Towards better prediction of ash related problems in biomass combustion via improved fuel analysis

Future biomass combustion systems should be fuel flexible. To avoid operational problems such as fouling, slagging, fluidized bed agglomeration, and corrosion, the relation between ash behaviour and fuel composition needs to be known. Ash-related problems are strongly coupled to phase transitions. Molten or partly molten ash is sticky and cause furnace-wall slagging, heat exchanger corrosion, or severe bed agglomeration. Vaporisation of ash species leads to fouling in the boiler when ash condenses on heat exchanger tubes. The extent of these phase transitions can be determined based on the chemical composition of the ash. Our search for better prediction tools includes the following four steps. Conventional fuel analyses give the fuel element composition. The bulk fuel element concentrations determine the amount of ash formed and its average composition (1). Element analysis of fuel fractions (i.e. different particle sizes, seams, plant tissues or other distinct fuel components) gives information on the composition of individual ash particles and helps identifying problematic fuel fractions (2). Global equilibrium calculations allow quantification of the problematic ash fractions (3). Kinetic restrictions to the equilibrium (4) can be considered by applying advanced fuel analyses, such as chemical fraction analysis (CFA) or computer-controlled scanning electron microscopy (CC-SEM). Advanced fuel analyses determine the ash-forming matter in the fuel: the mineral, salts and organically associated ash-forming elements. In this paper, these predictions methods are applied on a forest residue fuel fired in Scandinavia.

T o w a r d s b e tte r p r e d ic tio n o f a s h r e la te d p r o b le m s in b io m a s s c o m b u s tio n v ia im p ro v e d fuel an a ly sis Johan Werkelin, Mikko Hupa Abo Akademi University, Process Chemistry Centre, Piispankatu 8, FI-20500 Turku, Finland ABSTRACT Future biomass combustion systems should be fuel flexible. To avoid operational problems such as fouling, slagging, fluidized bed agglomeration, and corrosion, the relation between ash behaviour and fuel composition needs to be known. Ash-related problems are strongly coupled to phase transitions. Molten or partly molten ash is sticky and cause furnace-wall slagging, heat exchanger corrosion, or severe bed agglomeration. Vaporisation of ash species leads to fouling in the boiler when ash condenses on heat exchanger tubes. The extent o f these phase transitions can be determined based on the chemical composition of the ash. Our search for better prediction tools includes the following four steps. Conventional fuel analyses give the fuel element composition. The bulk fuel element concentrations determine the amount of ash formed and its average composition (1). Element analysis o f fuel fractions (i.e. different particle sizes, seams, plant tissues or other distinct fuel components) gives information on the composition o f individual ash particles and helps identifying problematic fuel fractions (2). Global equilibrium calculations allow quantification o f the problematic ash fractions (3). Kinetic restrictions to the equilibrium (4) can be considered by applying advanced fuel analyses, such as chemical fraction analysis (CFA) or computer-controlled scanning electron microscopy (CC-SEM). Advanced fuel analyses determine the ash-forming matter in the fuel: the mineral, salts and organically associated ash-forming elements. In this paper, these predictions methods are applied on a forest residue fuel fired in Scandinavia. 1 INTRODUCTION The combustion of industrial, municipal and agricultural by-products (waste fractions) will reduce the greenhouse gas emissions: less organic matter in landfills reduces the methane emissions and the recovered energy reduces the carbon dioxide emissions from fossil fuel combustion. There are several examples o f this development in Europe. Figure 1 shows the fuel energy input to the district heating network in the municipality o f Norrkoping, Sweden. Due to high CO2-taxes in Sweden over the last decades, the use o f fossil fuels has been replaced by forest residue (tree tops and branches), sorted wood waste and municipal solid waste. However, combustion systems that fire these fuels must be robust and fuel flexible; the price, quality and availability o f these materials is quite variable. 2000 1800 1600 1400 1200 1000 800 600 400 200 0 GWh/year Oil Bitum ou ;oal Scrap ires res res idue as ■ ■ S3 inicipal id /as te 91 92 93 94 95 96 97 98 99 00 01 02 03 04 05 06 Figure 1. Fuel energy input to the district heating network of Norrkoping, Sweden (approx. 90000 inhabitants) over the last two decades Solid fuel combustion renders ash-related problems in the combustion device. The ash causes fouling and corrosion in the convection pass and slagging or bed agglomeration in the furnace. Ash-related problems continue to be the most important cause o f unscheduled shut­ downs in large-scale combustion systems [1, 2], and the costs involved are enormous. To meet the challenge o f utilizing all potential biomass and waste streams in old and new energy conversion systems, a sound scientific model for ash behaviour is needed that can be applied universally. Ash-related problems are strongly coupled to the presence o f liquid and/or gaseous phases in the ash. The fouling mechanism involves the physical transport o f ash to the heat exchanger surfaces [3, 4]. A molten or partly molten ash causes slagging and fouling by inertial impaction; it has been suggested that 15 wt-% molten phase is enough for the particle to become "sticky" [5, 6] although at high dust loads in the convective pass, also solid ash may cause severe fouling by rapid sintering [7]. Gaseous ash causes fouling upon condensation on heat exchanger surfaces [8, 9]. The mechanisms o f bed agglomeration include both molten and gaseous phases [10]. Corrosion is caused by molten salts and gaseous chlorides [11, 12]. The extent o f ash-related problems in the combustion process should therefore correlate with the presence o f sticky ash particles and the amount o f volatilized ash formed in the furnace. The aim o f this work is to increase the knowledge o f how to introduce new types o f fuels and avoid severe technical problems. This knowledge will allow better use o f biomass and waste for energy production and decrease the global emissions o f greenhouse gases. The objective of this paper is to show the increased value o f improved fuel analyses and suggest how it can be interpreted for better prediction o f ash behaviour in combustion systems. This will be demonstrated by a detailed study o f a forest residue fuel, the tree tops and branches collected after lumbering. Fuel analysis includes measurement o f heat content, density, and particle size distribution for accurate furnace design, proximate and ultimate analysis for prediction o f burning characteristics and flue gas composition, and element analysis for prediction o f the ash behaviour. Special focus is on the problematic ash-forming elements, such as alkali metals (Na, and K) and non-metals (Si, P, S, and Cl). Improved fuel analysis includes (1) a systematic division o f the fuel into homogeneous fractions and 2) analyses of the ash-forming elements and ash-forming matter in these fractions. The ash-forming matter (AFM) is the chemical form o f the ash-forming elements in the fuel. There are four categories o f AFM in solid fuels: water-soluble salts, organically associated elements, included minerals, and excluded minerals. It can be studied by computercontrolled SEM (CC-SEM) or chemical fraction analysis (CFA), which is a step-wise leaching procedure in which the fuel is leached in increasingly aggressive solvents. 2 METHOD This paper uses fuel analysis data from earlier publications to predict the ash behaviour in combustion o f forest residue fuels. Table 1 shows the bulk element composition o f six forest residue fuel samples. The data are selected from the Abo Akademi Finland database for solid fuel characterization [13]. The samples are chosen randomly and are placed in the order of decreasing Si content. The Si and Al contents in these samples correlates, which indicates that the two elements have the same origin, i.e. as aluminium silicates in excluded minerals. Forest residue Forest residue Forest residue Forest residue Forest residue Forest residue #29 #77 #27 #84 #73 #97 Al Mg Mn K P Si Fe Ca Na 7030 1050 495 4577 546 408 55 2270 6700 1200 520 5800 690 440 59 2500 4290 598 281 4240 502 270 37 2000 3000 480 260 4500 590 290 45 1900 2300 350 200 6500 850 440 40 2600 1200 370 720 5400 710 300 170 1700 S 631 540 518 480 700 490 Table 1. Bulk element composition o f six fuel samples o f forest residue Cl 437 430 335 380 570 550 109 260 1830 110 140 140 Table 2 shows the element composition of different types of biomass present in a forest residue fuel: the wood and bark o f branches, whole twigs (thin branches), needles, leaves and shoots from four trees o f different species. The data are selected from a paper presenting the element content in four trees o f summer harvest [14]. Table 3 shows the dry weight o f these fractions in each tree crown o f the four species [14]. Al Si 59 2 Spruce Wood 123 4 Pine Wood Birch Wood 77 2 Aspen Wood 63 6 171 98 Spruce Bark 60 908 Pine Bark Birch Bark 114 19 Aspen Bark 94 20 982 221 Spruce Twigs 312 332 Pine Twigs Birch Twigs 69 23 Aspen Twigs 188 35 6640 83 Spruce Needles 549 374 Pine Needles 300 27 Spruce Shoots 747 331 Pine Shoots 318 40 Birch Leaves 133 20 Aspen Leaves Table 2. Element composition of Mn K P Ca Na S Cl Mg 4 724 95 98 6 215 4 50 51 8 641 189 81 15 407 41 94 85 6 636 92 102 4 315 49 82 40 5 998 286 49 15 1370 191 125 35 39 8350 865 714 26 2030 452 367 260 52 6350 874 343 22 3180 1260 311 147 24 7860 323 534 14 1710 428 329 149 27 11700 1370 256 12 4730 663 520 40 167 4320 909 496 97 3560 1080 776 317 73 5300 715 244 40 3040 848 587 200 40 4730 448 354 43 3020 820 493 120 35 10400 644 183 19 5870 708 479 87 45 8030 1050 1390 48 4270 1540 704 504 40 4140 804 839 28 4770 1270 845 407 43 1670 907 245 13 14600 3830 1320 1090 113 2370 1020 193 36 8790 2590 1250 538 83 9120 2030 1600 32 9420 3140 1690 181 56 9800 2940 667 9 24000 5140 2560 511 18 samples of plant tissues of four trees of different species Fe Spruce Pine Wood o f branches (> 0.5 cm) 31 kg 22 kg Bark o f branches (> 0.5 cm) 10 kg 5 kg Twigs (< 0.5 cm) 16 kg 23 kg Perennial foliage (needles) 20 kg 24 kg Juvenile foliage (shoots or leaves) 3 kg 5 kg 80 kg 79 kg Whole tree crown Table 3. Mass o f each fraction o f biomass in the Birch Aspen 22 kg 13 kg 7 kg 6 kg 8 kg 11 kg 4 kg 3 kg 43 kg 31 kg four trees Figure 2. The first step in the prediction tool: bulk fuel composition (1) Figure 2 shows the process model o f the first step in the prediction tool: conventional fuel sampling together with element analyses (typically Si, Al, Fe, Ca, Mg, Mn, Na, K, P, S, and Cl). It allows predicting the amount of ash formed in combustion and its average composition. The chemical form o f the ash is assumed as the elements oxides: SiO2, Al2O3, Fe2O3, CaO. MgO, MnO, Na2O, K2O, P2O5, SO3, and Cl. This can be calculated only via stoichiometry. The step concerns only the bulk fuel element composition. The bulk, however, is the average composition of all the components and their ash-forming matter present in the fuel, also the excluded minerals such as sand, dust and soil. 2 i- ►T T - H q -" 3 -* - Figure 3. The second step o f the prediction tool: division into fractions (2) The chemical composition o f individual ash particles in combustion can be predicted by element analysis o f selected fuel fractions, i.e., different fuel particles, i.e., plant tissues or other distinct fuel components. Such analysis also helps identifying problematic fuel fractions, i.e. fuel fractions having high concentrations o f problematic ash-forming elements. The fuel can be improved by removing such fractions by fuel pre-treatment. Figure 4. The third step o f the prediction tool: equilibrium for a) bulk and b) fractions (T = temperature and p = pressure) (3) Global equilibrium calculations allow quantification o f the problematic ash fractions. The commercial software package FactSageTM5.5 [15] was used to model a combustion system operating under atmospheric conditions with excess air (5 % O2 in wet flue gas) at 800 - 1300 °C. It calculates the equilibrium composition o f the system by minimizing the Gibbs free energy for more than 200 gaseous species, two immiscible liquid phases (melts), ten solid solutions (ss), and more than 300 pure solid compounds. Thermodynamic data were taken from the FACT and SGTE databases (FACT was chosen in cases o f parallel data). Reduced species and solid phases with structural water were omitted in order not to exceed the maximum allowable output o f phases. 4 Figure 5. The fourth step o f the prediction tool: kinetic restrictions to the equilibrium (T = temperature, p = pressure, and t = time) (4) Kinetic restrictions to the equilibrium can be considered only when knowing the chemical form o f the ash-forming matter in the fuel. The fourth step takes into account the time needed for the transformations o f the ash-forming matter into ash during combustion. Some o f the equilibrium phases might not be present because o f too short residence time. Table 4 shows the ash-forming matter present in forest residue fuels. The data is collected from a submitted article series to the Fuel journal [16-18]. The ash-forming matter has been determined and quantified using chemical fraction analysis (CFA), a step-wise leaching procedure in water, ammonium acetate and hydrochloric acid. This was combined successfully with ion chromatography (IC) [19] and methylene blue sorption (MB) [20, 21] to determine the anionic matter of the biomass. The ash-forming matter has been determined by the use of CFA for the determination and quantification of various modes of ash-forming elements in woody biomass fuels. Water-soluble salts Potassium hydrogen phosphate K2HPO4 and KH2PO4 (3/4 o f all P) Potassium chloride KCl (all Cl) Potassium sulphate K2SO4 (1/4 o f all S) Organically associated elements K+, Na+, Mg2+, Mn2+, Ca2+(> 50%), Al3+ and Fe3+ Ion-exchangeable metal ions Covalently bonded non-metals S (3/4 o f all S) and P (1/4 o f all P) Included minerals Calcium oxalate CaC2O4 2H2O (2/3 o f all Ca in bark, twigs and foliage) SiO2 (H2O)x (6000 - 10000 mg/kg D.S. in spruce needles) Opal Excluded minerals from rocks, sand, clay, earth etc. Aluminum silicates (Al2O3)x(SiO2)y also with Fe 2O3, CaO, K2O, and Na2O Table 4. Types of ash-forming matter in forest residue fuels and their occurence 3 RESULTS Figure 1 shows the results of global chemical equilibrium calculations for the six fuel samples o f forest residue, which correspond to the prediction tool 3a. The weight percent o f melt in the equilibrium ash is plotted for discrete temperatures ranging from 800 °C to 1300 °C. These results however, correspond to a state of the process assuming complete mixing and complete reaction o f all the ash-forming matter in the fuel. Then, all ash particles would have the same chemical composition: the ash particles o f fuel #29 are sticky in the whole temperature range, but the ash particles o f fuel #73 are always non-sticky. The melt is a slag that consists largely o f SiO2. The four samples with the highest Si content (see Table 1) are almost completely molten at 1300 °C. ash of six forest residue samples (dashed line is stickiness criteria: 15 wt-% melt) The state shown in Figure 5 does not represent any real process for combustion of forest residue. Only very homogeneous fuels could be modelled in this way. For the forest residue, it assumes that all the Si from the excluded minerals has mixed and reacted completely with the ash-forming matter in the biomass. This is hardly the case since the particles are suspended in the flue gas and meet physically only via collisions during the second-long residence time in the furnace. Figure 6 - 9 show a more likely scenario, where instead the chemical equilibrium o f distinct fuel particles has been calculated. The weight percent o f melt in the ash of 18 different plant tissues are plotted for discrete temperatures ranging from 800 °C to 1300 °C (missing data where the equilibrium calculation failed). The fuel particles consist o f the branch wood or the branch bark, twigs or different types foliage o f the four wood species. wt-% melt 60 % n spruce wood A spruce bark -X - spruce twigs 50 % - 40 % spruce needles spruce shoots wt-% melt 60 % 50 % 40 % 30 % 30 % 20 % 20 % 10 % 10 % ht- 0% 0% 800 □ pine wood A pine bark -X - pine twigs pine needles - pine shoots 900 1000 1100 1200 1300 800 900 1000 1100 1200 1300 Figure 6. Percent melt in the ash o f spruce Figure 7. Percent melt in the ash of pine Figure 8. Percent melt in the ash o f birch Figure 9. Percent melt in the ash of aspen The ash o f foliage (needles, shoots and leaves) contains the largest portions o f melt, and their ash is sticky at most temperatures. Also the ash o f twigs (thin branches) is sticky at some temperatures. The ash particles o f pine bark and spruce needles are sticky over the whole temperature range. A melt in the equilibrium ash has different composition at different temperatures: a melt o f approximately 50% potassium carbonate and 50% potassium sulphate is stable around 900 °C, but is not stable at higher temperatures due to the decomposition of the carbonate; a melt o f potassium sulphate is stable at 1100 °C, but it decomposes also at higher temperatures. Spruce needles contain a silica-rich melt o f 2/3 SiO2 and 1/3 K2O. Stick y Ash (wt -% o f dry fuel) 800°C 900°C 1000°C 1100°C 1200°C 1300°C 1,0 % 1,0 % 1,3 % 1,0 % 0,9 % 0,9 % Spruce residue 0,7 % 0,6 % 0,6 % 0,4 % 0,4 % Pine residue 1,1 % 0,0 % 0,4 % 0,0 % 0,6 % 0,0 % 0,0 % Birch residue 0,0 % 1,8 % 0,5 % 0,4 % 0,4 % 0,0 % Aspen residue Table 5. Amount of sticky ash at equilibrium as weight percent o f dry fuel Table 5 shows the amount of sticky ash per kilogram dry forest residue fuel of the four wood species. It has been calculated by assuming that all the ash of a certain tree part is sticky if the amount o f melt exceeds 15% and by calculating how much this is per kilogram o f fuel. Sticky ash may constitute as much as two percent of the dry weight of the forest residue fuel. Total as] i content (wt-% of iry fuel) 800°C 900°C 1000°C 1100°C 1200°C 1300°C 1,8 % 1,8 % 1,7 % 1,6 % 1,5 % 1,5 % Spruce residue 1,5 % 1,4 % 1,3 % 1,2 % 1,0 % 1,0 % Pine residue 1,2 % 1,0 % 0,9 % 0,9 % 0,8 % 0,8 % Birch residue 2,4 % 2,0 % 1,7 % 1,6 % 1,4 % 1,4 % Aspen residue Table 6. Total equilibrium ash content at different temperatures Table 6 shows how much ash is formed in total from the forest residue of the four species. The total amount o f ash decreases with temperature due to volatilisation o f ash components. The common phases in the equilibrium ash including Ca are wollastonite, CaSiO3, hydroxyapatite, Ca5(PO4)3OH, and lime, CaO. The Ca in lime is often in solid solution with other divalent metals such as Mg and Mn. The common phases including K in the equilibrium ash are arcanite, K2SO4, potassium carbonate, K2CO3, and potassium orthophosphate (K3PO4), or a melt of these compounds. Comparing Table 5 and Table 6 gives the fraction o f sticky ash o f the total amount o f ash. The highest fraction of sticky ash comes from the combustion of aspen forest residue at 900 °C, and pine forest residue at 1100 °C where more than 90 % of the ash is sticky. Table 7 shows the amount o f volatilized ash at equilibrium during combustion o f forest residue o f the four species. It has been calculated for each fuel fraction separately and weighed together as an average for the whole tree. The juvenile foliage o f all four species (shoot and leaves) renders the highest amount of vaporized ash at all temperatures: from some 0.1 % o f the dry fuel at 800 °C to well over 1.0 % o f the dry fuel at 1300 °C. Aspen leaves renders even more: 3.5 % of its dry weight is vaporized ash at 1300 °C. The wood, bark and twigs render much less vaporized ash in combustion: 0.01 - 0.05 % at 800 °C and 0.1 - 0.5 % at 1300 °C. The volatilized ash species present at equilibrium are mainly KCl(g), KOH(g), and K2SO4(g). Volatili zed Ash (wt-% o f c ry fuel) 800°C 900°C 1000°C 1100°C 1200°C 1300°C 0,1 % 0,1 % 0,2 % 0,2 % 0,3 % 0,3 % Spruce residue 0,1 % 0,2 % 0,3 % 0,4 % 0,5 % 0,5 % Pine residue 0,0 % 0,1 % 0,2 % 0,3 % 0,3 % 0,3 % Birch residue 0,1 % 0,3 % 0,6 % 0,7 % 0,8 % 0,8 % Aspen residue Table 7. Amount of volati ized ash at equilibrium (800 - 1300 °C, 1 atm) 4 DISCUSSION Both vaporised ash and sticky ash particles are harmful to the combustion device. The equilibrium results show that less than one percent o f the fuel's weight is vaporised ash at the studied temperatures. The amount o f sticky ash is around one percent o f the fuel's weight. This quantification is based on the first three steps in the prediction tool suggested here, which assumes chemical equilibrium for individual fuel particles. The fourth step discusses kinetic restrictions to this equilibrium. Previous work on ash predictions excludes the less reactive ash-forming matter from the equilibrium calculations to obtain better predictions [22-24]. It has proved successful to combine equilibrium calculations with CFA, a fuel analysis using step-wise leaching in water, ammonium acetate, and hydrochloric acid. The excluded minerals, like sand and soil, are mainly acid-soluble or non-soluble, and these fractions are omitted in the calculations. This prediction tool also omits the excluded minerals by calculating the equilibrium for the clean fuel particles, and it accounts for the inhomogeneous character of the fuels. Fuel particles may contain less reactive or slowly reacting ash-forming matter. If these do not react, it will shift the equilibrium composition. This is considered with the fourth step of this prediction tool. The two main elements in the biomass ash are Ca and K. The kinetic consideration here will concentrate on these two elements; in which chemical form they are present in the biomass, and in which chemical form they occur in the equilibrium ash. The reactions involved coming from the first form to the last form may have kinetic restrictions. Some o f the potassium in the biomass is organically associated ions (K+). The remaining fraction is water-soluble salts of mainly K2HPO4/KH 2PO4, some KCl, and small amounts of K2SO4. The two latter compounds are the dominating K-containing compounds in the equilibrium ash or vapours. The sulphur is mainly organically associated in the biomass and reacts first to gaseous sulphur oxides during combustion. The potassium sulphate also forms via reactions with gaseous sulphur oxides. This reaction competes with other reactions of sulphur oxides and metal oxides. It is possible that the formation o f K2SO4 is kinetically restricted, and that the equilibrium modelling exaggerates the occurrence of this compound in the ash. This would shift the equilibrium to a more abundant occurrence o f K2CO3 in the ash or KOH(g) as vapour. The Ca-containing species in the equilibrium ash are Ca5(PO4)3OH, CaO, and CaSiO3. None of these compounds exist in the biomass: Ca occurs either as organically associated metal ions (Ca2+) or as oxalate, CaC2O4. The CaC2O4 decomposes upon heating and forms lime, CaO. Lime may further react with organically associated phosphorus or potassium hydrogen phosphate and form Ca5(PO4)3OH and calcium potassium phosphates, respectively [25]. Spruce needles have high concentrations of biogenic silica. It will react with the lime and form CaSiO3 to some extent [26], but unless a liquid phase appears, this reaction is probably also kinetically restricted. 5 CONCLUSION The residues after harvesting lumber, the tree tops and branches, are collected and fired in large-scale combustion. It comprises the wood, bark and foliage of several wood species with different ash content and composition. This paper predicts the ash behaviour in combustion via improved fuel analyses and interpretation with theoretical tools such as equilibrium calculations and kinetic considerations. The key improvements are 1) a systematic division of the fuel into fractions and 2) analyses of their ash-forming elements and ash-forming matter. This allows a thorough and fundamental approach to predict ash-related problems quantitatively, such as sticky particles of partly molten ash or volatilised ash that causes fouling in the convection pass. Fuel-flexible combustion o f the future needs a sound scientific model that successfully couples ash-related problems with the fuel being fired. 6 ACKNOWLEDGEMENTS This project was part o f the activities o f the Abo Akademi Process Chemistry Centre, funded by the Academy of Finland through its Centre o f Excellence program. We gratefully acknowledge the financial support o f the Finnish Graduate School of Chemical Engineering and the TEKES project ChemCom and its industrial partners. REFERENCES [1] Harding N. S., O'Connor D. C., Fuel Processing Technology, 88:11-12 1082 (2007) [2] Demirbas A., Progress in Energy and Combustion Science, 31:2 171 (2005) [3] Raask E., Mineral impurities in coal combustion : behavior, problems, and remedial measures, Hemisphere Pub. Corp., Washington, 1985 [4] Valmari T., Potassium behavior during combustion o f w ood in circulating fluidized bed power plants, VTT Publications, 2000 [5] Tran H. N ., M ao X., Kuhn D. C. S., Backman R., Hupa M., Pulp & Paper Canada, 103:9 29 (2002) [6] M ueller C., Selenius M., Theis M., Skrifvars B.-J., Backman R., Hupa M. et al., Proceedings o f the Combustion Institute, 30:2 2991 (2005) [7] Skrifvars B. J., Hupa M., H yoty P., Proceedings - Annual International Pittsburgh Coal Conference, 7th:33 (1990) [8] Flagan R. C., Sarofim A. F., Progress in Energy and Combustion Science, 10:2 170 (1984) [9] Brostrom M., Kassman H., H elgesson A., Berg M., Andersson C., Backman R. et al., Fuel Processing Technology, 88:11-12 1171 (2007) [10] Bartels M., Lin W., Nijenhuis J., Kapteijn F., van Ommen J. R., Progress in Energy and Combustion Science, In Press, Corrected Proof:(2008) [11] N ielsen H. P., Frandsen F. J., Dam-Johansen K., Baxter L. L., Progress in Energy and Combustion Science, 26:3 283 (2000) [12] Skrifvars B. J., Backman R., Hupa M., Salmenoja K., Vakkilainen E., Corrosion Science, 50:5 1274 (2008) [13] Zevenhoven M., Yrjas P., Backman R., Skrifvars B.-J., Hupa M., Proceedings o f the International Conference on Fluidized B ed Combustion, 18th:667 (2005) [14] Werkelin J., Skrifvars B.-J., Hupa M., B iom ass and Bioenergy, 29:6 451 (2005) [15] B ale C. W., Chartrand P., Degterov S. A., Eriksson G., Hack K., Ben Mahfoud R. et al., Calphad, 26:2 189 (2002) [16] Werkelin J., Skrifvars B.-J., Holm bom B., Hupa M., Fuel (submitted), (2008) [17] W erkelin J., Skrifvars B.-J., Hupa M., Fuel (submitted), (2008) [18] W erkelin J., Zevenhoven M., Skrifvars B.-J., Fuel (submitted), (2008) [19] IC Application N ote N o. S-127, Metrohm, [20] Fardim P., H olm bom B., Tappi Journal, 2:10 28 (2003) [21] Fardim P., H olm bom B., Ivaska A., Karhu J., Mortha G., Laine J., Nordic Pulp & Paper Research Journal, 17:3 346 (2002) [22] Skrifvars B.-J., Blom quist J.-P., Hupa M., Backman R., Proceedings - Annual International Pittsburgh Coal Conference, 15th:652 (1998) [23] Zevenhoven M., Yrjas P., Backman R., Skrifvars B.-J., Hupa M., Proceedings o f the International Technical Conference on Coal Utilization & Fuel Systems, 27th:Vol. 1 281 (2002) [24] Zevenhoven-Onderwater M., Blom quist J. P., Skrifvars B. J., Backman R., Hupa M., Fuel, 79:11 1353 (2000) [25] Lindstrom E., Sandstrom M., Bostrom D ., Ohman M., Energy & Fuels, 21:2 710 (2007) [26] W erkelin J., Lindberg D ., Skrifvars B.-J., Hupa M., Bostrom D ., Biom ass and Bioenergy, submitted in June 2008:(2008)