EUROSIM 2016 & SIMS 2016

Agglomeration Detection in Circulating Fluidized Bed Boilers

using Refuse Derived Fuels
Nathan Zimmerman1 Konstantinos Kyprianidis1 Carl-Fredrik Lindberg1,2
1 School of Business, Society and Engineering, Mälardalen University, Box 883, 72123 Västerås, Sweden,
{nathan.zimmerman,konstantinos.kyprianidis}@mdh.se
2 ABB Corporate Research, Forskargränd 8, 72178 Västerås, Sweden, carl-fredrik.lindberg@se.abb.com

Abstract low for a high degree in fuel flexibility, high combustion

efficiency, lower emissions, and relatively fast response to
The formation of agglomerates in a refuse derived fuel load change (Basu and Fraser, 2015; Hernandez-Atonal
(RDF) fired circulating fluidized bed (CFB) boiler has et al., 2007; Gungor, 2009), and therefore can handle the
been investigated by implementing a dynamic model of complexities of RDF.
the combustion process. The nature of refuse derived fuel, Typically, CFBs operate at a combustion temperature
which is complex in composition, leads to an increased between 750 ◦ C and 900 ◦ C and use sand to fluidize the
tendency for agglomerate formation. Notwithstanding the bed material. However, a large portion of the combustible
fact that a robust control scheme is essential in prevent- material in RDF contain high levels of alkali and chlorine
ing the decrease in boiler efficiency from accelerated ag- compounds, which both have low melting and low vapor-
glomerate formation. Therefore, a mechanism for detect- ization temperatures (Pettersson et al., 2013), and can re-
ing agglomeration through a physical model by looking at act with the fluidizied sand to form low melting eutectics.
the minimum fluidization is presented. As agglomerates It is crucial to maintain the operating temperature within
form between the fuel ash and bed sand the average diam- limits. Else there is a significant risk of sand-ash reac-
eter of the sand will increase and therefore the minimum tions which can lead to agglomeration, slagging, and in the
fluidization velocity. Samples of bed material have been worse case scenario complete defluidization. The variabil-
sieved and measured from a 160 MW circulating fluidized ity of RDF’s composition, and therefore heating value, can
bed boiler fired with refuse derived fuel to determine bed lead to the temperature deviating above the boiler’s lim-
material size distribution. The findings have been corre- its for short periods, in the form of hot zones. The latter
lated and match an increase in the minimum fluidization can increase the propensity of agglomerate formation ei-
velocity during a seven day sampling period where the bed ther through ash melting or the ash coating of sand. Once
material size distribution increases above the average sand agglomerates form they reduce the boiler’s efficiency be-
diameter. cause the fluidized median is hindered and requires in-
Keywords: circulating fluidized bed, agglomeration, de- creasingly more air to properly combust the incoming fuel.
tection, RDF A dynamic model of an RDF fired CFB has been mod-
eled after a 160 MW industrial installation. Bed samples
1 Introduction for a one week period have been taken and analyzed for
The role of waste management in the 21st century is es- any increase in the bed material particle diameter. The
sential in maintaining the excessive amount of waste pro- model has proven to be capable of detecting agglomerate
duced each year in the world. Most developed coun- formation off-line, showing an increase in the minimum
tries have modern routines and guidelines for dealing with fluidization during a period of time when there was a cor-
municipal solid waste (MSW) by sorting out recyclable, responding significant increase in particle diameter.
toxic, and organic materials. Some of these countries are
utilizing MSW for producing refuse derived fuels (RDFs). 2 Literature Review
These are used in waste incinerators for the purpose of
producing heat and power. When compared to other alter- 2.1 Circulating Fluidized Bed Boilers
natives, such as land-filling, waste incineration is more re- The schematic illustrated in Figure 1 is that of a CFB.
liable and environmentally friendlier (Hernandez-Atonal Its a type of boiler that uses sand as the fluidizing me-
et al., 2007). dian, where a degree of solid reflux is achieved allow-
Due to the complex nature of RDF, a viable option ing for a more uniform temperature throughout the boiler.
for thermal waste treatment is to use them as a source Major advantages include stable combustion at low tem-
of fuel in circulating fluidized bed (CFB) boilers to pro- peratures (750-900◦ C), uniform temperature distribution,
duce heat and power, simultaneously solving waste and large solid-gas exchange area, ability to handle problem-
energy issues. CFB boilers are unique in design and al- atic fuels with a large variability in size, moisture content,

DOI: 10.3384/ecp17142148 Proceedings of the 9th EUROSIM & the 57th SIMS 148
September 12th-16th, 2016, Oulu, Finland
EUROSIM 2016 & SIMS 2016

Method, The Balance Method, and the The 14 C-Method

(Staber et al., 2008). The downside is that these methods
are impractical for real-time applications.
2.3 Agglomeration
The development of agglomerates, the binding of bed par-
ticles, is highly dependent upon the characteristics of the
fuel’s ash content, particle-to-particle interactions and the
hydrodynamics in the boiler. As the bed material is flu-
idized ash can melt and bridge bed material together, as
illustrated in Figure 2a. Alternatively, ash can lead to a
buildup on bed material and create a sticky coating, as il-
lustrated in Figure 2b, which causes more particles to stick
together. Therefore, the likelihood of agglomeration for-
Figure 1. Circulating Fluidized Bed Boiler.
mation is dependent upon ash characteristics and has been
the topic in many studies looking at biomass and waste
fuels (Elled et al., 2013; Lin et al., 2003; Scala and Chi-
and heating value. They are capable of high heat trans- rone, 2008; Bajamundi et al., 2015; Ryabov et al., 2003).
fer coefficients between bed and heat exchanger surfaces Proper fluidization is also needed to inhibit the likelihood
(Werther, 2007; Scala and Chirone, 2004). of agglomeration. If particle mobility and mixing are be-
When the fuel enters the boiler it begins to heat, where low optimal it will not only cause a reduction in heat and
water is evaporated, volatiles begin to be released through mass transfer of the bed material (Liu et al., 2012), but
devolatilization, ignition begins, and fragments begin to the viscous materials can bond together and can lead to
form. During this process more char becomes available permanent bonds depending upon the residence time.
for combustion. Primary air enters the bed of the boiler In a study by (Lindberg et al., 2013) they stated that the
and entrains the sand and fuel. Above the bed is the free- predominant ash forming elements are K, Na, Ca, Mg, Fe,
board, where secondary air is injected to allow for a higher Al, Si, P, S, Cl, C, H, and O. The main ash characteris-
combustion efficiency and it has been shown to reduce CO tics that can lead to agglomeration come from Na and K,
concentrations (Hernandez-Atonal et al., 2007). At the top alkali metals. When agglomerates form they disrupt the
of the boiler the flue gas along with a net solid flux of dynamics in the boiler and if the temperature exceeds the
solids are carried to the cyclone which allows for the flue melting temperature of the particles for too long the sticky
gas to be utilized in the convective section and separates outer layer of bed material can form permanent bonds (sin-
fly ash from hot solids (sand and ash). The hot solids then tering), which can lead to slagging or in worst case sce-
pass through a heat exchanger to utilize their high enthalpy narios complete defluidization. In a study by (Chirone
content, and then the cooled solids which contain unburnt et al., 2006) they report that bed agglomerates begin to
char are recirculated back into the bed via the loop seal to form near burning char because of the higher temperatures
be fully combusted. and this increases the formation of melt, or the particles
stickiness. This is especially problematic because it can
2.2 Characteristics of RDF lead to the formation of eutectics and the escalation of ad-
RDF is a byproduct of home and industrial waste, but hesive forces during sintering. Eutectics formed can have
before it can be fired it needs to be sorted and treated. melting points as low as 401 ◦ C and 552 ◦ C for Na2 S2 O7
The waste comes into the sorting facility were it is first and Na3 K2 Fe2 (SO4 )6 respectively (Dunnu et al., 2010).
chopped and shredded into credit card size pieces. Metals
are removed by using high power magnets and the remain-
ing mixture passes over a large wind-sifter, which allows
for heavier objects such as glass and ceramics to dropout
of the mixture. The separated items are sent to recycling,
where what is left can be classified as RDF, and is ready
for transport into the boiler. (a) Agglomeration from ash melting.
We can characterize what remains in the fuel as either
biomass-based (climate neutral) or fossil-based, where the
fossil-based constituents are typically plastics and textiles.
Due to the sorting process, determining the proportions of
biomass vs. fossil based portions through manual sorting
is impractical. There are three established methods for (b) Agglomeration from ash coating.
determining the biomass-based portion, and therefore the
fossil based portions of RDF: The Selective Dissolution Figure 2. Prominent modes of Agglomeration.

DOI: 10.3384/ecp17142148 Proceedings of the 9th EUROSIM & the 57th SIMS 149
September 12th-16th, 2016, Oulu, Finland
EUROSIM 2016 & SIMS 2016

Agglomerates are a common occurrence in fluidized The method proposed by the authors in this paper for
beds, but a well controlled boiler can control the fluidaza- the detection of agglomerates is based on the minimum
tion velocity or temperature well enough to break these fluidization needed to fluidize the bed material. As the di-
clusters apart before sintering takes place. Presented in ameter of the bed material increases so will the amount of
(Skrifvars et al., 1994; Yan et al., 2003), they discuss how air need to fluidized the bed material in order to operate the
the oxygen concentration in the boiler is linked to the risk boiler within the proper temperature limits. Hence model-
of sintering, where regions of higher oxygen content can ing the minimum fluidization can be used in early warning
lead to hot spots. Two mechanisms, flue gas recirculation detection of minimal to extreme agglomerate formation.
and air flow, can be used to prevent the likelihood of hot
spots. Flue gas recirculation into the boiler bed will help 3 Methodology
in reducing the oxygen content as well as bed temperature.
Maintaining a consistent air flow rate, depending upon the The dynamic model used has been calibrated and val-
quality of the fuel, for fluidizing the bed material can also idated using data from a RDF fired CFB, boiler 6, at
help in facilitating a well mixed combustion median in or- MälarEnergi, in Västerås, Sweden. It was possible to
der to reduce hot spots. back calculate the mass flow rate of fuel by conducting
A wide range of methods for agglomeration detection, a heat balance on the boiler’s heat exchangers. By this
when fired with biomass, have been compared and can be method it was determined that 16.8 kg/s of RDF are fed
grouped into three categories: on-line detection, experi- into the boiler, taking the thermal efficiency of the boiler
mental methods (controlled agglomeration tests), and the- into consideration, this corresponds to MälarEnergi’s re-
oretical evaluations (fuel ash analysis), (Gatternig, 2015). ported value of 30 tonne/hour. The model is designed to
The last two methods seem reasonable when considering allow for the real input of primary air, secondary air, and
a fuel that is relatively homogeneous, but RDF is a com- flue gas recirculate mass flows and corresponding temper-
plex and difficult fuel to model and predict melting points, atures. It is assumed that the mass flow rate of the fuel and
and ash composition because of its composition variabil- its respective LHV are constant.
ity. In most industrial fluidized bed boilers it is routine to 3.1 Description of Model
take random fuel samples to check for composition con-
sistency. These samples could be used in agglomeration A model has been designed in DYMOLA using Modelica
indexes (Visser, 2004; Vamvuka et al., 2008) that look at, programming language. The reason for having a dynamic
among other elements, Na and K in the fuel to determine model is to be able to capture the transient behavior of
the likelihood of agglomerate formation. However, this RDF through the combustion process. With the end goal
approach assumes that the fuel characteristics will remain of being able to have a model that has the ability to not
the same year after year, but in reality RDF composition is only monitor agglomeration, but to also be used for emis-
based on the consumers and such an assumption is unre- sions tracking, decision support, and fault detection. The
alistic. Also, this method only considers a fraction of the Modelica modeling language allows users to build model
predictability of agglomeration by looking at fuel compo- libraries with ability to reuse component blocks and to eas-
sition and neglects fuel-ash-bed material interactions, and ily change parameters to match any complex dynamic sys-
therefore can give a bit of insight into agglomeration ten- tem.
dencies and should be used cautiously. Another method Individual component blocks were made for the CFB-
for agglomeration detection is through advanced multi- loop, also represented in Figure 1, for the boiler bed, free-
component/multiphase thermodynamic modeling (Lind- board, cyclone, and hot cycle recirculate (super heater
berg et al., 2013), but there is currently a lack of com- (SH)) respectively. Multiple functions have been been
prehensive thermodynamic databases for ash compounds written in order to accurately represent the thermody-
and the phases formed during combustion. namic properties for all in-coming streams, bottom and
Therefore, online detection is the most suitable way to fly ash, bed material,and flue gas.
predict agglomeration when RDF is used. Current meth-
ods used for the early detection of agglomeration have 3.2 Mass and Energy Balances
been explored by looking at pressure drops and tempera- The model is based on mass and energy balances (equa-
ture fluctuations. However, at this point, the probability of tions 1 and 2) used for the freeboard and bed of the boiler,
defluidization is already prevalent and leaves no options cyclone, and superheater following that of a similar ap-
for the operator but to shutdown the plant, and is not a proach to that presented in (Basu and Fraser, 2015; Gun-
suitable form of early detection. An alternative method, gor, 2009). Where i represents the control volume in the
suitable for early detection, of agglomerates on a small- CFB-loop (A, B, C, D) in Figure 1, m is the mass, H is the
scale in fluidized beds has been proposed by (Nijenhuis enthalpy, α is the percentage of combustion, and Q̇ is the
et al., 2007). They were able to develop an early agglom- heat released during combustion.
eration recognition system that detects very small changes
in hydrodynamic multiphase systems, and allow detection d(mi )
of agglomeration up to 60 minutes before occurring. = Σṁin,i − Σṁout,i (1)
dt
DOI: 10.3384/ecp17142148 Proceedings of the 9th EUROSIM & the 57th SIMS 150
September 12th-16th, 2016, Oulu, Finland
EUROSIM 2016 & SIMS 2016

therefore temperature dependent ash specific heat values

could be estimated, equation 7.
d(mi Hi )
= ṁin,i Hin,i − ṁout,i Hin,i + α Q̇released,i (2)  
dt
C p,ash,i = Σ ξ j ·C p, j (7)
It was possible to calculate the enthalpies of all con-
stituent parts in the energy balance, equation 3, where Ti where i represents either fly or bottom ash, ξ is the com-
is the temperature in the control volume and Tre f is refer- pound percentage in the ash, and j is the compound. With
ence temperature taken at 289.15K. the calculation of Cpgases and Cp and corresponding
solids
mass flows equations 2 and 3 can be used to determine the
d(Hi )
= Cpi (Ti − Tre f ) (3) boiler temperature.
dt
The fuel composition and flow rate is constantly chang- 3.3 Hydrodynamics
ing, and in turn so is the boiler bed and flue gas tempera- CFBs operate at a higher gas velocity and therefore the
ture, so a series of functions have been developed to cal- gas and solids within the boiler act like a fluid. In the bed
culate the specific heat for each element in the control vol- of a CFB boiler there is a dense emulsion phase, where
ume in the energy balance. In this way, instead of assum- the gas moves through the inventory as large bubbles. At
ing constant values, the model has the capability of more the top of the bed, where the secondary air input is, the
accurately simulating temperature profiles for the bed and bubbles burst dispersing inventory and unburnt char into
freeboard of the boiler. The function developed for calcu- the freeboard. The addition of this secondary air, theoreti-
lating the specific heat (Cp) of the gas follows the practices cally, allows for a dilute well-mixed phase throughout the
of (Wester, 1987), equation 4. freeboard. The freeboard is typically further segregated
Z T n=7  n into a core and annulus. The core is where the gas and
1 T solids flow upward, fine particles are carried out of the
Cp = ∑ anc (4)
T − Tre f Tre f n=−1 1000 bed, and coarser particles tend to form clusters and then
fall back down in the annulus region as a thin film, where
The coefficients for anc represent air, N2 , O2 , CO2 , and the further up in the freeboard the smaller the annulus re-
water vapor. The gas thermophysical properties and com- gion becomes, as shown in Figure 1.
position were calculated by using flue gas stoichiometric
The minimum amount of air velocity required to flu-
calculations. The flue gas composition is found to primar-
idized the bed material is called the minimum fluidization
ily be comprised of CO2 , H2 O,O2 , and N2 . Where the
velocity um f and is the velocity at which the drag of the
model uses the ultimate analysis of the fuel as input, cal-
fluidized median (sand) is equal to the weight of the bed
culates the amount of excess air in the system, and then
material. Following the method used in (Kunii and Leven-
the flue gas composition can be used to calculate the spe-
spiel, 2013) the minimum fluidization can be determined,
cific heat, equation 5, as well as viscosity and density at
equation 8.
the operating temperature.

d pUm f ρg 2 150(1 − εm f ) d pUm f ρg

 
Cpgas = [N2 ]CpN2 + [O2 ]CpO2 1.75
(5) · + · =
+ [CO2 ]CpCO2 + [H2 O]CpH2 O εm3 f φ µg εm3 f φ 2 µg
(8)
d 3p ρg (ρ p − ρg )g
The solids contained within each control volume con-
sist of fuel, sand, ash, and char, as shown in equation 6. µg2
Where α, β , γ, and δ represent the percentage of fuel,
sand, ash, and char respectively that are in the control vol- Where εm f is the void fraction at minimum fluidizing
ume i. conditions, φ is the sphericity of the sand, d p is the aver-
age diameter of sand particles, ρ is either the density of
C p,solids,i = α · C p, f uel + β · C p,sand + γ · C p,ash + δ · C p,char the fluidizing median or sand, µg is the viscosity of the
(6) fluidizing median, and g is gravity.
As the amount of air is increased into the boiler the min-
The functions for fly and bottom ash were developed imum fluidazation velocity of the gas will reach the super-
by using chemical compositions found in (Chang et al., ficial velocity (Us ) and is affected by the amount of gas,
1997). Where they analyzed the chemical composition density and size in a given control area. Typically val-
and determined the compounds and corresponding per- ues for industrial CFBs is in the range of 5-10 m/s, (Basu
cent composition of ash, and found that CaO, SiO2 , Al2 O3 , and Fraser, 2015). This velocity carries bed material, ash,
Fe2 O3 , ZnO, MgO, Cr2 O3 were the major components (in and unburnt char up into the freeboard, where a portion
decreasing order). Specific heat correlations for the com- of these circulate within the boiler but there is also a net
pounds were developed from (Abu-Eishah et al., 2004; solid flux, Gs that leaves the boiler to be recirculated back
Karditsas and Baptiste, 1995; Madelung et al., 1999), through the loop-seal. Where Gs is equal to the amount

DOI: 10.3384/ecp17142148 Proceedings of the 9th EUROSIM & the 57th SIMS 151
September 12th-16th, 2016, Oulu, Finland
EUROSIM 2016 & SIMS 2016

of solids going up in the boiler minus the solids circulat- standard procedure in industry for predicting agglomera-
ing within the boiler. A correlation between Gs and op- tion is determined from pressure drops and temperature
erational conditions was presented by (Guan et al., 2010), changes (Gatternig, 2015), but it has already been men-
equation 9, where Gs is in the rang of 200 and 400 kg/m2 , tioned that this is not suitable for early detection. For early
which is within the operating limits of the designed CFB. detection, using plant process parameters, it is possible to
detect agglomeration by looking at the minimum fluidiza-
tion velocity, where a small increase shows that the diam-
Us 0.375 D 0.195
   
Gs d p eter of the circulating bed material is increasing.
= 547Ar0.248 p (9)
µg gD H This is because with the onset of agglomerates and then
sintering, the minimum fluidization required to fluidized
Where Ar is Archimedes number, D is the diameter of the bed material will slowly increase. Therefore, agglom-
the freeboard, and H is the height of the freeboard. eration can be monitored by modeling process parameters,
while operators can still keep an eye on pressure drops in
4 Results and Discussion the boiler bed and temperature fluctuations.
As mentioned the composition of RDF can vary from
4.1 Validation hour to hour. However, the composition of sand, with
an average diameter in the range of 0.40 < d p < 0.63,
The model has been constructed to predict the perfor-
is known and as agglomerates form this will require a
mance of a 160 MW CFB boiler. The model has been
higher minimum fluidization. Figure 4 confirms the pres-
designed to predict the bed and flue gas temperatures. The
ence of a relationship between the bed particle size in-
validity of the model has been determined by comparing
creasing and the required increase in the minimum flu-
data from boiler 6 at MälarEnergi. Figure 3 illustrates a
idization. Where the bed material size distribution was ob-
simulation, for a week, of the flue gas temperature as it
tained from Mälarenergi, daily average, and the minimum
exits to the cyclone. The accuracy of the model is quite
fluidization is the model’s prediction, taken on a daily av-
good initially, less than a few percent. However, it can
erage. This method allows for the detection of a a small
be seen that when there is sudden increase or decrease the
change in the bed material average diameter with a suit-
model tends to underestimate, or overestimate, the respec-
able amount of time for the operator to make the decision
tive temperature. Since the model is currently designed
to add fresh sand to the boiler before there is a subsequent
using a constant fuel mass flow rate and heating value it
formation of further agglomeration, slagging, or possibly
is reasonable to believe that this is attributing to the devi-
complete defluidization.
ation in the predicted temperature. The model’s profile is
able to follow that of the actual profile, but if the quality Figure 5 illustrates the main elemental composition in
of the fuel coming into the boiler is poor this would be g/kg TS of the bed material samples over the week in ques-
reflected in a substantial drop in the temperature profile. tion. Because the ash melts are coating the sand particles it
is reasonable that the main composition comes from silica,
4.2 Agglomerate Prediction with an average value of 301 g/kg TS. However, a portion
of the silica could come from glass fines in the fuel mix,
There is a lot in the literature about methods used to pre-
but this value is unknown. It should also be noted that all
dict CFB failure from agglomeration, but typically these
of these elements, as stated before, are predominant ash
do not take into account the impact of the combustion en-
forming elements. Therefore, it can be quantified that the
vironment (Yan et al., 2003) like gas to particle interac-
agglomeration of the sand has occurred due to either ash
tions and alkali vapor condensing.
melting or ash coating.
Agglomeration prediction has been studied to a lengthy
extent where fossil and biomass fuels are concerned. The

Figure 3. Temperature profile of the fluegas during the study Figure 4. Bed Material Size Distribution 0.4 < d p < 0.63 mm
period. Correlates to the Minimum Fluidization.

DOI: 10.3384/ecp17142148 Proceedings of the 9th EUROSIM & the 57th SIMS 152
September 12th-16th, 2016, Oulu, Finland
EUROSIM 2016 & SIMS 2016

search Profile at Mälardalen University.

Nomenclature
Acronyms
CFB circulating fluidized bed
RDF Refuse derived fuel
Greek Symbols
εm f Void fraction at minimum fluidization
kg
µ Viscosity, ms
φ Sphericity
Figure 5. Primary elemental composition of the bed material
kg
during the study period. ρ density, m3
Roman Symbols
ρ p (ρ p −ρg )gd 3p
5 Conclusions Ar Archimedes number, µ2
kJ
The model presented shows the ability to determine the Cp Specific heat, kgK
agglomeration of bed material while off-line. This has dp Particle diameter, m
been accomplished by building a dynamic model us- m
ing process parameters as input to model and calculate g Gravity, s2
the minimum fluidization velocity required to maintain a kg
Gs Net solid flux, m2s
boiler operating temperature within the range of 750 ◦ C
kJ
and 900 ◦ C. It should be noted that during the seven days H Enthalpy, kg
there were periods where the fluegas temperature dropped h Height of the freeboard, m
below 750 ◦ C and can be attributed to the possibility of
poor fuel. The simulated results show that an increase in m mass, kg
minimum fluidization velocity corresponds to an increase Q Heat released, KW
in bed material share that is greater than the average di- T Temperature, K
ameter of the sand used, 0.40 < d p < 0.63. Since agglom- m
Um f Minimum fluidization velocity, s
eration is prevalent no mater what the fuel source is, it is
m
possible to implement this model as a means for real-time Us Superficial Velocity, s
detection of agglomerate formation as a means for deci- Subscripts
sion support to operators.
g Gas
Compared to other CFB models. The library devel-
oped in this study can potentially be reused for any CFB i Corresponding control volume
installation through the ease of the drag-and-drop nature p Particle
of object-oriented programming. The model presented re f Reference temperature, ambient
has the ability to handle the transient behavior of RDF,
not only for modeling agglomeration, but to also capture References
the dynamics of the temperature fluctuation in the boiler,
hence the possibility for real-time emissions monitoring. S. I. Abu-Eishah, Y. Haddad, A. Soliman, and A. Bajbouj. A
Decision support and fault detection can also be imple- new correlation for the specific heat of metals, metal oxides
and metal fluorides as a function of temperature. Latin Amer-
mented by formulating a probabilistic distribution through
ican Applied Research, 34(OCTOBER 2004):257–265, 2004.
Bayesian nets. Only one model is needed to monitor mul- ISSN 03270793.
tiple aspects of the combustion process in CFBs.
Cyril Jose E. Bajamundi, Pasi Vainikka, Merja Hedman, Jaani
Acknowledgment Silvennoinen, Teemu Heinanen, Raili Taipale, and Jukka
Konttinen. Searching for a robust strategy for minimizing al-
The author would like to thank Linda Svensson, Allmyr kali chlorides in fluidized bed boilers during burning of high
Marianne, and Lisa Granström for their assistance in SRF-energy-share fuel. Fuel, 155(2015):25–36, 2015. ISSN
data acquisition and process information on block 6 at 00162361. doi:10.1016/j.fuel.2015.03.087.
Mälarenergi, Västerås. The authors would also like to Prabir Basu and S A Fraser. Circulating Fluidized Bed Boil-
thank Erik Dahlquist at Mälardalen University for his ers: Design, Operation and Maintenance. Springer Inter-
leadership and guidance. Funding for this work comes national Publishing, Switzerland, doi: 10.10 edition, 2015.
from The PolyPo Project within the Future Energy Re- ISBN 9783319061726. doi:10.1007/978-3319061733.

DOI: 10.3384/ecp17142148 Proceedings of the 9th EUROSIM & the 57th SIMS 153
September 12th-16th, 2016, Oulu, Finland
EUROSIM 2016 & SIMS 2016

Ni-Bin Chang, H.P. Wang, and K.S. Lin. Comparison between O. Madelung, U. Rössler, and M. Schulz. In II-VI and I-
MSW ash and RDF ash from incineration process. Dec 1997. VII Compounds; Semimagnetic Compounds, volume 41B of
Landolt-BÃűrnstein - Group III Condensed Matter. 1999.
Riccardo Chirone, Francesco Miccio, and Fabrizio Scala. Mech- ISBN 978-3-540-64964-9.
anism and prediction of bed agglomeration during fluidized
bed combustion of a biomass fuel: Effect of the reactor scale. J. Nijenhuis, R. Korbee, J. Lensselink, J. H a Kiel, and J. R. van
Chemical Engineering Journal, 123(3):71–80, 2006. ISSN Ommen. A method for agglomeration detection and control
13858947. doi:10.1016/j.cej.2006.07.004. in full-scale biomass fired fluidized beds. Chemical Engi-
neering Science, 62(1-2):644–654, 2007. ISSN 00092509.
Gregory Dunnu, Jörg Maier, and Günter Scheffknecht. doi:10.1016/j.ces.2006.09.041.
Ash fusibility and compositional data of solid re-
covered fuels. Fuel, 89(7):1534–1540, 2010. Anita Pettersson, Fredrik Niklasson, and Farzad Moradian. Re-
doi:10.1016/j.fuel.2009.09.008. duced bed temperature in a commercial waste to energy
boiler - Impact on ash and deposit formation. Fuel Pro-
A. L. Elled, L. E. Åmand, and B. M. Steenari. Composition cessing Technology, 105:28–36, 2013. ISSN 03783820.
of agglomerates in fluidized bed reactors for thermochemical doi:10.1016/j.fuproc.2011.09.001.
conversion of biomass and waste fuels: Experimental data
in comparison with predictions by a thermodynamic equi- Georgy Ryabov, Dmitry Litoun, and Eduard Dik. Agglomera-
librium model. Fuel, 111:696–708, 2013. ISSN 00162361. tion of bed material: Influence on efficiency of biofuel flu-
doi:10.1016/j.fuel.2013.03.018. idized bed boiler. Thermal Science, 7(1):5–16, 2003. ISSN
0354-9836. doi:10.2298/TSCI0301005R.
Bernhard Gatternig. Predicting Agglomeration in Biomass Fired
Fluidized Beds. PhD thesis, 2015. Fabrizio Scala and Riccardo Chirone. Fluidized bed com-
bustion of alternative solid fuels. Experimental Thermal
Guoqing Guan, Chihiro Fushimi, and Atsushi Tsutsumi. Pre- and Fluid Science, 28:691–699, 2004. ISSN 08941777.
diction of flow behavior of the riser in a novel high solids doi:10.1016/j.expthermflusci.2003.12.005.
flux circulating fluidized bed for steam gasification of coal
or biomass. Chemical Engineering Journal, 164(1):221–229, Fabrizio Scala and Riccardo Chirone. An SEM/EDX study of
2010. ISSN 13858947. doi:10.1016/j.cej.2010.08.005. bed agglomerates formed during fluidized bed combustion of
three biomass fuels. Biomass and Bioenergy, 32(3):252–266,
Afsin Gungor. One dimensional numerical simulation of 2008. ISSN 09619534. doi:10.1016/j.biombioe.2007.09.009.
small scale CFB combustors. Energy Conversion and
Management, 50(3):711–722, 2009. ISSN 01968904. Bengt-Johan Skrifvars, Mikko Hupa, Rainer Backman, and
doi:10.1016/j.enconman.2008.10.003. Matti Hiltunen. Sintering mechanisms of FBC ashes.
Fuel, 73(2):171–176, 1994.
Francisco D. Hernandez-Atonal, Changkook Ryu, Vida N.
Sharifi, and Jim Swithenbank. Combustion of refuse- Wolfgang Staber, Sabine Flamme, and Johann Feltner.
derived fuel in a fluidised bed. Chemical Engineer- Methods for determining the biomass content of waste.
ing Science, 62(1-2):627–635, 2007. ISSN 00092509. Waste management & research : the journal of the
doi:10.1016/j.ces.2006.09.025. International Solid Wastes and Public Cleansing Asso-
ciation, ISWA, 26(1):78–87, 2008. ISSN 0734-242X.
Panayiotis Karditsas and Marc-Jean Baptiste. Thermal and doi:10.1177/0734242X07087313.
structural properties of fusion related materials, 1995.
D Vamvuka, D Zografos, and G Alevizos. Control methods for
Daizo Kunii and Octave Levenspiel. Fluidization engineering. mitigating biomass ash-related problems in fluidized beds.
Elsevier, 2013. Bioresource technology, 99(9):3534–44, 2008. ISSN 0960-
8524. doi:10.1016/j.biortech.2007.07.049.
Weigang Lin, Kim Dam-Johansen, and Flemming Frand-
sen. Agglomeration in bio-fuel fired fluidized bed combus- H J M Visser. The influence of Fuel Composition on Agglomera-
tors. Chemical Engineering Journal, 96(1-3):171–185, 2003. tion Behaviour in Fluidised-Bed Combustion. ECN Biomass,
ISSN 13858947. doi:10.1016/j.cej.2003.08.008. (September):44, 2004.

Daniel Lindberg, Rainer Backman, Patrice Chartrand, and Joachim Werther. Fluidized-Bed Reactors. In Ullmann’s Ency-
Mikko Hupa. Towards a comprehensive thermody- clopedia of Industrial Chemistry. 2007. ISBN 3527306730.
namic database for ash-forming elements in biomass and doi:10.1002/14356007.b04_239.pub2.
waste combustion: Current situation and future develop-
L. Wester. Tabeller och diagram för energitekniska beräkningar.
ments. Fuel Processing Technology, 105:129–141, 2013.
L. Wester, 1987.
doi:10.1016/j.fuproc.2011.08.008.
Rong Yan, David Tee Liang, Karin Laursen, Ying Li, Leslie
Zhen-Shu Liu, Tzu-Huan Peng, and Chiou-Liang Lin. Ef- Tsen, and Joo Hwa Tay. Formation of bed agglomeration
fects of bed material size distribution, operating conditions in a fluidized multi-waste incinerator. Fuel, 82(7):843–851,
and agglomeration phenomenon on heavy metal emission 2003. doi:10.1016/S0016-2361(02)00351-4.
in fluidized bed combustion process. Waste management
(New York, N.Y.), 32(3):417–25, 2012. ISSN 1879-2456.
doi:10.1016/j.wasman.2011.10.033.

DOI: 10.3384/ecp17142148 Proceedings of the 9th EUROSIM & the 57th SIMS 154
September 12th-16th, 2016, Oulu, Finland