutes earlier (in this particular case) the
temperatures in the bed start to deviate. Even earlier, the temperature just underneath the air
distributor plate started to decrease. So, the moment pressures drop over the bed starts to change
and agglomeration is considered to be a fact is preceded by changing temperatures.
ECN-C-99-090
28
Standard agglomeration test, detail of last minutes

wheat straw (bed material:silica sand), 23 December 1998
200
bed temperatures
850
150
temperature [C]
agglomeration
(end of test)
650
100
50
550
pressure drop
0
450
first signs of
agglomeration
temperature under
air distributor
pressure drop over bed [mbar]
750
-50
350
250
-100
-20
-15
-10
-5
10
time to agglomeration [minutes]
Figure 3.12 Detail of last 20 minutes of X-test with Dutch wheat straw in standard
agglomeration test in BFBG at ECN.
Tables 3.9 and 3.10 sum up the bed materials and additives used in the different experiments
and the different parameters and data relevant for this project. Table 3.11 shows the
agglomeration temperatures for the tests done.
Table 3.9
Bed material data used for tests in BFBG facility at ECN

particle size [m]
bed material
main
components
silica (500 m) SiO2 (>96.5%)
silica (250 m) SiO2 (>98.5%)
alumina
Al2 O3
mullite
Al2 O3 .SiO2
D10 D50
D90
480
120
120
120
520
320
250
250
500
250
200
200
total outer
min. fluidization velocity
density surface for 1 kg for median size particles

[kg/m3 ] material [m2 ]
[m/s]
2600
4.6
0.11
2600
9.2
0.028
4000
7.5
0.027
3200
9.4
0.022
Table 3.10 The additives used in the experiments in the BFBG at ECN. The mass flow of all
additives was 2 wt% of the fuel mass flow (approx. 1 kg/h)
additive
formula
way of feeding density

[kg/m3 ]
kaolin
magnesite
dolomite
gibbsite
Al2 SiO5 (OH)4

MgO
MgCO3 .CaCO3
Al(OH)3
mixed with fuel

mixed with fuel
separate feeding
mixed with fuel
ECN-C-99-090
2600
3600
1400
2400
size
[m]
< 250
< 75
700-1000
30-150
remarks
Euroclay, extruder dried DB-kote II
M.A.F., MgO no 12, mesh 200
Marches les Dames
Pecheney Belgium NV, SH950
29
When comparing tests 1 with 2 and test 20 with 5 it is striking that the temperatures are (about)
the same but the time to agglomeration is very different. The time to agglomeration for an Xtest is relatively long. Apparently, in an X-test the temperature reached in the last phase of the
test is the most crucial parameter for agglomeration. Even though the amount of ash in the
system is much higher for the X-tests than for the Y-tests mentioned, this does not result in
significantly quicker agglomeration. This observation leads to the conclusion that the
agglomeration temperature as measured in an X-test is the limiting temperature even for shortterm experiments.
Table 3.11 Overview of tests performed (marked grey) with the lab-scale BFBG (WOB) at
ECN and the measured agglomeration temperature and, if applicable, the time-toagglomeration. All fuels were sieved to 700-2000 m. Bed material is silica (SiO2 ),
unless stated otherwise
test type of
no. test
1
2
3
4
5
6
7
8
9
10
11
13
15
16
18
20
X
Y
Y
X
Y
Y
X
X
X
X
X
X
X
X
X
X
fuel #1
verge grass
fuel #2
Dutch straw
fuel #4
fuel #6
Danish straw cacao shells
(weathered)
fuel #7
willow
bed material
or
additive (*)
825
825/20 minutes
800/1 hour
810 and 810
835/2 hours
810/8.4 hours
840
840
>930
>960
840
bed: alumina
bed: mullite
add: kaolin
add: magnesite
add: gibbsite
800
900
875
>900
840 and 840
add: dolomite
bed material
size
[m]
500
500
500
500
250
250
200
200
250
250
250
500
500
500
200
250
(*) only mentioned if test has not been done using silica as bed material with no additive, when additives
were used, 2wt% of fuel input has been added, see tables 3.9 and 3.10.
Figure 3.13 shows the relation between time and agglomeration temperature. It shows that
agglomeration occurs at lower temperature if more fuel has been added to the system. In other
words, it illustrates that accumulation of elements has the expected effect on agglomeration. For
the tests performed, an extension of time (amount of fuel and ash fed to the system) with a
factor of 3 decreases the agglomeration temperature with about 25C.
ECN-C-99-090
30
agglomeration temperature [C]
860
NL straw (fuel #2)
840
verge grass (fuel #1)
820
800
780
0
4
6
time to agglomeration [h]
10
Figure 3.13 The relation between time to agglomeration and agglomeration temperature for
two fuels, the data on the y-axis are the results of a standard X-test (starting at
750C and gradually increasing
Furthermore, smaller bed material increases agglomeration temperature (test 20 compared to
test 4) and extends the time to agglomeration at a given temperature (test 6 compared to test 3
and test 5 compared to test 2). This might be the result of two effects: (1) smaller bed material
means also more bed material surface available for melting components and thus thinner sticky
layers less capable of sticking together particles and (2) more rigorous mixing and more friction
between particles because of the larger gap between minimum fluidization velocity and actual
velocity.
Cacao shells seem to show low tendency for agglomeration compared to the other fuels. This is
not what would be predicted when considering K-content as the most useful prediction tool.
Cacao shells have the highest K-content of all fuels tested. However, cacao shells are very low
in chlorine. The high K/Cl ratio, also given in table 3.5, indicate that theres little chlorine to
make the K mobile. The low chlorine content is a plausible explanation, other reasons should
not be excluded.
The use of alumina and mullite (tests 7 and 8) as bed material instead of silica (test 20) does not
change the agglomeration temperature. This might mean that either the silicon of the sand does
not play a significant role in agglomeration or the silicon already present in the straw is more
than the minimum amount necessary for the process of agglomeration. Since the amount of
silicon in the straw is rather high (3.4 wt% Si in the dry fuel which is 7.3 wt% if present as
SiO2 ) the latter possibility seems plausible.
Kaolin, magnesite and dolomite used as additives (tests 9, 10 and 18 versus test 20) turn out to
increase the agglomeration temperature by at least 60C. Gibbsite however does not seem to
improve the process. Apparently, gibbsite does either not form the porous particles in the reactor
as expected or the extra surface created by the porous particles does not operate as a physical
absorber of any relevance. It may also be the case that the gibbsite, either before or after the
conversion to porous particles, has a too low retention time in the bed to be able to physically
absorb the (potentially) melting components. Since the particle size of the gibbsite is only
around 100 m, this might well be the case.
Surprisingly, willow showed agglomeration behaviour during a standard test at a temperature as
low as 875C. Willow generally is considered to show no tendency for agglomeration during
thermal conversion. Its not clear what might be the reason for this.
ECN-C-99-090
31
Chemical analysis were carried out on bed material collected during gasification (after 4 hours
at 750C) of test 9 and test 20. From test 1 and 13 material was collected after the experiment
where agglomeration was created. The analyses were done in order to see possible enrichment
of certain elements of the coating of the bed material compared to the ash of the fuel. Table
shows the main results focussed on the concentration of K.
Table 3.12 Chemical analyses, expressed as ratios, of bed material during and after several
controlled agglomeration experiments
fuel
test#
sample
fuel #2
Dutch straw
fuel #2
Dutch straw
fuel #1
verge grass
9*
after 4 hours at
750C
20
after 4 hours at
750C
1
after controlled
agglomeration
fuel #4
Danish straw
(weathered)
13
after controlled
agglomeration
(K/Na)bed / (K/Na) fuel

0.68
0.63
0.81
0.14
(K/Mg)bed / (K/Mg) fuel
0.98
0.84
0.96
0.48
(K/Al)bed / (K/Al) fuel
0.73
0.24
0.15
0.19
(K/P)bed / (K/P) fuel
0.86
0.75
0.57
0.38
(K/Ca)bed / (K/Ca) fuel
1.01
0.83
0.89
0.85
(K/S)bed / (K/S) fuel
72
45
14
17
* kaolin (2 wt%) used as additive, composition of fuel plus additive has been used for calculating the
ratios
It turns out that the ratios presented in the above table are close to unity. This might indicate that
the bed materials coating has roughly the same composition as the fuel used in the experiment.
The expected enrichment of K in the coating compared to the fuel is not found. The high ratios
shown in the lower row are caused by the conversion of sulphur to H2 S during the test, taking
away S to be used in the bed material coating.
3.6.3
BFBC: bubbling fluidized bed combustion facility (5 kWth) by Ris (ETC) (c)
Figure 3.14 shows a typical recording of a test performed with the bubbling fluidised bed
combustion facility at ETC. Tables 3.13 and 3.14 summarise the main results of the
agglomeration tests. It turned out that the ashing temperature (the temperature at which the fuel
is combusted during the first phase of every experiment prior to increasing the temperature of
the bed) is the main parameter influencing the agglomeration temperature. Other operational
variables i.e. air flow, oxygen content, and size of sand particles have less influence on the
agglomeration temperature. The influence of the ash concentration (determined by the amount
of fuel feed into the system) has also a limited influence as long as there is sufficient amount of
ash available to form an agglomeration mixture. Figure 3.15 shows the relation between the
amount of fuel (ash) in the system and the agglomeration temperature.
ECN-C-99-090
32
Agglomeration Experiment 970623_ris_0006: Tashing= 694 C, O2= 5.6 %, Taggl_init= 687 C,

Taggl_final= 855 C
36
1000 C
33
950
30
900
27
850
temperatures
24
800
21
750
BED_PRESSURE_E
18
700
dp (N1)
15
650
12
dp (N)
oxygen
6
3
0
dp (W)
17: 17: 17: 17: 17: 17: 17: 17: 18: 18: 18: 18: 18: 18: 18: 18: 18: 19: 19: 19: 19: 19: 19: 19: 19: 20: 20: 20: 20: 20: 20: 20:
03: 10: 17: 25: 32: 39: 46: 54: 01: 08: 15: 23: 30: 37: 44: 52: 59: 06: 13: 21: 28: 35: 42: 50: 57: 04: 11: 19: 26: 33: 40: 48:
15 30 45 00 15 30 45 00 15 30 45 00 15 30 45 00 15 30 45 00 15 30 45 00 15 30 45 00 15 30 45 00
BED_PRESSURE_N
BED_PRESSURE_N1
600
BED_PRESSURE_W
550
O2
500
BED_TEMP_E
450
BED_TEMP_S
400
2151 g Dry Wheat Straw_ris-1 (6.3 % ash), 542 g Silversand (200-250 m), 20 % fuel ash
Figure 3.14 Typical output from experiment with BFBC at ETC. In this case DK straw is used
as a fuel. The experiment ends with a sudden decrease of pressure drop and a
increase of temperature differences in the bed.
900
DK straw weathered (fuel #4)
880
NL straw (fuel#2)
860
840
820
800
780
0
1000
2000
3000
4000
fuel fed in system [g]
Figure 3.15 The effect of the amount of fuel whcih was fed into the BFBC and the observed
agglomeration temperature
ECN-C-99-090
33
Table 3.13 Agglomeration test results of three different straws as well as willow in ETC
fluidised bed combustion facility
test
no.
1
9
6
10
8
33
34
35
36
55
56
43
47
46
45
48
49
50
51
fuel
fuel fed [g]
DK straw (fuel #3)

DK straw (fuel #3)
DK straw (fuel #3)
DK straw (fuel #3)
DK straw (fuel #3)
DK straw weathered (fuel #4) with 3 wt% kaolin
DK straw weathered (fuel #4) with 1 wt% kaolin
NL straw (fuel #2)
NL straw (fuel #2)
NL straw (fuel #2)
NL straw (fuel #2)
willow (fuel #7)
willow (fuel #7)
willow (fuel #7)
willow (fuel #7)
2151
2151
2151
2151
2150
2151
2151
3225
1075
1075
1075
538
807
807
1076
2000
2000
964
3500
ashing temp.
[C]
659
693
694
702
727
745
748
750
747
746
748
749
747
753
745
691
772
774
774
agglomeration
temp. [C]
891
860
855
785
732
852
844
832
863
897
849
876
839
838
801
898
882
892
892
Table 3.14 Agglomeration test results of pure sewage sludge and straw/sewage sludge
mixtures (two kinds, a: Egaa, b: stdeponi) in ETC fluidised bed combustion
facility
test fuels
no.
18
37
21
22
23
24
25
26
27
28
29
30
38
39
40
41
sewage sludgea (fuel #5)

sewage sludgeb (fuel #5)
DK straw (fuel #3) and sewage sludge a (fuel #5)
DK straw (fuel #3) and sewage sludge b (fuel #5)
straw in mix total fuel

[wt%]
fed [g]
0
0
9
9
34
50
59
71
50
75
86
86
67
80
89
86
525
525
578
578
790
1050
1292
1838
1050
1050
918
918
800
1338
2414
1876
ashing temp.
[C]
agglomeration
temp. [C]
885
798
802
891
804
804
800
799
798
800
799
750
798
800
800
795
970
939
974
956
968
915
899
942
937
974
944
956
932
919
914
920
From the experiments with fuel #3, the relation between ashing temperature and agglomeration
temperature becomes clear, see figure 3.16.
ECN-C-99-090
34
950
900
850
800
750
700
650
675
700
725
750
ashing temperature [C]
Figure 3.16 The relation between agglomeration temperature and ashing temperature in BFBC
experiments with DK straw (fuel #3). The dotted line represents the minimum
agglomeration temperature (equal to ashing temperature)
Furthermore, sewage sludge seems to have a positive effect on preventing agglomeration. In all
cases the addition of sewage sludge increases the agglomeration temperature as can be seen in
figure 3.17. Also kaolin as an additive improves the process, 3% kaolin increases the
agglomeration temperature with about 30C.
1000
950
900
850
800
750
700
0
20
40
60
80
100
straw in straw/sludge mix [wt%]
Figure 3.17 The effect of sewage sludge in a mixture of straw and sewage sludge
The FTIR pyrometric measurements turn out to clearly show the temperature distribution of the
particles in a fluidized bed during combustion. Figure 3.18 shows the temperature distribution
during test 6 (DK straw, fuel #3) during the combustion at 694C. It is clear that the range of
temperature broadens because of the combustion process. Particles have temperatures up to
100C higher than the median value. On the other side, colder particles exist, probably freshly
(cold) fuel particles.
ECN-C-99-090
35
0.1
Sand
Probability
Probability
0.1
0.01
0.001
0.0001
Sand
+
Straw pellets
0.01
0.001
0.0001
-50
0
D
T
50 100 150
-50
0
50 100 150
DT oC
Figure 3.18 Measured temperature distribution in BFBC at ETC during test 6 (DK straw, fuel
#3) at 694C bed average temperature at 80 mm above distributor plate (approx.
70 mm above bed)
Table 3.15 shows the measured particle temperatures during several tests. It can be seen that
particles can have 50-100C higher temperatures than the bed temperature. Peaks up to 130C
have been measured. Needless to say that this can have significant effects on the agglomeration
behaviour. This might be one of the main reasons for the relation shown in figure 3.16 where
the agglomeration temperature drops when the ashing temperature increases. At high ashing
temperatures there is a relative high probability that some particles have temperatures high
enough to initiate agglomeration.
The over-temperature apparently existing during combustion, is probably dependent on the fuel
material. Not only the type of material, but also physical characteristics as moisture content and
particle size. In the above-described experiments, all fuel were palletised. Because this might
decrease the overall reaction rate because of limitations of heat and mass transfer, excess
temperatures are probably less than in cases where fuel is fed in a non densified form.
Table 3.15 Overview of measured particle temperatures during the combustion tests in the
BFBC at ETC. Tpar is the average of the 10 hottest events, Tpar, max is the highest
value measured during the test, both temperatures shown as difference with bed
temperature
test
no
bed temp.
[C]
O2 in gas
[vol%]
fuel
probe position [mm

above distributor]
2
3
4
5
6
7
8
9
10
16
17
18
19
784
739
737
730
694
735
727
693
702
694
686
885
703
7.2
6.5
10.2
3.8
5.6
11.4
4.2
5.4
5.6
7.6
5.7
5.3
5.7
DK straw, fuel #3
DK straw, fuel #3
DK straw, fuel #3
DK straw, fuel #3
DK straw, fuel #3
DK straw, fuel #3
DK straw, fuel #3
DK straw, fuel #3
DK straw, fuel #3
sewage sludge, fuel #5
80
103
103
58
80
80
80
80
80
80
103
103
80
3.6.4
T par
[C]
63
42
53
42
58
85
90
75
98
62
80
35
43
T par, max
[C]
96
64
67
58
95
119
123
120
120
67
129
69
67
CFBG: circulating fluidized bed gasification facility (500 kWth ) at ECN (d)
The 500 kWth circulating fluidized bed gasification facility at ECN has been used to convert
wheat straw pellets from the Netherlands (fuel number 2) during a 10 hour experiment. The
gasification temperature was chosen to be low: 750C. Table 3.17 sums up the main results of
the experiment. Also shown are some results obtained with other fuels in the same facility.
What is clear from the figures is that the carbon conversion of the straw during the lowtemperature test was low, as low as 75%. This is one of the penalties for the choice of a low
process temperature. This however is not always the case. It is known that K, considered as the
ECN-C-99-090
36
major evil element for agglomeration, can also have a catalytic function for gasification
reactions. In a confidential experiment in the same facility using chicken manure, very high
carbon conversion rates were obtained unless the low process temperature (750C). This was
expected to be the result of the relatively high content of calcium and potassium as catalytic
active elements for gasification.
The chlorine analysis of the second cyclone ash revealed that about 50% of the chlorine
originally present in the straw is retained in the ash. This is a very high fraction compared to
other experiments (at temperatures between 800 and 850C) where this fraction is 10-20%.
Probably, chlorine compounds like KCl are preferably attached to solid particles at temperatures
around 750C. Table 3.16 shows the transfer coefficient of some elements in the ash from the
second cyclone.
During the test, every hour bed material samples were taken from the sealpot and analysed for
K-content. Figure 3.19 shows the concentration profile.
Table 3.16 The fraction of some elements from fuel appearing in cyclone
ash (transfer coefficient)
verge grass
Al
As
B
Ba
Ca
Cd
Cl
Co
Cr
Cu
Fe
K
Mg
Mn
Mo
Na
Ni
P
Pb
S
Sb
Se
Si
Sn
Sr
Ti
V
Zn
0.93
1.00
< 0.04
0.50
0.84
0.93
0.34
0.49
0.47
0.47
2.62
0.41
< 0.12
0.34
0.51
< 0.44
0.42
0.10
0.67
0.71
ECN-C-99-090
Dutch
wheat straw
0.90
< 0.48
0.87
0.95
0.92
< 0.12
0.48
0.31
1.00
1.43
0.94
0.79
1.05
1.03
0.87
1.29
1.05
0.95
1.60
0.21
< 0.16
1.12
< 0.18
0.97
0.96
0.95
1.03
cacao shells
< 0.60
1.73
< 0.04
0.78
1.50
1.24
0.52
0.58
0.75
0.83
1.32
0.08
0.29
< 0.24
< 0.36
0.46
< 0.04
0.78
1.53
average value for

CFBG at ECN
0.9
0.6
0.8
0.9
0.8
0.1
0.4
0.5
0.6
0.7
0.7
0.6
1.2
0.7
1.1
0.6
0.8
0.7
0.3
0.3
1.3
1.3
0.6
0.8
0.8
1.0
0.7
0.6
37
Table 3.17 Numerical results of the gasification tests using different biomass fuels in the
CFBG at ECN. Abbreviations: ar: as received (wet basis), HHV: higher heating
value, CC: carbon conversion, CGE: cold gas efficiency, na: not analysed
fuel
fuel #1
verge grass
reference
PROCESS CONDITIONS
fuel
kg/h ar
fuel water content
wt% ar
energy input (HHV)
kW
temperature
C
FUEL GAS
CO
vol% dry
H2
vol% dry
CO2
vol% dry
CH4
vol% dry
C2 H4
vol% dry
C2 H6
vol% dry
benzene
vol% dry
toluene
vol% dry
xylene
vol% dry
H2 S
vol% dry
HCl
vol% dry
NH3
vol% dry
Ar (calculated)
vol% dry
N2 (calculated)
vol% dry
H2 O (calculated)
vol% wet
light tars
mg/mn 3
heavy tars
mg/mn 3
dust
g/m n3
HHV fuel gas (calc.)
MJ/mn 3 dry
dry fuel gas flow (calc.)
m n3 /h dry
wet fuel gas flow (calc.)
m n3 /h wet
BALANCES and OTHER VALUES (calculated)
total mass balance
%
C-balance
%
O-balance
%
inert-balance
%
E-balance (HHV)
%
CC
%
GCE
%
ER
linear gas velocity
m/s
NH3 /fuel-N
%
H2 S/fuel-S
%
ECN-C-99-090
fuel #2
Dutch wheat straw
fuel #6
cacao shells
fuel #7
willow
[1]
this project
[1]
[1]
102.6
7.3
450
815
105
9.2
447
750
59.8
13.4
275
822
69
17.5
323
827
10.64
8.45
15.18
2.79
1.04
0.04
0.13
0.041
0.005
na
0.0002
1.25
0.60
59.5
8.4
na
273
6.0
4.90
206
224
12.85
4.68
14.06
3.56
1.39
0.18
0.18
0.063
0.012
na
0.0007
0.29
0.57
62.2
22.7
5400
na
22.9
5.12
155
190
8.00
9.02
16.02
2.34
1.13
0.05
0.16
0.06
0.03
0.002
0.0001
0.81
0.58
61.5
10.3
na
78
13.1
4.61
131
145
9.40
7.20
17.10
3.30
1.10
0.10
0.21
0.065
0.013
0.005
na
0.180
0.58
60.5
14.3
na
510
1.7
4.92
151
172
96.2
99.5
93.9
66
97.5
92.7
64
0.36
7.9
83
102.7
97.7
104.7
131
98.5
75.2
50
0.24
6.3
45
99.3
103.3
98.6
77
109.5
85.9
62
0.34
5.2
42
4
101.0
104.3
101.6
88
102.2
97.0
66
0.37
6.1
73
49
38
K in bed material [mg/kg]
3500
3000
2500
2000
1500
1000
0
10
time [hours]
Figure 3.19 The K-concentration in the bed material taken from the sealpot during the
gasification test with Dutch wheat straw (fuel #2)
From figure 3.19 it is clear that the K-concentration in the bed does not increase significantly
during the test: the increase of concentration of 1000 mg/kg in 10 hours corresponds to a
retention in the bed of less than 0.5% of the K from the fuel. So, most of the K leaves the
gasifier without some kind of bonding to bed material.
The absence of an increase of K-content is not what has been observed during the chicken
manure test already mentioned earlier. In this test (also at 750C), the K-content of the bed
material increased from about 2000 to 10000 mg/kg after 10 hours of testing. The difference
may be related to the difference in carbon conversion. In the case of straw, much carbon having
mineral components attached to it, are removed from the system. This way, elements like K are
removed from the system partially, preventing a fast build-up in the system. The gasification of
chicken manure showed high carbon conversion and therefore probably didnt remove the K
from the system fast enough.
Another difference between straw and chicken manure is the molar ratio K/Cl. This is 3.5 for
the manure and 1.7 for the straw indicating that the K in straw is more mobile. This might be
one of the reasons for the build-up of K when firing chicken manure.
3.6.5
DTA/TGA: differential thermal analysis and thermo-gravimetric analysis at

ECN and Ris (e)
The ECN and Ris results show good agreement, especially concerning the characteristic
temperatures of mass losses. Some minor disagreement is seen concerning the extent of mass
loss. This is probably reflecting the problem of obtaining fully representative fuel samples.
The difference between inert (N2 ) and oxidising atmosphere turns out to be small. Furthermore,
no heat effects were noticed during cooling and second heating cycles, which indicates that all
processes were non-reversible. Melting is associated with volatile potassium which evaporates
from the melt at the high temperatures of 1000C and 1200C, thereby preventing substantial
solidification effects in the cooling cycle. The high mass loss (>12 %) for pure ashes ashed at
450 or 550C and low mass loss (a few %) for ashes ashed at 650C supports this.
ECN-C-99-090
39
Visual inspection revealed that nearly all of the pure straw ash samples were completely fused
at 1000C while willow ash was typically only partly sintered. The tests performed on various
mixtures of ash and (100-125 micron) quarts sand revealed that the sand mainly behaves as
inactive and inert.
Figure 3.20 shows a typical result. It can be seen clearly that large endothermic effects take
place between 700 and 800C probably due to melting of potassium chloride (melting point
770C) or potassium silicates (melting point 740-780C).
Tables 3.18 to 3.20 summarise the main results of the experiments in nitrogen atmosphere.
Table 3.21 gives the results concerning experiments with straw ash (550C) mixed with sand
particles (100-125 micron) with varying ratio.
Table 3.18 Summary of DTA/TGA-measurements with ash from unweathered wheat straw
(fuel #3) in nitrogen atmosphere. The bold numbers indicate where melting may
start (large heat effect)
ashing
temp.
mass loss (TG)
mass loss rate (DTG)
heat effects (DTA)
450C
2-4% moisture
150 - 690C: 5-5.5 %
690 - 1000C: 6.5-8.5 %
550C
0.5-3% moisture
150 680C: 2-2.7%
680 - 870C: 7.2-9.2%
870 - 1200C: 1.5-4.1%
increase from:
650 - 875C
max. 725C
max. 835C
increase from:
680-870C
max. 720-735C
max. 840-845C
650C
1-2% moisture
150 680 C: 1.7-2%
680 - 1200 C: 9-12%
simultaneous large endotherms:

690 - 875C
750 - 800C
max. 780C
small and broad endotherm (Ris):
max. 725C
large endotherm (ECN):
700 - 800C
max. 770C
small and broad endotherm (Ris):
max. ~ 750C
simultaneous large endotherms (ECN):
680 - 875C
770 - 800C
max. 780C
increase from:
680 - 875C
max. 730-760C
max. 815-825C
Table 3.19 Summary of DTA/TGA-measurements with ash from weathered wheat straw (fuel
#4) in nitrogen atmosphere. The bold numbers indicate where melting may start
(large heat effect)
ashing
temp.
mass loss (TG)
450C
4.5-5.5% moisture
150 - 550C: 4-5%
550-680C: 5-6.5%
680-1000C: 6-8%
continuous from 150C with a clear

increase between 530 and 680C
max. 675C
550C
5 % moisture
150 - 570C: 5.5%
570 - 670C: 3%
670 - 1000C: 11.5%
total *: 20%
1.5 % moisture
150 - ~ 800C: 3%
800 1000C: 5.5%
650C
ECN-C-99-090
heat effects (DTA)
two large endotherms:

625-700C
740 - 950C
max. 685C
max. 880C
continuous from 150C with a small endotherms:
increase between 570 and 670C and 635C
a larger twin peak
770C
between 710 - 885C
max. 635C
max. 760C
max. 770C
continuous from 150 - 800C
small endotherms:
increase between:
580C
800 - 990C
930C (?)
max. 950C
max. ~ 580C
max. ~ 930C (?)
40
Table 3.20 Summary of DTA/TGA-measurements with ash from willow (fuel #7) in nitrogen
atmosphere
ashing
temp.
mass loss (TG)
heat effects (DTA)
450C
0.5-1% moisture
150 - 640C: 1-2%
640 - 740C: 9-13%
1.5% moisture
150 - 520C: 2%
520 - 710C: 7%
1.0% moisture
150 - 520C: 1.5%
520 - 660C: 1.5%
strong increase:
620 - 740C
max. 700C
strong increase:
520 - 710C
max. 650C
increase from
520 - 660C
max. 630C
endotherm coinciding with mass loss: 620 730 C

max. 700 C
endotherm coinciding with mass loss: 610 710C
max. 670C
small endotherm coinciding with mass loss:
610 - 660C
max. 645C
550C
650C
Table 3.21 Summary of DTA/TGA-measurements with ash from unweathered straw (ashed at
550C, fuel #3) mixed with silica sand (100-125 micron) in nitrogen atmosphere
wt%
sand
0%
mass loss (TG)
heat effects (DTA)
3% moisture
150 - 650C: 4%
650 - 1000C: 11.5%
strong endotherm in very broad range around

720C
max. 720C coinciding with mass loss peak
50%
1.5% moisture
150 - 650C: 2%
650 - 900C: 5%
75%
1.5% moisture
150 - 650C: 2%
650 - 900C: 5%
90%
0.5% moisture
150 650C: 0.5%
650 - 830C: 1%
0% moisture
two peaks between:

650 - 980C
max. 730C
max. 875C
two peaks between:
650 - 900C
max. 730C
max. 840C
two peaks between:
650 - 900C
max. 730C
max. 840C
small increase between:
650 - 830C
max. ~ 780C
essentially no increases
100%
ECN-C-99-090
endotherm in broad range around 730C

endotherm in broad range around 730C

very small endotherm at 570C

max. 570C
very small endotherm at 570C
max. 570C
41
ECN DTA/TG, Unweathered Wheat

Atmosphere N2 (100 %)
100
Weight in %
95
90
Ashing temperature
85
450 C
550 C
80
650 C
75
0
200
400
600
800
1000
1200
DTA in 10E-6 V
10.00
5.00
0.00
-5.00
-10.00
-15.00
-20.00
-25.00
-30.00
Figure 3.20 DTA and TGA signal of unweathered wheat straw in a nitrogen atmosphere, top:
weight as function of temperature, bottom: thermal effect (dip is endothermic)
Table 3.22 summarises the dangerous temperatures for the different biomass materials. It is
the result of a combination of both the DTA/TGA-results as shown above and the chemical
analysis of the samples.
Table 3.22 Temperatures at which agglomeration might occur according to DTA/TGAexperiments
fuel # and name
dangerous
temperature [C]
remarks
fuel #3: unweathered Danish wheat

straw
700C
fuel #4: weathered Danish

wheat straw
750C
fuel #7: willow
no signs of melting
below 1000C
substantial melting around 770C, probably due to melting of

potassium chloride (melting point 770C) or potassium
silicates (melting point 740-780C)
substantial melting around 850C, probably due to melting of
potassium silicates (melting point 740-780C) considering the
low chlorine content
only calcination occurs
Mixing wheat straw ash with sand resulted in roughly the same pattern as found in case where
no sand was added. No new peaks appear which indicates that silica sand has not a very big
influence on melting behaviour. This might be expected when considering that the pure ash
already contains considerable amounts of silica.
ECN-C-99-090
42
3.6.6
Compression strength measurement facility by Ris (TNO) (f)
Five materials were selected for compression strength measurements according to the standard
method described in chapter 3.3.6. The total carbon after ashing the samples was between 1 and
3% by weight, see table 3.23. It turned out that more than 80% of the carbon was inorganic
except for the straw where only 40% or less was inorganic carbon.
Table 3.23 Carbon content of different ash samples
prepared for compression strength measurements
ash sample
verge grass
Dutch straw
Danish straw
Danish straw (weathered)
cacao shells
chicken manure
fuel #
carbon content in
ash [wt%]
1
2
3
4
6
1.7
0.8
2.2
2.3
2.7
1.1
Figure 3.21 shows the compression strength of ashed samples of the materials as indicated. Also
shown is the curve for chicken manure. The results for Danish straw (both dry and weathered)
turned out to be impossible to obtain. The ash pellets changed to a sort of spongy structure
during heat treatment. The outer surface turned out to a smooth and molten structure with greybrown colour. SEM-analysis revealed a enrichment of silicon, potassium and sulphur in the
outer layer. Similar behaviour has been reported by [8].
The only explanation for the swelling behaviour during heating of the straw pellets is the
following. Because the outer surface melts, an impermeable layer is formed. During heating,
CO2 by the dissociation of carbonates, and possibly other gaseous products, are formed. The
impermeable outer layer prevents the gases to escape resulting in a swelling of the pellets. If this
is the reason this indicates that the temperature where melting occurs is lower than the
temperature where inorganic carbonates dissociate (650-750C).
From the figure it can be seen that the curves for verge grass and chicken manure show the
expected increase of compression strength with increasing temperature. It can be concluded that
sintering occurs between 800 and 850C for verge grass and between 900 and 950C for
chicken manure. For cacao shells, temperatures probably were not increased to high enough
values to see any sintering. This means that cacao shells show sintering, according to
compression strength measurements, above 900C. Dutch straw shows strange behaviour. This
probably is caused by the same effect as mentioned earlier for the other straw samples. In this
case however, the effect is not so dramatic. This might well be related to the relative low carbon
content of the Dutch straw ash, see table 3.23. Table 3.24 summarises the results of the
compression strength measurements.
ECN-C-99-090
43
compression strength [N/mm2] .
80
70
Dutch
straw
60
50
chicken
manure
40
30
verge grass
20
10
cacao
shells
0
600
700
800
900
1000
temperature [C]
Figure 3.21 Compression strength measurements on different materials after ashing and
heating to indicated temperature
Table 3.24 Sintering temperatures according to compression strength
measurements with ash
ash sample
verge grass
Dutch straw
Danish straw
Danish straw (weathered)
cacao shells
chicken manure
3.6.7
fuel #
sintering temperature
[C]
1
2
3
4
6
800-850
< 750
< 750
< 750
> 900
900-950
Standard facilities for measuring standard ash melting temperatures by ECN

(NMi) and Ris ( DK-TEKNIK) (g)
Using standard techniques the melting temperatures were determined of ashes of three materials
ashed at three temperatures each. Figure 3.22 shows the different temperatures as measured in a
reducing atmosphere. Figure 3.23 shows the difference between reducing and oxidising
atmosphere for two samples.
ECN-C-99-090
44
1600
temperature [C]
1400
1200
1000
800
65
0
w
as
he
d)
w
(ra
he
45
in
at
0
w
st
a
sh
ra
w
ed
(ra
)5
in
50
w
as
he
d)
65
0
55
0
st
ra
w
(ra
in
w
he
at
w
he
at
st
ra
w
w
he
at
st
ra
w
45
0
st
ra
w
w
he
at
st
ra
w
65
0
w
he
at
wi
llo
w
55
0
wi
llo
w
wi
llo
w
45
0
600
IDT
ST
HST
FT
Figure 3.22 Melting temperatures for three fuels ashed at three temperatures (450,
550 and 650C) in a reducing atmosphere (20% CO in nitrogen)
1400
wheat straw (red)
wheat straw (ox)
temperature [C]
1200
wheat straw (rain washed) (red)

wheat straw (rain washed) (ox)
1000
800
600
IDT (C)
HST (C)
FT (C)
Figure 3.23 The influence of atmosphere on standard melting

temperature of two wheat straw ash samples ashed
at 550C, red: reducing (20% CO in nitrogen),
ox: oxidising (air)
3.6.8
SEM and HTM: high temperature microscope at Ris (h)
A comprehensive set of SEM measurements has been performed on the main part of the
samples taken from the ETC fluidised bed agglomeration experiments and they confirm that the
particles in general are covered with a more or less homogeneous surface layer. The
measurements are taken both at different locations on the surface of the particles before and
after agglomeration, and if possible also at the bridge between agglomerated particles. These
measurements will be very useful in modelling of agglomeration when fundamental
thermodynamic data for intermediate phases become available. It is interesting to notice that the
spectroscopic data from all the sludge experiments are nearly independent of experimental
conditions including the mass ratio between straw and sludge.
ECN-C-99-090
45
Ash taken from the ETC fluidized bed at the end of the ashing process and before agglomeration
has been used in the HTM microscope. The samples were heated in the hot stage but it was not
possible to detect any change in the picture as a function of stage temperature. When not
agglomerated samples, which had proved in the fluid bed that they could agglomerate, were
heated in the HTM to a temperature where they definitely should agglomerate they sometimes
agglomerate and sometimes not and there was no difference in the pictures. Several samples
taken from different fluidized bed experiments gave the same negative response. The method
could therefore not be used to observe at which temperature agglomeration occurred.
A reason for the negative result may be that the atmosphere (vacuum) in the ESEM instrument
during the heating of the samples is very different from the condition in a fluid bed. In particular
volatile components may be removed from the sample by the pumping system of the
microscope.
3.6.9
BFBC: bubbling fluidized bed combustion facility by Ris (VTT) (i)
Both the dry Danish straw and the weathered Danish straw turn out to result in defluidization
at 775C while at 750C no sintering occurred. In both cases no visual agglomerates could be
detected afterwards. However, by microscopy the agglomerates could be seen clearly. The
sintering behaviour seems to be more pronounced using fuel #3 than the weathered straw (fuel
#4).
3.7
TASK 7: Description of the total methodology for prediction of ash

agglomeration and related problems, and identification of potential
solutions
In the present project, the focus has been on methods for the prediction of agglomeration and
related problems. This however is only useful if measures can be taken to avoid the problem. In
this chapter several solution methods will be shortly described. In table 3.25 the different
methods are given.
ECN-C-99-090
46
Table 3.25 Solution methods, divided into different categories

-
solution category
growing and harvesting
fuel pre-treatment
blending
fluid dynamic control
bed particle management
temperature control
other measures
3.7.1
method
using fertiliser with low content of chlorine
postponing the year of harvest in short rotation forestry
avoid contamination with soil particles
leave some of the growth layer biomass components behind in the field/forest
allow weathering
removing non fuel components
degrading oversize fuel material
removing silica contamination
washing out alkaline and chlorine
homogenisation
blending with coal and other fuels
avoiding certain fuel combinations
combinations of high alkaline and high chlorine fuels
high sulphur coal and high alkaline biomass
high alkaline biomass and high silica coal/biomass
choose proper gas velocity with given particle size
staged air or increased cross section at bottom
secure good air distribution at all loads
high fluidisation velocity
frequently/occasionally increasing fluidization velocity
limiting the turn down ratio
remove oversize material from the bed
addition of fresh bed material
limiting the retention of small particles
selectively draining problematic components
additives
avoiding temperature excursions
limiting the reaction temperature, special reactor concepts
limiting single particle temperatures
avoiding hot internal surfaces in critical areas
choosing a sufficiently high temperature
avoid condition of re-carbonation of CaO
avoid condition of sulfidation of static Ca-compounds
avoid condition of sulfidation of K/Na-compounds
Growing and harvesting
The elements considered to be (partly) responsible for agglomeration are: K, Na, Cl, P, S and Si.
Chlorine is not essential for growth. The chlorine content of biomass therefore is mainly the
result of soil condition (location near the sea) and fertilisers used. Most fertilisers contain
chlorine because these (KCl) generally are cheaper than e.g. sulphates.
Alkalis (mainly K) is essential for growth and therefore is concentrated in fast growing parts.
This explains the high concentration in annual crops but also the relative high amounts in young
branches, leaves and needles from trees. One way to reduce the share of young biomass is to
postpone the harvest or leave those materials in the wood as natural fertilisers.
The presence of silica, generally soil components, in the biomass fuel can be avoided by careful
harvesting. It may be worthwhile accepting some waste by not sampling too close to the ground.
The concentration of alkalis and chlorine in biomass can be reduced considerably by allowing
rain to leach K and Cl. Large reductions can be achieved. Some disadvantages are the delaying
effect on potential subsequent crops, a decrease in biomass yield and the mentioned tendency
for higher contamination with soil material. Also the material has to be dried.
ECN-C-99-090
47
3.7.2
Fuel pre-treatment
In practice stones, pieces of iron and other tramp materials will find their way into the
feedstock. Such heavy components may accumulate in the bottom of the reactor/bed and
obstruct the normal flow conditions, which can limit the local mass and heat transport and lead
to severe agglomeration. Therefore whenever there is a risk of heavy non-fuel components in
the fuel supply the fuel preparation and/or the feeding system should comprise means for
trapping the unacceptable components.
A too large fraction of coarse fuel particles will negatively impact the quality of fluidisation.
Especially under slow reaction conditions coarse char particles may accumulate to a
concentration that negatively influences and eventually destabilise the fluidisation. Such
destabilisation easily leads to hot spots that promotes local sintering due to the obstructed heat
transport and within a short time a large part of the whole bed may be involved in this
irreversible process. Paradoxically, this risk is especially present in fluidised bed systems
operated at low temperature, where the reactivity is low. Mechanical breaking and crushing may
be a solution. It should be mentioned that the resulting fuel should not contain too much fines
because of the risk of excessive carbon losses in the process caused by the relatively low
residence time of fine particles.
Soil (mainly silica) attached to the fuel may be separated from the fuel by washing, sieving or
fluid dynamic separation. The attached sand can be removed from the soil more easily if the fuel
is dry, which might be relevant when sieving is the pre-treatment. However the mentioned
precaution is probably not worthwhile considering if not also limiting or completely avoiding
the addition of silica along with a potential addition of supplementing bed material. This means
that the benefit from removing (non-fuel inherent) silica contamination is also doubtful when
the inherent fuel ash contains a high concentration of silica. In such cases the best strategy
might as well be the opposite, i.e. to try to flush the alkaline/trouble components out of the
bed by allowing a large addition of silica with the fuel.
Alkalis and chlorine can be washed out using hot water as a fuel pre-treatment. This solution is
evaluated by both of the Danish power companies ELSAM and ELKRAFT and the Institute of
Chemical Engineering at the Danish Technical University and a demonstration facility is
expected to be build at the power plant Studstrupvrket.
Agglomeration and deposition problems are sometimes seen even though fuel analyses indicate
an unproblematic fuel. The reason may be that the fuel quality varies beyond what was indicated
by even frequent fuel analyses. Besides this, changing fuel composition leads to changes in
process temperature, which increases the risk of agglomeration. In order to limit the risk of
frequent agglomeration and deposition problems due to such fluctuation in fuel quality,
homogenization of the feedstock may be worthwhile.
3.7.3
Blending
Gasification tests conducted by the Danish power companies ELSAM and ELKRAFT at VTT
and Enviropower test facilities showed that even straw with a high content of alkaline and
chlorine may be efficiently gasified without agglomeration and deposition problems if mixed
with coal. The mechanism primarily considered responsible for the positive influence of cofiring with coal is that a major part of alkaline from straw are bound into high melting
aluminium silicates found in the coal. Furthermore a lower content of alkaline and chlorine in
coal gives a diluting effect that may also be obtained by co-firing other less problematic types of
biomass such as wood.
ECN-C-99-090
48
Furthermore there may exist some lucky combinations of biomass-biomass and biomasswaste fuels, i.e. combinations that give less trouble than either of the fuels used separately.
From the present study, sewage sludge and straw seem to profit from each other: agglomeration
risk seems low.
On the other hand, combination of fuels having high alkali and high chlorine concentrations
should be avoided. Chlorine mobilises the alkalis which otherwise might not have been active
as far as agglomeration processes are concerned. In this study, this phenomenon has been
demonstrated by the fact that cacao shells (high K, low Cl) do not show the expected
agglomeration problems.
Also the combination of fuels rich in KCl (e.g. straw) and sulphur (some coals) should be
avoided. As experienced at the Grenaa CFBC plant, large depositions form on heat absorption
surfaces. The main mechanism is suspected to be KCl condensation and the subsequent reaction
between KCl and SO x forming KSO x.
Furthermore, the combination of high alkali and high silica fuels should be avoided. This
however is only worthwhile when alternative bed material (not silica) is used to solve the
agglomeration problem.
3.7.4
Fluid dynamic control
In general the phenomenas designated as deposition, bed sintering and agglomeration can be
seen as the result of a competition between adhesion of particles to surfaces of the construction
and/or neighbour particles at one hand and on the other hand wear on the adhered particles due
to mechanical forces induced by e.g. other particles. Furthermore the movement of particles is
also important because of the corresponding transport of energy. In many fluidised bed reactors
this transport is so effective that temperature differences are kept down to only a few C across
the square section of the bed. However, an ineffective/obstructed particle movement can easily
lead to local temperature derivations resulting in a further and potentially self-enforcing
obstruction of the particle movement. So, two good reasons exist (mechanical forces on
particles and temperature homogeneity) why fluid dynamics are playing a major role in the
process of agglomeration. More specific points are described below.
Especially for small particles and large cross sections the superficial gas velocity should always
be kept several times the theoretical minimum fluidisation. Important to realise that the particle
size (and distribution) of the bed material does change during operation. This means that
choosing proper gas velocities must be seen together with what is often called particle
management, see later.
In CFB-reactors, usually a nominal gas velocity in the range of 5 10 m/s is chosen. In the
resulting flow field it is only the sufficiently small particles (e.g. < 0.2-0.3 mm) that can be
efficiently suspended by the upwards drag forces. Larger particles concentrate at the bottom of
the reactor, possibly creating temperature inhomogeneity and obstructions, increasing the risk of
agglomeration. On the other hand, in bubbling beds the additional effect of buoyancy forces
allows also large particles to be suspended. It is possible (and also very usual) to benefit from
this effect in CFBs, either by adding only a part of the reactant gas in the very bottom or by
enlarging the square section of the bottom of a CFB reactor. In the first case this means that a
considerable part of the reactant gas is added as secondary or tertiary gas some distance above
the openings for adding the primary gas.
The risk of uneven gas distribution through the different nozzles becomes real when the flow of
gas (usually air) is reduced. This might induce inhomogeneous distribution of material and
ECN-C-99-090
49
temperature, finally increasing the risk of agglomeration. Therefore in cases where a high turn
down ratio is wanted, gas should be added via individually controllable lines.
High fluidisation velocities always lower the tendency for agglomeration and deposition
problems because of the already mentioned competition between particle adhesion and friction
forces.
In some occasions it is seen that agglomerates already formed can be disintegrated by
temporarily increasing the gas velocity. Increasing the gas velocity will mainly work on weakly
bounded agglomerates that are typically formed rather fast at relatively high temperatures [9].
Agglomerates formed in the BFBG at ECN, used in the present project as a result of the
standard test, turn out to be weak (like a sandcastle). Generally, emptying the bed by removing
the bottom creates enough forces to render the major part of the reactor contents into individual
particles.
Increasing the gas velocity may also work in a negative direction. This is when the agglomerates
are stronger and the settlement on the base of the bed is already pronounced. Strong and slowly
settling agglomerates are more typically formed in the lower temperature range. Increasing the
gas velocity in the primary zone may then give a worse situation by reducing the bulk density in
the still fluidised areas adjacent to the agglomerates and thereby reducing the buoyancy forces
acting on the agglomerates. In this case a better strategy is to frequently or occasionally
drain/remove the agglomerates from the bed even if the plant has to be stopped. It should be
mentioned that a sudden increase of air flow can also result in an increase of temperature, which
might worsen the situation.
Fluidised bed reactors used for energy production are often operated at part load conditions i.e.
at reduced gas throughput. Due to reasons already mentioned, this situation typically constitutes
the worst case concerning proper distribution of the fluidising gas and the quality of fluidisation.
If problems occur and the already suggested measures shows insufficient, it may be necessary to
consider a less ambitious turn down ratio.
3.7.5
Bed particle management
In lab-scale (short time) experiments, steady state is hardly obtained with respect to the
composition of the bed material. In fact the bed composition during the test period may be close
to the composition of the particles initially loaded. When running a real world plant on a more
continuous basis, the bed material (composition and size) may after some days be very different
from the particles initially loaded. The bed is more the result of the choice of fuel and other
design and the operating parameters such as type and extent of bed additives, the efficiency
characteristic of particle separators, the re-circulated share of separated particles, the location of
bed particle drains and rate of draining and the classification and partially re-introduction of
drained particles. The importance of such parameters was recognised during the 80s and the
cowering phrase particle management was invented. Some related potential solutions to
agglomerating problems are mentioned below.
In cases where there is a risk that agglomeration takes place and/or it is not possible to remove
oversize particles by fuel preparation (as earlier mentioned) the reactor should be designed in
such a way that even the worst possible types of tramp material passing through the preparation
and feed system as well as agglomerates formed in the bed can be effectively drained out. Some
of the major design means for effective draining is:
designing for a heavy state of fluidisation in the bed bottom (choice of low fluidisation
velocity and/or coarse particles) in order to make even relatively coarse particle float by
buoyancy.
ECN-C-99-090
50
on the other hand the gas flow in the reactor bottom should be high enough to keep the
fluidisation vigorous and stable at all times.
providing overall particle flow conditions that provides upwards (gulf-) streams in critical
areas
inclining the bed bottom against draining openings/means
using directional air/gas nozzles
using nozzles shaped in order not to hold up irregular bed components
providing an open type of bed bottom (e.g. a so-called sparge pipe distributor) in stead of a
base plate with nozzles.
The addition of new bed material can keep the bed content sufficiently well sized and
chemically correct. In this way, the accumulation of certain elements is prevented. This method
is used at the CFB combustion plant in the Danish town Grenaa, where the fuel is low sulphur
coal in combination with straw (maximum 60 % straw by energy). By adding approximately 20
t of coal ash and 4 t of silica sand every 24 hours, the concentration of water soluble K
compounds in loop seal particles are kept below 2000 ppm. In this case the limit is also set in
order to avoid excessive deposition on the superheater tubes in the back pass.
The potential needs for adding fresh bed particles is sometimes fully or partly covered by
adding particles that are chemically active in absorbing particularly sulphur or chlorine,
catalytic cracking of tars in gasifiers or absorbing alkaline components in high melting
components usually called getter materials.
Different identified or anticipated effects of various additives have been reported or can be
suspected:
physical absorption of melted phases within porosity of added particles
chemical absorption of alkali metals into stable components with high melting points
powdering that makes melted phases act less sticky to the surfaces of other particles
interaction that change the melting/strength characteristics of problematic components in a
positive direction
The table below includes additives that at least in some cases are suspected to offer the above
mentioned positive effects.
Table 3.26 List of possible additives preventing agglomeration
type of effect
physical absorption
chemical absorption
powdering
additive
limestone
pre-calcined dolomite
sintered magnesite
gibbsite
coal ash
aluminosilicates
bauxite
diatomaceous earth
kaolin
coal fly ash
principal components
CaCO3
CaO.MgO
MgO
Al(OH)3
AlxSiy Oz
SiO 2 7.0-10.0 %, Al2O3 81.5-88.0%
SiO 2 92 %, Al2O3 5%
Al2O3(SiO 2)2(H2O)2
AlxSiy Oz, Ca-/MgO, etc
CaCO3
CaO
MgO
CaMg(CO3)2
CaO.MgO
Limiting the retention of small particles reduces the risk of agglomeration. The retention of
especially small particles/fines may enhance sintering and agglomeration due to various
reasons:
the large specific surface area generally enhances surface phenomenas and e.g. adhesion
forces dominate compared to inertia and gravity forces
ECN-C-99-090
51
volatile ash components such as many alkaline components tend to concentrate on the large
surface area of fine particles by condensation and absorption
also less volatile ash components that includes problem components tends to be liberated
mainly as small particles, i.e. these components tends (conveniently) to be elutriated and
lost if not re-circulated
A high degree of particle re-circulation has also in BFB reactors been emphasised due to the
positive influence on carbon conversion, tendencies for lower boiler emissions and decreased
needs for adding fresh bed material. However, and as noticed by [10], the consequence has often
been agglomeration problems. Limited retention of ash components is probably also part of the
explanation for some mainly successful tests conducted at the HTW gasifier at RWTH, Aachen
in Germany [11]. In this case miscanthus, triticale, barley and even highly chlorine containing
ray grass are efficiently gasified.
3.7.6
Temperature control
Agglomeration is in most cases caused by melted ash or ash derived components condensed at
particle surfaces and acting as to glue particles together. It is therefore obvious to try to avoid
agglomeration by keeping a sufficiently low temperature. However, there are several other
temperature-related mechanisms to consider.
Temperature overshooting during start up in combustors and gasifiers (start up and shut down
usually in combustion mode) is a serious problem. The delayed reaching of the set point
temperature often results in a major increase in fuel feed rate due to the very natural impatience
of the operator. The result may be a situation with an unrecognised excessive amount of char in
the bed. Typically this char start to react at a much faster rate shortly before reaching the pre-set
operation temperature. The response to a decreased fuel feed rate at this stage may be an even
faster temperature increase due to the lack off the temperature moderating effect of moisture
evaporation and pyrolysis reactions. A similar risk exists when shifting to a less reactive fuel. In
this case the natural reaction of bringing in more fuel when the temperature initially falls
(during building up the new balance concentration) may lead to more or less the same situation.
A good test during start up is to see the temperature response on a short feeding stop e.g. 50C
before the operation temperature is reached. If the temperatures do not start to level out within a
few minutes, this can be an indication of a dangerously large char accumulation. It is also a
good idea not to exceed a previously experienced or calculated feed rate when approaching the
operation temperature. So, avoid temperature excursions during start-up and shut down by
simply being patient and careful.
Simply lowering the process temperature generally results in unacceptable carbon losses due to
the low reaction rates. Special concepts can solve this dilemma. One of the possibilities is the
concept by DFBT where char is converted at low temperature and long residence times [12].
Local temperature peaks should be avoided. Several measures can be mentioned:
improving the distribution of fuel and oxygen
moderating the temperature near gas nozzles e.g. by adding steam to gasifiers or recirculating flue gas to combustor
better mixing by e.g. providing a more aggressive state of fluidisation in critical areas
inducing in-bed golf streaming through critical in-bed areas
allowing a higher particle flux through the freeboard or other critical parts of a CFB reactor
addition of properly distributed heat absorption surfaces. E.g. providing wing/panel walls
in the free board of a BFB or the upper part of a CFB reaction chamber
ECN-C-99-090
52
It is well known that especially in areas of a reaction chamber with a high oxygen concentration
the temperature of burning char particles may be considerably higher than the mean temperature
of the surroundings of the particles. In this report this is reported as well. Some counteractions
against the above problems would be to avoid high concentrations of oxygen in areas with high
concentrations of char. Measures such as improved distribution of combustion air or
gasification agent (improved grid and nozzle design e.g. staged oxidation) may be a solution.
Furthermore, it could be considered to oxidise the char in areas with low concentrations of silica
particles that may contribute to the formation of eutectics. Eventually oxidation at low (mean)
temperatures should limit the mentioned problem due to the generally slowed down reactions
kinetics and higher margin from bed temperature to temperatures of melting and eutectics
formation.
It is well known that sticky particles have much higher tendency to stick to high temperature
surfaces than to surfaces at low temperatures. If sticky particles can not be avoided in certain
areas in a reaction chamber or downstream passages it may therefore be a solution to use a
sufficiently cooled wall construction instead of insulating refractory liners.
At high temperatures, salts might not condense and result in sticky substances. It is therefore
possible that agglomeration can be avoided not only by decreasing but also by increasing the
process temperature. Some evidence for this is seen from experimental results presented by
[13]. In several cases during slowly increasing the temperature of beds of various compositions
some self-dissolving thickening tendencies of the fluidisation are observed at temperatures
above 800C. However, the alkalis evaporated may condense and course problems at lower
temperatures downstream and during the temperature increase some of the alkalis may
recombine into other sticky components with lower vapour pressure.
3.7.7
Other measures
One of the main types of agglomeration mechanisms is not based on the presence of melted
phases and is some times called sintering/agglomeration by chemical reaction. Occurring on
the interface between adjacent particles some reactions can form a solid bridge between the
particles. Such a reaction is the re-carbonation of CaO to CaCO3 :
CaO + CO2 CaCO3 (exothermic)
It can be seen that the potential for this reaction to take place can be increased by:
the presence of a high concentration of CaO
a high partial pressure of CO2
decreasing temperature
The risk of sintering/agglomeration due to re-carbonation is particularly present when using
highly Ca-containing fuels, bed materials and/or additives and in locations of the process where
the temperature is decreased enough to make the carbonate the thermodynamically stable Ca
component (below approximately 700C in atmospheric reactors). However practical
experiences shows that a further important (necessary) circumstance is particle stagnation, i.e.
the problem tend to appear in slowly / badly fluidised areas or as deposition on cooled surfaces.
Severe problems are seen in especially the loop seals of petro coke fired boilers. The risk for
problems due to re-carbonation are high in a pressurised plant which has clearly been
experienced at the bio-IGCC plant in Varnamo [14], where depositions in the raw gas cooler
have been experienced when gasifying woody fuels at up to 18 bar. However, these problems
seem to be solved by using magnetite as bed material [15]. Depending on the application /
choice of fuel some solutions that may be worth considering are:
ECN-C-99-090
53
avoid combinations of temperatures total pressures that enhance the tendency for Cacompounds to alternate between calcinating and re-carbonating conditions.
avoid stagnant zones near gas distribution grids and in particle return ducts and particle
coolers.
limit the CO2 concentration in critical areas (e.g. limiting the char concentration and/or
adding steam or nitrogen in stead of air/oxygen).
choose proper bed materials/additives.
precautions against deposition in design of flue gas and product gas coolers.
A further example on sintering/agglomeration by chemical reaction is based on the sulphating

of Ca compounds, e.g. by reactions:
CaO + SO 2 + O2 CaSO4 (exothermic)
CaCO3 + SO3 CaSO 4
(exothermic)
Except for the replacement of CO 2 with SO 2 /SO 3 , it can be seen that the nature of the potential
agglomeration mechanism is similar to the mechanism based on re-carbonation. Avoiding
problems due to sulphating of Ca/CaCO3 may involve:
avoid stagnant zones near gas distribution grids and in particle return ducts and particle
coolers.
limit the SO x concentration in critical areas (e.g. designing the reactor in such a way that the
liberation of SO x as well as its subsequent absorption are primarily taking place e.g. in the
vigorously fluidised main reaction chamber.
provide sufficient clearance and cleaning in heat absorption surfaces.
One of the risks in biomass fired FB/CFB boilers is the formation of agglomerates/deposits
based on condensation of KCl on heat absorption surfaces and the subsequent formation of
K2SO4 according to the reaction:
2KCl + SO 2 + 2H2 O K2 SO4 + 2HCl
Very severe deposition problems experienced at the Grenaa CFB boiler when using high
alkaline straw in combination with a high sulphur coal have been explained by this. It is
reasonable to assume that the same reaction goes on in relation to KCl contained in/on
fluidised/suspended particles and perhaps also from KCl in the gas phase, but it is not clear to
what extend such reactions are causing agglomeration problems. Some possible solutions are:
avoid the combination of large inputs to the process of K/Na, chlorine and sulphur (e.g.
avoid co-firing straws/grasses and high sulphur coal).
limit the release of K/Na to the gas phase and/or promote its rebinding into non-sticky solid
phases inside the reaction chamber.
provide some rows of so called screening tubes in front of superheater surfaces in the
convection pass (deposits will form on the screening tubes but tend to peal off due to
thermal stresses induced by a relatively low surface temperature.
provide sufficient clearance and cleaning in/of heat absorption surfaces.
It should be mentioned that concerns about corrosion unfortunately points away from the first of
the stated solutions (avoid combination of K and S). The sulfation of KCl on superheater
surfaces helps to avoid the risk of severe Cl-corrosion by removing chlorine from the deposits
[16]. According to this some input of sulphur may be an advantage.
ECN-C-99-090
54
3.8
TASK 8: Conclusive evaluation
In the following table, the resulting agglomeration temperatures are summarised as determined
using the methods as described in chapter 3.3 and 3.6 of this report. Figure 3.24 shows the
results in a graphical way.
Table 3.27 Agglomeration temperatures (or range if more tests are done) measured within the
project with different methods for different fuels
fuel#:
verge
grass
Dutch
straw
Danish Danish sewage

straw weathered sludge
straw
test method
cacao
shells
willow remarks
a chemical analysis
no temperature output
b BFBG, gasification (5 kWth)

c BFBC, combustion and controlled
agglomeration (5 kWth)
d CFBG, gasification (500 kWth)
825
810
> 810
800880
> 750
e DTA/TGA, melting temperature of

ash and ash/bed
f compression strength of sintered ash
pellets
g standard ash melting temperature
800850
800
730890
900
830860
940970
> 820
< 750
h HTM, heating while observing

ash/sand
i BFBC, combustion and controlled
agglomeration (5 kWth)
700
750
< 750
< 750
930
905
775
775
1370
1100
1100
1000
1000
1000
900
900
900
800
800
800
verge grass
d e
f
NL straw
1100
1100
1100
1000
1000
1000
1000
900
900
900
900
800
800
800
800
700
b
c
d e f
g
DK weathered straw
700
b
d e
f
g
sewage sludge
DK straw
1100
700
IDT, average value for

the three ashes
not possible
700
700
b
0.5 mm sand, standard

X-test
dependent on ashing
temperature
> 860 no agglomeration
occurred
> 1000
> 900
1100
700
875
700
b
d e
f
g
cacao shells
e
f
willow
Figure 3.24 Graphic representation of results from table 3.27, agglomeration temperatures
measured within the project with different techniques. (b): BFBG, lab-scale
gasification at ECN, (c): BFBC, lab-scale combustion at ETC, (d): CFBG, benchscale gasification at ECN (e): DTA/TGA, (f): compression strength of sintered ash
pellets, (g): standard ash melting temperature, (i): BFBC, lab-scale combustion at
VTT. Large error bars are caused by the measurement of only a minimum or
maximum value.
ECN-C-99-090
55
Chemical analysis of the fuel can give a first indication of whether there might be an
agglomeration problem during thermal conversion. In general a high K-content means a risk for
agglomeration. However, cacao shells (very high K-content, roughly twice that of straw) do
have significant less tendency for agglomeration than straw as found using the BFBG at ECN
(b). Probably the very low Cl-content of cacao shells renders the K less mobile and therefore
less reactive. The result of this relative low K-mobility is that agglomerates are not formed by
the melting of K-salts (like KCl) but by the formation and melting of K-silicates (with e.g. bed
material), which means that the agglomeration temperature shifts to (50-100C) higher
temperatures.
If high-K / high-Cl fuels show agglomeration by simply forming and melting salts, this
implicate that bed material is not playing a major role. This is shown in experiments with the
BFBG at ECN (b) with different bed materials (silica, alumina and mullite) with straw as fuel.
The agglomeration temperature in all cases was 840C.
place. However care must be used when comparing results from different facilities as both
experimental conditions (including fuel preparation) and the design of the fluid bed will
influence the result. Fluid bed systems offer the possibility to easily compare different measures
like additives and bed materials. Contrary to methods like DTA, compression strength and ash
melting temperatures, fluidized bed experiments combine all processes that might be relevant
for agglomeration.
From particle temperature measurements in a BFB combustion facility at ETC (c) it turns out
that burning particles can easily have temperatures of up to 100C in excess of the mean bed
temperature. This has a large influence on agglomeration behaviour. This relation also is
illustrated by the relation that is found between ashing temperature (at which the fuel is
combusted prior to an increase of temperature up to the agglomeration) and agglomeration
temperature, see chapter 3.6.3. Because gasification, contrary to combustion, is a process where
peak temperatures are lower or even absent, one might expect that agglomeration during
gasification will occur at higher temperatures than during combustion. This however can not be
concluded from the experiments, once again illustrating the importance of other factors like the
design of fluid beds (the gas distribution, the type of nozzles, etc.).
The use of additives is one solution which might be used to mitigate agglomeration problems. It
has been shown that kaolin, magnesite and dolomite (2 wt% of fuel flow) increase the
agglomeration temperature with at least 60C during gasification. During combustion tests a
30C improvement has been demonstrated for kaolin (3 wt%) added to straw. Also the addition
of sewage sludge shifts the agglomeration temperature to considerable higher ranges. This
already is significant at the lowest sludge/straw-ratio tested (14wt% sludge in mix). It might
well be that sludge slows down the combustion reactions and therefore decreases peak
temperatures. As sewage sludge is available in enormous quantities, and represents a waste
problem in itself, this may be one way to reduce/eliminate the agglomeration problems when
using alkaline rich biofuels. Gibbsite, thought to be a physical getter for alkalis, does not
change the agglomeration temperature.
In the lab-scale fluidised bed facilities at ECN (b) and ETC (c), the effect of accumulation of
elements is shown. More fuel/ash in the bed decreases the agglomeration temperature. An
increase of fuel (ash) fed to the system by a factor of 3 decreases the agglomeration temperature
with about 25C. For Dutch straw in the ETC combustion facility this was even circa 75C.
DTA/TG investigations (e) can at present only give an indication of a temperature range where
agglomeration might occur because the energy exchange from the agglomerating components is
to some extent hidden by other chemical reactions. However, the method might be improved by
ECN-C-99-090
56
including other possible chemical reaction mechanisms and their influence on the energy
balance and from this introduce a correction term to the measured temperatures.
Compression strength measurements (f) may give an indication of the temperature where
agglomeration may start to occur. In practice, some problems exist. For example unweathered
wheat straw was swelling during the heat treatment and compression strength measurement
could not be performed.
ESEM/HTM measurements (h) do not give any reliable indication of agglomeration during the
heating process in the microscope and a following examination of the samples was also
unsuccessful.
SEM measurements give very useful information of the composition of the surface layer of the
different bed particles and of how agglomeration is distributed. These measurements give vital
information for modelling a possible surface interaction between bed particles.
Thermo-chemical modelling of the processes taken place between interacting particles is at the
present still in an early stage, but fundamental work may be a realistic way to improve the
accuracy. Further investigations in e.g. fluid beds may not be the best way to improve the
understanding of the fundamental mechanisms, which are involved in the agglomeration
process.
Experiments in the CFBG facility at ECN (d) did not result in agglomeration. A CFB is less
suitable than a BFB as a tool for agglomeration research because temperatures in a CFB are less
homogeneous (differences of 10-20C are normal) and agglomeration therefore is not as sharply
recognisable as it is when using a BFB (temperature differences of only a few degrees). During
the experiment with straw, The K-content of the circulating sand reached a constant level of
about 3500 mg/kg, already after a few hours. The bulk of the K (80%) and many Cl (50%) was
found in the ash separated by the second cyclone. The rest is probably leaving the system with
the gas. Apparently the first cyclone of the CFB has a relatively low efficiency for small (ash)
particles, preventing these from circulating and accumulating in the reactor. This has been
demonstrated before by the fact that bed material with 270 m (instead of the normal 500 m)
was completely blown out of the system within a few hours. During the test with straw at low
temperature (750C), much carbon left the system unconverted. Because this means that
captured minerals are leaving the system as well, the (generally unwanted) low carbon
conversion has a positive effect. So, conversion at low temperature prevents agglomeration for
two reasons: the low temperature and the removal of significant amounts of minerals with badly
converted fuel.
Standard ash melting temperatures (g) proved to be not related to agglomeration in a way that
can be used when trying to predict agglomeration. The difference between reducing and
oxidising atmosphere couldnt be shown unambiguously.
As main conclusion it can be said that elemental analysis always should be performed since it
gives a good first indication of the risk of agglomeration. Fluidized bed experiments combine all
relevant processes taking place in real conversion plants and therefore suitable to be used as
prediction tool. Especially BFBs, because of the high temperature homogeneity, have shown to
be accurate and reproducible tools to compare different fuels and solutions for agglomeration.
However, the temperatures where agglomeration starts might de different from the temperature
in real plants since different hydrodynamics, nozzle design, etc. have large influence on the
agglomeration process. Furthermore, during combustion actual local temperatures can be
considerably (over 100C) higher than measured using normal thermocouples.
ECN-C-99-090
57
4.
COMPARISON OF PLANNED ACTIVITIES AND WORK

ACCOMPLISHED
The results achieved in the project are according to the planned activities with a few exceptions.
Within task 1, material should have been collected from full-scale facilities. When the project
started, the Grenaa plant had changed to a less problematic fuel combination and also had made
some design changes to reduce the agglomeration problems. It was therefore decided to use
problematic fuels in the lab scale experiments.
Within task 5, the modelling activities were less comprehensive than planned. This is mainly
due to a lack of thermodynamic data, see chapter 3.5. On the other hand, a total of 7 fuels have
been selected in the project, whereas only 3-4 have been planned in the original project.
Because of this, the amount of tests within (task 6: performance of lab-scale tests) was larger
than planned.
ECN-C-99-090
58
5.
CONCLUSIONS
Chemical analysis of the fuel can give a first indication of whether there might be an
agglomeration problem during thermal conversion. In general a high K-content means an
increased risk for agglomeration. However, the K-content alone is not a good indicator. Also the
amount of chlorine was found to be crucial. It has been shown in standard test that cacao shells,
having a K-content of roughly twice the K-content of straw, show agglomeration at circa 100C
higher temperatures than in case straw is used. The very low chlorine content of cacao shells
probably causes this.
place. Contrary to methods like DTA, compression strength and ash melting temperatures, all
processes that might be relevant for agglomeration actually can occur during fluid bed
experiments: fuel-bed material interactions, volatilization and condensation, shear forces,
temperature homogeneity and accumulation. Standard tests have been developed where process
temperature is gradually increasing until agglomeration. These tests have been applied in the
project at three different laboratories: ECN (NL), ETC (S) and VTT (FIN). They have proved to
be accurate and reproducible. Standard agglomeration tests therefore seem good tools for
agglomeration research.
It has been shown with standard tests in lab-scale fluidized bed facilities that the addition of
kaolin, magnesite, dolomite and sewage sludge significantly decrease the actual temperature of
agglomeration. The agglomeration temperatures increased with at least 60C in these cases.
During combustion experiments, measured particle temperatures appeared to have up to 100C
in excess of the bed temperature. This might have a large influence on agglomeration. Because
gasification, contrary to combustion, is a process where peak temperatures are lower or even
absent, one might expect that agglomeration during gasification will occur at higher
temperatures than during combustion. This however can not be concluded from the experiments,
illustrating the importance of other factors like the design of fluid beds (the gas distribution, the
type of nozzles, etc.).
Simply choosing a low temperature for the conversion in order to prevent agglomeration to
occur, has proven successful during a 10-hour test with straw in a CFB gasifier. Apart from
preventing melting and formation of eutectics to happen, also some important elements like K
are removed from the system with the relative high fraction (25%) of unconverted char.
Compression strength measurements, DTA/TG, standard ash melting temperature measurements
and high temperature microscopy as means to predict agglomeration do not seem worthwhile.
However, DTA/TG and electron microscopy can be useful as a research tool.
It can be concluded that not only the type of fuel and other chemical input is determining the
agglomeration temperature, also other factors like gas distribution, size of bed material, type of
nozzles, cyclone efficiency in CFBs, etc. can have an important role. This means that results
from lab-scale facilities can be interpreted in a relative way (comparing fuels and evaluate
possible solution) but should always be used with great care when trying to draw conclusions
for full-scale plants. Nevertheless, standardised lab-scale bubbling fluidized bed experiments, as
developed and used in this project, seem to be the most reliable tools for the prediction of
agglomeration.
ECN-C-99-090
59
6.
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
REFERENCES
A.van der Drift, J.J. Vermeulen: Ten residual biomass fuels for circulating fluidized bed
gasification, submitted for publication in Biomass and Bioenergy (1999).
G.C.A.M.Heemskerk: Best practice list for biomass fuel and ash analysis, Utrecht,
Novem-EWAB no.9820, 58 p. (1998).
Description of the CFB gasification plant at ECN on the internet:
http://www.ecn.nl/unit_fb/bivkin/index.html (1998).
M.hman, A.Nordin, B.Skrifvars, R.Backman and M.Hupa: Accepted for publication in
Energy & Fuels (1999).
M.hman and A.Nordin: Accepted for publication in Energy & Fuels (1999).
B.Skrifvars, M.Hupa, R.Backman and M.Hiltunen: Fuel 73 (2) 171-176 (1994).
Biomassa and waste composition database on the internet: http://www.ecn/phyllis (1999).
B.Olanders and B.M.Stenaari: Characterisation of ashes from wood and straw. Biomass
and Bioenergy 8 (2) 105-115 (1995).
W.Lin, G.Krusholm, K.Dam-Johansen, E.Musahl and L.Bank: Agglomeration
phenomena in fluidized bed combustion of straw. In: Proceedings of the 14th International
Conference on Fluidized Bed Combustion, 11-14 May 1997, Vancouver, Canada (Ed.
D.S.Preto), pp. 831-838 (1997).
A.Brown: Phase composition of sintered masses formed in fluidized bed combustors,
Studsvik Nuclear, report Studsvik/NS-90/73 (1990).
W.Hahn: Gasification of renewable feedstocks in a circulating fluidised bed.
Implementing CO2 reduction targets through power-and-heat cogeneration with
integrated biomass gasification, Technische Hochschule Aachen (Germany). Fakultaet
fur Bergbau, Huettenwesen und Geowissenschaften (1994).
P.Stoholm, R.G.Nielsen, H.Nygaard, L.Tobiasen , M.W.Fock, K.Richardt and
U.Henriksen: Low temperature CFB gasifier, conceptual ideas, applications and first test
results. In: submitted for presentation during the 1st World Conference and Technology
Exhibition, 5-9 June 2000 in Sevilla, Spain (2000).
F.Zintl and T.hman: Alkali-induced particle agglomeration and bed-fluidization on
CFB-combustion of biomass - Test of common and alternative bed materials. In: Biomass
for energy and industry. Proc. 10th European conference and technology exhibition (Eds.
H.Kopetz et al.), pp. 1348-1350, C.A.R.M.E.N., Rimpar (Germany) (1998).
K.Sthl and M.Neergaard: Experiences from the biomass fuelled IGCC plant at Vrnamo .
In: Biomass for energy and industry. Proc. 10th European conference and technology
exhibition (Eds. H.Kopetz et al.), pp. 291-294, C.A.R.M.E.N., Rimpar (Germany) (1998).
K.Sthl and M.Neergaard: Experiences from the Vrnamo IGCC demonstration plant. In:
VTT Symposium 192: Power production from biomass III. Gasification and pyrolysis
R&D&D for industry (Eds. K.Sipil and M.Korhonen), pp. 73-86, VTT Energy, Espoo
(Finland) (1998).
H.Junker, F.Fogh, L.Baxter and A.Robinson: Co-firing biomass and coal. Experimental
investigations of deposit formation. In: Biomass for energy and industry. Proc. 10th
European conference and technology exhibition (Eds. H.Kopetz et al.), pp. 1482-1485,
C.A.R.M.E.N., Rimpar (Germany) (1998).
B.-J.Skrifvars, M.Hupa, M.hman and A.Nordin: Predicting bed agglomeration
tendencies for biomass fuels. A comparison of three different techniques. submitted for
publication in Energy and Fuels (1999).
ECN-C-99-090
60
18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
32.
33.
M.hman: A new method to quantify fluidized bed agglomeration in the combustion of

biomass fuels. Licantiate thesis, Umea University, 23 p. (1997).
NREL: Alkali deposits found in biomass boilers, The behaviour of inorganic material in
biomass-fired power boilers - Field and laboratory experiences, Vol II, NREL/TP-4338142, SAND96-8225 (1996).
B.Skrifvars, J.P.lomquist, M.Hupa and R.Backman: Predicting the ash behaviour during
biomass combustion in FBC condition by combining advanced fuel analyses with
thermodynamic multi-component equilibrium calculations. In: Pittsburgh coal conference,
USA, 14-18 Sept. 1998.
E.A.Osman: A study of the effects of ash chemical composition and additives on fusion
temperature in relation to slag formation during gasification of biomass, PhD-thesis,
University of California (1982).
T.F.Wall, R.A.Creelman, S.Gupta, C.Coin and A.Lowe: Coal ash fusion temperaturesNew characterisation techniques and associations with phase equilibria . In: Applications
of Advanced Technology to Ash-Related Problems in Boilers (Eds. L.Baxter and
R.DeSollar), pp. 541-556 (1996).
B.M.Steenari: A practical and theoretical study of sintering of coal ash and limestone in
a fluidized bed combustor, Staten energiverk, FBT-87/34 (1987).
B.-J.Skrifvars, R.Backman and M.Hupa: Characterization of the sintering tendency of ten
biomass ashes in FBC conditions by a laboratory test and by phase equilibrium
calculations. Fuel Process. Technol. 56 (1) 55-67 (1998).
A.Nordin, M.hman, B.J.Skrifvars and M.Hupa: Agglomeration and defluidization in
FBC of biomass fuels- mechanisms and measures for prevention. In: Proc. 1995 Eng.
Found. Ash Conf., Waterville Valley, 16-22 July 1995 (1995).
A.Nordin and R.Backman: Heavy metal behaviour during combustion and gasification of
biomass fuels - comparison between chemical equilibrium and experimental results. In:
Ashes and particulate emissions from biomass combustion, Vol. 3, edited by I.
Obenberger, dbv-Verlag, Graz, Austria (1998).
A.Ljung and A.Nordin: Theoretical Feasibility for Ecological Biomass Ash
Recirculation: Chemical Equilibrium Behaviour of Nutrient Elements and Heavy Metals
during Combustion. Environmental Science & Technology 31 (9) 2499(1997).
R.Backman and A.Nordin: High temperature equilibrium calculations of ash forming
elements in biomass combustion / gasification systems - state-of-the-art, possibilities and
applications, Ashes and particulate emissions from biomass combustion, Vol. 3, edited by
I. Obenberger, dbv-Verlag, Graz, Austria (1998).
M.Blander, T.Milne, D.Dayton, R.Backman, D.Blake, V.Khnel, W.Linak, A.Nordin and
A.Ljung: Equilibrium chemistry of biomass combustion: A round-robin set of
calculations using available computer programs an data bases. In: Impact of mineral
impurities in solid fuel combustion (Eds. T.F.Wall and L.L.Baxter) (1997).
M.Blander and A.D.Pelton: The inorganic chemistry of the combustion of wheat straw.
Biomass and Bioenergy 12 (4) 295-298 (1997).
L.A.Hansen, H.P.Michelsen and K.D.Dam-Johansen: Alkali metals in a coal- and
biomass-fired CFBC - Measurements and thermodynamic modelling. In: Fluidized Bed
Combustion, Volume 4, ASME 1995, pp. 39-48 (1995).
P.Kallner, A.Nordin and R.Backman: Fate of ash forming elements in gas turbine
combustion of pulverized wood - chemical equilibrium model calculations. In:
ASME9620 (1996).
R.W.Bryers: Fireside slagging, fouling and high-temperature corrosion of heat-transfer
surface due to impurities in steam-raising fuels. Progress in Energy and Combustion
Science 22 (1) 29-120 (1996).
ECN-C-99-090
61
34.
35.
36.
37.
38.
39.
40.
41.
42.
43.
44.
45.
46.
47.
48.
49.
B.Olanders and B.-M.Steenari: Characterization of ashes from wood and straw. Biomass
and Bioenergy 8 (2) 105-115 (1995).
M.K.Misra, K.W.Ragland, L.L.Baxter, B.M.Jenkins and L.L.Oden: Alkalis in alternative
biofuels. FACT 18 211-220 (1998).
W.R.Livingston: Straw ash characteristics, Babcock Energy Limited , DE92 519748, 23
p. (1991).
N.Padban, Z.Ye and I.Bjerle: Alkali removal and bed material agglomeration studies
applied to biomass gasification. In: Developments in thermochemical biomass
conversion, Vol 2 (Eds. A.V.Bridgwater and D.G.B.Boocock), pp. 1031-1044, Blackie
Academic & Professional, London (1997).
B.J.Skrifvars, M.Hupa, A.Moilanen and R.Lundqvist: Characterization of biomass ashes.
In: Proc. Applications of advanced technology to ash-related problems in boilers (Eds.
L.Baxter and R.DeSollar), pp. 383-398, Plenum Press, New York (1996).
E.J.Raask: Thermal Anal. 16 91(1979).
E.J.D.Smith: Journal of the Institute of Fuel 29 253(1956).
M.J.Gluckman, J.Yersuhalmi and A.M.Squires: Defluidization characteristics of sticky or
agglomerating beds. In: Proc. of the Int. Fluidization Conf., California, U.S.A, pp. 395422 (1976).
A.R.Manzoori: Role of the inorganic matter in agglomeration and defluidization during
the circulating fluid bed combustion of low-rank coals, PhD-thesis, University of
Adelaide (1990).
I.W.Cumming, W.I.Joyce and J.H.Kyle: J. Inst. E. 58 (1985).
I.W.Cumming and A.Sanyal. In: Proceedings of the Engineering Foundation Conference,
12-17 July 1981 (1981).
R.E.Conn: Laboratory techniques for evaluating ash agglomeration potential in
petroleum coke fired circulating fluidized bed combustors. Fuel Process. Technol. 44
(1/3) 95-103 (1995).
J.R.Gibson and W.R.Livingston. In: Engineering Foundation Conference on Inorganic
Transformations and ash deposition during combustion, Palm Coast, Florida, pp. 425-447
(1991).
N.Padban, S.Kiuru and A.L.Hallgren: Bed material agglomeration in PFB biomass
gasification. Preprints of Papers, American Chemical Society, Division of Fuel
Chemistry 40 (3) 743-747 (1995).
E.Natarajan, M.Ohman, M.Gabra, A.Nordin, T.Liliedah and A.N.Rao: Experimental
determination of bed agglomeration tendencies of some common agricultural residues in
fluidized bed combustion and gasification. Biomass and Bioenergy 15 (2) 163-169
(1998).
M.hman: A new method for Quantification of fluidized bed agglomeration tendencies: A
sensitivity analysis. Energy & Fuels 12 90-94 (1998).
ECN-C-99-090
62
ANNEX A. A LITERATURE SURVEY ON AGGLOMERATION,

MECHANISM, PREDICTION AND PREVENTION
Contents of Annex 1
SUMMARY
1. INORGANIC MATTER IN BIOMASS AND ITS TRANSFORMATIONS

1.1. Chemical composition of biomass-ash
1.2. Forms of inorganics in biomass
1.3. Mineral matter transformations
1.4. Volatilisation of alkalis
3
3
3
4
5
2. BED-MATERIAL AGGLOMERATION DURING GASIFICATION AND

COMBUSTION
2.1. Agglomeration mechanisms
2.1.1. Particle-to-particle bonding mechanisms relevant to bed-agglomeration
in fluidised-bed processes
2.1.2. Bed-agglomeration mechanisms in fluidised bed combustion and
gasification
2.1.3. Defluidisation of fluidised-beds due to ash chemistry
2.1.4. Combustion versus gasification: some important differences
2.2. Influence factors promoting agglomeration
2.2.1. Ash chemistry
2.2.2. Bed material
2.2.3. Fluidisation conditions
2.3. How to prevent agglomeration?
2.3.1. Chemical additives
2.3.2. Fuel blending
2.3.3. Periodical refreshment bed material
2.3.4. Lowering the bed temperature
2.4. How to predict agglomeration?
2.4.1. Fuel analyses
2.4.2. Ash melting, sintering and agglomeration temperatures
9
11
12
12
12
13
14
14
14
16
17
17
18
18
19
3. REFERENCES TO ANNEX A
21
ECN-C-99-090
8
8
8
Annex A; page 1
SUMMARY
Agglomeration mechanisms
Agglomeration problems in fluidised-beds are associated with a high content of organicallybound (volatile) alkali metals in the fuel. In biomass, alkali metals (esp. potassium) concentrate
in regions of most vigorous growth. Likely problematic fuels are therefore residues of
agricultural crops, young energy crops and all other fuels containing high amounts of young
organic material. Problems are more severe when a high alkali content is combined with high
contents of sulphur (in combustion), chlorine, silica and phosphorus. These elements form lowmelting compounds or eutectics, which become deposited on the bed particles, coating them
with a sticky ash layer.
Influence factors promoting agglomeration
Apart from ash chemistry, system design and operational conditions are very important factors
determining the time-scale of the process. Problems are sometimes associated with temperature
excursions, and it seems that the higher the temperature the more severe the problems.
However, for high-sulphur fuels there appears to be a temperature window where in-bed sulphur
retention is less and agglomeration problems are reduced. The use of silica-sand bed material
promotes agglomeration due to reaction of alkali compounds with the silica to form low-melting
surfaces on the particles. The time-scale of the agglomeration process and maximum allowable
temperatures in the bed are furthermore strongly influenced by fluidisation conditions, gas
atmosphere, bed and fuel particle size, and ash re-circulation from cyclones (in CFBs).
How to prevent agglomeration?
Adding inert high-melting minerals (CaO, MgO), sulphur-sorbents (limestone, dolomite) and
alkali-sorbents (clay minerals) to the bed are reported to reduce agglomeration problems in
fluidised-beds, but mechanisms are not fully understood. In some reported cases, sulphursorbents even decreased ash deformation temperatures and increased fouling. Blending of straw
with wood and coal is reported to be successful to some extent. In full-scale practice, the bedmaterial is refreshed regularly before a critical ash-accumulation (alkalis) or agglomeration
level is reached.
How to predict agglomeration?
A lot of indices based on the ash chemical composition are reported to asses the fouling and
slagging propensity of coals in pulverised fuel boilers. These are fuel specific and often do not
include system design parameters and operational conditions. No such indices were found for
fluidised-bed gasification of biomass. Some attempts have been undertaken to determine a
critical alkali accumulation in the bed on laboratory-scale for different bed materials, but values
were substantially different from full-scale values. A lot of laboratory sintering methods were
developed to predict the agglomeration and sintering behaviour of coal and biomass ashes. All
suffer from the limitation that gas-phase reactions, reactions between gas components and the
ash, condensation of alkali vapours, the influence of particle-to-particle collisions in fluidisedbeds, and partitioning of elements can not be adequately simulated in the laboratory.
Furthermore, thermal ashing to prepare a laboratory ash can create artefacts in the ash and, more
important, gives a substantial alkali-loss, artificially increasing the ash-fusion temperatures.
These methods can therefore not give absolute agglomeration temperatures, but some seem
useful for gaining a more fundamental understanding of ash behaviour in thermal conversion.
ECN-C-99-090
Annex A; page 2
1.
1.1
INORGANIC
MATTER
TRANSFORMATIONS
IN
BIOMASS
AND
ITS
Chemical composition of biomass-ash
Reported ash contents and compositions for biomass differ considerably with fuel-type and for
the same fuel-type with soil, fertilisers, climate, maturity, harvesting, handling, etc., and also
with ashing temperature and analysis method employed. Important inorganics in biomass, from
a point of view of agglomeration problems, are the refractory element silicon, the alkali
elements potassium and sodium, the alkaline earth element calcium and the non-metallic
(anionic) elements chlorine, sulphur and phosphorus. For indicative purposes, some general
trends are described below for woody fuels and agricultural residues.
A database of biomass composition, ash composition and ash fusion temperatures is available
on the internet: http://www.ecn.nl/phyllis. The database is made by ECN and is called Phyllis,
the Greek goddess of trees and wisdom. Some important information: wood generally has a very
low ash content (1-3 %), of which 50 - 60 % is calcium (predominantly as CaO), 5 - 10 % is
potassium and 10 % or less is silica. Some extraneous clay and quartz may be present from
contamination with soil. Ash melting temperatures are generally very high. Sulphur and
chlorine are very low, giving rise to little problems in combustion. Agricultural residues have a
higher ash content, consisting largely of silicon, potassium and phosphorus. Their ash melting
temperatures are therefore lower than for wood, usually between 800 and 1000C. Straws and
grasses may also contain relatively high concentrations of chlorine and sulphur. The ash content
of these fuels shows a large variance, ranging from 2 % to sometimes more than 20 %.
Bryers (1996) distinguishes three different biomass-ash systems:
high-silica/high-potassium/low-calcium ash with low fusion temperatures (agricultural
residues)
low-silica/low-potassium/high-calcium ash with high fusion temperatures (wood)
high-calcium/high-phosphorus ash with low fusion temperatures (manure)
1.2
Forms of inorganics in biomass
The form in which an inorganic element occurs in a solid fuel has a decisive influence on its
reactivity and volatility in combustion and gasification. According to Baxter (1993) inorganic
material in solid low-rank fuels (lignites and biomass) can be roughly divided in two fractions:
1) atomically dispersed inherent inorganics and 2) extraneous discrete minerals added through
geologic of processing steps. Inherent inorganics in biomass and low-rank coals are largely
associated with oxygen-containing functional groups (hydroxyl, carboxyl, ether and ketone
groups) as chelates or cations, or are in solution in inherent moisture as salt or precipitated as a
crystalline salt (like sodium chloride).
Bryers (1996) gives a short overview on the modes of occurrence and origins of the main
inorganic elements in biomass. Sulphur is in the form of sulphates and organic sulphur.
Chlorine occurs as a chloride ion. Silicon is from absorption of silicic acid from the soil and
deposits as hydrated oxide in amorphous form. Potassium occurs as univalent ion and has an
important function in osmotic processes. It concentrates in the regions of most vigorous plant
growth and is highly mobile. Sodium is not an essential element in plants, but may be
ECN-C-99-090
Annex A; page 3
introduced through intrusion of salt water during transport or trough fuel processing. Calcium is
essential for cell wall stiffening and forms exchangeable bonds with cell walls. Aluminium is
toxic and its presence is usually a marker of contamination via dust, dirt or other soil
constituents.
Miles et al. (1996) determined the modes of occurrence of inorganics in very different biofuels
in terms of their solubility in increasingly aggressive solvents. The fuels analysed included rice
and wheat straws, grasses, urban wood fuel, Almond hulls and shells, olive pits and nonrecyclable paper. The analysis procedure distinguishes between the following four modes of
occurrence:
solubility
-
likely form in the fuel
water soluble
water soluble salts
soluble in ammonium acetate acid (ion-exchangeable) organically bound as cations
soluble in hydrochloric acid
acid soluble salts as carbonates and sulphates
insoluble
mineral form as oxides, silicates or sulphides
All of the refractory elements (Si, Ti and Al) are in the form of oxides and were insoluble.
Titanium and aluminium were present only in trace concentrations. The alkali and alkaline earth
elements (K, Na, Ca and Mg) were predominantly in a soluble form. In non-contaminated (no
soil) fuels, over 90 % of potassium was in a water-soluble or ion-exchangeable form and thus
highly volatile. Calcium was largely in an ion-exchangeable or acid soluble form and thus less
volatile than potassium. Sodium and magnesium were present only in trace concentrations. In
the non-recyclable paper and urban wood fuel, clay contamination gave rise to a large fraction
of insoluble and thus non-volatile potassium. Almost all the chlorine was in a water soluble and
thus highly volatile form. Sulphur can be a significant component in straws and grasses and
occurs in variable forms, ranging from completely water soluble in wheat straw to completely
insoluble in switch grass. However, the volatility of sulphur is not so much depended on its
form in the fuel, as it is oxidised in combustion to gaseous SO 2 and released as H2 S in
gasification. Phosphorus only occurs in significant concentrations in manure and litters and its
form is variable. Iron in biomass is present in very low concentration and does not play an
important role in slagging, fouling or bed-agglomeration.
1.3
Mineral matter transformations
The physical and chemical transformations of mineral matter in combustion of biomass in

conventional boilers as elucidated by the Alkali Deposit Investigation of NREL, U.S. (Miles et
al., 1996) are illustrated in figure A1.1
ECN-C-99-090
Annex A; page 4
Figure A1.1 Mineral matter transformations in biomass combustion (Miles et al., 1996)
This mechanism for ash deposition in biomass-fired boilers is based on analysis of deposits
from different locations in full-scale boilers, including travelling grates and fluidised beds.
Phase transformations occurring in ash from wood and straw were studied by Olanders and
Steenari (1995). The change in crystalline phases on heating in air of low temperature ash are
summarised in table A1.1. Wood ash from combustion consists largely of high melting oxides
of calcium (melting point 2572C) and magnesium (melting point 2500C) and some silicates of
potassium, calcium and magnesium. Straw ash from combustion consists largely of low-melting
potassium sulphate (melting point 1069C) and potassium chloride (melting point 770C), that
can form even lower melting eutectics. The main difference in mineralogy after slow heating of
low-temperature ash and direct combustion at 1000 to 1200C was the form of silica, which was
thermodynamically more stable in low-melting silicates than as Quartz. Straw ash from fixedbed combustion showed large sintered lumps, with the dominant structure in the sintered
material solidified melt. Thermal (DTA/TGA) analysis of the ashes showed softening of the
straw ash at 700C and melting between 975 and 1025C, while the wood/bark ashes only
started to soften at 1000 - 1200C and melting only occurred when heated to 1470C.
1.4
Volatilisation of alkalis
Volatilisation of potassium and re-condensation on silicate fly ash plays a key role in fouling in
biomass-fired boilers. It is therefore of importance to be able to predict the amount of potassium
that can be released as function of the fuel type and temperature. However, there is no general
agreement on the temperature at which volatilisation of alkalis starts and how much of the
alkali material is released into the gas. Obviously this is due to the many factors influencing the
release: form of the alkali in the fuel, heating rate and final temperature of the fuel particle,
residence time, furnace temperature, amount of volatile organic matter released during
pyrolysis, presence of silica, sulphur, chlorine and gettering alumino-silicates, local oxygen
partial pressure, bed material, etc. (Ligthart, 1995).
ECN-C-99-090
Annex A; page 5
Table A1.1 Change in crystalline (no amorphous phases) phases on slow heating in air
(equilibrium composition) of low temperature ash (LTA) of straw and wood
mixtures (Olanders and Steenari, 1995)
temperature crystalline phases in wood ashes
[C]
400
700
CaCO3 , SiO2 (quartz), traces of K2 SO4

and carbonates
CaCO3 , SiO2 (quartz), traces of K2 SO4
and carbonates
CaO, SiO2 (quartz)
750
CaO, SiO2 (quartz), MgO
650
temperature crystalline phases in straw ashes

[C]
400
CaCO3 , SiO2 (quartz), K2 SO4 , KCl,

KAlSi3 O8
650
K2 SO4 , KCl, KAlSi3 O8 , traces of
SiO2
700-750 K2 SO4 , KCl, K2 CaSiO4 , KAlSi3 O8 ,
traces of SiO2
800-1000 K2 SO4 , KCl, K2 CaSiO4 , KAlSi3 O8 ,
traces of SiO2
800
CaO, MgO, traces of SiO2 (quartz) and

silicates
850
CaO, MgO, traces of SiO2 (quartz) and
silicates (K2 CaSiO4 )
900
CaO, MgO, K2 CaSiO4 , Ca-silicates
1000-1150 CaO, MgO, Ca-(Mg)-silicates
Fundamental investigations on alkali release have been undertaken by Campisi (1988),

Manzoori (1990), Livingstone (1991), Misra et al. (1993) and French et al. (1994). In short, a
large part of the soluble potassium will be released in the gas phase above 800C. How much
depends on many factors, as mentioned earlier.
Low-rank coals
Campisi (1988) compared the release of sodium from lignite (organically bound sodium) and
brown coal (sodium chloride) with direct evaporation of sodium chloride, by heating small
samples at different heating rates (25 to 1667C/s) in a reducing atmosphere and analysing the
evolving gas with AAS. Heating rate did not seem to have an effect on the release profiles.
Vaporisation of sodium from the coals started at 800C and was complete at 1700C, while the
release rate was highest at 1000C. Release profiles were markedly different from that of
evaporation of pure sodium chloride.
Manzoori (1990) studied the ash forming process in fluidised-bed combustion of high-sodium,
high-sulphur low-rank coal, by conducting both pyrolysis and combustion experiments with
single coal particles. Sodium in this coal was partly organically bound and partly as sodium
chloride. In pyrolysis (700 - 830C), a large fraction of the chlorine (40 - 70 %) was lost from
the char after 120 seconds, while only less than 10 % of the sodium was lost. Combustion at
700C showed more or less the same picture, although results were interfered by char
fragmentation. He suggested sodium chloride dissociates or reacts with other compounds to
form less volatile sodium and hydrogen chloride.
Wood and straw
Livingstone (1991) investigated the volatilisation of alkalis from several straw types by rapidly
heating (500 - 1000C/s) small samples in air to temperatures between 800 and 1200C. The
release depended on the straw type. Wheat and barley straws showed larger alkali loss than rape
straw. For example, combustion at 800 and 1000C of wheat straw gave a 20 and 50 % total
potassium loss respectively, which corresponded with 30 and 80 % respectively of the water
soluble and ion-exchangeable potassium. Sodium loss was relatively low below 1000C, but
increased up to 95 % of the soluble sodium (55 % of total) at 1200C.
Misra et al. (1993) studied the composition of wood and bark ashes as function of temperature.
On heating low temperature ash from 600 up to 1300C they noticed a significant loss in
potassium from the ash, starting between 800 to 900C and resulting in 63 to 90 % of the total
potassium released into the gas phase at 1300C.
ECN-C-99-090
Annex A; page 6
French et al. (1994) studied the gas phase release and speciation of alkalis during combustion of
woody and herbaceous fuels, at 800 and 1100C. Almost all the chlorine was released as HCl
during devolatilisation, while a much smaller part of the alkalis were released during char
combustion as chlorides or hydroxides. Alkali release was higher for the herbaceous fuels than
for wood, due to both higher chlorine and alkali concentrations, and increased markedly with
temperature. They estimated the release of potassium in switch grass combustion at 1100C to
be 20 %.
ECN-C-99-090
Annex A; page 7
2.
BED-MATERIAL AGGLOMERATION DURING GASIFICATION

AND COMBUSTION
Despite the relatively low temperature in fluidised-bed thermal conversion and combustion,
usually below the ASTM ash deformation temperature, problems with bed-agglomeration are
reported, especially with agricultural residues and lignite. Most of these problems are associated
with a high alkali content of the fuel, operational set-ups (temperature excursions), or systems
originally designed for high-rank (low-alkali, low-sulphur) fuels.
2.1
2.1.1
Agglomeration mechanisms
Particle-to-particle bonding mechanisms relevant to bed-agglomeration in

fluidised-bed processes
Particle-to-particle bonding or sintering starts when a neck is formed between particles,

resulting in a initially porous network. Given time, the neck regions grow until individual
particles can no longer be identified. Finally, a dense mass forms. The driving force for this
process is reduction of the surface energy of the particles (Skrifvars et al., 1992a). Skrifvars et
al. (1989, 1992a, 1992b and 1996) have studied the sintering behaviour of ashes from coal, peat
and biomass under FBC conditions and consider three sintering mechanisms to be important:
partial melting, viscous flow and chemical reaction. Which one occurs depends on the ash
chemical and mineralogical composition, the temperature history of the ash particles and gas
components.
Partial melting: sintering in the presence of a reactive non-viscous liquid phase consisting of
molten alkali salts, where the solid phase is partly soluble in the liquid at the sintering
temperature. The amount of liquid controls the neck formation. Cooling below the solidus
temperature gives crystallisation of the bonding agent and densification of the agglomerate. This
mechanism is predominant for fuels with ashes high in alkali, sulphur and chlorine. These
elements form low-melting eutectics, causing sintering and agglomeration at temperatures even
below 700C. Skrifvars et al. (1992a) mention that sintering by partial melting can be predicted
by thermodynamic equilibrium calculation of the amount of molten phase in the ash as function
of temperature. Theoretically predicted initial melting temperatures of ashes correlated fairly
well with experimentally determined initial sintering temperatures.
Viscous flow sintering or vitrification: sintering due to viscous flow of a vitreous silicate phase.
When a silicate ash is heated above the solidus temperature, a highly viscous liquid phase
forms. Due to its high viscosity, the liquid remains viscous on rapid cooling below the solidus
temperature, forming a glassy phase. Bryers (1996) states that absorption of alkalis on silicate
fly ash particles gives rise to formation of a low-melting silicate on the surface of the particles.
When the temperature in a ash-deposit rises, due to reduced heat transfer, flowing of this
viscous layer forms a neck between adjacent particles. On cooling this neck freezes to a glassy
phase.
Vitrification of peat ash in an oxygen-blown HTW gasifier (750-950C) gave rise to unexpected
formation of deposits, hampering operation by clogging the gasifier outlet and cyclones
(Moilanen, 1993). Ash fusion temperatures were above 1000C. Sintering was ascribed to rapid
heat up of the ash particles in flames of the oxygen-blown bed, to temperatures exceeding the
ECN-C-99-090
Annex A; page 8
solidus temperature. On cooling the slag formed in the flames remained glassy with low enough
viscosity to give rapid sintering in the freeboard zone by viscous flow at temperatures as low as
700 to 800C.
The dependence of this mechanism on properties of fly ash particles is given by the Frenkel
formula:
x2 =
3 r t
2
where r
x2
=
=
=
=
=
radius of particles
equivalent of contact area (neck) between particles
surface tension of the vitreous phase
viscosity
time
Neck growth (x(t)) is thus controlled by time and viscosity, the latter varying with temperature
and chemical composition of the slag. Moilanen (1991) mentions viscosities in the range of 107
to 109 Poise to be critical for agglomeration in fluidised-beds. Moilanen (1989) also studied the
effects of temperature, ash chemical composition and atmosphere on sintering of peat ash.
Sintering of 100 m premelted (to obtain slag) ash particles was measurable between 700 and
800C and increased with temperature and the basic-to-acid ratio of the ash. Surface
crystallisation in the slag inhibited sintering, which was achieved by adding sodium and
phosphorus. Normally, these elements have a strong fluxing effect. Sintering was much stronger
in a reducing than in a oxidising atmosphere. This was ascribed to oxidation and crystallisation
of iron to hematite.
Chemical reaction: sintering due to reaction between the particles or the particles and the gas, to
form a third compound binding the particles together. Skrifvars et al. (1992a and 1996)
concluded that this mechanism is responsible for sintering of coal and biomass ashes rich in
calcium. These ashes show increased sintering in CO2 atmosphere at temperature between 600
and 800C and with SO 2 in the gas at all temperatures (above 500C), when compared to
sintering in dry air. The effect of CO2 was only noticeable for ashes in which calcium was as
oxide and not when it was as carbonate. It was concluded that carbonation of CaO in a gas with
high CO2 partial pressure gives particle-to-particle bonding via CaCO3 crystals. Above 800C
these again decompose to CaO and CO2 . They mention that this sintering mechanism could be
relevant to deposit hardening in combustion, where ash is first formed at high temperatures
above 800C, decomposing the carbonates to oxides, and than cooled in the post combustion
zone in a CO2 -rich environment. Sulfation of CaO to CaSO4 , due to SO 2 in the gas, gives the
same kind of bonding by crystals, stable over a wider temperature range.
Bryers (1996) also mentions calcium sulphate bonding to be an important deposit hardening
mechanism when firing coals rich in organically bound calcium and sulphur or when adding
finely ground limestone for sulphur capture. Submicron CaO particles deposit on tube surface
through thermophoresis and then become sulphated by reaction with SO 2 or SO 3 . Local low
oxygen partial pressures at the tube surface could additionally give rise to formation of the low
melting eutectic CaSO 4 /CaS (eutectic point 830C).
2.1.2
Bed-agglomeration mechanisms in fluidised bed combustion and gasification
Fluidised-bed combustion of low-rank coals

Brekke and Karner (1982), Bobman, Hajicek and Zobeck (1985), Mann, Galbreath and
Kalmanovitch (1992), and Manzoori (1990) all experimentally studied the mechanisms of
ECN-C-99-090
Annex A; page 9
agglomeration in FBC of high-sodium, high-sulphur low-rank coals. They all identified an

initial stage in which highly sulphated ash, rich in sodium and calcium, becomes deposited on
the bed particles forming a sticky ash coating. The sodium in the coating is derived from the
organically bound sodium and sodium chloride. This ash coating gives rise to particle
agglomeration due to sulphate-to-sulphate sintering (chemical reaction) or liquid phase sintering
(partial melting) due to the presence of low-melting point eutectics of sulphates of sodium,
calcium and possibly magnesium. In later stages the ash reacts with the quartz sand bed material
to form low-melting silicates on the surfaces of the particles. Viscous flow sintering then
accelerates the process.
Brekke and Karner (1982) mention that problems are promoted by the use of high-sodium
lignite, increased operation time, ash re-injection, temperature excursions, quartz sand bed
material and ash accumulation in the bed. They report that the use of limestone as additive
inhibits to some extent agglomeration. The agglomeration process occurred over a period of
about 100 hours, when further operation was hampered by substantial defluidisation. Severe
agglomeration was evident after 60 hours of .
Bobman, Hajicek and Zobeck (1985) noticed that there is an intermediate temperature range
where agglomeration is reduced. Agglomeration was severe at high temperatures above 980C
and lower temperatures around 800-830C, causing shut down after 50 to 80 hours operation. At
900C agglomeration was however minimal even after 100 hours of operation. They noticed a
correlation between agglomeration and the sulphur retention in he bed, but did not speculate on
a cause/effect relation.
Mann, Galbreath and Kalmanovitch (1992) reviewed the results of the experimental work
conducted on agglomeration at EERC of the University of North Dakota, U.S. and full-scale
data and came to the following general mechanism for the formation of agglomerates in AFBC
of high-sodium, high-sulphur low-rank coals:
1. Initial deposition/condensation of volatile sodium ash species on the surfaces of bed
particles, followed by deposition of submicron calcium species (from organically bound
calcium) on the initial sticky layer. Condensation is a gas-to-solid condensation, which may
occur in combustion as the temperature of the burning char is higher than the temperature of
the bed particles.
2. Reaction of the sodium and calcium with SO2 in the gas to form sulphates, before,
during or after deposition on the bed particles.
3. Entrapment of other larger ash particles in this initial sulphated ash layer.
4. Cohesion of bed particles due to sulphate-to-sulphate sintering (chemical reaction) or
formation of low-melting eutectics of sulphates (partial melting).
5. Formation of larger agglomerates.
6. Partial defluidisation due to increasing fraction oversized particles in the bed, causing
hot spots.
7. In the high temperature zones in the bed, low-melting-point silicates form due to the
reaction of sodium and calcium with the quartz bed material. Temperatures exceeding
the solidus temperature of the newly formed silicates give rise to viscous flow
sintering, forming hard agglomerates and accelerating the agglomeration process.
A same kind of mechanism is proposed for the deposition of ash on in-bed heat exchange
surfaces in FBC by Oakey, Minchener and Stringer (1990). They also mention the inherent
propensity of very fine particles (< 5 m) to sinter, independent of their composition. They
claim therefore that some ash deposition on surfaces above 540C is unavoidable.
Manzoori (1990, 1992, 1993 and 1994) reports that deposition of sticky ash on bed particles and
subsequent agglomeration is reduced by a higher content in the fuel of fine-sized inert minerals,
like silica, alumina and aluminosilicates, a lower volatile sodium content of the fuel and a lower
ECN-C-99-090
Annex A; page 10
bed temperature. Agglomeration was accelerated by impregnation of the fuel with sodium
chloride solution, which resulted in chlorine in the ash coating lowering its melting temperature.
Below the initial sintering temperature of the spent bed material (700C), as determined by
thermal analysis, no bed-agglomeration occurred. Manzoori (1990) derived an expression for
the rate of agglomeration in FBC in terms of the thickness and physical properties of the ash
coating and the operational conditions.
Fluidised-bed gasification of low-rank coals
Mason et al. (1980), Marinov et al. (1992) and Williamson, et al. (1994) all discuss the
important role of pyrite in the formation of agglomerates in FBG of iron-rich coal. Marinov et
al. (1992) report that the pressurised fluidised bed gasification of a high-sulphur lignite at 930C
was hampered due to intensive ash agglomeration. The low-melting (924C) eutectic of FeS and
FeO was considered to be the agglutinating agent for coarse ash particles. Mason et al. (1980)
state that this is due to rapid decomposition of pyrite (FeS2 ) to pyrrhotite (FeS), followed by
oxidation of FeS to FeO even at very low oxygen partial pressure. Williamson (1994) noticed
that pyrite forms very fine sized ash particles which become attached to the larger clay-derived
ash particles, coating them with low-melting material.
Fluidised-bed combustion and gasification of high-alkali biomass
A number of investigators report that agglomeration in FBC and FBG of high-alkali biomass is
due to reaction of the alkalis with the silica sand forming low-melting eutectics on the bed
particle surfaces (Grubor et al. 1995, Paisley and Litt 1993, Ghaly and Ergudenler 1993a,
1993b, 1994a and 1994b, Padban et al. 1996). Agglomeration occurred at temperatures as low
as 750C.
Representative of agglomeration in FBC of high-alkali biomass are the results obtained by
Nordin et al. (1996). Combustion of Lucerne and olive flesh in a laboratory-scale fluidised-bed
reactor at 850C resulted for both fuels in severe agglomeration within less than 30 minutes.
The process from initial agglomeration to total defluidisation was found to be very rapid. They
noticed that the fuel particle temperature was much higher than the average bed temperature and
suggest that ash melts at the surface of the burning fuel particle and then deposits on the bed
particles. Controlled increase of the temperature of a air-fluidised mixture of ash and quartz
sand followed by analysis of the spent bed particles revealed initial agglomeration was indeed
due to a melt. In the case of Lucerne (low-silica, high-potassium) the initial agglomeration
temperature was as low as 680C and the bonding agent was a alkali salt melt (Ca, K, S, P, CO3
and some Cl). For olive flesh (high-silica, high-potassium) agglomeration started at 940C and a
potassium-silicate melt (K, Ca and Si) was the glue. Analysis of agglomerates from real
combustion showed agglomeration in the end stage of the process to be due to a molten
potassium-silicate layer on the bed particles, likely formed by reaction of quartz sand with alkali
compounds.
From these results it follows that deposition of low-melting high-alkali ash on bed particles is
the initial step in agglomeration. This ash consists of alkali-silicates or alkali-salts (chlorides,
carbonates and sulphates), depending on the ash chemical composition and temperature. Hot
spots in the bed accelerate the process due to reaction of quartz particles with alkalis forming
low-melting silicates on their surfaces.
2.1.3
Defluidisation of fluidised-beds due to ash chemistry
Basu and Sarka (1983) noticed that the minimum fluidisation velocity of a bed of coal ash
minerals linearly increases with the bed temperature when the temperature rises above the initial
sintering temperature of the bed (thermal analysis). They state that clinker formation is preceded
by defluidisation. Reduced heat transfer gives rise to hot spots in the bed exceeding the
deformation temperature of the coal ash. Defluidisation of the bed is a consequence of an extra
ECN-C-99-090
Annex A; page 11
adhesive force between the bed particles, which become sticky above the sintering temperature.
When the fluidisation velocity falls below the true minimum fluidisation velocity, the bed
defluidises instantaneously. Smaller particles showed sintering and defluidisation at lower
temperatures, which was ascribed to the larger contact area. Also according to Manzoori (1990
and 1994) defluidisation is an instantaneous process which may occur without preceding
agglomeration of the bed material. He states that the ash chemistry causing defluidisation is the
same as for agglomeration, but that both phenomena are not the same. He derived an expression
for the onset of fluidisation in terms of the thickness and physical properties of the ash coating
and the operational conditions.
2.1.4
Combustion versus gasification: some important differences
Major differences between combustion and gasification of high-sulphur fuels is that sulphates
do not form in gasification and the much higher fuel particle temperature in combustion,
sometimes several hundred degrees higher then the furnace temperature. Manzoori (1990) and
Nordin et al. (1996) mention that volatilisation or melting of ash at the high fuel particle
temperature may be the initiating step in agglomeration.
2.2
Influence factors promoting agglomeration
2.2.1
Ash chemistry
The important role ash chemistry is largely described in section 2.1. Most undesirable elements
in the ash are potassium and sodium, especially in combination with high amounts of sulphur
(in combustion), chlorine, phosphorus and silica. Calcium gives problems in the special case of
co-combustion of a high-sulphur fuel with a fuel rich in organically bound calcium. Calcium is
also an important bulk component in ash deposits, especially when adding limestone to the bed,
but is mostly not responsible for the initiation of deposits.
2.2.1.1
Potassium
Potassium is by far the most important element in biomass responsible for ash-related problems.
In oxidising conditions the thermodynamic stability of potassium compounds with increasing
temperature is as follows: sulphates (< 1000C) < silicates < chlorides. In reducing conditions
sulphates do not form and silicates and chlorides are the stable compounds. These compounds
all have relatively low melting points and can form even lower-melting eutectics. A high
amount of volatile (organically bound) potassium in biomass combustion is therefore associated
with slagging in the furnace due to fluxing of silicates, fouling and corrosion in convective
passes due to condensation on the heat transfer surfaces as chlorides, sulphates, hydroxides or
oxides and subsequent sulfation to sticky and corrosive sulphates, and agglomeration in
fluidised beds due to formation of low-melting compounds or eutectics by reacting with silica
(from the fuel or bed media), sulphur and chlorine. Volatilisation of potassium starts to be
significant at temperature above 800C, depending on the fuel-type.
2.2.1.2
Chlorine
Chlorine is almost completely released as HCl in the gas during devolatilisation. Chlorine
greatly facilitates the release of alkali into the gas phase, due to the reaction of HCl with alkalis
forming stable alkali chloride vapours. This can even release alkalis from aluminosilicate
ECN-C-99-090
Annex A; page 12
minerals. The amount of alkali released into the gas phase is more related to the total chlorine
than total alkali content. Chlorine also increases the danger of bed-agglomeration due to the low
melting points of chlorides and the formation of low-melting eutectics. Manzoori (1990) noticed
agglomeration was greatly accelerated by impregnation of coal with sodium-chloride solution.
Furthermore, chlorine can accelerate high temperature corrosion.
2.2.1.3
Sulphur
In combustion sulphur is released as SO2 , which reacts with alkalis to form sulphates. Sulphates
have very low vapour pressures and condense readily on cooler surfaces, such as heat transfer
tubes and bed particles. In gasification, sulphur is largely converted in hydrogen sulphide and
sulfation does not occur. Formation of low-melting silicates and stable alkali chloride vapours is
then more likely.
2.2.1.4
Silica
Silica particles absorb alkalis creating low-melting surfaces due to the fluxing action of alkalis.
Paisley and Litt (1993) report that agglomeration in gasification of switch grass and poplar in
the Battelle pilot plant was due to formation of the low-melting eutectic K2 Si4 O9 -SiO2 (eutectic
point 760C) by absorption of fuel potassium by the sand.
2.2.2
Bed material
From the literature it becomes clear that silica sand is not a suitable bed material in FBC and
FBG of high-alkali fuels, due to formation of low-melting eutectics on the surfaces of bed
particles by the reaction of silica with alkalis. Alumina does not react with the ash and may
make higher temperatures possible. Pure hematite is reported as the most suitable bed material
in combustion of high-alkalis biomass. Dolomite, sometimes used for its catalytic effect on
thermal cracking of tars, can give rise to strong calcium-based ash deposition in the freeboard.
The same goes for limestone, which is used for in-bed sulphur removal. Addition to the bed of
limestone may in some cases even lower the ash melting temperature. Some illustrative
experiences are reported below.
Grubor et al. (1995) conducted long term combustion tests with high-alkali corn cobs in a 150
kW fluidised bed with silica sand as bed material. Operation at 820C had to be stopped after 60
hours due to extensive agglomeration, while this was necessary after 180 hours at 750C.
Severe fouling of the freeboard walls was also noticed, possibly due to the increase in freeboard
temperature as a result of combustion of corn cobs on the top of the defluidised bed. There was
a build up of sodium, potassium and iron in the bed with time, corresponding with deposition of
40 % of the alkalis fed with the fuel on the bed particles. Silica bed particles showed a glassy
amorphous phase on their surface, resulting from the fluxing action of the alkalis. Heat
treatment of mixtures of alkali salts and bed materials, including corundum (Al2 O3 ), silica,
hematite and hematite ore, revealed that hematite is the only suitable bed material for the FBC
of high-alkali biomass fuels. Silica already showed sintering at 700C, corundum at 800C,
while hematite did not show any sintering even at 900C and a alkali content of 20 %. Feeding
alkali salt solutions into a laboratory-scale fluidised-bed of hematite did not give any
agglomeration at temperatures around 950C. There was however a strong increase in fine
particles elutriated from the bed, caused by attrition of the porous scale formed on the hematite
particles by reaction with alkalis.
Ghaly and Ergudenler (1993a, 1993b, 1994a and 1994b) studied agglomeration of silica and
alumina bed materials in the fluidised-bed gasification of straw. The silica bed agglomerated
ECN-C-99-090
Annex A; page 13
once the temperature reached 800C, resulting in defluidisation. No agglomeration was noticed
at 750C. The alumina bed agglomerated at 920C, corresponding with the softening point of
the straw ash of 921C. Below 900C, gasification of straw in alumina did not give any
agglomeration problems. They mention that temperature is the key variable and not the ash
content of the bed or the gas velocity. However, test runs were relatively short so that no real
consideration was given to ash accumulation in the bed with increasing operating time. Heat
treatment in air of mixtures of straw ash and bed material at different temperatures gave similar
agglomeration temperatures, indicating this kind of simple tests can be valuable in assessing the
agglomeration propensity of a bed material/fuel ash combination.
At VTT (Madsen and Christensen, 1995) 100% straw was gasified in a PDU-scale bubblingfluidised-bed reactor in beds of alumina and dolomite. Sintering was especially noticed in the
freeboard at temperatures around 800C, which was severe with dolomite in the bed. Sintering
in the bed at temperature around 750 was ascribed to the re-circulation of cyclone ash giving
accumulation of low-melting straw ash in the bed.
2.2.3
Fluidisation conditions
Atakul and Ekinci (1989) studied the effects of fluidisation conditions, particle size, bed
material and lignite type on agglomeration and defluidisation in FBC. They found that the
agglomeration temperature could be lowered by:
1. Larger fuel particles: agglomeration temperature increased 60-70C when the fuel particle
size was increased from 1.4 to 4.5 mm.
2. Smaller bed particles: agglomeration temperature increased 50-70C when the bed particle
size was increased from 0.85 to 0.4 mm.
3. Higher fluidisation velocity.
4. The use of limestone and dolomite instead of quartz as bed material. This was ascribed to
the lower density and thus lower minimum fluidisation velocity of these materials.
The static bed height did not have an effect, due to the concurring effects of poorer fluidisation
and higher bed material to ash ratio.
2.3
How to prevent agglomeration?
Some measures to prevent or delay ash related problems for an existing installation include:
- chemical additives to the bed to increase the ash melting temperature,
- blending with wood or coal,
- lowering the bed-temperature below the ash sintering temperature,
- periodical refreshment of the bed inventory before critical ash build-up in the bed.
2.3.1
Chemical additives
Additives mentioned to reduce or delay bed-agglomeration and defluidisation in combustion

and gasification are:
- Periclase (MgO)
- Calcite (CaO)
- Dolomite (CaCO3 .MgCO 3 )
- Limestone (CaCO3 )
- Clay minerals (aluminosilicates), like kaolin
ECN-C-99-090
Annex A; page 14
The mechanisms are not fully understood. Some fine-grained additives, like limestone, can also
have adverse effects by forming deposits on heat transfer surfaces and reactor walls (Miles et
al., 1996) and are even reported to decrease ash deformation temperatures (Osman, 1982).
Above mentioned additives are also reported to reduce problems with corrosion and erosion in
biomass-fired fluidised-beds (Coe, 1996).
Mechanisms to prevent agglomeration by additives appear to be:
Increasing the melting temperature of the ash or reducing stickiness of ash coatings on bed particles
periclase
calcite
alumina and clay minerals
Fixation of volatile alkali metals in high melting minerals
kaolin
Removing sulphur before it reacts with the ash to form low melting compounds
limestone
dolomite
Reported experience
Skrifvars et al. (1992a) studied the effect of limestone and clay mineral on the sintering
behaviour of coal ashes. Limestone and clay mineral decreased the sintering tendency of brown
coal ash at 500 to 600C, but increased the sintering tendency at 700C. Further addition at
700C above a Ca/S ratio of 1.3 again decreased the sintering tendency below that of the ash
alone. This effect was slightly greater for the clay mineral. The results were not explained by
chemical reactions taking place, but simply by dilution of the ash with high melting material.
Osman (1982) investigated the effect of CaCO3 , CaO and MgO on the standard deformation and
fusion temperatures of biomass ashes. Addition of CaCO3 increased the fusion temperature by
100C, but the deformation temperature was reduced by 100C. Addition of MgO increased
both the deformation and fusion temperatures effectively above normal FBC and FBG
temperatures. CaO gave only a moderate increase of these temperatures. Use of CaCO3 , MgCO3
and CaO in a downdraft gasifier did not reduce slagging problems, as the ash melting
temperature could not be increased above the temperature in the hot spots of the reactor. CaO
was added to the raw straw as dry particulate matter, followed by pelletization. CaCO3 and
MgCO3 were added by spraying a solution of these compounds on the fuel.
Manzoori (1990) noticed that a high content of fine-sized extraneous inert minerals in a highsodium, high-sulphur low-rank coal, inhibited agglomeration in fluidised-bed combustion at
850C. The same coal with similar contents of sulphur, organically bound sodium and sodium
chloride, but almost no inert mineral inclusions, gave total defluidisation of the bed after only
3.5 hours. The difference was ascribed to the fact that the fine inert minerals become embedded
in the ash coating deposited on the bed particles, making it less sticky.
Paisley and Litt (1993) report that gasification of switch grass and poplar lead to formation of
sand agglomerates in the combustor of the Battelle pilot plant, hampering circulation of the
sand. The bonding agent was the low melting eutectic K2 Si4 O9 -SiO2 (eutectic point 760C)
formed by reaction of fuel potassium with the sand. Addition of 2 % by weight of MgO with the
feed prevented agglomeration problems with poplar, even at normal high combustor
temperatures around 1050C. Thermomechanical analysis of mixtures of sand, ash and chemical
additives showed kaolinite is not effective in decreasing the sintering tendency, calcite (CaO)
can only give a slight reduction in sintering, and periclase (MgO) can eliminate sintering.
Mann et al. (1992) report that addition of limestone in fluidised-bed combustion of low-rank
coals reduces problems with ash deposition and bed-agglomeration. They ascribe this to
removal of sulphur, so that it no longer reacts with the alkalis and calcium from the fuel.
ECN-C-99-090
Annex A; page 15
Particle size seems to be important as they noticed that deposition of fine-grained CaO from the
fuel on heat transfer surface followed by sulfation is an important deposit initiating mechanism.
Addition of limestone can have adverse effects, which follows from the following reported
experience. Miles et al. (1996) report that limestone addition in fluidised-bed combustion of
biomass blends gives rise to deposits consisting largely of calcium sulphates. The deposit
initiating step is unclear and may be related to fuel inorganics. Delvinquier et al. (1995) noticed
that addition of fine-grained limestone to a fluidised bed of silica sand at 850C caused
defluidisation after only 10 minutes. This was ascribed to reaction of very fine calcined lime
particles (3 m) with potassium-aluminosilicate impurities in the surface of the silica sand bed
particles, binding the particles together. When SO 2 was added to the air this no longer occurred.
Sulfation of CaO makes it less reactive towards the aluminosilicate impurities. Anthoney et al.
(1995) report that sulfation of added lime particles caused agglomeration and fouling in CFBC
of petroleum coke. Petroleum coke is a fuel with very low ash and very high sulphur content.
Agglomerates consisted entirely of sintered limestone-derived particles which were highly
sulphated (CaSO4). Addition of MgO reduced the problems. MgO does not become sulphated
and forms inert barriers between the limestone particles which prevents good contact and so
extensive chemical hardening.
Mann et al. (1995) report that kaolin is suitable for in-bed removal of sodium in FBC of Beulah
lignite. They noticed a 93 % reduction, from 3.6 ppm to 0.27 ppm, in the vapour phase alkali
concentration in the flue gas with addition of fine-sized kaolin to the bed. The particle size of
the kaolin had a significant effect on the alkali capture efficiency. Alkali absorption lowers the
melting point of the clay mineral, which may be of some concern. A frequently mentioned
reaction product of kaolin is nepheline, which has a melting point of 1500C. However, it is
likely that also non-stoichiometric phases form with lower melting points. Padban et al. (1996)
tested Quartz, Baskarp and Fyle sands, all based on silica but with different impurities, for their
alkali absorbing potential by heat treatment of mixtures of sand and different alkali salts in a
TGA. Alkali capture depended on both sand type and alkali salt (KCl, NaCl and KNO3 ). They
noticed an optimum temperature for alkali capture around 890C, ascribed to the opposite
effects of temperature on physisorption and chemical reaction. At temperatures between 680
and 1100C the alkalis always caused considerable agglomeration, making these materials
unsuitable for alkali capture in FBG of high-alkali biomass. Alkali gettering by aluminosilicates
in down stream fixed-beds has been studied quite extensively for the purpose to remove alkalis
from hot fuel gas (Scandrett and Clift 1984, Uberoi 1990, McLaughlin 1990). Alkali gettering
by minerals has also been studied with respect to alkali loss in catalytic coal steam gasification
(Kuhn and Plogmann 1983, Formella et al. 1986, Bruno 1986). Conditions were however quite
remote from those in fluid-bed gasification. Potential getters according to McLaughlin (1990)
are characterised by a open layered crystal structure. Kaolin, emathlite and calcium
montmorillonite were found to be very good getters. Water vapour in the gas increases
gettering, while HCl substantially decreases gettering. The latter effect may limit the use of
aluminosilicates to low-chlorine fuels. Furthermore, the getter efficiency decreases with
temperature. Waste paper contains a high concentration of kaolin, making blending with this
fuel possibly attractive.
2.3.2
Fuel blending
One way to achieve a higher ash melting temperature or lower alkali input into the system is to
blend a problematic fuel with coal or wood. When blending fuels one must always keep in mind
that it is the interaction between alkalis, alkaline earth elements and silica, chlorine, phosphorus
and sulphur that causes the problems. Bryers (1996) for example mentions that co-firing of nonproblematic wood with a high-sulphur fuel, like heavy oil, is almost sure to give serious fouling
due to the combination of organically bound calcium and sulphur, giving rise to calcium
sulphate bonded deposits.
ECN-C-99-090
Annex A; page 16
Fist rules biomass fired boilers

From full-scale experience and fuel compositions, Miles et al. (1996) give some fist rules to
avoid excessive fouling problems in biomass-fired boiler. One concerns the maximum alkali
input, (Na2O+K2O) in kg/GJ, into the boiler: 0-0.17 does not give problems, 0.17-0.34 becomes
problematic and would make it necessary to have frequent outages and lower furnace exit
temperatures, while > 0.34 gives rise to unmanageable problems.
Experience with co-gasification
Madsen and Christensen (1995) report experience at VTT (PDU-scale) and Enviropower (pilotscale) with co-gasification of straw and coal. At VTT blends with 54 % straw could be gasified
at temperatures between 960 and 970C in the bed and 970 and 990 in the freeboard without any
problems. The bed material consisted of limestone and char and ash from the coal. At
Enviropower blends of 50 % straw could be gasified at temperatures of 940C in the bed and
950C in the freeboard. Gasification at 980C of a blend with 25 % straw gave strong bedagglomeration, resulting in plant shutdown after only 12 hours. This indicates that temperature
is a very critical variable.
Experience with co-combustion
Salour et al. (1993) investigated co-combustion of wood and rice straw and report that bedagglomeration could be delayed by addition of at least 50 % wood. The reason for this high
percentage is the low ash content of wood. Nordin et al. (1996) noticed severe bedagglomeration within less than 30 minutes when combusting Lucerne and olive flesh in a
fluidised-bed at 850C. Blending with coal (50 wt. %) eliminated the problems, making stable
operation possible for the entire duration of the test runs (10 hours). Additionally, a high sulphur
retention in the ash of 80% was obtained with blending. Initial agglomeration temperatures for
the fuels were raised by blending from 670C in the case of Lucerne and 940C for olive flesh
to 950C and more than 1050C respectively. The agglomeration temperatures correlated very
well with initial sintering temperatures from a compression strength sintering method.
2.3.3
Periodical refreshment bed material
Regular sampling and analysis of the bed material during operation and refreshing the bed
inventory before the a critical alkali concentration or critical agglomeration level is reached may
be a necessary measure when firing of gasifying a high-alkali fuel alone. Grubor et al. (1995)
tried to determine a critical alkali content by spraying alkali salt solutions into a fluidised-bed
until defluidisation. However, they found a considerable difference between critical
concentrations determined on lab-scale and those found from larger scale combustion
experiments with corn cobs.
2.3.4
Lowering the bed temperature
Lowering the bed-temperature below the initial sintering temperature of the bed material/ash
mixture would be the most logical measure to prevent problems. Reducing the air to fuel ratio in
gasification is however undesirable, as this generally gives a lower carbon conversion and
higher concentrations of tar in the gas. Increasing the air excess ratio in combustion is also
difficult as it can cause extinction of the flame in the bottom of the bed. Furthermore, a low bedtemperature may give high CO emissions (Grubor et al 1995). Better ways to achieve a lower
bed temperature for a given air to fuel ratio may be addition of steam to the fluidising gas or
additional heat transfer surface in the bed or freeboard. Freeboard temperatures in biomass
combustion are generally 100C higher than in the bed, due to combustion of volatiles. Osman
(1982) studied the effect of lowering the bed temperature by steam on slagging in downdraft
gasification of agricultural wastes. By mixing steam with the combustion air he was successful
ECN-C-99-090
Annex A; page 17
in keeping the bed temperature below the determined standard ash deformation temperature,
which resulted in slagging free operation.
2.4
How to predict agglomeration?
Many laboratory methods are mentioned in the literature for assessing the slagging, fouling or
agglomerating propensity of a fuel. All suffer from the fact that ash produced in the laboratory
is significantly different from the ash formed in real combustion and gasification, due to the
many influence factors in ash formation that can not be adequately simulated in the laboratory.
Important in this respect are partitioning, gas-solid reactions and re-condensation of alkali
vapours on fly ash. The predictive value of below mentioned laboratory methods is therefore
hard to assess, also because results were not always compared to full-scale experience. Below
mentioned methods should be combined with larger-scale tests in fluidised beds and modelling
to gain a more fundamental understanding of ash formation and agglomeration.
2.4.1
Fuel analyses
Slagging/fouling indices derived from standard ash chemical composition

Osman (1982) discusses some empirical indices correlating the ash chemical composition to the
standard deformation and fusion temperatures of coal ashes. Bryers (1996) gives several
empirical correlations for slagging and fouling. He mentions hat this kind of indices are fuel
specific and do not take into account important operational and design parameters.
Reactive alkali and chlorine content: chemical fractionation
Miles et al. (1996) give a procedure to determine the modes of occurrence of inorganics in
biomass in terms of their solubility in increasingly aggressive solvents. Leaching a biomass
sample with water extracts the loosely bound inorganics and simple inorganic salts. Leaching
with 1 M ammonium acetate extracts all ion-exchangeable inorganics (organically bound).
Finally, leaching with 1 M hydrogen chloride extracts the acid soluble salts, like carbonates and
sulphates. Inorganics in the residue are typically in silicates, oxides or sulphides. This procedure
can yield 5 to 15 % higher total alkali concentrations then analysis of ash prepared at 600C.
Because water soluble and ion-exchangeable alkalis are in a volatile form, their concentration
together with the chlorine concentration gives more information about the fouling propensity of
a fuel than the total alkali content or standard fusion temperatures. This method of chemical
fractionation is also used for analysing spent bed material and agglomerates to determine
whether a reaction has taken place between volatile alkalis and bed material that would have
changed the form of the alkalis (Manzoori 1990, Padban et al. 1996).
Recommended ashing procedures to produce laboratory ash for analysis
Results of ash characterisation tests to determine the sintering and melting behaviour of ash are
strongly influenced by the ashing temperature. Ashing temperatures above 600C give
significant alkali loss, artificially increasing the fusion temperatures. Miles et al. (1995)
therefore recommend an ashing temperature of 600C. A lot of investigators on ash behaviour
have used ashing procedures with temperatures around 500C. Also oxygen plasma ashing is
mentioned as a method to burn away the organic matter, while leaving the mineral matter intact.
This may work for a high rank coal with all the inorganics in minerals, but in biomass the
inorganics are largely associated with the organic matrix. Bryers (1996) warns that low
temperature ashing with long oxidation times gives rise to the formation of artefacts from the
organically bound inorganics, e.g. potassium nitrates and perchlorates, not normally found in
biomass ash.
ECN-C-99-090
Annex A; page 18
2.4.2
Ash melting, sintering and agglomeration temperatures
Standard ASTM ash fusion test

Standard ash fusion temperatures are determined according to ASTM-D1857. This method is
developed to assess the slagging propensity of coal and coke in stoker fired boilers. It is based
on visual observation of the deformation of a ash cone during heat treatment. This method is
well known and many investigators report ash fusion temperatures of biomass ashes (Bining
1988, Miles et al. 1996, Bryers 1994, Skrifvars et al. 1996, Osman 1982). Apart from the
difficulty of visual observation, there are some strong arguments against the predictive value of
this method. Huffman (1981) reports that coal ash already substantially melts at temperatures
200 to 400C below the visually determined initial deformation temperature. Furthermore, that
further deformation of the cone is largely due a decreasing viscosity with temperature, rather
than increased melting. Osman (1982) determined the ash deformation and fusion temperatures
in air of 26 different biomass fuels. He mentions that the melting process is not a sharp
phenomenon, but a gradual process. The change in cone shape differed markedly from sample
to sample. Especially the detection of the fusion temperature was difficult, as the ash did not
always deform to a lump. Osman further mentions that the temperatures determined by this
method are affected by 1) the size, shape and inclination of the cone, fineness of the ash, the rate
of heating and the nature of the atmosphere.
Empirical correlations between standard ash fusion temperatures and ash chemical
composition
Osman (1982) concludes that phase diagrams can be used to predict measured deformation and
fusion temperatures of biomass ash only when the ash consists of a two or three component
system. Fusion temperatures of biomass ashes with four or more components can not be
determined from phase diagrams. Wall et al. (1996) discuss correlations between fusion
temperatures of coal ashes and eutectic points from phase diagrams and find a relatively good
correlation. Steenari (1987) tried to explain sintering behaviour of coal ash and limestone in
fluidised-bed combustion from the eutectic melting points of the oxide systems. However, ash is
not only made of two to four oxides and furthermore these systems are often not fully
investigated.
Differential thermal analysis/thermogravimetric analysis
A number of investigators (Skrifvars et al. 1996, Bryers 1994, Olanders and Steenari 1994,
Misra et al. 1993, Livingstone 1991) have used DTA/TGA to characterise the melting and
volatilisation behaviour of biomass ashes, or to study sorbents for alkali capture and reactions
between bed material and alkali salts (Padban et al. 1996). The method is based on measuring
heat effects in the ash sample (endothermic or exothermic processes) and simultaneous
measurement of the weight loss. It can be a useful tool for determining initial melting
temperatures and to estimate the volatilisation of elements. However, interpretation of the
results without knowledge of the mineralogical composition of the ash is very difficult due to
the many processes that can occur simultaneously in ash (decomposition, volatilisation,
evaporation, reactions with the gas, melting) and the disturbing effects from absorption of gases
and oxidation during storage, combustion of residual carbon, artefacts from low temperature
ashing, etc.
Electrical resistance and shrinkage (thermo-mechanical analysis) method
Wall et al. (1996) discuss different methods based on electrical resistance and shrinkage for
measuring the sintering tendency and initial melting temperatures of coal ashes and interpreted
their results form phase diagrams. The different methods gave different results, already
indicating the sensitivity of these methods for factors like ashing. Manzoori (1990) used
thermo-mechanical analysis to identify the initial sintering temperature of spent bed material
from CFBC and found a good correlation between this temperature and the temperature where
agglomeration started in CFBC of high-sodium, high-sulphur low-rank coals. The initial
sintering temperature from TMA was the same as the temperature where melting started in
ECN-C-99-090
Annex A; page 19
DTA/TGA, so that partial melting was probably the sintering mechanism. Khan (1989)
describes a apparatus for direct measurement of ash fusion and sintering by simultaneous
measurement of electrical resistance and shrinkage at high temperatures and pressures. Sintering
temperatures measured were considerably less than the standard initial deformation
temperatures for the ashes.
Laboratory sintering method: pressure strength measurement of heat-treated ash pellets
Skrifvars et al. (1989) describe a laboratory method to measure the sintering tendency of ashes
based on the compression strength of heat treated ash pellets. This method has been used by a
number of investigators (Ots et al. 1995, Nordin et al. 1996, Skrifvars et al. 1996, 1992a and
1992b). Pellets of laboratory ash or mixtures of ash and bed material are heat treated in a
relevant gas atmosphere. After cooling the crushing strength of the pellets is measured. The
temperature where this crushing strength deviates from the base strength is then referred to as
the initial sintering temperature. Comparison of lab-results with full-scale experience by
Skrifvars et al. (1989, 1992a, 1992b and 1996) shows that this method is very useful in
revealing the mechanism behind agglomeration (f.e. effect gas components) and comparing
different fuels for their sintering and agglomerating propensity, but that it does not give absolute
initial agglomeration temperatures due to the many other important factors besides ash
chemistry.
Heat treatment mixtures of laboratory-ash and bed material and other similar agglomeration
tests
Ghaly and Ergudenler (1994a and 1994b) used heat-treatment of mixtures of straw-ash and bed
material to determine their agglomeration characteristics. The agglomeration temperatures
determined corresponded very well with experience from lab-scale fluidised-bed gasification of
straw. Padban et al. (1996) used heat treatment of low temperature ash (550C) and silica sand
to study agglomeration of biomass ashes with SEM. Moilanen et al. (1996) used heat treatment
of low temperature ash to investigate the agglomerating propensity of a large number of Finnish
and Italian biomass fuels. These kind of tests can be used to obtain a first indication and
comparison of the agglomerating propensity of mixtures of bed material and ash, but do not
give absolute initial agglomeration temperatures again due to the many other important factors
besides ash chemistry.
Flow properties heat treated ash (relevant to loop seal)
Conn (1995) describes a method to evaluate the agglomerating propensity of ash from
petroleum coke in CFBC, based on the hot angels of repose and internal friction. Ash from
the j-valve of a CFBC was heat treated in a vertical stainless steel cylinder. When the desired
temperature was reached the ash was allowed to flow from the cylinder through a small hole in
the bottom. The repose and internal friction angles were then taken as an indication of the ease
of flow of the ash in a real j-valve at the given temperature. These results were compared to
results from the compression strength sintering method, electrical resistance method and SEM
analysis of deposits from a full-scale CFBC to determine the mechanism of sintering. This
allowed determination of the agglomeration mechanism for petroleum coke in CFBC.
ECN-C-99-090
Annex A; page 20
3.
-
REFERENCES OF ANNEX A
Anthoney, E.J., Iribarne, A.P., and Iribarne, J.V. A new mechanism for FBC agglomeration
and fouling when firing 100% petroleum coke. Fluidized bed combustion, ASME, 1:523533, 1995.
Atakul, H. and Ekinci, E. Agglomeration of Turkish lignites in fluidised-bed combustion.
Journal of the Institute of Energy (March):56-61, 1989.
Basu, P. and Sarka, A. Agglomeration of coal ash in fluidized beds. Fuel 62(August): 924926, 1983.
Baxter, L.L. Ash deposition during biomass and coal combustion: a mechanistic approach.
Biomass and Bioenergy 4(2):85-102, 1993.
Baxter, L.L. Ash deposition and corrosion during biomass combustion. TNO-report. TNOMEP-R96/307, 1996.
Bining, A.S. Cereal straw ashes compositional analyses and their fusibility characteristics.
The Combustion Institute Canadian Section Spring Technical Meeting. :40-43, 1988.
Bitowft, B.K. and I.Bjerle, A generic study of the sintering aspects of biomass in a fluid bed
gasifier. Proc. of 11th Annual Symposium on Energy from Biomass and Wastes, Orlando,
16 March 1987.
Bobman, M.H., Hajicek, D.R., and Zobeck, B.J. A study of bed agglomeration resulting
from the AFBC of low-rank coals. 8th International Conference on Fluidized Bed
Combustion. Houston, Texas, DOE/METC-85/6021, 1985.
Brekke, D.W. and Karner, F.R. Analysis and characterisation of atmospheric fluidized-bedcombustion agglomerates. DOE/FC/10120--T7, 1982.
Brooker, D.D. and Oh, M.S. Iron sulfide deposition during coal gasification. Fuel
processing technology 44:181-190, 1995.
Brown, A. Phase composition of sintered masses formed in fluidized bed combustors.
Studsvik Nuclear. Studsvik/NS-90/73, 1990.
Bruno, G. et al. Correlation between potassium losses and mineral matter composition in
catalytic coal steam gasification. Fuel(65): October 1986.
Bryers, R.W. Fireside slagging, Fouling, and High-Temperature Corrosion of heat transfer
surface due to impurities in steam-raising fuels. Prog. Energy Combust. Sci. 22:29-120,
1996.
Campisi, A. A microcomputer-controlled AAS and electrothermal atomizer. State
Electricity Comission of Victoria, Research and Development department, Report
LO/88/003, 1988
Coe, D.R,Successful experience with limestone and other sorbents for combustion of
biomass in fluid bed power boilers. In: Proc. 1st biomass conference of the Americas, Vol 1
(Ed. D.L. Klass), pp. 382-392, National Renewable Energy Lab. Golden, CO (USA) (1993).
Conn, R.E. Laboratory techniques for evaluating ash agglomeration potential in petroleum
coke fired circulating fluidized bed combustors. Fuel processing technology 44:95-103,
1995.
Delvinquer, V. Defluidization at high temperature in fluidized bed of sand by addition of
calcium carbonate. Fluidized Bed Combustion. ASME. 2:801-806, 1995.
Formella, K. et al. Interaction of mineral matter in coal and potassium during gasification.
Fuel(65): October 1986.
French, R.J., et al. The direct observation of alkali vapour species in biomass combustion
and gasification. NREL: NREL/TP-430-5597, 1994.
Ghaly, A.E. and Ergudenler, A. Agglomeration of silica sand in a fluidized bed gasifier
operating on wheat straw. Biomass and Bioenergy 4(2):135-147, 1993a.
Ghaly, A.E. and Ergudenler, A. Agglomeration of alumina sand in a fluidized bed straw
gasifier at elevated temperatures. Bioresource Technology 43:259-268, 1993b.
ECN-C-99-090
Annex A; page 21
Ghaly, A.E., Ergudenler, A., and Laufer, E. Study of agglomeration characteristics of silica
sand-straw ash mixtures using scanning electron microscopy and energy dispersive x-ray
techniques. Bioresource Technology 48:127-134, 1994a.
Ghaly, A.E., Ergudenler, A., and Laufer, E. Agglomeration characteristics of alumina sandstraw ash mixtures at elevated temperatures. Biomass and Bioenergy 5(6):467-480, 1994b.
Grubor, B.D. et al. Biomass FBC combustion- bed agglomeration problems. Fluidized Bed
Combustion. ASME. 1:515-522, 1995.
Hald, P. Alkali metals at combustion and gasification -equilibrium calculations and gas
phase measurings-. The Technical University of Denmark, Department of Chemical
Engineering. PhD thesis, 1994.
Huffman, G.P. et al. Investigation of the high-temperature behaviour of coal-ash in reducing
and oxidising atmosphere. Fuel(60): July 1981.
Khan, M.R. Apparatus for direct measurement of ash fusion and sintering behaviour at
elevated temperatures and pressures. Rev. Sci. Instrum. 60(10):3304-3309, 1989.
Kuhn, L. and H. Plogmann. Reaction of catalyst with mineral matter during coal
gasification. Fuel(62): Februari 1983.
Ligthart, F.S. Fate of sodium and potassium in fluidised-bed gasification and hot gas clean
up. Delft University of Technology, Laboratory of Thermal Power Engineering: EV-report
1865, 1995
Livingstone, W.R. Straw ash characteristics. Department of Energy: DE92-519748, 1991.
Madsen, M. and E. Christensen. Combines gasification of coal and straw. In: Cogasification of coal/biomass and coal/waste mixtures, APAS. Clean coal technology
programme 1992-1994. European Comission.
Mann, M.D., Swanson, M.L., and Yagla, S.L. Characterization of alkali and sulfur sorbents
for pressurized fluidized-bed combustion. Fluidized Bed Combustion. ASME. 1:333-340,
1995.
Mann, M.D., Galbreath, K.C., and Kalmanovitch, D.P. The role of ash chemistry and
operating parameters on ash agglomeration and deposition in FBC systems. In proceedings
of conference Inorganic transformations and ash deposition during combustion :773-789,
1992
Manzoori, A.R. Role of the inorganic matter in agglomeration and defluidization during the
circulating fluid bed combustion of low-rank coals. PhD thesis, Department of Chemical
Engineering, University of Adelaide, 1990.
Manzoori, A.R. and Agarwal, P.K. The fate of organically bound inorganic elements and
sodium chloride during fluidized bed combustion of high sodium, high sulphur low rank
coals. Fuel 71(May):513-522, 1992.
Manzoori, A.R. and Agarwal, P.K. The role of inorganic matter in coal in the formation of
agglomerates in circulating fluid bed combustors. Fuel 72(7):1069-1075, 1993.
Manzoori, A.R. and Agarwal, P.K. Agglomeration and defluidization under simulated
circulating fluidized-bed combustion conditions. Fuel 73(4):563-568, 1994.
Marinov, V., Marinov, S.P., Lazarov, L., and Stefanova, M. Ash agglomeration during
fluidized bed gasification of high sulphur content lignites. Fuel processing technology
31:181-191, 1992.
Mason, D.M. and Patel, J.G. Chemistry of ash agglomeration in the U-GAS process. Fuel
processing technology 3:181-206, 1980.
McLaughlin, J. The removal of alkali salt vapours from hot coal-derived gases. Guildford
(UK): University of Surrey, Dep. of Chemical and Process Engineering. PhD thesis, 1990.
Miles, T.R., et al. Alkali deposits found in biomass power plants: A preliminary
investigation of their extent and nature. NREL: NREL/TP-433-8142, 1996.
Miles, T.R., Bryers, R.W., Baxter, L.L., Jenkins, B.M., and Oden, L.L. Alkalis in alternative
biofuels. FACT 18:211-220, 1994.
Misra, M.K., Ragland, K.W., and Baker, A.J. Wood ash composition as function of furnace
temperature. Biomass and Bioenergy 4(2):103-116, 1993.
Moilanen A., The effect of temperature, chemical composition and gas atmosphere on
sintering of peat fly ash. Proc. Int. Coal Science. Tokyo, Japan, 1: 397-400, 1989.
ECN-C-99-090
Annex A; page 22
Moilanen, A. The effect of ash viscosity on the sintering of peat fly ash, J. Inst. of Energy,
64: 21-25, 1991.
Moilanen, A. Studies of peat properties for fluidized-bed gasification. Espoo:Technical
Research Centre of Finland. VTT publications 149, 1993.
Moilanen, A., Nieminen, M., Sipila, K., and Kurkela, E. Ash behaviour in thermal fluidisedbed conversion processes of woody and herbaceous biomass. 9th European Bioenergy
Conference & 1st European Energy from BiomassTechnology exhibition. Copenhagen,
1996.
Moseley, J.L. and O'Brien, T.J. A model for agglomeration in a fluidized bed. Chemical
engineering Science 48(17):3043-3050, 1993.
Muthukrishnan, M. Salient features and operating experience with world's first rice straw
fired fluidised bed boiler in a 10 MW power plant. Fluidized Bed Combustion. ASME.
1:609-614, 1995.
Nordin, A. et al. Agglomeration and defluidization in FBC of biomass fuels- mechanisms
and measures for prevention. Proc. of Conf. on Applications of Advanced Technology to
Ash-related Problems in Boilers. Baxter, L. and DeSollar, R., 1996.
Oakey, J.E., Minchener, A.J., and Stringer, J. Ash deposition in atmospheric-pressure
fluidised-bed combustion systems. Journal of the Institute of Energy (March):208-219,
1990.
Olanders, B. and Steenari, B.M. Characterisation of ashes from wood and straw. Biomass
and Bioenergy 8(2), 1995.
Osman, E.A. A study of the effects of ash chemical composition and additives on fusion
temperature in relation to slag formation during gasification of biomass. PhD thesis,
University of California, 1982.
Ots, A., Skrifvars, B.J., Hupa, M., and Parve, T. Sintering and fouling characterization of
ashes with high calcium oxide content. Fluidized Bed Combustion. ASME. 2:1393-1405,
1995.
Overend R. P. Thermochemistry: the key to minerals separation from biomass for fuel use
in high performance systems. In: VTT Symposium 164: Power production from biomass II
with special emphasis on gasification and pyrolysis R and DD (Eds. K. Sipilae and M.
Korhonen), pp. 197-214, VTT Energy, Espoo (Finland) (1996).
Padban, N., Ye, Z., and Bjerle, I. Alkali removal and bed material agglomeration studies
applied to biomass gasification. Development in Thermochemical Biomass Conversion.
Banoff, Canada, 1996.
Padban, N., Kiuru, S., and Hallgren, A.L. Bed material agglomeration in PFB biomass
gasification. American Chemical Society, Division of Fuel Chemistry, 40 (3), pp 743-747,
1995.
Paisley, M.A. and R.D. Litt. Evaluation of energy plantation crops in a high-throughput
indirectly heated biomass gasifier. Battelle, 1993
Salour, D., Jenkins, B.M., Vafaei, M., and Kayhanian, M. Control of in-bed agglomeration
by fuel blending in a pilot scale straw and wood fueled AFBC. Biomass and Bioenergy
4(2):117-133, 1993.
Scandrett, L.A. and R. Clift. The thermodynamics of alkali removal from coal-derived
gases. J. Inst. Energy. Vol. 1, No. 433, pp. 391-397, 1984.
Skrifvars, B.J., Hupa, M., and Moilanen, A. Measuring the sintering tendency of ash by a
laboratory method. Journal of the Institute of Energy (September):131-137, 1989.
Skrifvars, B.J., Hupa, M., and Hiltunen, M. Sintering of ash during fluidized bed
combustion. In proceedings of conference Inorganic transformations and ash deposition
during combustion 1992
Skrifvars, B.J. and Hupa, M. Sintering of ash during combustion and gasification. Liekki
progress report. 2:531-565, 1992a.
Skrifvars, B.J., Hupa, M., and Hiltunen, M. Sintering of ash during fluidized bed
combustion. Ind. Eng. Chem. Res. 31:1026-1030, 1992b.
Skrifvars, B.J, M. Hupa, R. Backman and M. Hiltunen: Sintering mechanisms of FBC
ashes. Fuel 73 (2) 171-176 (1994).
ECN-C-99-090
Annex A; page 23
Skrifvars, B.J., Sintering tendency of different fuel ashes in combustion and gasification
conditions, PhD Thesis, Abo Akademi, Department of Chemical Engineering, 1994
Skrifvars, B.J., Hupa, M., Moilanen, A., and Lundqvist, R. Characterization of biomass
ashes. Applications of Advanced Technology to Ash-related Problems in Boilers. Baxter, L.
and DeSollar, R. pp. 383-398, 1996.
Steenari, B.M. A practical and theoretical study of sintering of coal ash and limestone in a
fluidized bed combustor. Staten Energiverk. FBT-87/34, 1987.
Uberoi, M. et al. The kinetics and mechanism of alkali removal from flue gases by solid
sorbents. Progress in Energy Combustion Science. Vol. 1, pp. 205-211, 1990.
Vassilev, S.V., Kitano, K., Takeda, S., and Tsurue, T. Influence of mineral and chemical
composition of coal ashes on their fusibility. Fuel processing technology 45:27-51, 1995.
Wall, T.F., Creelman, R.A., Gupta, S., Coin, C., and Lowe, A. Coal ash fusion
temperatures- New characterisation techniques and associations with phase equilibria.
Applications of Advanced Technology to Ash-Related Problems in Boilers. Baxter, L. and
DeSollar, R. :541-556, 1996.
Williamson, J., West, S.S., and Laughlin, M.K. The behaviour of bed material during
fluidized bed gasification: The effects of mineral matter interactions. Fuel 73(7):1039-1045,
1994.
ECN-C-99-090
Annex A; page 24
ANNEX B. OVERVIEW OF PREDICTION METHODS REPORTED IN LITERATURE

Method
Method description
fuel analysis
ASTM D1857, standard ash Four characteristic temperatures are identified from an ash cone according to a
fusion test
standardised description (visual observation of the ash cone during heating in
an oven). The characteristic temperatures are: initial deformation temperature
IDT, softening temperature ST, hemispherical temperature HT, fluid
temperature FT.
Chemical fractionation
Increasingly aggressive solvents leach the same sample in a series of three
sequential leachings, producing four samples (incl. raw material) for
characterisation.
static procedures
Empirical correlation
Phase diagrams and thermodynamical calculations (eutectic melting points) are
between standard fusion
used to predict the deformation and fusion temperatures of biomass ash.
temperatures and chemical Skrifvars: thermodynamically calculated temperature at which 15% of the ashes
ash composition
is melted (Tmelt 15% ) is a measure for the sintering temperature.
DTA/TGA: differential
thermal analysis/
thermogravimetric analysis
Dilatometric shrinkage test
Thermal- and electrical

conductance test
Electrical conductance test
and shrinkage test
The heat effects (endothermic or exothermic processes) in the ash sample and
the weight loss of the sample are measured simultaneously. Detectable
processes are melting processes, decomposition processes and chemical
reactions.
The change in length of a pulverised ash sample is measured during heating.
Shrinkage indicates sintering.
The thermal or electrical conductance of an ash sample is measured. As the
cross sectional area of sinter bonds between the particles grows, the
conductance path is increased.
Simultaneous measurements
Compression strength based The ash sample is crushed and pelletised to a cylindrical pellet and heated in a
sintering test
controlled gas atmosphere. After cooling, the pellets are crushed in standard
compression-testing equipment, and the compression strength is taken as
measure of the degree of sintering.
ECN-C-99-090
Predictability
Remarks
References
(chapter 6)
Predictability is very low. It predicts too high agglomeration

temperatures and the predicted trends are false as well. Skrifvars:
ASTM failed in all cases to predict the bed agglomeration temperature.
[17,18] and this

report
NREL: laboratory to laboratory variations can be large. Skrifvars:

predictability is moderate.
[19,20]
It is still hard to predict the agglomeration temperature with a

thermodynamical model. However, models and programs are improving
fast. Even with a simple model the predicted temperatures are not worse
than a laboratory experiment. Skrifvars: good correlation of Tsint with
Tmelt15% for 6 of 10 ashes. Nordin: a complete non-ideal data-set was not
available but the general results, as well as the trends obtained agree
well with the experimental results obtained (compression strength
sintering test and lab-scale FBC).
The DTA measurements do not clearly support the melt formation
which is measured with the compression strength based sintering test
and/or thermodynamically calculated.
ash-mixtures are very complex [21,22,23,24,20,25,26

and hard to model (esp. silica), ,27,28,29,30,31,32]
unknown which part of the ash and this report
has to be melted before
agglomeration is a fact, kinetic
parameters are not considered
but probably relevant
Can be used to reveal the
mechanism behind sintering
[33,34,35,36,37,38]
and this report
hman: The shrinkage method is superior to the ash fusion tests, but
should be supplemented with continuous measurements of the
conductance.
unknown
Gluckman relates sintering to

ration of velocity and minimum
fluidisation velocity
Satisfactory contact between
ash and electrodes is essential
but hard to achieve.
Good contact between ash and
electrodes is essential but hard
to achieve. This method is in
the literature used for coal
ashes only.
Very useful in revealing the
mechanism behind
agglomeration
[18,39,40,41,42]
unknown
This test is good for predicting bed agglomeration trends except for
cases at which interaction between the bed material and the ash takes
place. Skrifvars investigated 10 biomass-ashes: for 5 the result
correlated with fluidized bed tests.
[18]
[18,43,44,45,46]
[33,17,18,25] and this

report
Annex B; page 1
Method
Method description
Predictability
static procedures (cont.)

Heat treatment and
subsequent SEM analysis.
Mixtures of ash and bed material are heated in an oven for a certain period. The Not used to predict agglomeration temperatures.
heat-treated samples are chemical analysed (X-ray fluorescence spectroscopy,
atomic adsorption spectro-photometry, ion chromatography) and the particle
surfaces are analysed by the SEM.
Flow properties heat treated Ash is heat treated in a vertical stainless steel cylinder. When the desired
No temperature is predicted and the predictability of this method is not
ash
temperature is reached the ash is allowed to flow from the cylinder through a
very promising.
small hole in the bottom. The repose and internal friction angels are then taken
as an indication of the ease of flow of ash in a L-valve of a CFBC.
Viscosity measurements
A critical viscosity is required for adhesion of ash particles, this fact can be
used to predict an agglomeration temperature. At high temperatures viscosity
measurements are mostly carried out using a rotating crucible viscometer.
Viscosity of less fluid slags can be measured by a rod penetration viscometer.
dynamic procedures
Bubbling Fluidized Bed
Combust a certain amount of fuel at fixed temperature ("ashing temperature")
Combustion Facility: BFBC and subsequently stop the biomass fuel feeding and raise the bed temperature
(at ETC)
externally (by supplying propane flue gas with the same oxygen concentration
as the biomass flue gas) with 2-3C/min until agglomeration occurs (sudden
drop in pressure and a sudden change in temperature). The particle temperature
(both sand and biomass/char) is measured using a FTIR pyrometer of which the
probe is mounted in the bubbling bed.
The biomass material is converted by air-blown gasification at 750 during 4
Gasification Facility: WOB hours while feeding the material with about 1 kg/h. Subsequently the
(at ECN)
temperature of the bed is raised stepwise (by adjusting the electrical heating of
the gasification vessel) with 25 every 30 minutes until agglomeration occurs.
Both the fuel feeding rate and airflow remain unchanged. The 4 hours of
stationary operation makes up the biggest part of the total duration of the
experiment. This ensures a certain ash accumulation in the system, which is
independent of the agglomeration behaviour of the fuel tested.
At a fixed temperature, a fixed amount of fuel (50 gram) is combusted in the
Combustion Facility: BFBC bed containing 65 gram of bed material. After the fuel batch is fed, the reactor
(at VTT)
is emptied and a new experiment starts at a higher temperature (steps of 25C)
until signs of ash sintering are observed.
ECN-C-99-090
No prediction of the agglomeration temperature is mentioned.
Remarks
References
(chapter 6)
Used to reveal the mechanism

behind
sintering/agglomeration.
[47] and this report
In combination with other

methods this method
contributes to the
determination of the
agglomeration mechanism of
petroleum coke.
This method is in the literature
used for coal ashes only.
[45]
[18]
Method includes interaction between bed material and ash. Skrifvars: It Particle temperatures can be
is excepted to give accurate predictions of possible bed agglomeration 100 higher in temperature
but no full-scale tests are carried out.
than the bed during
combustion.
[17,9,48,25,18,49]
Method includes interaction between bed material and ash. Fuel supply
until agglomeration ensures constant presence of volatile components
(like alkali's). Gasification generally leads to better temperature
homogeneity (lower peak temperatures).
this report
this report
Annex B; page 2
ANNEX C. PUBLICATIONS, CONFERENCE PRESENTATIONS

AND PATENTS
Publication:
F. S. Ligthart, A. van der Drift and A. Olsen, Bed-agglomeration in fluidised bed conversion of
biomass, Proc. 10th European conference and technology exhibition: Biomass for energy and
industry, Wrzburg (Germany), 8-11 june 1998 (Eds. H. Kopetz et al.), pp. 1765-1768.
Conference presentation:
A. van der Drift and A. Olsen, Prediction and solution methods for ash agglomeration and
related problems during biomass conversion, submitted for presentation at 1st World Conference
and Technology Exhibition, 5-9 June 2000 in Sevilla, Spain (submitted).
Patent:
none.
ECN-C-99-090
Annex C; page 1

Ash Fusion Temperatures For Biomass

Uploaded by

Copyright:

Available Formats

Ash Fusion Temperatures For Biomass

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Ash Fusion Temperatures For Biomass

Uploaded by

Copyright:

Available Formats

November 1999

ECN, Holland; 2) Ris, Denmark

A. van der Drift

C.A.M. van der Klein

ECN Fuels Conversion & Environment

COMPARISON OF PLANNED ACTIVITIES AND WORK ACCOMPLISHED

ANNEX A. A literature survey on agglomeration, mechanism, prediction and prevention

TASK 1: Information and material collection

CFB: circulating fluidised bed

TASK 2: Selection of biomass fuels, bed materials and additives

Experimental programme with selected biomass materials, marked grey are

verge Dutch Danish Danish sewage cacao willow

combustion and controlled aggl. (5 kWth)

gasification (500 kWth)

melting temp. of ash and ash/bed

f. compression strength compr. strength of sintered ash pellets

standard ash melting temperature

heating while observing ash/sand

combustion and controlled aggl. (5 kWth)

TASK 3: Description of the lab-scale apparatus and test procedures

Chemical laboratories at ECN and Ris (DK-TEKNIK) (a)

BFBG: bubbling fluidized bed gasification facility (5 kWth ) at ECN (b)

Figure 3.1 BFB gasifier at ECN

Figure 3.4 Sketch of fluidised bed combustion facility at ETC in Sweden

DTA/TGA: differential thermal analysis and thermo-gravimetric analysis at

Ris and ECN test apparatus and conditions for DTA/TGA-analysis

ash and sand

Ris: N2 and N2 /O2

Ris: N2 and N2 /O2

Compression strength measurement facility by Ris (TNO) (f)

Compression strength measurements were performed at TNO in the Netherlands as a

Standard facilities for measuring standard ash melting temperatures by ECN

SEM and HTM (high temperature microscope) at Ris (h)

BFBC: bubbling fluidized bed combustion facility by Ris (VTT) (i)

TASK 4: Construction and modification of lab-scale fluidised bed

BFBG: bubbling fluidized bed gasification facility (5 kWth ) at ECN (b)

TASK 5: Development and description of an agglomeration model

wt% molten material .

TASK 6: Performance of lab-scale and bench-scale tests

Chemical laboratories at ECN and Ris (DK-TEKNIK) (a)

Analyses of biomass fuels

Chemical composition of the seven selected biomass/waste fuels as used in the

------ wheat straw -----Dutch

average values from Phyllis

Some indices for the fuels used in the project

average values from Phyllis

Ash composition [wt%] of willow, ashed at different temperatures

Ash composition [wt%] of wheat straw (dry), ashed at different temperatures

wheat straw (dry)

Ash composition [wt%] of weathered wheat straw (wet), ashed at different

weathered wheat straw (wet)

BFBG: bubbling fluidized bed gasification facility (5 kWth ) at ECN (b)

temperature under air distributor

pressure drop over bed [mbar]

time to agglomeration [minutes]

Standard agglomeration test, detail of last minutes

pressure drop over bed [mbar]

time to agglomeration [minutes]