REPORT 01-03 - Åbo Akademiusers.abo.fi/mzevenho/portfolj/publikationer/PhD MZ.pdf · Process Chemistry Group REPORT 01-03 ... in this thesis was done within the Process Chemistry

ÅBO AKADEMI

KEMISK-TEKNISKA FAKULTETEN Processkemiska forskargruppen

FACULTY OFCHEMICAL ENGINEERING

Process ChemistryGroup

REPORT 01-03

Ash Forming Matter in Biomass Fuels

Maria F.J. Zevenhoven-Onderwater

Academic Dissertation

Combustion and Materials Chemistry Lemminkäinengatan 14-18 B

FIN-20520 Åbo, Finland http://www.abo.fi/instut/pcg/

Åbo Akademi


by

Maria Zevenhoven-Onderwater

Akademisk avhandling som för avläggande av teknisk doktorsexamen vid Åbo Akademiförsvaras vid offentlig disputation .......

Fakultetsopponent:

Avhandlingen försvaras på engelska

Department of Chemical EngineeringProcess Chemistry Group

Biskopsgatan 820500 Åbo, Finland

Åbo Akademi


by

Maria Zevenhoven-Onderwater

Department of Chemical EngineeringProcess Chemistry Group

Biskopsgatan 820500 Åbo, Finland

Preface

I

PREFACE

The work as described in this thesis was done within the Process Chemistry Group at Åbo

Akademi University.

5

I especially want to thank my supervisor, Prof. Mikko Hupa, for the “challenge and

opportunity” to be a member of his group. Bengt-Johan Skrifvars and Rainer Backman are

thanked for their patience and enthusiasm and encouragement to continue and finish this

work. Dr. Flemming Frandsen is gratefully acknowledged for never-ending encouragement.

10

Prof. Brian Scarlett from the University of Technology in Delft is gratefully acknowledged for

teaching me “thinking”. Prof. Kaj Karlsson is warmly thanked for introducing me in the

world of inorganic chemistry. He taught me “to walk on the ice of research” without hurting

myself badly.

15

I gratefully want to thank all my coauthors for their contribution to this work.

Prof. Mikko Hupa is thanked for carefully reading and discussing the manuscripts; Dr. Rainer

Backman is thanked for providing a solid basis for the thermodynamic equilibrium

calculations as used in this work; Dr. Bengt-Johan Skrifvars is thanked for reading and20

discussing manuscripts as well as developing the “star” system for ranking fuels (See paper

II); Dr. Patrik Yrjas is thanked for discussing and commenting paper I; J-P Blomqvist is

thanked to carry out part of the experiments as described in paper II.

My coauthors Prof. Risto Laitinen, from the University of Oulu, is acknowledged for carefully25

reading and commenting paper I; Laura Nuutinen is thanked for the lion part of the analysis

of the chemical fractionation and SEM analysis as described in paper I

My coauthors Dr. Marcus Öhman and Dr. Anders Nordin from the Energy Technology

Center, Piteå, Sweden are thanked for taking care of the lab- scale experiments and SEM30

Preface

II

analysis as described in paper VI;

Truls Liliendahl, Christer Rosén, Dr. Krister Sjöström from The Royal Institute of Technology

(Sweden) and Klas Engvall and Dr. Anders Hallgren from TPS Termiska Prosesser AB

(Sweden)are thanked for taking care of the gasification experiments as described in paper5

V.

Fortum Power and Heat oy is acknowledged for kindly sharing fractionation results.

My special thanks go to Linda Fröberg, Johan Werkelin and Clifford Ekholm who carried10

out almost all of the experimental work considering fractionation and SEM analysis as

described in this thesis.

Dr. Don Hardesty, Prof. Larry Baxter ,currently at Brigham Young University in Utah, and

Gian Sclippa from Sandia National Laboratories, Livermore, California are gratefully15

acknowledged for their hospitality during the winter 1999-2000.

Sonja Enestam and Kristoffer Sandelin are acknowledged for sharing their thoughts about

“ash chemistry at equilibrium”. All remaining group members have contributed in several

ways to the completion of this thesis. Especially my colleagues in “Axelia” have been20

supporting me during these long years.

Further I want to acknowledge my financial supporters, without whom this thesis never could

be finalised. I gratefully thank the Finnish National Combustion and Gasification Research

Programme LIEKKI 2; The CODE programme; The Academy of Finland; the Finnish25

Technology Agency (TEKES); the Värmeforsk Project No B8-803; EU/Joule III “Improved

technologies for the gasification of energy crops” JOR3-CT97-0125 (DG12-WSMN) project;

Foster Wheeler Energia OY, Karhula, Finland; Brista Kraft Sigtuna Energi, Rävsta, Märsta,

Sweden; Skellefteå Kraft AB, Skellefteå, Sweden; Falun Energi, Falun, Sweden; Söderenergi,

Södertälje, Sweden, C-4 Energi, Kristianstad, Sweden; Fortum Power and Heat oy, Vantaa,30

Preface

III

Finland and Kvaerner Pulping oy, Power division, Tampere, Finland.

Finally, I want to thank Pia and Liza, my daughters, for welcome distraction and love and

my parents, although living a long way south, always interested and supporting my life and

work here in the north.5

And last but not least there is Ron......

”ilman sinua elämäni olisi hävinnyt kuin tuhka tuuleen...”

10

Turku, ........2001

15

Maria Zevenhoven

Abstract

IVIV

ABSTRACT

Ash-forming matter in biomass fuels can be present in different ways, i.e. as soluble ions,

organically associated, as included minerals or as excluded minerals. The way ash-forming

matter is present can have consequences for the behaviour of a fuel in a fluidised bed boiler.5

Deposit formation and possible bed agglomeration are dependent on the release of ash-

forming matter from the fuel.

In this thesis the ash-forming matter in biomass fuels has been studied by means of

traditional fuel analysis and an extended fuel characterisation consisting of chemical

fractionation analysis and in some cases SEM/EDX analysis.

The extended fuel characterisation was carried out for 16 fuels. Clear differences have been

shown in the distribution of ash-forming elements in the different fuels. In the older fuels

more ash-forming elements were present as included or excluded minerals. In relatively15

young fuels up to half of the ash-forming elements was organically associated or present as

easily soluble salts or as included minerals.

The results of the extended fuel analyses were used to predict fuel behaviour in fluidised bed

combustion and gasification. Deposit formation tendencies were predicted for 23 fuels using

the combination of chemical fractionation with Thermodynamic multi-Phase multi-

Component Equilibrium(TPCE) calculations. This work shows that easily leached elements

form the main constituents in the fine fly ash, and are consequently reasonable

approximations of the fly ash compounds. The ranking of the fuels as studied in this work

was presented as less-problematic < problematic: coal< peat < wood derived fuels <25

annual crops < agricultural waste. This corresponds well to the general practical experiences

with these fuels.

TPCE calculations as presented in this work were used as guidelines for predicting bed

agglomeration. Modelling showed that the presence of an excess of dolomite/calcite

Abstract

VV

decreases the amount of alkali components in the bed due to an increase in the amount

volatilised. An excess amount of silicates increased the amount of alkali retained in the bed,

forming low melting alkali silicates, leading to bed agglomeration. At atmospheric pressure

the amount of melt formed could be smaller, when compared to high pressures, indicating

a lower risk for bed agglomeration.5

Chemical fractionation results revealed that when firing woody biomass fuels potassium and

calcium present in a bed coating are originating from the reactive fraction in the fuel, i.e.

leachable with water or ammonium acetate and/ or present as included small minerals as

pointed out by SEM/EDX analysis.

Table of contents

VI

TABLE OF CONTENTS

1 Introduction 1-11.1 The use of biomass in heat and power generation 1-11.2 Producing heat and power from biomass fuels 1-21.3 Ash-related operational problems 1-31.4 The objective of this work 1-3

2. Literature review 2-12.1 Fuel characterisation 2-12.1.1 Differences in solid fuels 2-12.1.2 Traditional fuel analysis 2-32.1.3 Chemical fractionation 2-52.2 Fluidised bed reactors 2-82.2.1 Combustion in atmospheric systems 2-92.2.2 Combustion in pressurised systems 2-112.2.3 Gasification 2-122.3 Ash behaviour in fluidised beds 2-132.3.1 Conversion of fuel particles 2-132.3.2 Deposit formation 2-172.3.3 Deposit formation: Field and laboratory experiences 2-242.3.4 Bed agglomeration: The mechanism 2-292.3.5 Bed agglomeration: Field and laboratory experiences 2-31

3 Results 3-13.1 Biomass fuel characterisation 3-13.1.1 Traditional fuel analysis 3-13.1.2 Chemical fractionation 3-23.1.3 SEM/EDX analysis of biomass fuels 3-63.2 Ash behaviour in (P)FBG and FBC 3-83.2.1 Prediction of ash deposit formation 3-83.2.2 Prediction of bed agglomeration 3-16

4 Conclusions 4-14.1 Fuel characterisation 4-14.2 Deposit formation 4-14.3 Agglomeration 4-3

5 References 5-1

Table of contents

VII

PUBLICATIONS

I. “Searching for improved characterization of ash forming matter in biomass”, MariaZevenhoven, Bengt-Johan Skrifvars, Patrik Yrjas , Mikko Hupa,Laura Nuutinen,RistoLaitinen, (Paper 73) accepted for presentation at the 16th International Conferenceon Fluidised Bed Combustion, May 2001, Reno, Nevada, USA

II. “The prediction of behaviour of ashes from five different solid fuels in fluidised bedcombustion, Maria Zevenhoven, Jan-Peter Blomqvist, Bengt-Johan Skrifvars, RainerBackman, Mikko Hupa, Fuel, 79(9), (2000), 1353-1361

III. “Predicting the ash behaviour of different fuels in fluidised bed combustion”, Bengt-Johan Skrifvars, Maria Zevenhoven, Rainer Backman, Mikko Hupa,(Paper 113)accepted for presentation at the 16th International Conference on Fluidised BedCombustion, May 2001, Reno, Nevada, USA

IV. “The ash Chemistry in Fluidised bed gasification of biomass fuels: Part I- Predictingthe chemistry of melting ashes and ash-bed material interaction”, Maria Zevenhoven,Rainer Backman, Bengt-Johan Skrifvars, Mikko Hupa, Fuel in press

V. “The ash Chemistry in Fluidised bed gasification of biomass fuels: Part III-Ashbehaviour prediction versus bench scale agglomeration tests”, Maria Zevenhoven,Rainer Backman, Bengt-Johan Skrifvars, Mikko Hupa, Truls Liliendahl, ChristerRosén, Krister Sjöström, Klas Engvall and Anders Hallgren, Fuel in press

VI. “Effect of fuel quality on the bed agglomeration tendency in a biomass fired fluidisedbed boiler”, Maria Zevenhoven, Marcus Öhman, Bengt-Johan Skrifvars, RainerBackman, Anders Nordin, Mikko Hupa, to be submitted

VII. “The prediction of deposit formation in combustion and gasification of biomassfuels”, Maria Zevenhoven, report 01-02, Åbo Akademi University, in press

Chapter 1: Introduction

1-1

1 INTRODUCTION

1.1 The use of biomass in heat and power generation

Biomass is the ecological term for organic material, both above and below ground and both5

living and dead, such as trees, crops, grasses, tree litter and roots. The types of biomass used

in power generation production include energy crops, agricultural and agro-industrial wastes,

sewage sludges, municipal solid wastes, black liquor and peat. This diversity and ready

availability make biomass a strong alternative to fossil fuels for future energy requirements

around the world.10

Efforts to develop ways of producing and using biomass resources for heat and power

generation are currently supported by various national and international stimulation

programmes. Interests in bioenergy vary from country to country, but can be generalised for

developed and developing countries. Governments of developed countries are searching for15

ways to reduce both the emissions (especially CO2) produced by combustion of traditional

fuels and the amount of municipal solid waste, sewage sludge, etc. requiring disposal.

Developing countries face pressures to build energy systems that supply heat and power to

rural areas. There is no shortage of biomass on a global scale. The total energy content of

biomass reserves equals the proven oil, coal and gas reserves combined; However, most20

of this biomass energy is held in trees and the biomass source is regenerated slower than as

it is harvested, thus depleting the resource. (Kendall et al.,1997)

The European Union aims at increasing the use of bioenergy to 90 million toe (tons of oil

equivalent) by the year 2010 (EU15).15 million toe should come from the exploitation of25

biogas which is emitted from landfills to the atmosphere today. 30 million toe should be

retrieved from agricultural and forest residues. The last 45 million toe should come from

cultivated energy crops such as Salix (willow), Eucalyptus and Miscanthus. (Communication

from the Commission, 1997)

30

The use of bioenergy in Finland has increased by more than 60% since 1980 and was


1-2

almost 7 million toe in 1996 . The use of peat is 2,1 million toe and the rest consists mainly

of wood and wood-based fuels. The share of bioenergy in the primary energy consumption

is more than 20%. The target of Finnish energy policy is to increase the bioenergy use by

25% by the year 2005.This increases to 1.5 million toe. The target for the year 2010 is to

increase with 3.5 million toe from the 1995 level. The interest in growing energy crops such5

as Reed Canary Grass, Turnip rape and Willow, or methanol and ethanol production from

barley and wheat in Finland is very low due to the abundant resource of woody biomass

and peat for producing heat and power. The potential of those two is far much higher than

that of energy crops (Communication from the Commission, 1997).

10

1.2 Producing heat and power from biomass fuels

The energy conversion technologies that are interesting from the energy crop’s point of view

are production of pyrolysis oils and combustion or gasification. Combustion means direct

burning in presence of oxygen to produce heat. Gasification means conversion under sub-15

stoichiometric conditions producing combustible gases. These can, for example, be burned

in a gas turbine producing heat and electricity. In power generation, the heat produced is

generally applied in boilers which produce steam to drive turbines. For coal and biomass,

combustion technology is well understood, fully commercial and widely used for district

heating and power production. Gasification technology for biomass fuels is in a more20

experimental stage.

The performance of bioenergy technologies depends on local circumstances such as power

production requirements, availability of fuel and delivery costs, as well as on the chemical

and physical characteristics of the fuel (Faaij, 1997). Research on the reduction of25

atmospheric emissions has come a long way in recent years. Although emissions are already

substantially lower than with fossil fuels, efforts to reduce them further have been successful.

Assuming complete regeneration, biomass is a more CO2-neutral fuel when compared to

coal, only releasing what it has taken up during growth when fired.

30


1-3

1.3 Ash-related operational problems

Conversion of biomass fuels in fluidised bed reactors (FB) seems a most promising way to

produce electricity due to their high fuel flexibility. The relatively long residence times and

good mixing can ensure high conversion efficiencies (Clean Coal Technologies, 1993).5

Technical inefficiencies, which are often ash-related and pollution-related still exist today.

Ash-related problems in fluidised bed combustion (FBC) and fluidised bed gasification

(FBG) could lead to deposit formation and defluidisation. Both types of problems depend

on geometry, process conditions and ash chemistry. These are often reasons for

unscheduled shut-down. The ash chemistry is fuel-related, whereas the other factors will10

depend strongly on boiler design and operation. Fluidised beds are useful due to the good

mixing and relative low process temperatures preventing many of the ash-related problems

that might occur in other furnace types. However, ash-related problems during combustion

or gasification of biomass fuels can still lead to operational problems.

15

Much research has been carried out on FBC and FBG failure due to ash-related problems.

The ASTM /DIN standard ash fusion test is often employed for predicting bed agglomeration

(DIN 51730, 1984). It has, however, been reported to be a poor indicator of ash related

operational problems with biomasses and energy crops. Another predictive method is the

ash pellet compression strength measurement as used, among others by Skrifvars (1994).20

Many FB test rig and pilot plant tests are carried out both under reducing and oxidising

conditions. Test rig experiments are time-consuming.

1.4 The objectives of this work

25

Ash-forming matter in biomass fuels can be present in different ways, i.e. easily soluble salts,

organically associated compounds, as included minerals or as excluded minerals. The way

ash-forming matter is present can have consequences for the behaviour of a fuel in a

fluidised bed boiler. Deposit formation and possible bed agglomeration are dependent on

the release of ash-forming matter from the fuel.30


1-4

In this thesis the ash-forming matter in biomass fuels has been studied by means of

traditional fuel analysis and an extended fuel characterisation consisting of chemical

fractionation analysis and in some cases SEM/EDX analysis.

The extended fuel characterisation of 16 fuels is described. The results of the extended fuel5

analyses were used to predict the behaviour of 23 fuels in fluidised bed combustion and

gasification. Deposit formation tendencies and agglomeration are predicted for 30 fuels

using the combination of chemical fractionation with Thermodynamic multi-Phase multi-

Component Equilibrium(TPCE) calculations.

10

This thesis consists of a short literature review on biomass fuel characterisation, ash deposit

formation and bed agglomeration in fluidised beds (Chapter 2); a summary of the results

obtained from studying 30 biomass fuels (Chapter 3); conclusions which could be drawn

from the work described in chapter 3 (Chapter 4); references (Chapter 5); six papers and

one report (referred to as I-VII)15

Paper I describes the use of the chemical fractionation procedure in the characterisation of

biomass fuels. SEM/EDX analyses of the untreated biomass fuels, leached fuels and

laboratory ashed fuels complete the fuel characterisation.

20

In paper II and III the combination of chemical fraction and TPCE calculations for the

prediction of deposit formation is described and demonstrated for several biomass fuels.

Predictions are compared to bench-scale and full-scale measurements.

Papers IV and V describe the use of TPCE calculations for prediction of bed agglomeration25

in biomass fired fluidised bed gasification; a comparison between 1) the thermodynamically

modelled chemical behaviour of four biomass fuels in pressurized and non-pressurised

gasification experiments as carried out in two test rigs and 2) the thermodynamic modelled

chemical behaviour and SEM/EDX analysis from bed material retrieved from these test rigs.

The small-scale experiments were carried out in the pressurized FBG at the Royal Institute30

of Technology and in an atmospheric FBG of Termiska Processer AB (TPS) both in Sweden.


1-5

In Paper VI chemical fractionation of biomass fuels is used to explain bed agglomeration in

fluidised bed combustion. SEM/EDX analyses of bed samples retrieved from bench scale

experiments are used to determine which kind of ash forming elements are responsible for

the formation of agglomerates. The bench scale experiments were carried out at ETC in

Umeå, Sweden.5

In Report VII the approach as described in Papers II through VI is used to predict ash

deposition and agglomeration of biomass fuels studied at Sandia National Laboratories,

Livermore, California. Chemical fractionation results for these fuels were described by Miles

et al. (1995a through c). This report summarises the work as carried out during winter10

1999-2000 at Sandia National Laboratories, Livermore, California.

Chapter 2: Literature review

2-1

2 LITERATURE REVIEW

2.1 Fuel characterisation

2.1.1 Differences in solid fuels5

Naturally occurring solid fuels include fuels such as biomass, peat, lignite, bituminous coal

and anthracite coal. In addition to carbon and hydrogen constituents solid fuels contain

significant amounts of oxygen, water, ash-forming elements, nitrogen and sulphur. The

oxygen is chemically bound in the fuel and varies from 45 wt% for wood to 2 wt% for10

anthracite coal on a dry ash-free basis. Moisture can exist in two forms in the fuel- as free

water between cell walls in wood or in the larger pores of low grade coal or as bound water

held by physical adsorption. Green wood typically consists of 50 wt% water. Lignite contains

between 20-40 wt% moisture most of which as free water. Bituminous coals contain about

5 wt% moisture as bound water (Borman and Garland 1998).15

Ash is the inorganic residue remaining after the fuel is completely burned. Wood usually has

only a few tenths percentage of ash, while coal typically contains 10 wt% or more ash. Ash

characteristics play an important role in boiler design in order to minimise deposit formation,

erosion and corrosion and defluidisation.20

Coal has a structure and composition which differ widely from that from biomass fuels.

However in both coal and biomass ash-forming matter can be present in four general forms

(see Figure 2.1):

1) As easily leachable salts25

2) As inorganic elements associated with the organic matter of the biomass. This is

defined as organic associated matter in this thesis.

3) As minerals, included in the fuel structure, so-called included minerals.

4) As inorganic material, typically sand or salt or clay from harvesting the biomass fuel

or deposited in plant debris as discrete particles of foreign material. According to30


2-2

Figure 2.1: Schematic of the different forms of ash-forming matter

in coal and biomass

Reid (1984) inorganic material

in coal may also have been

deposited later by mineral-

laden water percolating through

the coal seam. This ash-5

forming matter is called

excluded minerals.

In biomass the included and

organically associated ash-10

forming matter makes up for

the major part of the total ash-

forming matter

Bryers (1996) summarised the15

occurrence of ash-forming

elements in biomass in a review

article. The two principal forms

of sulphur in plants are

sulphates or organic sulphur.20

The former increases with

increasing sulphate in the nutrient supply. Chlorine in biomass appears as a chloride ion. Its

concentration is closely related to the nutrient composition of the soil. Phosphorous exists

in its most oxidised form in biomass fuels and is not reduced during plant’s metabolism. It

is primarily introduced as H3PO4 and either remains in the inorganic form or is incorporated25

in organic structures by forming esters or pyrophosphates. Silicon is introduced in the plant

by absorption of silicic acid from the soil. Silicon is deposited as a hydrated oxide usually

in an amorphous form, but occasionally in crystalline forms. Potassium is the next most

important element in the plants. Potassium occurs as ion that is highly mobile with little

structural function. Potassium uptake is highly selective and correlates with the plants30


2-3

Figure 2.2: Schematic of traditional fuel analysis after Hannes,

(1996)

metabolic activity. Thus, potassium is often found in regions where plant growth takes place.

2.1.2 Traditional fuel analysis

Since solid fuels are not homogeneous, an elementary description is not possible. Due to the5

different ages and origins of many fuels their composition will vary widely. Standard testing

and analysis of coal are prescribed by ASTM standards or DIN standards. The proximate

analysis, ultimate analysis and heating value will be described below shortly. (See Figure

2.2)

1) With the proximate analysis, the amount of moisture, mineral residues (ash), volatile10

matter and fixed carbon are determined (ASTM D3172-89, DIN 51718, DIN

51719). Although the determination is done quite accurately, the name “proximate”

indicates the empirical nature of the method; a change in procedure can change the

results. A sample of coal (or biomass fuel) is crushed and dried in an oven at 105 to

110/C to constant weight to determine residual moisture. The sample is then heated15

in a covered crucible (to prevent oxidation) at 900/C to constant weight. The weight

loss is referred to as volatile matter. The remaining sample is then placed in an oven

at 750/C with the cover off so that the sample is combusted. The weight loss upon

combustion is termed fixed carbon or char. The remaining residue is ash. The

components of a proximate20

analysis are rather arbitrary.

There is no sharp distinction

between free water and water

chemically bond to the fuel. The

split between volatile matter and25

fixed carbon depends on the rate

of heating as well as the final

temperature. Some of the ash

can be volatilised during the char

determination. Nevertheless, the30


2-4

Figure 2.3: Typical volatile matter, fixed carbon and higher heating

value for wood, peat and coal on a moisture and ash free basis (after

Borman and Garland, 1998)

proximate analysis provides a useful comparison between fuels. The proximate

analysis for biomass is limited to 600/C (Borman and Garland, 1998).

2) The ultimate analysis (ASTM D 3176) does not distinguish between the origin of an

element coming from fixed carbon or volatile matter. The analysis determines the

content of the elements, carbon, hydrogen, oxygen nitrogen and sulphur. The5

analysis provides the major elemental composition of the fuel, usually reported on

a dry ash-free basis. Carbon and hydrogen are determined by burning the sample

in oxygen in a closed system and quantitatively analysing the combustion products.

The carbon includes organic carbon as well as carbon from the mineral carbonates.

The hydrogen includes the organic hydrogen as well as any hydrogen from the10

moisture of the dried sample and mineral hydrates. The extraneous carbon and

hydrogen are usually negligible. Nitrogen and sulphur are determined chemically.

Oxygen is usually determined as the difference between 100 and the sum of the

percentages C, H, N and S. Sometimes chlorine is included in the ultimate analysis.

3) The ash analysis is mainly used to get a better inside in ash related problems15

Analysis takes place after ashing of the fuel sample. Today a lower ashing

temperature is used when analysing biomass when compared to coal in an attempt

t o a v o i d u n w a n t e d

volat i l i sat ion of alkali

components and subsequently20

an underestimation of the

alkali amount in the fuel

(ASTM E1755-95). Elemental

analysis of the ash-forming

matter often is expressed as25

oxides (DIN 51719-A, ASTM

D 271-68, Borman and

Garland, 1998)).

30


2-5

Table 2.1: Average fuel composition of wood, husk/shells, grass/plant, coal and Finnish peat

(after Phyllis, 2000) (daf = dry, ash free)

Figure 2.3 and Table 2.1 clearly show differences between the composition of fuels. Figure

2.3 shows a typical volatile matter, fixed carbon, and heating value for wood, peat and coal

on a moisture and ash free basis. Table 2.1 shows the composition of different groups of

biomass fuels, peat and coal as taken from ECN’s database Phyllis (2000)

5

2.1.3 Chemical fractionation

The chemical fractionation is a method based on selective leaching by water, ammonium

acetate and hydrochloric acid. The method was originally developed by Benson et al. (1985)

for the characterisation of coal. Baxter (1994) modified the method for the characterisation

of biomass fuels. The method can be used to distinguish how ash-forming elements are10

bound in the fuel. A simplified scheme is given in Figure 2.4, which shows that the chemical


2-6

Figure 2.4: A schematic of the fractionation procedure as

developed by Benson et al., (1985) and Baxter (1994)

f r a c t i ona t i o n t e c h n i q ue

distinguishes different types of

ash-forming matter in the fuel

according to their solubility in

different solvents. Increasingly5

aggressive solvents, i.e. water

(H2O), 1M ammonium acetate

(NH4Ac) and 1M hydrochloric

acid (HCl) leach samples into a

series of four fractions (including10

the unleached rest) for analysis.

T y p i c a l a s h - f o r m i n g

components, which are leached out by water are alkali sulfates, carbonates and chlorides.

Elements leached out by NH4Ac are believed to be organically associated, such as Mg, Ca15

as well as K and Na. HCl leaches the carbonates and -sulfates of alkaline-earth and other

metals. Silicates and other minerals remain in the insoluble rest.

The amounts of leaching agents as suggested by Baxter (1994) and Benson et al. (1985)

cannot be used in case of leaching dry biomass fuels. It is difficult to achieve proper mixing20

for most fuels. Instead, an excess of water should be used in the first step, firstly to achieve

proper wetting of the samples, secondly to achieve proper leaching. These problems are not

encountered in the other leaching steps, since the solid samples here are already thoroughly

wet from previous steps and extensive washing (See I). Miles et al. (1995a through c)studied

the chemical fractionation technique to characterise seven fuels and their inorganic25

composition. These fuels were almond hulls, almond shells, olive pits, paper, rice straw,

switch grass and wheat straw (See also 3.1 and VII).


2-7

c)

g)

Figure 2.5: Fractionation results as retrieved by

Miles et al. (1995 a through c)

a) almo nd hulls white = water leached

b) almo nd shells striped = acetate leached

c) olive pits grey = hy droch loric

d) paper acid leached

e) rice straw black = residue

f) switch grass

g) wheat straw

e)

a) b)

f)

d)


2-8

Figure 2.6: Schematic of different stages of fluidisation

Figure 2.5 a through g show the distribution of ash-forming elements over the different

leaching agents. In all fuels the major part of the refractory materials, Si, Al, Fe, Ti was found

in the residual fraction only a minor part was leached by the other leaching agents. The

silica content of the grasses was high. It plays an important role in the sturdiness of the

plants. These elements are believed to play only a minor role in photosynthesis and are5

supposedly present as inorganic granules. According to Miles et al (1995a through c) the

alkali and alkaline earth materials (K, Na, Ca and Mg) occur in organic structures or very

mobile inorganic components. The major part of the alkali materials was found in the water

soluble fraction. Miles states calcium is present in the cell wall and will be organically

bonded as easily ion exchangeable material, thus found in the acetate fraction. Potassium10

and sodium were also found in the residual fraction as components of mineral soil

contamination of the fuels. Non-metallic materials (S, P, Cl) occur as plant nutrients. The

fractionation results showed that all chlorine could be present in an easily soluble form.

2.2 Fluidised bed reactors (FB)15

When a gas is passed upwards through a bed of particles, the degree of disturbance is

determined by the velocity of the gas. At low velocities there is only little particle movement.

As the velocity increases, individual particles begin to be forced upwards until they reach the

point at which they remain suspended in the gas stream (the minimum fluidisation velocity).20

Any further increase in gas velocity causes turbulence, with rapid mixing of the particles. A

particle bed in this state can be

described as ”fluidised”. (See

Figure 2.6)

25

Fluidised bed combustion or

gasification of coal or biomass

fuels uses a constant stream of

air in super- or sub-

stochiometric ratios, i.e.30


2-9

combustion or gasification conditions, which creates the necessary turbulence. The bed

particles are initially heated by a start-up fuel (mostly natural gas) after which the solid fuel

is fed to the bed continuously. The fuel ignites and releases heat allowing for the start-up fuel

to be shut down. The mixing of the particles encourages complete combustion or gasification

and allows a constant temperature to be maintained in the conversion zone. Part of the ash5

accumulates in the bed. In case of coal these ash particles together with sorbent material for

sulphur capture will form the bulk of the bed particles. In case of biomass fuels, which

contain much less ash the bed material consists mostly of sand. Ash from fuel conversion

and bed material are drawn from the bed at regular intervals and replaced when necessary

to maintain the bed at a correct level and to maintain bed properties. Advantages of fluidised10

bed conversion are:

• The bed can be operated at temperatures below 900°C. This temperature is low

enough to prevent defluidisation of the bed and unwanted shut-down of the furnace.

• The scrubbing action of moving particles on immersed water tubes increases the rate

of heat transfer.15

• The bed has a substantial thermal capacity. This allows a variety of fuels to be

burned, including those with a high moisture content.

• Adding crushed limestone ensures that more than 90% of any sulphur dioxide

released during combustion of sulphur bearing fuels, such as coal, peat and lignite,

can be retained in the bed.20

2.2.1 Combustion in atmospheric systems

From the two alternatives, FB combustion (FBC) and FB gasification (FBG), FBC is the most

developed and commercially available. Some hundred either bubbling (BFBC) or circulating25

FBC’s (CFBC) operate in the 50-300MWth size range worldwide.


2-10

Figure 2.7: BFBC process and residue sources (after Clarke,

1992)

Figure 2.8: CFBC process and residue sources (after Clarke,

1992)

The present status of BFBC in

the world has been reviewed by

Salmenoja (2000). The BFBC

(see Figure 2.7) is the oldest

application and is offered as a5

standard option by many

manufacturers for biomass. In

Finland, the first BFBC’s were

delivered to the Finnish pulp and

paper industry and were rather small in size, ranging from 15 to 50 MWth. Today the largest10

units are close to 300 MWth. The world’s largest BFBC for biofuels will have a capacity of

267 Mwth and will be built at Pietersaari, Finland. One important characteristic of the

bubbling bed is that it retains a more or less defined surface bed level. The velocity of the

primary air introduced at the bottom of the combustion chamber is limited to 1-2 m/s to

prevent excessive turbulence and loss of bed material due to bed entrainment. A typical bed15

temperature in BFBC boilers is in the range of 750-900°C. A major advantage of BFBC’s

is that a variety of fuels, many of them low-grade, can be used in a standard boiler.

However, the feed systems must be appropriate for the individual fuels. BFBC’s can be used

in both new and retrofit applications (Salmenoja, 2000).

20

The bed in a CFBC (see Figure 2.8) has no clearly definable bed surface. The fuel is injected

near the base of the combustion chamber. Together with the existing bed material, the fuel

is converted by the velocity of the

fluidising gas (typically around 8

m/s) into a turbulent cloud of25

solids which fills the whole

primary combustor. The most

dense particles remain near the

base of the furnace. The finer

particles occupy the upper part of30


2-11

Figure 2.9: PFBC process and residue sources (after Clarke,

1992)

the furnace and are entrained by the gas stream. Cyclones recover the major part of these

particles and return them to the bed. This ensures an effective high residence time of fuel

particles in the system, ensuring complete burn-out. The finest particles are not recirculated.

They escape through the cyclone and are extracted from the gas stream in a filter system

just before the gases are emitted into the atmosphere. CFBC’s for biomass such as peat,5

sludge, wood waste and coal have been built up to a range of 250 MWe. A basic difference

between BFB and CFB is the heat transfer from the bed material. In a bubbling bed most

of the heat released from the fuel to the bed material is kept in the lower part of the furnace,

while in CFB the heat is released from the circulating bed material to the furnace water

walls. This means that CFB’s can operate nearly isothermal, mostly at some 870°C10

(Salmenoja, 2000).

2.2.2 Combustion in pressurised systems

While pressurised FBC systems are already commercially applied for the conversion of coal15

and lignite, systems for biomass are still under development. Operating temperatures are

similar to those maintained in an atmospheric bubbling bed and the pressurised FBC

(PFBC) could be used for a wide range of fuels (See Figure 2.9).

Pressurised systems offer two distinct advantages compared with atmospheric systems:20

• The potential for

c o m b i n e d c y c l e

application

• A smaller unit size;

however, since the heat25

transfer tubes are

immersed within the bed,

the degree of size

reduc t i on w i l l be

determined by the total30


2-12

Figure 2.10: IGCC process showing residue sources (after

Clarke, 1992)

heat transfer requirements of the system.

PFBC is mainly demonstrated in coal fired power plants. Some coal-biomass mixtures have

been tested in a 1 MWth test facility in Sweden (Anderson et al., 1999).

5

2.2.3. Gasification

Nowadays, for gasification the

most important FBG concepts,

that are demonstrated in full scale10

are the airblown FBG’s, i.e. for

example an atmospheric pressure

FBG (Rensfelt, 1997), or an

atmospheric pressure circulating

FBG as demonstrated in Lahti,15

Finland. This is a gasifier (40

MWth) firing low calorific waste

derived and biofuels such as recycled fuel, peat, demolition wood waste and shredded tires

with moisture contents up to 60 wt%. The raw gas produced is directly fired in an existing

pulverised coal-fired boiler. With the gasification of biofuels up to 30% of the coal fed into20

the main boiler is replaced (Nieminen et al., 1999). A pressurised BFBG has been

demonstrated for biofuels (Salo et al., 1998). A pressurised CFBG was demonstrated in

Värnämo (Stahl et al., 1996). Although higher efficiencies and lower emissions have been

achieved than in conventional technologies, biomass integrated gasification combine cycle

(IGCC) concepts cannot compete at the present with natural gas fired combined cycles and25

low cost conventional CFBC. Furthermore a secured fuel supply for a large scale biomass

IGCC over its lifetime is questionable (Engström, 1999). Figure 2.10 shows a schematic of

an IGCC process. The gasifier product gases are burned in a gas turbine, producing

electricity. Excess heat is recovered in a steam cycle producing electricity with a classical

steam turbine.30


2-13

Figure 2.11: Schematic of ash-particle formation (partly after Flagan 1984)

2.3 Ash behaviour in fluidised beds

2.3.1 Conversion of fuel particles

When a fuel is burned in a fluidised bed, the ash-forming elements that are released from5

the fuel undergo different reaction paths. After entering the furnace fuel particles will heat

up rapidly and dry at first. After this the pyrolysis will start, i.e. organic volatile species will

be released from the fuel and the fuel particles will burn with a visible flame. During this

stage some reactive ash-forming elements will be released together with the gases. After the

volatile species left the particle char burning will start. A schematic of the fate of ash-forming10

matter in coal is shown in Figure 2.11. Most of the ash-forming elements will end up in the

residual ash in the case of burning coal by the coalescence of the fused included minerals

in coal during char burnout. The particle size of the residual ash depends on many factors.

A single fuel particle may fragment during combustion and each fragment may produce an

15


2-14

ash particle. The size of the ash particle will depend on the initial fuel diameter, its mineral

content, the uniformity of its distribution, and the number and size of fragments produced

during combustion. The fragmentation of a fuel particle is only partly understood. As

reaction occurs in depth within the porous char, a point is reached where the pores merge

and undermine segments of the partially burned char which are then released as fragments.5

(Flagan et al.1984) This description of the formation of ash is also valid for biomass fuels.

As described above, ash-forming matter in solid fuels can be divided in four fractions, three

that are inherent in the fuel (included minerals, organically associated, easily water soluble)

and one added to the fuel through geologic or processing steps (excluded minerals).10

According to Baxter (1993) a large fraction of the inherent ash-forming matter in lignites and

probably the dominant part in biomass fuels is associated with oxygen containing functional

groups. These functional groups provide sites for ash-forming matter to become

incorporated in the fuel matrix as, for example, chelates and cations. The release of this

atomically dispersed material from a fuel particle is influenced both by its volatility and the15

reactions of the organic portions of the fuel. Material that is volatile at combustion

temperatures includes derivates of the alkali and alkaline-earth metals, most probably Na

and K. Other non-volatile material can be released by convective transport during rapid

pyrolysis. The amount of fuel lost during the pyrolysis stage of combustion increases with

increasing hydrogen-to-carbon ratio, and to a lesser extend, with increasing oxygen-to-20

carbon ratio. Lignite, peat and biomass can lose more than 90% of their mass at this first

stage of combustion. Typically, the loss of volatiles during pyrolysis of biomass is some

75%. The large quantities of tars leaving the fuel can convectively carry ash-forming matter

out of the fuel, even if the inorganic material itself is non-volatile (Baxter, 1993).

25

The other class of ash-forming matter in solid fuels includes material that is added to the fuel

from extrageneous sources, the excluded minerals. In the case of coal, geological processes

and mining techniques contribute much of this material. In the case of biomass, fuel

processing in the field is likely to contribute the majority of it. This material is often

particulate by nature. Components of the minerals may be released from the fuel by either30


2-15

Figure 2.12: Fly ash particle mass size distribution downstream of

the process cyclone at flue gas temperatures 810-850/C (After

Valmari et al., 1999a)

thermal decomposition or vaporisation (Baxter, 1993).

A typical ash size distribution at the outlet of a fluidised bed boiler is bimodal. The vast

majority of the mineral inclusion, fuel contamination and entrained bed material are found

in the large particle mode of the size distribution. The smaller particle mode represents5

particles in sizes between 0.01 and 0.2 :m. These particles are smaller than can be

explained from the fragmentation of the fuel particles during conversion. The source of this

fume is volatilised ash-forming matter which nucleates homogeneously as the vapours

diffuse from the hot reducing atmosphere near the surface of burning char particles into the

cooler oxidising atmosphere (in case of combustion). The nuclei are initially very small but10

grow rapidly by condensation of additional ash-forming vapours and by coagulation (Flagan

et al., 1984).

Lind and Valmari studied the behaviour of the most important ash-forming elements firing

willow and forest residue in a CFBC (Lind 1999, Lind et al. 1999a, Valmari et al. 1998,15

1999a and b, Valmari 2000). During combustion of forest residue 30-40% of the ash was

retained in the bed, attached to quartz sand bed particles. Ca and P were believed to be

retained via particle collisions. K was believed to be retained due to a reaction between the

vapour phase potassium and the

sand bed particle. The fly ash20

size distributions consisted of two

distinct modes (see Figure 2.12)

The coarse mode contained

more than 90% of the mass in

both cases. Fine mode particles25

contributed 2% of the total fly

ash mass with forest residue and

8% with willow. When firing

forest residue potassium and

sodium were mainly present in a30


2-16

water insoluble form in de fly ash, indicating presence as silicates. Sulphur and chlorine were

volatilised. During combustion of a forest residue the sulphur had reacted to CaSO4. When

combusting willow 50% of the fuel sulphur remained as SO2. The particle sulphur was

found in the fine particle mode as alkali sulphates. These were not detected when firing a

forest residue. Half of the alkali chlorides was found to condense on coarser particles,5

whereas the other half condensed in the fine particle mode (See Figure 2.13).

Figure 2.13: Elemental mass size distributions downstream of the cyclone in CFBC (after Valm ari et al.,

1999a)


2-17

Figure 2.14: Typical slagging and fouling areas

as found in a boiler

2.3.2 Deposit formation

The mechanism

Deposit formation in a boiler can be divided

in slagging and fouling. Bryers (1996) defined5

slagging as deposition of fly ash on heat

transfer surface and refractory material in the

furnace volume primarily subjected to radiant

heat transfer. Fouling is defined as deposition

in the heat-recovery section of the boiler (See10

Figure 2.14).

Both the ash deposition rate and the

properties of the ash deposits are important

considerations in the operation in a15

combustor. The properties of ash deposits

most important to the successful operation of

a boiler include (see Baxter 1993):

1) The ease of removal from a combustor wall or heat exchanger surface

2) Viscosity20

3) Effective thermal conductivity

4) Effective emissivity

5) Deposit strength.

Additional deposit properties are:25

6) Elemental composition

7) Morphology

8) Porosity

9) Chemical species composition (Baxter 1993)


2-18

The formation of a hard deposit can be described by four most relevant steps:

1) Formation of an ash particle

2) Transport of the ash particle or ash-forming compound to a surface

3) Adhesion to the surface

4) Consolidation of the deposit5

After formation of an ash particle as described in 2.2.1 it is transported to a heat transfer or

boiler surface before deposition can take place. Typical transport processes are diffusion,

thermophoresis and inertial impaction. Diffusion and thermophoresis are processes of

particle transport in a gas due to local concentration and temperature gradients, respectively.10

In case of Fick diffusion molecules will move to a surface due to a concentration gradient

present. Brownian diffusion describes the random movement of small particles. Eddy

diffusion describes the diffusion in turbulent systems. In case of thermophoresis a particle

suspended in a fluid with a strong temperature gradient interacts with molecules that have

higher average kinetic energies on the side with the hot fluid than on the side with the cold15

fluid. The collisions of the high energy molecules on the hot side of the particle have more

impact than those on the colder side. This gives rise to a net force on the particle. In general,

these forces act in the direction opposite to that of the temperature gradient.

Inertial impaction is usually the process by which the bulk of the ash deposit is transported20

to a heat exchanger surface. The rate of inertial impaction depends on target geometry,

particle size distribution and density and gas flow properties. Inertial impaction is important

for large particles (10:m and larger).

After an ash particle hits the surface it may adhere to it. Adhesion can take place through25

glueing to the surface. This is possible when a partly molten phase is present that can act as

glue between a particle and a surface. When a gaseous ash-forming compound is formed,

this could diffuse to the surface and condensate directly.

30


2-19

Figure 2.15: A sche matic v iew of stic kyne ss criteria

The stickiness of ash particles is

strongly dependent on

temperature and physical state.

It has been shown that the

presence of a melt in an ash5

particle acts as a sticking agent

for the particle (Backman et al.,

1987). The physical state, i.e.

the share of melt vs. solid material in the particle is dependent on chemistry and elemental

composition of the particle and temperature. Inorganic compounds like those found in ash,10

do not melt at a certain temperature but have a temperature range, where both a melt and

a solid phase coexists. This temperature range between the first melting temperature (T0) and

the complete melting temperature (T100), also referred to as the liquidus temperature, may

be several 100 degrees of Celsius. From an ash stickiness point of view the temperature at

which enough melt is present is of major importance. Backman et al., (1987) defined the15

temperature at which 15% of the condensed phase, i.e. the sum of liquid and solid phases,

is molten as the critical stickiness temperature in recovery boiler deposits. This limit works

well for simple ionic salts, but for deposits containing silicon, leading to viscous melts,

another criterion is required. Here, the viscosity of a melt could be a criterion for stickyness

(See Figure 2.15). 20

When a deposit layer is formed, chemical sintering reactions may lead to consolidation of

the deposit. Condensation of ash compounds or ash-forming elements can influence the

efficiency of capture on a surface. The amount of condensate in a deposit depends strongly

on the mode of occurrence of the inorganic matter in the fuel. Low rank coals, lignites,25

biomass and similar fuels have the potential of producing large quantities of condensable

material. Furthermore the role of the condensate in determining deposit properties can be

substantially greater than the mass fraction of the condensate in the deposit might suggest.

Condensation increases the contacting area between an otherwise granular deposit

increasing the difficulty of removal by several orders of magnitude, influencing bulk strength,30


2-20

Figure 2.16: Mechanisms controlling the deposition and

maturation of ash deposits (after Laursen et al., 1998)

thermal conductivity mass

diffusivity etc. of the ash

deposit (Baxter, 1993).

L a u r s e n e t a l . ( 1 9 9 8)5

summarised the mechanisms

controlling deposition and

maturation of ash deposits as

summarised in Figure 2.16.

Apart f rom impact ion,10

thermophoresis, condensation

and chemical reaction, eddy

deposition is mentioned as

well. Small particles will be able

to follow small turbulent whirls,15

eddies, around the tube

surface. This might lead to

loose deposits on the lee side of

the heat exchanger tubes as well.

20

Figure 2.17 shows a schematic of the deposit present on a heat exchanger surface. Initially

condensation of volatile species on the heat exchanger surface will increase the inertial

impaction efficiency. Further nucleation of volatile ash components will increase the density

and strength of the deposit. The deposit thickness will be dependent on erosion and

shedding of the deposit.25


2-21

Figure 2.17: A schematic from deposit formation on a heat

exchanger pipe (after Frandsen 2000)

The role of compounds

formed during combustion

leading to deposit formation

Biomass contains potassium in

organic form, which will vaporise5

and decompose dur ing

combustion to form carbonates,

hydroxides, chlorides and

sulphates, depending on the local

composition and residence time10

of the products of combustion.

These compounds all exhibit very

low initial melting temperatures.

Their impact on fireside deposits

depends on their vapour pressure15

and whether they condense

homogeneously, on tube surfaces

or on other fly ash particles

making these sticky. When

potassium condenses on a fly ash particle, it forms a particle, which surface is enriched with20

potassium (Miles et al., 1996).

It was observed that most types of woods (containing high levels of calcium and low levels

of sulphur)hardly cause deposits when burned alone. When burned together with straw,

deposits become enriched with alkali sulphates and alkaline-earth sulphates. Calcium25

sulphate only partially prevented deposit build-up of potassium sulphates. Instead it was

found to act as a binder between particles on superheater tubes (Miles et al., 1996).

Stable chlorine compounds generated during combustion include alkali chlorides and

hydrogen chloride. Chlorine will increase the volatility of alkali compounds. The alkali30


2-22

chlorides formed tend to condense further downstream in the flue gas channel. The chloride

content is indicative for the volatility of the alkali compounds. Deposit formation tends to

increase with increasing degree of vaporisation of alkali compounds and thus with an

increase of the chlorine content in the fuel. Thus fuels high in alkali but low in chlorine show

less severe deposit formation than fuels with a higher amount of chlorine (Miles et al. 1996).5

As described above important compounds with respect to deposit formation often contain

potassium, chloride and sulphur. This might be the main reason why fundamental studies

have been carried out on the release of chlorine and potassium from the fuel. As stated

above already, during the pyrolysis stage potassium and chlorine containing compounds are10

released into the gas phase. Jensen et al. (2000) studied the release of potassium and

chlorine during pyrolysis of straw. They proposed a five-step release mechanism:

1) At a temperature of 200-400/C much of the organic matrix is destroyed, releasing

K and Cl to a liquid tar phase. Cl is further released as HCl or reacts with K.

Potassium could be present in de condensed phase as KCl or K2CO3. Some further15

release of HCl could take place caused to reaction of KCl with the char oxygen

containing functional groups, whereby K is bound to the char matrix.

2) At 400-700/C no significant amounts of K or Cl are released to the gas phase.

3) From 700-830/C all KCl evaporates. In this temperature range K also reacted with

silicon forming K-silicates.20

4) From 830-1000/C, K2CO3 decomposes and potassium is released as KOH or as free

K-atoms. Possibly potassium can also be released from the char matrix.

5) Above 1000/C potassium may be released to the gas phase from the char matrix and

from potassium silicates.

25

Yu et al. (2000) described the alkali release during pyrolysis of straw as taking place in two

steps, i.e. organic bonded release of potassium and inorganic potassium compound

vaporisation. These are in accordance with the extensive description of Jensen et al. (2000).

30


2-23

Corrosion

According to Salmenoja et al. (1996) chlorine induced corrosion is usually related to either

formation of gaseous hydrochloric acid or the deposition of alkali chloride on tube surfaces.

Corrosion from gaseous HCl is restricted mainly to high temperatures, high HCl

concentrations and reducing conditions. Alkali and metal chlorides reduce the melting5

temperature range and hence worsen corrosion. If the fuel contains also sulphur, and the

residence time is long enough and the temperature in the combustion zone is high enough,

alkali chlorides are converted to sulphates before being deposited on the tubes. Corrosion

is limited in this case because the formation of a melt is small at typical tube metal

temperatures due to the higher melting temperature of the alkali sulphates. The situation10

becomes more complicated when the alkali chlorides do not have enough time to become

sulphated completely before reaching the tubes. Sulphation of alkali chlorides deposited on

the tubes leads to the liberation of chlorine near the tube surface, thus making the chlorine

available for corrosion reactions.

15

Predicting ash deposition

There are many approaches to predicting ash deposition. One is the use of indices calculated

from the fuel composition. Indices are mainly based on the ash analysis presented as oxides

and were developed to predict deposit formation in coal fired boilers (Winegartner, 1974;

Skopurska et al., 1993; Reid 1984). An index based on alkaline and alkaline-earth metals20

was developed by Hupa et al. (1983) for boilers co-fired by bark, coal and oil. The portion

of sulphate-forming compounds (water soluble CaO+MgO+Na2O+K2O) is expressed as

a percentage of the total ash in the fuel fed to the boiler.

Another method is the determination of the fusibility of ash. With this technique a cone of25

(laboratory made) ash is heated and the temperature is recorded when the tip of the cone

is first deformed (IT); when the cone ash has molten into a spherical lump. With the height

equal to the width (ST); when the height is equal to half the diameter at the base (HT); and

when the cone is melted into a layer not more than 1.6 mm high (FT). The method is

completely empirical and interpretation of the values obtained is difficult (DIN 51730, 1978;30


2-24

ASTM D-1857-68, 1970). These predictors are the basis for prediction of deposit formation

and agglomeration in FBC today. It has, however, been reported to be poor indicators of

ash related operational problems with biomasses and energy crops (Juniper, 1995; Nordin

et al., 1995; Wall et al., 1995). Ash fusion temperatures as determined from the “whole” fuel

ash poorly predict ash deposit behaviour when assuming that reactive, mobile ash-forming5

elements form the initial deposits.

Another approach was introduced by Skrifvars et al., (1998). This approach was purely

based on chemistry and melting behaviour of ash-forming compounds. TPCE calculations

in combination with chemical fractionation were used to predict deposit formation. The10

same approach is used in this work. (See also 3.2)

2.3.3 Deposit formation: Field and laboratory experiences

Deposit formation in biomass fired FBC installations, i.e. in full-scale, pilot-scale and lab-15

scale equipment is studied all over the world. Only a few examples relevant for this work will

be reported here.

Wood derived fuels

According to Bryers (1996) pure wood does not contain a measurable amount of sulphur.20

Consequently pure wood is considered not to be a problematic fuel. However, sand

inclusions partitioned from the calcium-rich wood ash may react independently during

combustion with any potassium present as volatile species. The potassium absorbed on a

surface of the quartz surface produces low melting potassium silicates.

25

Miles et al. (1995a) found the same trend. When firing wood in an FBC with potentially

large amounts of contaminations (soil), the role of potassium may be reduced when

compared to agricultural wastes high in potassium. The same was the case when co-firing

wood-derived fuels with coal. (Skrifvars et al., 1997c). The role of alkaline-earth metals, i.e.

calcium was more pronounced. Miles believed initial deposition takes place by deposition30


2-25

of hydroxides, followed by sulphation of alkaline and alkaline-earth elements. If large

amounts of soil contamination or clay are present, the role of silicon may still be quite

pronounced in secondary deposit growth by particle impaction following the initial formation

of condensed layers on surfaces. Complex alkali-alkaline earth-alumino silicates form or are

incorporated into superheater deposits in this manner (Miles et al. 1995a).5

Skrifvars et al. (1997c) found that alkali sulphates dominate the deposit composition in the

hotter part of a CFBC firing forest residue. Alkali sulphates and chlorides were the two major

components in the deposits collected in the colder region in the flue gas channel. The same

trend was found in a wood fired CFBC. Experiments with a semi-full scale CFBC firing coal,10

peat or wood showed the same trends (Skrifvars et al., 1998a). Deposits from wood were

enriched in potassium chlorine and sulphur when compared to coal and peat. These

deposits were predicted to be sticky at temperatures around 730-750/C. Even if coal or peat

contained an order of magnitude higher amount of ash-forming elements in the fuel the

resulting compounds were well-behaving and fairly non-sticky.15

As mentioned by Baxter et al. (1998) deposits from agricultural residues showed differences

in composition when comparing the wind and lee side of a deposit probe. Alkali compounds

deposited through condensation or thermophoresis were found equally distributed around

the probes, whereas silicate, i.e. the larger particles were enriched at the wind side, indicating20

inertial impaction as deposit mechanism. At the same time Skrifvars et al. (1998) presented

the approach used in this work. A combination of chemical fractionation and

thermodynamic multi-phase multi-component equilibrium calculations was used to predict

deposit formation of coal, forest residue and wood chips. The possibility of formation of

silicates was omitted from the calculations. Skrifvars’ prediction suggests that a mix of wood25

chips fired together with construction residue could cause more problematic deposit

formation than when firing a forest residue.

Skrifvars et al. (1999b) summarised the measurements as carried out and described by

Skrifvars et al. (1997b, 1998a) and Hansen et al. (1997, 1998). After a comparison between30


2-26

lee and wind side deposits it is concluded that chlorine and sulphur play a dominant role in

deposit formation in CFBC firing biomass fuels, i.e. wood, bark, forest residue and wheat

straw with or without co-firing with coal. An indication of sulphation of the deposits

containing alkali chlorides was shown in deposits from a 18 MWth CFBC. Sulphation could

not inhibit chlorine to reach deposits. Only switching to pure coal or peat firing could5

achieve this.

Steenari et al. (1999) studied the composition of fly ash deposits from a CFB firing a mixture

of 70 wt% wood and 30 wt% coal or 30-40 wt% wood and 60-70 wt% peat. Compared to

fly ash from combustion of wood and bark, the co-combustion ashes studied had lower10

contents of calcium, potassium, manganese and chlorine. The admixture of coal or peat to

the wood fuels added aluminium, iron, magnesium and sulphur to the ash. The same was

shown by Dayton et al. (1999) studying the formation of chlorides and alkali metals in case

of co-firing wood (oak) and straw with coal. The amount of HCl detected during the

combustion of a coal/wheat straw sample was higher than expected based on the15

combustion results for the pure fuels. The amounts of KCl(g) and NaCl(g) detected during

the combustion were lower than expected. Chemical equilibrium analysis indicated that the

amount of condensed alkali was enhanced, due to the formation of potassium silicates.

Energy crops20

Skrifvars et al., (1997b) studied the behaviour of Salix (moisture content up to 50 wt%) and

a mixture of forest residue with Salix in a semi full-scale CFBC. Deposit samples were taken

at the cyclone inlet (850/C) and at two different locations in the convective part

(temperature 680 and 250/C, respectively). Sulphur and chlorine were found in all samples

locations and sampled deposits. Skrifvars states that even if no severe deposit build-up was25

noticed in the test run, a molten phase may be present in the fly ash, that eventually can be

responsible for the formation of a deposit. Calcium oxide in the fly ash is supposed to have

recarbonised at lower temperatures causing deposits in the sampling location downstream

of heat-exchangers in the convective part of the flue gas channel.

30


2-27

Agricultural waste

A wide range of agricultural wastes fired as such or together with wood in full-scale and

laboratory-scale equipment were studied by Miles et al. (1995a through c) and Baxter et al.

(1998). The results of the full-scale experiments showed the influences of fuel composition

on the deposits formed. A bubbling fluidised bed unit burning wood and almond shells5

developed superheater deposits enriched in potassium and sulphate, chlorine and

carbonates, indicating that mechanisms of condensation and chemical reaction were

significant in the deposit formation. The mechanisms of condensation and sulphation of the

deposit, depend on mass transfer rates. This means that at the front side of the tubes

sulphation and alkali enrichment was high. The concentrations of sulphates, chlorides and10

carbonates along the convective pass varied with the stability of the compounds. As the

temperature decreased less sulphates and carbonates were found. Many deposits in the

superheater region were enriched in calcium as well due to lime stone addition to the

fluidised bed. Superheater deposits from the bubbling fluidised bed had a higher

concentration of chlorides than those from the circulating fluidised beds, which may be15

related to differences in fuel composition but also indicative of the differences due to

recirculation. The composition of the fire wall deposits reflected the composition from

impacting particles as expected.

Laboratory experiments showed that wheat straw deposits were in many respects similar to20

rice straw deposits and illustrated many of the mechanisms of ash deposition. Probe deposits

from straw firing were enriched in potassium and chlorine directly at the probe surface,

developing outwards into a matrix of sintered silicate-rich flyash particles. Phosphorous was

enriched in the outer layers as well. Switch grass deposits also showed potassium enrichment

and a greater enrichment of sulphate when compared to straw. Initial deposits containing25

alkali chloride were found (Baxter et al., 1998).

The almond shells and hulls, high in potassium, formed fine-textured deposits, rich in

potassium, more than could be accounted for by sulphates and chlorides. Baxter suggests

this is due to the deposition of hydroxides or carbonates. The deposits from the almond30


2-28

shells-wood blend were potassium and sulphate enriched, as in the deposits from full-scale

facilities, but contained more silicon and less calcium. These differences probably accounted

for the addition of limestone in the full-scale unit (Baxter et al., 1998).

The fouling tendency of straw and bagasse was found to decrease when the fuel had been5

exposed to rain in the field. The high alkali content was thereby lowered reducing the

amount of deposit formed. Also rice straw is known for its high fouling tendency and might

be only acceptable for existing boilers, when rain-leached and spring-harvested, at a low

concentration with more conventional fuels (Miles et al., 1995a).

10

Jensen et al. (1997) measured deposit formation in a wheat and barley straw fired grate

boiler. As found by Miles et al. deposits were enriched in potassium. Jensen finds an

enrichment of chlorine as well. Submicron aerosol particles with a mean particle diameter

of approximately 0.3:m were generated. Vaporised potassium compounds are supposed

to condense on the boiler walls and superheater surfaces acting as glue for impacting silicon-15

, calcium-rich particles. The surfaces are probably dominated by elements from the aerosols

or from condensation of vapours, while larger particles are covered with a layer of KCl

(Jensen et al., 1997). An increase in the local gas temperature and straw potassium content

gives an increase in the amount of hard deposit formed which has a higher chlorine- to

potassium ratio than the loose deposits found.20

Hansen et al. (1998) described similar studies in co-combustion of coal and straw in full

scale power plants in Denmark. In an 80MWth CFBC quartz was used as bed material and

limestone added for sulphur capture. The boiler was designed for firing 50-50 wt% coal and

straw. Mature deposits from the superheater tubes situated in the convective pass contained25

mainly potassium, sulphur and some SiO2, Al2O3, and CaO. The chlorine content was very

low. It was suggested that KCl was deposited initially on the tubes after which sulphation

took place with subsequent release of gas phase chlorine. Corrosion was found to be 5-25

times higher than when firing coal alone, probably due to the role of chlorine released after

sulphation of the initial deposit.30


2-29

2.3.4 Bed agglomeration: the mechanism

In literature terms such as agglomeration, sintering and defluidisation are used freely in

different ways to describe the unwanted collapse of a fluidised bed. In this work

agglomeration is defined as the phenomenon where particles gather into clusters of larger5

size than the original particles. Sintering is defined as the process in which fine particles

become chemically bonded at a temperature that is sufficient for atomic diffusion. Since in

fluidised bed conversion particles can be held together by a molten phase both the terms

agglomeration and sintering can be used to describe the same phenomena. The initial

agglomeration temperature is defined as the temperature where the first molten phases10

appear that are able to “glue” bed particles together into agglomerates. Defluidisation is

defined as the total collapse of the fluidised bed resulting in a rapidly decreasing pressure

drop or erratic behaviour of the bed, which results in substantial temperature changes. The

presence of agglomerates does not, by definition cause total defluidisation of the bed.

Instead the defluidisation temperature will still be dependent on boiler-specific conditions as15

well.

Bed agglomeration is a quite complex phenomenon that can take place in fluidised bed

boilers under certain circumstances. Agglomeration cannot only be explained by looking at

physical phenomena, such as temperature, particle size distribution, mixing processes with20

resulting shear stresses between particles and particle attrition, abrasion, fragementation and

cleavage, but also by chemical phenomena. An example is the reaction between ash-forming

components and bed particles, leading to coating build-up and possible agglomeration when

molten phases are formed. These molten phases could glue particles together. In practice

both physical and chemical phenomena will play together leading to bed agglomeration or25

not.

Bed agglomeration is tightly connected to the release of ash-forming matter. It is possible

that easily released ash-forming elements would rather condense on bed particle surfaces

than form a submicron fume and hence be transported to the flue gas channel. This kind of30


2-30

Figure 2.18:Schematic of bed agglomeration as described by

Öhman (2000)

coating formation has been

detected when firing biomass in

a quartz bed. It was found that

the coating usually does not

exceed a thickness of some 10-5

50 :m for a mean bed particle

diameter of 350-500 :m

(Skrifvars et al., 1997d).

Ö h m a n d e f i n e d t h e10

agglomeration sub-processes as

follows (See Figure 2.18):

1) Ash deposition on the bed material is probably dominated by i) an attachment of

small particles to the bed particle surface ii) condensation of gaseous alkali species

on bed particles and iii) chemical reaction of the gaseous alkali on the particle15

surfaces.

2) As the continuous deposition on the bed particles proceeds, the inner layer of the

coating is homogenised and strengthened via sintering.

3) The melting behaviour of the homogeneous silicate layer controls the adhesive

forces, which are responsible for the final temperature-controlled agglomeration20

process (Öhman 1999, Öhman et al. 2000).

The formation of a coating may initiate:

1) The formation of agglomerates when sticky, i.e. partially molten. In this way two

different particles can become glued together. The molten phase could be either a25

viscous glassy silicate melt or a non-viscous melt. The former has been found to be

the common reason for bed agglomeration in FBC’s using quartz as bed material

(Skrifvars et al., 1997a, Nordin 1993). A non viscous melt has been identified

occasionally when high sulphur lignites were combusted (Manzoori, 1990).


2-31

2) Chemical reaction sintering the bed particles together forming an agglomerate. This

mechanism has been identified in fluidised bed boilers using limestone as bed

material. The agglomeration had mainly occurred as a pluggage of the cyclone return

leg in an CBC (Skrifvars, 1997a).

5

2.3.5 Bed agglomeration: Field and laboratory experiences

Fluidised bed combustion power plants often employ limestone injection for control of

sulphur emissions, with additional benefits of reduced bed agglomeration. Other additives

such as, kaolin, dolomite or magnesium oxide, have been used to reduce bed agglomeration10

as well (Miles et al. 1995a). Miles et al. (1996) stated that bed agglomeration in fluidised bed

combustors at temperatures above some 760/C is dominated by silica. Examples of

experimental studies as shown below show the same trend. Either silica-rich bed material

or silica-rich fuels are involved in bed agglomeration processes due to formation of low

melting potassium silicates (See also paper IV and V).15

Agglomeration under reducing conditions

Maniatis et al. (1993) studied the FBG of wood and bark. They showed that an operation

temperature above 900/C resulted in sand agglomeration due to formation of low-melting

alkali silicates. 20

Ergudenler et al. (1993) found that a fluidised bed agglomerated under reducing conditions

when firing wheat straw. The sand bed agglomerated at some 800/C in the presence of

straw ash, resulting in channelling and defluidisation. Changes in air velocity and/or fuel

feeding rate did not improve agglomeration characteristics. Potassium was believed to be25

the major contributor to the agglomeration process. Ergudenler speculates that potassium-

oxide forms, melts and penetrates through the voids of the silica sand and then forms low

melting potassium silicates. However, in practice it is believed that volatile reactive

compounds such as alkali chlorides, sulphates and carbonates will react with the bed

material forming low-melting alkali silicates (See also 3.3)30


2-32

Four biomasses gasified in a pilot scale FBG with a dolomite bed, being pine sawdust, bark,

straw and pine bark were studied by Moilanen et al. (1995). It was shown that in the

gasification of pine sawdust, very high carbon conversions could be achieved already at

relatively low temperatures, whilst with bark and straw gasification high conversion

efficiencies could be achieved at above 850/C, but sintering of the straw ash caused severe5

operational problems. This could be accounted for by the potassium and silicon present in

the fuel, forming a silica-containing glassy melt, that caused bed agglomeration at these

temperatures (See also IV).

Liliendahl et al. (1999) studied defluidisation through multivariate methods. The fuels were10

Miscanthus, Reed Canary Grass, Salix, and olive waste in a lab-scale (P)FBG with sand as

bed material (See also V). Evaluation indicated that the temperature and the potassium and

sodium contents, and also to some degree the pressure seems to enhance defluidisation,

whereas the presence of calcium seems to have the opposite effect.

15

Bed agglomeration under oxidising conditions

Öhman et al. (1998) studied the behaviour of olive flesh, Lucerne, wheat, wood, Reed

Canary Grass, wood residue, bark, RDF and cane trash in a laboratory scale FBC using

quartz as bed material. With this FBC onset of defluidisation is measured with an accuracy

of ±5/C. (See also VI).20

Lin et al. (1997) found that during particle sampling agglomerates in a straw fired FBC had

been formed before the pressure drop could start to decline. This means that the time at

which the pressure drop begins to fall is not the time corresponding to the initial

agglomeration temperature (as described by Öhman). 25

Lin stated that small agglomerates will segregate to the lower part of an FBC. When the

number of agglomerates exceeds a critical value, a layer of defluidised agglomerates will be

formed. The height of this defluidised layer increases with an increase in number and size

of the agglomerates. This mechanism could explain the defluidisation at rather low30


2-33

Figure 2.18: Defluid isation te mpe rature s of som e biom ass fue ls

as determined by Öhman et al.(1998) and Skrifvars et al.(1999a)

temperatures, i.e. some 800/C.

At h igher tempera tu res

defluidisation is believed to be

caused by molten phases glueing

particles together. 5

A qualitative comparison

between the defluidisation

temperatures of different biomass

fuels is possible with Öhmans10

standardised method (see Figure

2.18). Skrifvars et al. (1999)

described the same experiments and compared the defluidisation temperatures as

determined by Öhman with the ASTM ash fusion initial deformation temperature (Tint) and

sintering temperature (Tsint) as determined from compression strength experiments. It was15

shown that in all cases the fusion temperature fails to predict bed agglomeration. The

sintering test predicted problematic behaviour for Lucerne, wheat straw and Reed Canary

Grass. Moderate behaviour is expected for olive flesh, cane trash and RDF. A bark and forest

residue were considered to be non-problematic fuels.

20

The compression strength tests seemed to be able to predict defluidisation fairly well when

compared to the lab-scale FBC. However, no data for comparison with full-scale equipment

were available.

Turn et al. (1998) carried out similar tests under gasification conditions and using alumina25

silicate as bed material firing bagasse and bananagrass as fuels. The experiments showed

as before that the silica in the bed captures potassium released from the fuel. Chlorine was

not detected in the bed material samples. Ca, K and P together with Si were the main

elements found in the coating on bed particles. Turn set up an element balance and found

that most of the potassium and roughly half of the calcium was retained in the bed.30


2-34

Werther et al. (2000) mentioned in an overview on combustion of agricultural residues that

an FBC firing various coals, sewage sludge and wood chips could be operated without

agglomeration problems. Firing coffee husk, however, led to agglomeration after six hours

when fed above the bed, when feeding within the bed agglomeration occurred after four

hours. Similar observations were reported by others who fired sunflower husk, cotton husk,5

cotton stalk, coffee husk, palm fibre, soy husk, pepper waste, groundnut shells and coconut

shell (Babat et al., 1997).

The role of additives for prevention of bed agglomeration

Moilanen et al. (1996) studied the agglomeration in FBC of among others rape and wheat10

straw, Reed Canary Grass, pine saw dust, pine bark, willow, and Miscanthus. As bed

material Al2O3 was used. Also, laboratory ash was made at different temperatures. The ashes

of wheat straw, willow and Miscanthus prepared at 500/C showed much higher sintering

than when heat treated at 850/C. In the fluidised bed tests the bed material was not totally

agglomerated in any of the cases in spite of the clear sintering observed in the bed samples.15

Öhman et al. (2000a) showed that addition of kaolin can shift agglomeration of bed material

to higher temperatures when firing biomass in a lab-scale FBC. The results showed that

kaolin was transformed to meta-kaolinite particles, which adsorbed potassium species. The

increased agglomeration temperature was explained by the decreased fraction of melt in the20

bed particle coatings.

Steenari et al. (1998) studied the role of additives such as kaolin and dolomite in sintering

of straw ash. As Öhman she found that potassium was captured by kaolin leading to a

higher first melting point. Dolomite added to wheat and barley ash reacted with silica to form25

calcium and magnesium silicates. No reactions between potassium and dolomite could be

detected.

Chapter 3: Results

3-1

Figure 3.1:The a sh com position of the stu died fu els

3 RESULTS

3.1 Biomass fuel characterisation

3.1.1 Traditional fuel analysis5

The three main ash-forming elements of the 30 biomass fuels as described in this thesis are

K, Ca and Si. The different biomasses could be presented in the triangular composition

diagram as shown above in Figure 3.1. In this diagram the main ash-forming components

are normalised to 100% as if they were the only ones present in the fuel. Fuel #1 through10

#23 represent the fuels that were fractionated. Fuel #1 through #16 are discussed in papers

I through III and VI; Fuel #17 through #23 represent the fuels as discussed in report VII and

by Miles et al.(1995a through c). Fuel #24 through #30 represent the fuels for which only

Chapter 3: Results

3-2

TPCE calculations were carried out (see chapter 3.2.2)[IV-V]. The Figure shows the wide

range of fuels studied. The amount of Si , presented as SiO2, in the fuels varied from 1 to 90

%wt, Ca, presented as CaO, varied between 5 and 77 %wt and K, presented as K2O, varied

between 2 and 72 %wt. Fuels low in Si, i.e. containing less than 10 %wt SiO2 are Wood

chips, Eucalyptus, Forest residue, Salix (#1), Lucerne, Almond hulls and Almond shells.5

Fuels high in Si are Reed Canary Grass, Rice straw and Peat (#3). Fuels low in Ca, but high

in K are Almond hulls and Almond shells. Eucalytus and Wood chips are the fuels highest

in Ca.

3.1.2 Chemical fractionation10

Ash-forming elements can be present in a reactive form or a less-reactive form. Reactive ash-

forming elements can be present as easily soluble salts or organically associated compounds.

These can volatilise, leading to deposit formation in the flue gas channel after condensation,

or interact with bed material in a FB causing agglomeration. Included minerals could be15

released from the fuel during the char burning stage and be entrained from the bed. Less-

reactive ash-forming elements can be present either as external minerals in the fuel. Thus,

knowledge of the structure of the fuel, i.e. the way ash-forming elements are bound is

important for understanding how fuels behave in boilers and furnaces. For this purpose the

chemical fractionation technique or SEM/EDX analysis can be used. Papers I, II, III, VI and20

VII show results for the fractionation of a bituminous coal, two types of peat, Salix (willow),

several types of Scandinavian forest residue, wood and wood chips, two types of pine bark,

construction residue wood, two wheat straws, almond hulls, almond shells, olive pits, paper,

rice straw and switch grass, respectively.

25

In all fuels the major part of Si, Al, and Fe was found in the residual fraction; only a minor

part is leached by the three leaching agents used. The alkali and alkaline-earth materials (K,

Na, Ca and Mg) occur in organic structures or reactive mobile inorganic components. The

fractionation results seem to be consistent with this biological function. The major part of the

alkali materials was found in the water-soluble fraction. Hence, they are believed to be easily30

Chapter 3: Results

3-3

Figure 3.2: Fractionation result of Bark #6 Figure 3.3: Fractionation result of Wood #12

volatilised, making them easily available for chemical reaction with other components

leading to possible deposition, corrosion and agglomeration. This means that fuels with

lower alkali content should be less problematic when fired in a boiler. Potassium and sodium

were also found in the residual fraction as components of mineral soil contamination of the

fuels. Ca was supposed to be present organically bonded in the cell wall as easily ion-5

exchangeable material, thus found in the acetate fraction. Alkaline-earth materials are less

volatile. Non-metallic materials (S, P, Cl) occur as plant nutrients. Chlorine plays an

important part in the transformation of inorganic alkali compounds during combustion and

gasification. The fractionation results suggest that all chlorine is present in an easy

vaporisable form.10

A large portion of the ash-forming elements in the studied coal and peat were found in the

acidic and rest fraction, indicating that most of the ash forming elements were associated

with either the earth alkali carbonates and -sulphates or silicates and clay minerals.

15

The results for the biomass fuels showed a different trend. In biomass fuels a major part of

the inorganic matter was associated with either water-soluble compounds or with the organic

matter. The relatively large amount of inorganic compounds leached out by the acid or

remaining in the rest fraction was probably originating from mineral inclusions or from

impurities included as a result of fuel sampling by foresting and handling the biomass fuels.20

In case of Bark #6 [I] the high amount of Ca leached out by the HCl could be accounted

for by calcium containing internal minerals, most probably calcite or calcium oxalate.

Chapter 3: Results

3-4

Figure 3.4a: Amount of element leached after

leaching Bark #6 with water

Figure 3.4c: Amount of element leached after

leachin g Bark #6 w ith HC l.

Figure 3.4b: Amount of element leached after

leaching Bark #6 with NH4AC

The sum of all fractions, i.e. H2O, NH4Ac, HCl and rest fraction was not always the same as

the amount as determined in the untreated solid. Especially for Cl a large difference was

noticed.

Figures 3-2 through 3-4c show the fractionation results for Bark #6 [I, III] and Wood #125

[I]. The results for the ash analysis of the untreated fuels were obtained from triple

measurements. In the figures the 95% confidentiality limits are also shown (Davies et al.,

1988). The largest interval was found for the elements which can be associated with soil

contamination of the fuel, such as Si, Al and Fe. This could be related to fuel sampling

before analysis10

The fractionation of Bark #6 �I, III� has been carried out three times. The bark was analysed

Chapter 3: Results

3-5

Figure 3.5a: Wood #12 (I)”forest residue” Figure 3.5b: Forest residue #14 (VII) “brun

grot”

Figure 3.5c: Fores t residu e #1 5(I) “grö n grot” Figure 3.5d: Forest resid ue #7 (II,III)

at three different laboratories, i.e. in triple (obtained from KCL), in quadruple (obtained from

Fortum Power and Heat oy) and two analysed once, (obtained from the University of Oulu)

respectively. The Figures 3.4a through 3.4c show the amount of leached elements as

obtained from the analysis of the liquid samples. Where possible the 95% confidentiality

limits are shown as well. The two first laboratories obtain results which are comparable. The5

last mentioned laboratory used partially non-standardised techniques, which led to

uncomparable results, such as in the case of K and Ca analyses.

Differences in fractionation results in one fuel group

Apart from analysis problems, the fractionation results for one fuel group such as “Forest10

residue” can be very different. Soil contamination, presence or absence of roots, leaves and

branches and seasonal effects greatly influence the outcome of the traditional fuel analysis

and the chemical fractionation. This indicates that for every fuel to be fired a representative

sample should be taken and analysed which makes the chemical fractionation as used today,

very time-consuming and expensive. However, an extensive database of fractionation results15

Chapter 3: Results

3-6

for different classes of fuels could give some insight of what to expect when planning to fire

a certain type of fuel.

Wrongly categorising a fuel can lead to misunderstanding of the fractionation results. Figure

3-5 gives an example of different fractionation results of samples of “forest residue”. Wood5

#12 is a “forest residue” low in ash, i.e. some 0.5 %wt and should be called “wood”. The

other three forest residues, #7,14 and 15 are better comparable.

3.1.3 SEM/EDX analysis biomass fuels

10

SEM/EDX analyses of biofuels before and after fractionation and after gentle ashing made

understanding and better interpretation of the fractionation results and fuel behaviour under

combustion conditions possible. Here only the results for Bark #6 will be presented as an

example. In paper I also SEM/EDX analysis of Wood #12 is presented.

15

Solid samples were taken from Bark #6 after each fractionation step. In addition 0.5 g of

Bark I, and Wood I, was ashed in a laboratory furnace during 15 minutes at a temperature

of 450, 650 or 850/C. All these samples were subjected to SEM/EDX analysis in an attempt

to verify the fractionation results and to distinguish between internal and external minerals

in the fuel. Hereto they were immobilized on a metal plate and coated with carbon, after20

which direct analysis of the samples took place. The samples were analysed for Si, Al, Fe,

Ti, Mn, Mg, Ca, Na, K, S, P, and Cl by point analysis and by use of X-ray maps.

The structure of Bark #6 after fractionation

The SEM/EDX images from samples of Bark #6 retrieved before and after each leaching25

step are shown in Figures 3.6a through 3.6d. The Figures a through c show the embedded

minerals which were rich in Ca (white). These do not leach out in the water or the acetate

step. Only after the total break up from the organic structure, which happens during leaching

with HCl the calcium containing minerals dissolve.

30

Chapter 3: Results

3-7

Figure 3.6c:The structure of Bark leached

with Ac etate

Figure 3.6d: The structure of Bark leached

with HCl

Figure 3.6a:The structure of untreated Bark Figure 3.6b: The structure of Bark leached

with water

The structure of Bark after laboratory ashing

Figures 3.7a and b show the structure of Bark #6 after laboratory ashing. Figure 3.7a shows

that at low temperatures the structure of the fuel remains intact. The white arrows point at

the calcium containing minerals still embedded in the fuel.

Chapter 3: Results

3-8

Figure 3.7a: The structure of Bark after

ashing at 450/C during 15 minutes; the bar

indicates 200:m; The arrows point at calcium

contain ing m inerals

Figure 3.7b: The structure of Bark after

ashing at 850/C during 15 minutes; The bar

indicates 200:m; The arrows point at calcium

containing minerals released from the fuel

At 650/C the fuels start to disintegrate and part of the calcium-containing minerals is

released from the fuel and could be pointed out by SEM/EDX. At even higher temperatures

the fuel structure disintegrates into small particles and the calcium-containing minerals are

completely released from the organic matter. They can be seen as small white “needles” on

the SEM/EDX picture as shown in Figure 3.7b. Some are pointed out by the white arrows.5

3.2. Ash behaviour in (P)FBG and FBC

3.2.1 Prediction of deposit formation

The fuel-specific characterisation method presented here combines advanced fuel10

characterisation tests with thermodynamic multi-component, multi-phase equilibrium

calculations (TPCE), which describe the chemical conversions of the ash components in an

FBC.

A principle sketch of the method is presented in Figure 3.8. The method is divided into three15

separate steps being:

Chapter 3: Results

3-9

Figure 3.8: Simplified schedule of the deposit predictor

1) chemical fractionation,

2) TPCE calculations, and

3) melting range estimations.

It was assumed that the ash-forming elements leached by water and ammonium acetate5

represent the more reactive species, many of which are volatile could form fine reactive

components. Therefore, the results from the analysis of the water and acetate fractions were

combined. This combined fraction was assumed to form one fraction of reactive

components. These could form deposits in the superheater and economiser surfaces of an

FBC/FBG (Skrifvars et al., 1998) or Stoker fired combustor or they could interact with less10

reactive components in the fuel or react with the bed particles, forming a coating. This last

event could lead to bed agglomeration under certain process conditions (see 3.2.2).

Ash-forming elements leached by hydrochloric acid and those present in the residual fraction

were assumed to represent the less reactive ash-forming elements that could remain in the

bed and form the more coarse ash particles. Part of the less-reactive components could be15

entrained from the bed together with bed material.

Figure 3.9 shows how the combined fractions, i.e. the reactive fraction and the less-reactive

fractions were distributed in the fuels. The vast majority of the ash-forming elements in coal

Chapter 3: Results

3-10

Figure 3.9: The distribution of leached elements over the reactive and less-reactive fractions

and peat are found in the fraction assumed to form the bottom ash, while the share of the

ash-forming matter forming the fine ash fraction is much higher in the biomasses. Grasses

such as wheat straw, switch grass and rice straw have a skeleton rich in silicon which will be

found in the less-reactive fraction. Scandinavian bark contains calcium-rich minerals which

were also found in the less-reactive fraction (See 3.1.3).5

The analysis results from the reactive fractions of the fuels were used as input in the

thermodynamic calculations. With this method, the chemical interaction of the two ash

fractions with the combustion or gasification gases were modelled. The interaction between

the reactive and less-reactive fraction has not been considered. In the calculations the fuel10

composition was taken from the ultimate fuel analysis to make the formation of a realistic

gas phase possible. An air factor of 1.2 �II, III �for the combustion cases and 0.3 �VII� for the

gasification cases and atmospheric pressure were assumed. The calculations were carried

out for a temperature range 500-1200/C. The presence of aluminium in the calculations

could lower the first melting point of silicates to below 500/C due to uncertainties in the15

Chapter 3: Results

3-11

calculations, introduced by extrapolation of the thermodynamic data. In the calculations the

presence of aluminum was therefore omitted to obtain a realistic first melting temperature.

When interpreting the results of the analysis, its limitations should be recognised. The

equilibrium approach implies that no reaction kinetics, transport limitations (diffusion etc.)5

or fluid dynamic effects are taken into account. Also many deposit formation mechanisms

are highly species-specific, resulting in deposit compositions that are not easily related to the

fuel ash composition. Nevertheless, a careful interpretation of the results gave useful

information about the systems studied. The assumption that all reactive ash forming

elements leave the bed could lead to an overestimation of the amount of deposit formation10

due to neglected interaction of reactive species with bed material. However, a qualitative

estimation of how a deposit is expected to behave can be made.

For ash-related problems in an FBC it was found that the amount of melt present in the

condensed phases was of major importance. In order to deposit on a surface or agglomerate15

ash particles should contain a certain amount of liquid melt. Based on experience with black

liquor recovery boilers it was assumed that an amount of 15% wt of the condensed phases

molten at a certain temperature (T15) enables deposit formation in the flue gas channel

(Backman et al., 1987)(see Figure 2.15). This limit was used in this investigation as well.

Whenever the amount of melt, from the reactive fraction, exceeded 15%, fly ash deposition20

in the flue gas channel was predicted at a certain temperature. Thus from the equilibrium

analysis, the amount of melt (as %wt) was calculated as a function of temperature.

Table 3.1 summarises the sticky temperatures, T15, as predicted by the above-mentioned

method for the fractionated fuels for combustion. In this same table the percentage of25

“reactives” compared to the total ash, the highest amount of a liquid phase predicted, and

the range at which the melt is predicted to appear are summarised. With the percentage of

“reactives” compared to the total ash the ability of the less-reactive coarse ash to function

as a cleaning agent in the flue-gas channel is characterised. Certain coals, for example, are

known not only to keep heat exchanger surfaces clean but even to cause erosion of tubes.30

Chapter 3: Results

3-12

Coal and peat as well as additives are also known to decrease fouling tendencies in co-

combustion with problematic fuels (Skrifvars et al.,1999; Hupa et al., 1983).

In Table 3.3 the fuels are ranked according to selected data in Table 3.1. A qualitative “star

ranking” is used where one star means a bad ash behaviour and three stars a good ash5

behaviour. In Table 3.2 the ranges for the rankings are identified. Table 3.3 shows that the

coal could be the less problematic fuel to be combusted. Peats, barks, and woods are also

fuels in the higher “star ranking”. Forest residues can be found both in a higher and lower-

ranking range indicating there are forest residues that cause little problems and forest

residues that form deposits when combusted, depending on fuel composition and10

distribution of ash-forming elements in the fuel. Salix, straws, grasses, almond shells and

hulls are in the lower range and should be expected to be rather troublesome to fire causing

deposit formation.

When looking at measured deposit formation (See also 2.3)as taken from some different15

sources, the same trends can be seen. Coal (Lind 1999) and peat (Skrifvars et al., 1998) are

found to be less problematic when combusted, whereas Salix (Lind et al., 1999a)and forest

residues (Skrifvars et al., 1997, Valmari et al., 1999) can form deposits (II).

In paper III the amount of collected deposits on hotter short-term probes at a flue gas20

temperature of 820-880/C is presented. The measurements confirm the ranking as presented

above as well. Adding a forest residue to a wood type fuel decreased the amount of deposit

on the probes (Peltola et al., 1999). Adding coal or forest residue to a wood chips fired

boiler decreased the deposit formation (Skrifvars et al., 1999) and firing straw or adding

straw to a coal fired boiler increased the deposit formation on the probe dramatically25

(Hansen, 1997).

Chapter 3: Results

3-13

Table 3.1: Summarised results from the TPCE for the deposit prediction (8=1.2)

Fuel* corrected values

Reactive

amount

(g/kg)

Reactive

% of total

ash

Total

fractions

(g/kg)

T15 (/C) Melt

range

(/C)

Max m elt

(%wt)

#4: Coal* 0,9 2,9 30,0 >1200 >1200 -

#3: Peat*5 3,4 13,6 25,0 >1200 550-750 5

#10 : Peat 6,9 11,0 63,2 >1200 550-600 ,04

#8: Wood 2,2 29,6 7,3 620 575-750 27

#9,#12: Wood 1,9 79,6 2,4 >1200 775-850 3

#13 : Wood 2,5 54,8 4,6 >1200 750- 22

#6: Bark10 7,3 55,9 13,1 >1200 >1200 -

#11 : Bark 6,6 27,1 33,7 >1200 600-650

850-900

3

#7: For.res.* 8,3 67,2 12,3 >1200 650-700 10

#14 : For.res 7,8 71,5 10,9 780 600-650

800-950

19

#15 : For.res 6,6 59,0 11,2 >1200 >1200 -

#16: Cons.res.15 6,5 16,7 39,0 675 650-725 15

#1: Salix* 5,8 62,2 9,3 860 825-1000 18

#2: Salix* 9,0 57,4 15,7 880 825-1000 24

#5: Wheat Straw 13,0 66,4 19,7 880 875-1080 48

#23: Wheat Straw 6,2 34,0 18,1 1000 980-1180 58

#22: Switch grass 20 3,6 22,3 15,9 1060 1050-1190 31

#17: Almo nd hulls 28,4 81,3 34,9 870 750-1200 63

#18: Almo nd shells 40,7 7,5 91,0 950 680-1160 27

#21: Rice straw 23,4 24,9 93,9 600 >600 92

Chapter 3: Results

3-14

Table 3.2: The used “star ranking” [III]

Fuel Reactive

amount

(g/kg)

Reactive

% of total

ash

Total ash

(g/kg db)

T15

(/C)

Melt range

(/C)

Max m elt

(%wt)

q >9 >60 >90 <700 >150 >15

qq5 2-9 30-60 20-90 700-900 50-150 5-15

qqq <2 <30 <90 >900 <50 <15

q=bad, qqq=good

The fuels as described in Paper VII were also fired in full-scale equipment and a laboratory

scale multi-fuel combustor (Miles et al., 1995a through c). It was found that silicon-rich10

deposits were formed firing rice straw and wheat straw and switch grass. Deposit

characteristics for almond shells and hulls were similar. Potassium compounds from almond

hulls were supposed to be typical bonding agents between silica or media particles in

superheater deposits. In some cases a glass was formed.

15

The composition of deposits found in full-scale or laboratory scale boilers will be different

from the composition as calculated with the TPCE calculations. The thermodynamics

assumes equilibrium, which will not be reached in practise. This means that in deposits often

different layers can be detected, each with its own composition and its own melting

behaviour. In the calculations it is assumed that these layers have interacted and reacted20

with each other, leading to an overall mean composition and subsequently melting

behaviour. However, the calculations as carried out in this work have shown to give useful

results in predicting deposit formation independent from boiler geometry

25

Chapter 3: Results

3-15

Table 3.3: Summarised results from the TPCE for the deposit prediction (8=1.2)

Fuel Reactive

amount

(g/kg)

Reactive

% of total

ash

Total

fractions

(g/kg)

T15 (/C) Melt

range

(/C)

Max

melt

(%wt)

Total

q

#4: Coal qqq qqq qq qqq qqq qqq 17

#3: Peat qq qqq qq qqq q qq 13

#10: Peat5 qq qqq qq qqq qqq qqq 16

#8: Wood qq qqq qqq q qq q 12

#9,1 2: Wood qqq q qqq qqq qq qqq 15

#13: Wood qq qq qqq qqq q q 12

#6: Bark qq qq qqq qqq qqq qqq 16

#11: Bark10 qq qqq qq qqq qqq qqq 16

#7: For.res. qq q qqq qqq qqq qq 14

#14: For.res qq q qqq qq qqq q 12

#15: For.res qq qq qqq qqq qqq qqq 16

#16 : Cons.res. qq qqq qq q qq qq 12

#1: Salix15 qq q qqq qq q q 10

#2: Salix qq qq qqq qq q q 11

#5: Wheat Straw q q qqq qq q q 9

#23: Wheat straw qq qq qqq qqq q q 12

#22: Switch grass qq q qqq qqq qq q 12

#17: Almo nd hulls20 q q qq qq q q 8

#18: Almo nd shells q q q qqq q q 8

#21: Rice straw q q q q q q 6

Chapter 3: Results

3-16

Figure 3.10. SEM/EDX image from sand bed material taken from

a PFBG firing Lucerne at 15 bar and 800/C

Figure 3.11. SEM/EDX image from sand bed material taken

from a PFBG firing Miscanthus at 5 bar and 900/C

3.2.2 Prediction of bed agglomeration

Bed agglomeration is a quite complex phenomenon that can take place in a fluidised bed

boiler under certain circumstances. It cannot be stressed enough that agglomeration cannot

be explained by looking only at physical phenomena, such as temperature, particle size5

distribution, mixing processes, shear stresses between particles and attrition of particles. Also,

chemical phenomena should

be taken into account.

In case of predicting bed

agglomeration behaviour of the10

sole fuels and the interaction of

the sole fuels with bed material

are of major importance. When

considering bed agglomeration

all ash-forming elements15

present in the fuel should be

taken into account. The

reactive elements could play a

role in the formation of a

coating on the bed material20

(VI), whereas less-reactive

material could become trapped

into a sticky coating. Thus,

w h e n m o d e l l i n g b ed

agglomeration the entire fuel25

should be taken into account

and not only a reactive or less-

reactive fraction as used for

predicting deposit formation. In

this way the thermodynamical30

calculations could simulate the

Chapter 3: Results

3-17

interaction of all ash-forming elements with the bed material.

The composition of “bridges” and “coatings” of the agglomerates as found with SEM/EDX

analysis showed that the main elements involved were the elements Si, K and Ca when a

silica bed was considered. Figures 3.10 and 3.11 show some examples of SEM images of5

agglomerates found. This is in agreement with other studies (see 2.2.4)

Papers IV, V and report VII describe the modelling of the agglomeration tendency of biomass

fuels under reducing conditions (IV, V, VII) and oxidising conditions(VII).

Large differences between combustion and gasification cases could not be noted. In case of10

gasification reduced species are formed which could have a higher volatility and lower first

melting point when compared to the combustion cases (VII). However the trends as

described here count for both combustion and gasification conditions.

The role of reactive ash forming elements in bed agglomeration(VI)15

Neither the chemical interaction between bed material and reactive fine particles nor the less-

reactive coarse fraction alone can explain the formation of agglomerates. Physical

phenomena will take their role as well. However, the formation of a coating on bed material

is a prerequisite for agglomeration and defluidisation. The SEM/EDX analyses such as shown

above can show which elements are involved, but not where they come from.20

In case of a sand bed, silicon found in the coating on the bed particles could originate both

from the fuel and the bed particles. Calcium and potassium could originate from the fuels

reactive or less-reactive particle fractions. The thickness of the coating formed is a function

of chemistry and erosion of the coating due to physical processes. There are two possible25

ways to obtain a typical 10:m thick coating:

1) The coating grows outwards onto the particle, assuming the bed particle act as an

inert carrier for the coating material. In this case all elements in the coating destine

from the fuel and coating formation should occur independent of bed material

2) The coating grows inwards into the particle. In this case reactive elements could react30

with the bed particle.

Chapter 3: Results

3-18

In both cases growth is limited due to either erosion of the coating, diffusion limitation or

both.

By comparing the amount of bed material after controlled combustion experiments as carried

out at ETC, (i.e. with the amount of bed material before the experiment started and the5

amount of fuel fed to the combustor) the total weight of coating present can be calculated

and distributed over the three main coating-forming elements. After this, the distribution of

these elements over the reactive and less-reactive fraction of the ash forming elements could

be combined with the quantity of the elements needed to build up the coating.

10

Up to 60 %wt of the potassium fed to the bed was found in the coatings. This could be

introduced by the potassium present in the reactive fraction alone. The rest of the potassium

could escape the bed as gaseous KCl.

Some 8-30 %wt of the calcium present in the fuel ended in the coatings. This means that up15

to 92 % of the calcium might react to form other components such as CaSO4. Calcium is

divided evenly over the reactive and less-reactive fractions in most biomass fuels. This makes

it difficult to determine which fraction is responsible for the calcium found in the coatings.

As described in section 3.1 calcium is possibly present as included calcite or as oxalate

minerals which are only leached by HCl. Even these calcium minerals could be considered20

reactive when released from the fuel. The calcium leached by water and acetate is supposed

to form submicron particles of for example CaSO4 or Ca3(PO4)2. Such components were

identified in coatings by SEM/EDX analysis. Even the included “reactive” minerals could be

involved in coating formation.

25

The amount of reactive silicon entering the boiler with the fuel is in most cases insufficient

to form a coating. The less-reactive particle fraction represents in most cases a contamination

of sand particles that enter the combustor together with the biomass fuel (or are present in

the skeleton of the fuel) The fuel contaminations could act as alternative bed material.

30

Chapter 3: Results

3-19

Coating formation could take place by reaction between Si, Ca and K. In practice

components containing these three elements will not coincide. There are two possibilities:

1) Silicon reacts with gaseous potassium forming potassium silicates with a first melting

point as low as 750/C. This could be the first sticky layer formed on bed particles,

which catches other small particles released from the fuel such as solid calcium5

components. After this first capture all components present in the coating could

interact forming a sticky coating raising the first melting point to some 800/C.

2) Silicon reacts with calcium first forming calcium silicates with a first melting point of

above 1500/C. This second route is considered unlikely due to the first melting point,

making capture of potassium components by glueing impossible at the low10

temperatures found in FBC. It is found that the first agglomerates occur at some

800/C in all cases studied (VI).

The modelling of bed agglomeration in (P)FBG (IV,V)

Agglomeration tendencies for four fuels, i.e. Salix #24, Miscanthus #26, Reed Canary Grass15

#27 and Lucerne #29 in (P)FBG could be compared to the calculations. Thirteen bed

samples from these tests were embedded in epoxy, cross-sectioned and polished for

SEM/EDX analysis. Experiences from the experiments were compared with the results

obtained from the SEM/EDX analysis and with the TPCE calculations.

20

Table 3.4 gives a summary of the results obtained. In this table the experiments are sorted

fuel wise. The column “presence of molten phases” summarises the results as obtained from

the TPCE calculations. Here the calculated amount of molten phases as %wt of total ash is

presented. As can be seen the amount of molten phases could exceed 100% as a result of

interaction of the fuel with bed material and subsequent melting of bed material. In the25

column “process experiences” experiences from the experiments are divided into three

categories, i.e.:

I) No signs of defluidisation or agglomeration of the bed;

II) Bed disturbances or presence of agglomerates after visual inspection of the bed;

III) Clear defluidisation of the bed. 30

The last column shows whether agglomerates were found with SEM/EDX analysis or not.

Chapter 3: Results

3-20

In six out of 13 cases formation of agglomerates was predicted and found with SEM/ED

analysis. In two cases no or little agglomeration was predicted, nor were agglomerates found

with SEM/EDX analysis. In two cases out of 13 the modelling predicted some degree of

agglomeration while no agglomerates could be detected with SEM/EDX analysis. However,

in these cases agglomerates were detected after visual inspection of the bed.5

Table 3.4: Summarised results from the TPCE calculations, experiments and SEM/EDX

analysis (IV,V)

Fuel P

(bar)

T (/C) bed

material

Presence of

molten

phases

(%wt of ash)

Process

experience

Agglomerates

found w ith

SEM

Salix #2410 5 892 sand 64 II Yes

R.C.G . #27 15 830 sand 11 III Yes

R.C.G . #27 1 900 dolom ite 0,3 I No

Miscanthus #26 15 807 sand 26 III No

Miscanthus #26 5 877 sand 26 III Yes

Miscanthus #2615 1 900 dolom ite - II No

Miscanthus #26 1 850 dolom ite - II Yes

Miscanthus #26 1 900 olivine - II Yes

Lucerne #29 5 875 sand 75 III Yes

Lucerne #29 15 797 sand 105 III Yes

Lucerne #2920 1 850 dolom ite 39 II No

Lucerne #29 1 900 dolom ite 36 II Yes

Lucerne #29 1 900 olivine - II Yes

R.C.G.= Reed Canary Grass

Chapter 3: Results

3-21

Figure 3.12: The formation of alkali components dependent on

the presence of Chlorine, Pottasium ans silicon in the gasifier

The first melting point of a fuel

interacting with bed material

and the amount of melt formed

will together determine whether

bed agglomeration will take5

place or not. TPCE calculations

carried out for the same process

conditions as the experiments

predict a substantial amount of

melt present in almost all cases.10

Exceptions were atmospheric

gasification of Miscanthus in a dolomite or olivine bed at around 900/C, atmospheric

gasification of Reed Canary Grass in a dolomite bed, and gasification of Lucerne in an

olivine bed. In case of gasification of the four fuels in a sand bed a molten silicon-rich melt

will determine the occurrence of bed agglomeration. The sand bed present may interact with15

the potassium present in the fuels, thus forming mainly potassium silicates melting below

800/C. In case of the atmospheric gasification of Miscanthus and Reed Canary Grass in a

dolomite bed the prediction indicates the occurrence of a melt free gap in a temperature

range between the molten salt phase, which occurs due to the presence of dolomite, and the

silicate molten phase which occurs due to the presence of the silica in the fuels. In case of the20

atmospheric gasification of Lucerne and Miscanthus in a magnesium olivine bed in the

prediction indicated the same, i.e. the presence of a melt free gap between the presence of

a molten salt phase due to the presence of an excess of magnesium, and a molten silicate

phase.

25

The modelling results showed that the amounts and types of alkali components formed are

dependent on the presence of components such as silica, calcium and chloride, as

summarized in Figure 3.12. The diagram represents the components, which could be

formed from different fuel types as a function of the chloride/potassium and potassium,

calcium/silica ratios. In cases of low silica but high potassium K2CO3 formed. If chloride was30

abundant as well, KCl formed. High silica contents in the fuel yielded potassium silicates,

Chapter 3: Results

3-22

Figure 3.14: The phases containing potassium under gasification

in presence of an excess of dolomite at 10 bar

Figure 3.13 The phases containing potassium under gasification

cond itions at 1 0 bar (n o bed mate rial pre sent)

whereas high silica and high chloride gave potassium silicate formation combined with the

release of HCl.

Figure 3.13 shows the phases

containing potassium for these

four fuels at 10 bar 700/C and5

900/C, respectively. It should be

noted that the fuels were very

different. Salix and Miscanthus

contained 1-2 %wt ash, whereas

the other fuels contain up to10

8.6%wt ash. At 700/C the major

part of potassium was present as

solid potassium salts or silicates.

When considering Salix or

Lucerne approximately 15% of15

the potassium was present as a salt melt. At 900/C all ash was molten. For the other fuels a

potassium silicate melt was found. Figure 3.13 shows that an increase in temperature when

firing Salix caused the alkali melt that was present at 700/C to volatilise. The rest of the alkali

was present in a molten silicate phase at 900/C. In case of Lucerne, which was low in silicon,

the solid alkali salt phase that was present at 700/C was molten at 900/C.20

Figure 3.14 shows the results for

these fuels when an excess of

dolomite was present as bed

material. The presence of25

dolomite decreased the relative

amount of silicate present

shifting even fuels with a high

silicon content, such as Reed

canary Grass, to the right side of30

Chapter 3: Results

3-23

Figure 3.16: The phases containing potassium under gasification

conditions in presence of an excess of sand at 10 bar

the diagram (Figure 3.12).

Instead of a potassium silicate

melt the major contribution to

the molten phases was now

accounted for by a molten5

potassium salt phase.

Figure 3.15 shows the results

for the fuels when an excess of

sand was present as bed10

material. In this case the

amount of silica available for

reaction with potassium is increased drastically, thereby shifting the fuels to the left side of

the diagram in Figure 3.12. Thus, potassium silicates were preferably formed with a first

melting point below 800/C.15

20

Chapter 4: Conclusions

4-1

4 CONCLUSIONS

4.1 Fuel characterisation

The extended fuel characterisation as used in this work provided useful information on the5

distribution of ash-forming elements in different fuels.

In this study clear differences have been shown in the distribution of the ash-forming

elements in the different fuels. In the geologically older fuels more ash-forming elements

were present as excluded and/or included minerals. In relatively young fuels up to half of the10

amount of ash-forming elements was present in the soluble fraction after leaching with water

and ammonium acetate.

At this stage, although informative, the chemical fractionation method in combination with

SEM/EDX analysis, is very time consuming and expensive. This means that it is an excellent15

research tool, but not available yet as a standard analysis. The method should be simplified

and standardised for biomass fuels.

4.2 Deposit formation

20

The combination of fractionation with TPCE calculation has shown to be more useful than

other traditional methods based on fuel ash analysis in the laboratory for predicting ash

behaviour in FBC.

The definition of reactive components as the sum of elements leached with water and25

ammonium acetate should be used with care when used as a basis for ash deposit

prediction. It was shown in this work that SEM/EDX could give valuable additional

information about the way ash-forming elements are present in the fuel. Hence, a smart use

of the combination of traditional fuel analysis, chemical fractionation and SEM/EDX analysis

could give a solid base for prediction of deposit formation with, for example, TPCE30

calculations.


4-2

This work shows the easily leached elements form the main constituents in the fine fly ash,

and are consequently a reasonable approximation of the fly ash compounds.

Prediction of the presence of a molten phase in the fly ash, together with the reactive

amount, total ash composition, melt range, T15, and maximum amount of melt, gives the5

possibility to rank fuels in order of deposit formation tendencies.

The ranking of the fuels as studied in this work was presented as less-problematic <

problematic was as follows: coal< peat < wood derived fuels < annual crops < agricultural

waste, which very well corresponds to the general practical experiences with these fuels.10

In the nearby future weighing factors should be used to express the relative importance of

the above-mentioned parameters on deposit formation.

A direct comparison between prediction models based on physical phenomena and the15

method as described in this work is impossible.

It can be assumed that thermodynamic equilibrium calculations can be used for reactive ash-

forming compounds such as easily soluble alkali salts. When composing the reactive fraction

containing calcium only from the water and acetate fraction, the amount of reactive calcium20

could be underestimated. Included calcium containing minerals should be included as well

in the reactive fraction and, thus, in the prediction for deposit formation. The same accounts

for silicon. In case silicon is present as soil contamination it should be omitted from the

reactive fraction. However, when present as included minerals, part of the silicon as leached

in the HCl fraction could be accounted for as reactive.25


4-3

4.3 Agglomeration

It is shown in this work that TPCE calculations increase our understanding of alkali

behaviour in (P)FBG/C. Interaction of bed material with alkali components released by the

fuels determines wether a fuel volatilises and interacts with the bed material.5

The calculations as presented in this work can be used as guidelines for predicting bed

agglomeration. A comparison between the TPCE calculations, SEM/EDX analysis of bed

material, lab- and bench-scale experiments showed good agreement.

10

Calculations showed that the presence of an excess of dolomite/calcite decreases the amount

of alkali components in the bed due to an increase in the amount volatilises. An excess of

amount of silicates increases the amount of alkali retained in the bed. This leads to formation

of low melting alkali silicates and subsequent bed agglomeration. At atmospheric pressure

the amount of melt formed could be smaller, when compared to high pressures, thereby15

decreasing the risk for bed agglomeration.

The mechanism leading to bed agglomeration as studied in this work is in agreement with

Öhman (1999) and Öhman et al. (2000). Chemical fractionation results revealed that when

firing woody biomass fuels potassium and calcium present in a bed coating are originating20

from the reactive fraction in the fuel, i.e. leachable with water or ammonium acetate and/

or present as included small minerals as pointed out by SEM/EDX analysis.

25

Chapter 5 References

5-1

5 REFERENCES

1. Anderson J., Anderson L., Power Gen. Europe, Frankfurt, Germany,http://www10.abb.se/carbon/pfbc7.html (1999).

2. ASTM D-1857-68: The fusibility of coal and coke ash, Annual book of ASTMstandards, part 19 (1970)

3. ASTM Standards D3172-89, Standard practice for proximate analysis of coal andcoke, Annual book of ASTM standards (1984).

4. ASTM standards D 3176-89e1 Standard practise for ultimate analysis of coal andcoke ash, Annual book of ASTM standards (1984).

5. ASTM D 271-68 : Laboratory sampling and analysis of coke and coal, Annualbook of ASTM standards (1984)

6. ASTM standard E1755-95 Standard test method for ash in biomass, Annual bookof ASTM standards (1984).

7. Babat D.W., Kulkunari S.V., Bhandarkar V.P., in proceedings of the 14th

International Conference on Fluidised bed Combustion, Preto A. eds.,Vancouver, ASME, New York, (1997), pp 165-174

8. Backman R., Hupa M., Uppstu E., Tappi J., (1987), 70(6), pp 123-

9. Backman R., Sodium and sulphur chemistry in combustion gases, Academicdissertation, Åbo Akademi university, Turku, Finland (1989)

10. Baxter L.L., Biomass and Bioenergy, (1993), 4(2), pp 85-102

11. Baxter L.L., A Task 2 . Pollutant emission and deposit formation duringcombustion of biomass fuels, Livermore, (CA), (1994)

12. Baxter L., Miles T., Miles jr. T., Jenkins B., Milne T. Dayton D., bryers R., OdenL., Fuel Processing and Technology (1998), 54, pp 47-48

13. Benson S.A., Holm P.L., Ind. Chem. Eng.Prod. Res.Dev., 24, (1985), pp 145-149.

14. Borman G.L., Lagland K.W., “Combustion Engineering”, WCB McGrawhill,(1998), §2.3.


5-2

15. Bryers R., Prog. Energy Combust. Sci, (1996), 22, pp 29-120

16. Clarke L.B., “Application for coal-use residues”, IEA Coal Research, UnitedKingdom, (1992), Chapter 2

17. Clean Coal Technology, “Options for the future”, OECD/IEA (1993).

18. Communication from the Commision, “ Energy for the future: renewable energy

sources of energy”, COM(97)599, (1997).

19. Davies O.L., Goldsmith P.L, Statistical methods in research and production” , 4th

edition, Longman Scientific and Science, Essex, England, (1988), Chapter 4

20. Dayton D., Belle-Oudry D., Energy and Fuels, (1999), 13, pp 1203-1211

21. DIN 51718 Bestimmung des wassergehaltes, Deutsche Normen, (1978).

22. DIN 51719 Bestimmung des Aschegehaltes, Deutsche Normen, (1978).

23. DIN 51730: Testing of solid fuels, determination of ash melting, Deutsche

Normen (1984)

24. Engström F., Gasification Technology Conference, San Fransisco, CA, (1999).

25. Ergudenler A., Ghaly A., Biomass and Bioenergy, (1993), 4(2), pp 135-147

26. Faaij A., Energy from biomass and waste, Academic dissertation, University of

Utrecht, Utrecht, The Netherlands, (1997).

27. Flagan R.C., Sarofim A.F., Prog. Energy Combust. Sci., (1984), 10, pp 171-175.

28. Frandsen F., in Chemistry in combustion processes, Part IIb, Turku, Finland


5-3

(2000)

29. Hansen P., Andersen K., Wieck-Hansen K., Overgaard P., Rasmussen I.,

Frandsen F., Hansen L., Dam-Johansen K., Fuel Processing Technology, (1998),

54, pp 207-225

30. Hansen P., “Deposit formation in coal/biomass fired CFB under load variations”,

ELSAM R&D project no 348, (1997)

31. Hannes J., “Mathematical modelling of circulating fluidised beds”, Academic

dissertation Delft University of Technology, Delft,(1996) Introduction Ch.5.

32. Hupa, M., Backman, R: Slagging and fouling during combined burning of bark

with oil, coal, gas, or peat, in Fouling of heat exchanger surfaces (Ed: R Bryers),

United Engineering Trustees, New York, NY, (1983), pp 419-432.

33. Jensen P.A., Stenholm M., Hald P., Energy and Fuels, (1997), 11, pp 1048-1055

34. Jensen P., Frandsen F., Dam-Johansen K., Sander B., Energy and Fuels, (2000),

14, pp 1280-1285

35. Juniper, L. Combustion News, Australian Combustion Technology Centre, (1995)

pp 1-4.

36. Kendall A., McDonald A, Williams A., “The power of biomass”,

http://ci.mond.org/9709/970913.html, (1997).

37. Laursen K., Frandsen F., Larsen O., Energy and Fuels, (1998), 12, pp 429-442

38. Liliedahl T., Kusar H., Rosén E., Sjöström K., in the proceedings of the 2nd Ole


5-4

Lindström Symposium, Royal Institute of Technology, Sweden (1999), pp 88-91

39. Lin W., Gitte K., kim D.J., Esther M., Bank L., in the proceedings of the 14th

International conference on Fluidised bed Combustion, Preto A. eds, Vancouver,

ASME, New York, (1997), pp 831-837

40. Lind T., “ Ash formation in circulating fluidised bed combustion of coal and solid

biomass”, Espoo, Finland, VTT Chemical technology, 1999, Academic

dissertation

41. Lind T., Kauppinen E., Sfiris g., Nilsson K., Maenhaut W., Energy &Fuels,

(1999a), 13(2), pp 379-89

42. Maniatis k., Bridgewater A., Buekens A., in the proceedings of an international

conference on pyrolysis and gasification, Ferrero G., Maniatis K., Bueskes A.,

Bridgewater A., (1993), pp 274-281

43. Manzoori A.R., “Role of inorganic matter in agglomeration and defluidisation

during the circulating fluidised bed combustion”, PhD thesis, University of

Adelaide, Australia, (1990)

44. Miles T.R., Miles T.R., Baxter L.L., Bryers R.W., Jenkins B.M., Oden L.L.,“Alkali

deposits, Found in biomass power plants”, summary report, NREL, (1995a).


deposits, Found in biomass power plants”,a preliminary investigation of their

extend and nature, NREL, (1995b).


deposits, Found in biomass power plants”, the behaviour of inorganic material in


5-5

biomass- fired power boilers- fiels and laboratory experiences, NREL, (1995c).

47. Miles T, Miles T., Baxter L., Bryers R., Jenkins B., Oden L., Biomass and

Bioenergy (1996), 10(2-3), pp 125-138

48. Moilanen A., Nieminen M., Sipilä K., Kurkela E., in the proceedings of the 9th

Europan Bioenergy Conference & 1st European Energy from Biomass technology

Exhibition, Denmark (1996).

49. Moilanen A., Kurkela E., Am. Chem. Soc. Div. Of Fuel Chem., (1995), 40(3), pp

668-693

50. Nieminen J., Palonen J., Kivelä M., VGB Towertech., (1999), 10, 69-74.

51. Nordin, A., Öhman, M., Skrifvars, B. J., and Hupa, M. Proc. of the Eng. Found.

Ash Conf., Waterville Valley, NH, (16-22, July 1995).

52. Phyllis, Netherlands Energy Research Foundation, ECN, “Database for biomass

and waste”, http://www.ecn.nl/phyllis/, (Dec 2000)

53. Reid W.T., Prog. Energy Comb. Sci., (1984), 10, pp 159-175

54. Rensfelt E.K.W., Int. Conf. on Gasification and Pyrolysis of Biomass, Stuttgart,

Germany, (1997).

55. Salmenoja K., Mäkelä K., Hupa M., Backman R., Journal of the Institute of

energy, (1996), 69, pp 155-162

56. Salmenoja K., Field and Laboratory studies on Chlorine induced superheater

corrosion in boilers fired with biofuels, Academic dissertation, Åbo Akademi


5-6

University, Turku Finland, (2000).

57. Salo K., Horvath A., Patel J., Pressurised Gasification of Biomass, Int. Gas

Turbine and Aeroengine Congress and Exhibition, Stockholm, Sweden, June

(1998).

58. Skrifvars B-J., Sintering tendency of different fuel ashes in combustion and

gasification conditions, Academic dissertation, Åbo Akdemi University, Turku

Finland (1994).

59. Skrifvars B-J., “Ash Chemistry and sintering-verification of the mechanisms, in

LIEKKI 2 annual Book, eds. Hupa M, Mattinlinna J., (1997a), Åbo Akademi

University, Turku, Finland,

60. Skrifvars B-J., Sfiris G., Backman R., Widegren K., Hupa M., Energy and Fuels

(1997b), 11, pp 843-848

61. Skrifvars B-J., Lauren T., Backman R., Hupa M., Binderup-Hansen P., in the

proceedings of The Engineering foundation Conference on the impact of mineral

impurities in solid fuel combustion, November HI, (1997c)

62. Skrifvars B-J., Hupa M., Backman R., in the proceedings of the 1st south east

European symposium on Fluidised beds in energy production, chemical and

process engineering and ecology, Macedonia, (1997d), pp 65-84

63. Skrifvars B-J., Blomquist J-P., Hupa M., Backman R., in the proceedings of the

15th Annual International Pittsburgh Coal Conference,, Pittsburgh, (PA), USA,

(1998)

64. Skrifvars B-J., Backman R., Hupa M., Sfiris G., Albyhammer T., Lyngfelt A., Fuel


5-7

(1998a),77(1/2),pp 65-70

65. Skrifvars, B-J., Backman, R., Hupa, M: Ash chemistry and behaviour in advanced

co-combustion, in Final report of the EU/JOULE 3 project Operational problems,

trace element emissions and by-product management for industrial biomass co-

combustion (JOF3-CT95-0010), (Ed:s H. Spliethoff, K. R G Hein). Brussels

(1999).

66. Skrifvars B-J., Öhman M., Nordin A., Hupa M., Energy and Fuels, (1999a),

13(2), pp 359-363

67. Skrifvars B-J., Laurén T., Backman R., Hupa M., in Impacts of mineral impurities

in solid fuel combustion, Gupta R., eds., Kluwer, New York, (1999), pp 525-539

68. Skopurska N., Couch N.,Coal characterization for predicting ash deposition; an

international perspective, presented at the Engineering Foundation Conference on

the impact of ash deposition in coal fired plants, Birmingham, UK, (1993)

69. Ståhl K., Neergaard M., VGB Kraftwerk, (1996), 4, pp 327-330.

70. Steenari B., Lindqvist O., Biomass and Bioenergy (1998), 14(1), pp 67-76

71. Steenari B., Lindqvist O., Fuel, (1999), 78, pp 479-488

72. Turn S., Kinoshita C., Ishimura D., Zhou J., Fuel (1998), 77(3), pp 135-146

73. Valmari T., Kauppinen E., Kurkela J., Jokiniemi J., Sfiris G., Revitzer H., J.

Aerosol. Sci., (1998), 9(4), pp 445-459

74. Valmari T, Lind T., Kauppinen E., Energy Fuels, (1999a), 13(2),pp 379-389


5-8

75. Valmari T, Lind T., Kauppinen E., Energy Fuels, (1999b), 13(2),pp 390-395

76. Valmari T., “Potassium behaviour during combustion of wood in circulating

fluidised bed power plants”, Academic dissertation, VTT Publications, Finland

(2000)

77. Wall, T. F., Creelman, R. A., Gupta, S., Coin, C., and Lowe, Proc. of the Eng.

Found. Ash Conf., Waterville Valley, NH, (July 1995).

78. Werther J., Saenger M., Hartge E-U., Ogada T., Siagi Z., Progress in Energy and

Combustion Science, (2000), 26, pp 1-27

79. Winegartner E.C., Coal fouling and slagging parameters, ASME Special

Publications, (1974).

80. Yu C., Zhang W., in the Proceedings of the International Conference of Progress

in Thermochemical Biomass Conversion, (2000), Austria

81. Öhman M., Nordin A., Energy and Fuels, (1998), 12, pp 90-94

82. Öhman M., “Experimental studies on bed agglomeration during fluidised bed

combustion of biomass fuels”, Academic Dissertation, Umeå, Sweden (1999)

83. Öhman M., Nordin A., Skrifvars B-J., Backman R., Hupa M., Energy and Fuels,

(2000), 14, pp 169-178

84. Öhman M., Nordin A.,, Energy and Fuels, (2000a), 14, pp 618-624

RECENT REPORTS FROM THE ÅBO AKADEMI PROCESS CHEMISTRY GROUP, COMBUSTION AND MATERIALS CHEMISTRY:

00-1 K. Salmenoja Field and Laboratory Studies on Chlorine-induced Superheater Corrosion in Boilers Fired with Biofuels

00-2 T. Norström Approaches for Prediction of NOx Emissions Using Computational Fluid Dynamics

00-3 L. Fröberg (Ed.) Proceedings of the 40th IEA FBC Meeting

00-4 T. Bergenwall, J. Konttinen, S. Kallio, P. Kilpinen

Oxidation of a Single Char Particle – Development and Testing of a Robust Numerical Solution Procedure

00-5 H. Ylänen Bone Ingrowth into Porous Bodies Made by Sintering Bioactive Glass Microspheres

00-6 B. O. Skrifvars Chemical Equilibrium Analysis in the Study of Corrosion

00-7 A. Brink, P. Kilpinen Modeling gas phase nitrogen chemistry at fuel rich conditions — An extension to the BKH-model for the range 0.1<λ<o.5 and 1200 K<T<1350 K

00-8 K. Sandelin, R. Backman Equilibrium Distribution of Arsenic, Chromium, and Copper when Burning Impregnated Wood

00-9 A. Brink, J. Keihäs, M. Hupa Specifying boundary conditons for CFD modeling of a slab reheating furnace

00-10 A. Brink, P. Kilpinen A simplified kinetic rate expression for describing the oxidation of fuel-N in biomass combustion

00-11 E. Nordström. H. Ylänen, M. Hupa, A. Itälä, H. Aro

Porous, Surface Pre-Treated Bioactive Glass for Initial Fixation of Hip Joint Implant, Final Report of the Tekes Project 40296/99

00-12 N. DeMartini Ammonia Formation Behavior in Green Liquor at 90°C

00-13 N. DeMartini The Effect of Oxidation on Black Liquor Nitrogen

RECENT REPORTS FROM THE ÅBO AKADEMI PROCESS CHEMISTRY GROUP,

COMBUSTION AND MATERIALS CHEMISTRY:

01-01 M. Zevenhoven, T. Laurén, B-J. Skrifvars, R. Backman

The Chemistry and Melting Behavior of Fly Ash Deposits in Co-combustion of Bark, Peat and Forest Residue

01-02 M. Zevenhoven The Prediction of Deposit Formation in Combustion and Gasification of Biomass Fuels Fractionation and Thermodynamic Multi-phase Multi-component Equilibrium (TCPE) Calculations

01-03 Maria F.J. Zevenhoven-Onderwater


ISSN 1457-7895 ISBN 952-12-0813-9

Åbo Akademis tryckeri Åbo, Finland, 2001

REPORT 01-03 - Åbo Akademiusers.abo.fi/mzevenho/portfolj/publikationer/PhD MZ.pdf · Process Chemistry Group REPORT 01-03 ... in this thesis was done within the Process Chemistry

Documents