DOLOMITE DESULFURIZATION BEHAVIOR IN A BUBBLING FLUIDIZED BED PILOT PLANT FOR HIGH ASH COAL

Although fluidized bed in situ desulphurization from coal combustion has been widely studied, there are aspects that remain under investigation. Additionally, few publications address Brazilian coal desulphurization via fluidized beds. This study used a 250 kWth bubbling fluidized bed pilot plant to analyze different aspects of the dolomite desulphurization of two Brazilian coals. Superficial velocities of 0.38 and 0.46 m/s, flue gas recycling, Ca/S molar ratios and elutriation were assessed. Results confirmed the influence of the Ca/S molar ratio and superficial velocity – SO2 conversion up to 60.5% was achieved for one coal type, and 70.9% was achieved for the other type. A recycling ratio of 54.6% could increase SO2 conversion up to 86.1%. Elutriation and collection of ashes and Ca-containing products did not present the same behavior because a lower wt. % of CaO was collected by the gas controlled mechanism compared to the ash.


INTRODUCTION
Fluidized bed coal combustion has the advantage of creating SO 2 treatment through a sulphation reaction from the in situ injection of dolomite or limestone. This procedure has been widely explored; however, some details and local behaviors based on coal characteristics are still under investigation. Moreover, it should be noted that there is a lack of information concerning Brazilian coal in fluidized beds. Furthermore, according to Anthony and Granatstein (2001), dolomitic stones have not found widespread use in atmospheric fluidized bed combustors because they are not particularly effective on a mass basis due to the inability of MgO to react with SO 2 .
However, the sulphation reaction is far from ideal. Typically, only a conversion of 30-40% of CaO is obtained. This relatively low utilization of limestone is one of the major limitations of the technology (Anthony and Granatstein, 2001). The presence of SO 2 affects limestone particle size because it generates a hard sulphate shell around the unreacted CaO core of the particle, which reduces its fragmentation (Scala et al., 2011). Therefore, substantial changes in the sorbents' particle size distribution can be achieved from particle attrition and fragmentation in fluidized bed combustors (Montanagro et al., 2010).
Because the molar volume of CaSO 4 is 52.2 cm 3 /mol, which is greater than that of CaCO 3 (36.9 cm 3 /mol), and no particle expansion occurs, the possible maximum practical conversion is low. Considering the bed temperature, it has been well established that the most favorable temperature for the sulphation reaction is located in the range of 800-850 °C, and the retention falls sharply as temperature increases (Yates, 1983). Moreover, excess air increases the changes in the gas composition in which the sorbent sulphates and the partial pressure of oxygen are increased; consequently, the partial pressure of CO 2 is reduced (Ulerich et al., 1980). Different aspects of the desulphurization mechanism have already been investigated and will be briefly discussed. Lyngfelt and Leckner (1989) investigated the SO 2 conversion relationship with process conditions, especially temperature. The oxygen concentration was considered in Collar's (2001) analysis, who evaluated pore structures in the reactivity of different limestones. It was observed that with an increase in the O 2 concentration, the desulphurization reaction was worse due to the additional SO 3 in the reaction. Tarelho et al. (2005) studied the influence of operational parameters for fluidized bed desulphurization. A constant Ca/S molar ratio of 3.5 was used. Furthermore, excess air values of 10, 25 and 50% were applied. Hlincik and Buryan (2013) reported that limestones with lower CaO content can provide a better desulphurization capacity. This result occurs because the desulphurization reactions of the combustion gases are significantly negatively affected by the reaction of CaO with the ballast oxides of ash and limestone.
The reaction kinetics were was studied by Irfan and Balci (2002). For temperatures of 850 °C and below, the reaction occurred with two mechanisms due to pore plugging and changes in the chemical composition of the solid. Bragança et al. (2003) studied the influence of limestone characteristics to determine the kinetic parameters of desulphurization using a batch fluidized bed reactor. Anthony and Granadstein (2001) investigated the fragmentation of sorbent particles. Furthermore, Bragança (1996) used dolomitic and magnesium limestone for the desulphurization of different particle sizes. Scala et al. (2000) used a bench-scale fluidized bed reactor to investigate the attrition behavior of two different limestones during calcination and sulphation. After sulphation of the pre-calcined lime, the attrition rate for both types of limestones tested decreased dramatically until a new steady-state value was reached.
Furthermore, models for the description of the conversion mechanism have been evaluated. Suyadal et al. (2005) explored the SO 2 breakthrough curves in the presence of O 2 , CO 2 , and H 2 O steam and tested one deactivation model description of these curves obtained in an integral fluidized bed reactor.
Additionally, and related to this study, due to the need for a minimum residence time of the calcinated particles, elutriation must be considered in the desulphurization process in fluidized bed systems. Elutriation is referred to the separation or removal of fines from a mixture, and it occurs to a lesser or greater extent at all freeboard heights. However, the solid carryover is strongly affected by gas velocity, and the fraction of fines in the bed is only slightly affected by changing the size of the coarse material and the minimum fluidization velocity (Kunii and Levenspiel, 1991). Altindag et al. (2004) demonstrated that the freeboard sulphur-capture was enhanced significantly by recycling the elutriated sorbent particles. A limestone with 93.21 wt. % CaCO 3 was used. In runs without flue gas recycling, a significant majority of the sulphur retention occured in the bed section. By introducing flue gas recycling, freeboard sulphur capture increased from 3.5% to 19.5%. This result occurred because, for fuels rich in volatile matter and combustible sulphur content, such as the ones used in this study, freeboard sulphur capture is enhanced significantly with recycling because the sulphur release to the freeboard is significant. Montanagro et al. (2010) investigated the influence of temperature on the attrition of two limestones during desulphurization in a fluidized bed reactor. The experimental procedure was conducted at 900 °C with the goal of assessing the extent of the primary fragmentation. Furthermore, results were compared with those obtained at 850 °C for the same reactor. The fluidized bed desulphurization experiments were conducted using a synthetic gas with an SO 2 -O 2 -N 2 mixture of 1800 ppmv-8.5%-N 2 balance. A temperature increase resulted in a larger amount of elutriation. The elutriation occurred on a time scale comparable with the sulphation. Therefore, after a harder sulphate shell is formed around the limestone particles, the attrition process dominates. Furthermore, in this report, the limestone particles are characterized by a rather limited propensity to yield elutriated fines as a result of surface wear. This issue should be considered in continuous industrial desulphurization runs, since such a limitation to yield elutriated fines tends to accumulate desulphurization products in the bed and change bed properties after long time runs.
Thus, Huda et al. (2006) demonstrated the influence of four coal types on the in situ desulphurization process in a 71 MW Pressurized Fluidized Bed Combustor (PFBC) demonstration plant. Important information on the formation of the bed material was discussed. White and yellow particles were formed in the bed after the desulphurization tests. The white particles were calcium carbonate or calcium carbonate coated with calcium sulphate while the yellow particles were calcium carbonate coated with calcium sulphate and Ca-aluminosilicate. Moreover, the CaSO 4 particles formed agglomerate. Thus, fly ash particles could be distinguished from one another through their morphologies. Scala et al. (2011) studied the primary fragmentation of two limestones in a 1-meter-height batch scale fluidized bed under air combustion and oxyfuel combustion. Synthetic gases were used to provide the atmospheres. Two types of limestone were used: CaCO 3 over 96% and CaCO 3 over 99%. Additionally, the limestone type was determined to be a more important variable with respect to the fragmentation tendency. The particle size, bed temperature, and simultaneous occurrence of the sulphation reaction were found to have an insignificant influence on the limestone primary fragmentation under either atmosphere. Again, the tendency of changes in the elutriation mechanism is one important result when controlling fragmentation.
However, according to Montanagro et al. (2010), an investigation of the effect of bed temperature on the limestone attrition and fragmentation phenomena, as well as their connection with the sulphation behavior and microstructural properties, is still lacking and appears to be of great practical interest. Scala et al. (2013) tested the fluidized bed desulphurization performance of lime particles obtained using a limestone slow calcination pre-treatment technique compared to the untreated material. Furthermore, the fragmentation and attrition processes were investigated. An externally heated stainless steel atmospheric bubbling fluidized bed reactor 40 mm in diameter and 1 m in height was used for the experiments. The pre-treatment technique appeared to induce a high mechanical strength in the sorbent particles (as opposed to the fragile lime generated after typical rapid fluidized bed calcination).
Thus, although this technology is regarded as a mature technology for in situ desulphurization, its dependence on local stone properties and interactions between the local stone/coal properties and the fluidization conditions may require further research (Altindag et al., 2004). Consequently, the use of coals from different regions is an aspect to be considered when studying desulphurization, especially Brazilian coals, which have poor data on fluidized bed combustion and desulphurization.
The primary objective of this study is to obtain fluidized bed operating parameters in a bubbling fluidized pilot plant for SO 2 conversion using an in situ dolomite feed. Different runs analyzed the fluid dynamic characteristics for two southern Brazilian coals, and the superficial velocity, flue gas recycling, elutriation behavior and stoichiometric analysis were investigated. Results may lead to the development of a fluidized bed desulphurization of Brazilian coals.   owchart of t rming the co e, solid fuels ontrolled by y into the fur generated ga one (7), w r the gases at t-combustion heat exchan enters into emperature re gases are con he sequence eased by the (not applied ed, the recy partially clos A forced drau y for the ga pressure co ed is rectang 7 cm and a le 0 m. Bed ch 1: Pilot pla ilo; (4) Coal 1; (9) Post co 2) Chimney 2 M. F. Gomes, C.

Materials
Brazilian Jou the 0.25 MW ombustion ru s from the si screws (2) a rnace (6) wh ases flow int which allows t an appropri n chamber ( ger (10)

Experimental Procedure
For the combustion runs, heating the bed began with an auxiliary fuel, ethanol, in the plant heater (13) until the bed reached 600-650 °C, which is when coal was added and the auxiliary fuel was not fed anymore. Fuel and air are adjusted to reach and stabilize the bed temperature at 850 ± 15 °C, when the steady-state condition was considered, which is controlled by two thermocouples located in the bed inlet, T04, and the bed outlet, T05. For each coal feed, after stabilizing the bed temperature, dolomite is added based on the applicable Ca/S relation.
For gas recycling, in order to provide a controlled transition, the following procedure was developed after reaching the steady-state condition on combustion: I. 100% recycling gas valve open (valve gas recycling is located after the chimney); II. Partial chimney closing via damper control; and III. Progressive recycling valve closed to maintain positive pressure on point P2 of Figure 1.
Recycled gases were injected into the reactor entrance in the same air duct and passed through the draught fan (14). For the gas air mass flow rate control, to maintain the same conditions as applied to the air combustion without recycling with the U s value, one valve located after the forced draught fan ( Figure  1) was used to control the flow rate.
It is important to maintain a positive pressure until point P2 to prevent air leakage in the section reactor-P2 as gas sampling is performed in this section. The bubbling fluidization condition is controlled by the pressure drop, dP, in the bed, which must be maintained between 4413-5197 Pa, and temperatures T04 and T05, that cannot present a difference higher than 5 °C between bed inlet and outlet. Furthermore, the fluidization conditions can be continuously visualized through one observation point located above the reactor.

Stoichiometric Analysis
For the desulphurization reactions from the dolomite calcination, CaO and SO 2 are converted to CaSO 4 as described by Reactions 1 and 2. (Yates, 1983;Anthony and Granatstein, 2001).
However, as also described by other authors, the desulphurization mechanism can be represented as the sum of Reactions 1 and 2, as expressed by Reaction 3 (Suyadal et al., 2005).

Average SO 2 Conversion Rate
Equation (4) expresses the rate of the SO 2 conversion, ξ, by taking its input and output and considering the reactor as the volume control.
where n iSO2 = number of moles of SO 2 into the system; n oSO2 = number of moles of SO 2 out of the system; and SO2 υ = SO 2 stoichiometric coefficient. SO 2 was considered to be reacting based on Equation (3).

Desulphurization Efficiency
The desulphurization efficiency, η, can be obtained from Equation (5) as follows:

Reactants Consumption
Himmenblau and Riggs (2006) defined yield as the quantity of desirable product, CaSO 4 , divided by the fed quantity of the reactant, CaO and SO 2 . Here, the yield is considered as the CaSO 4 molar formation and is related to the molar quantity of reactant consumed, as CaSO 4 /moles SO 2 feed and moles CaSO 4 / moles CaO feed.

Elutriation
Elutriation is considered as representing the wt.% of non-reacted CaO particles and ash particles carried out of the system during steady state conditions, as expressed by Equations (6) and (7). The CaO and ash particles were collected in Cyclone #1 of Figure 1 Because th e related to ts conversion onsidered in

RE
The steady ation data w oncentration ures, T04 an luidization co y controlling ure drop, wh nd the heigh nd 5197 Pa. onditions can ation from th

Combustion, ion Results
The results re presented ions were ta ial velocities o the pressur or higher ap etween the b   of Coal A, the Ca/S relation must be increased to decrease the SO 2 concentration around the sorbent particles.
The influence of the SO 2 concentration around the sorbent particles on the conversion can be confirmed by XRF analysis of the non-desulphurized ashes, A1 and B1, as indicated in Table 6. The alkaline and alkaline earth components of the coal ashes influence the desulphurization behavior by providing self-desulphurization (Cheng et al., 2003). Table 6 indicates that the alkaline components of Coal B are higher compared to those for Coal A, especially for CaO. By considering that the dolomite characteristics were the same for both coals and taking the same superficial velocity, the SO 2 concentration of the flue gases is confirmed to provide higher influence on conversion than the coal ash characteristics.
By using mean diameter and the density of dolomite, as expressed in Table 2, the terminal velocity can be calculated as 0.62 m/s. If we compare this terminal velocity with the superficial velocities applied in the experiments, the superficial velocities are 61.3% and 74.2% of the dolomite terminal velocity for U s = 0.38 and U s = 0.46 m/s, respectively. Consequently, since the dolomite is fed just above the bed, the particles are prone to reach the bed and start the fragmentation and calcination process in the bubbling bed. Moreover, it should be noted that a visual observation of the bed after the desulphurization runs indicated a concentration of white and yellow desulphurization product particles, which confirmed the high dolomite and Ca-containing products within the bed as a result of the change in its properties after long runs and will be discussed later. Thus, in industrial plants, this result indicates the necessity of controlling the changes in bed properties after long runs for bed reposition. However, an increase in the diameter of the dolomite is not necessary to increase the terminal velocity because it would increase the mass transfer re-sistance and decrease the SO 2 conversion, as well as increase the size and concentration of the desulphurization products that accumulate in the bed.
Furthermore, as seen in the XRF analysis in Table  6, self-desulphurization had already occurred without the sorbent feed due to the CaSO 4 formation because SO 3 appears significantly together with CaO. This result occurs because the coal ashes present CaO in their composition that can react with SO 2 . However, this self-desulphurization mechanism cannot only be predicted by the Ca in the ash composition due to the presence of other alkaline elements that are able to promote self-desulphurization (Cheng et al., 2003).

Desulphurization with Gas Recycling
The possibility of increasing the SO 2 conversion with flue gas recycling was analyzed. The same conditions applied for the desulphurization runs using U s = 0.38 m/s, and a Ca/S molar ratio of 2.0 was considered in the desulphurization with recycled gas. Table 7 presents the conditions and results obtained from these runs. To maintain the same superficial velocity as in condition B4, the total air plus gas flow rate was controlled by one valve located after both fluxes, i.e., just after the forced draught fan (Figure 1).  Table 7 indicates that the SO 2 conversion can be strongly increased from 48.0% to 86.1% when recycling part of the flue gases. Consequently, some elutriated and non-converted CaO particles are able to return to the reactor to react with SO 2 . Here, the percentage of gases recycled is referred to as the wt. % of gases returning to the fluidized bed from the total flue gas flow. In this way SO 2 emission, according to Brazilian regulations, can be reduced to 1143.4 g/10 6 kcal, far lower than the ones obtained in Table 5.
Furthermore, Altindag et al. (2004) applied fine recycling for lignite desulphurization. The introduction of recycling increased the sulphur retention efficiency from 68.5% to 81.1% as a result of the increased residence time of unreacted or partially sulphated recycled fine limestone-particles. According to the authors, for fuels that are rich in volatile matter and combustible sulphur content, the freeboard sulphur capture and release to freeboard was significantly enhanced with recycling. These observations could be applied to Brazilian coal due to its low fixed carbon content and high volatile matter.

Stoichiometric Analysis
The desulphurization reactions considered the relation of 1 mol CaSO 4 per mol of SO 2 reacted as well as 1 mol CaSO 4 per mol of CaO reacted. Consequently, considering the results before, Table 8 presents the values for a CaSO 4 yield based on a SO 2 and CaO feed. Table 8 also provides an average SO 2 conversion rate for the number of moles of SO 2 consumed/hour.
From Table 8, a comparison can be made between the runs obtained from one type of coal and different U s values, such as A4 x A7 and B4 x B7. It can also be made for the same U s and different coals, such as A4 x B4 and A7 x B7.
If we consider the reactant consumption, as expressed by moles CaSO 4 /moles SO 2 feed and moles CaSO 4 /moles CaO feed, few differences can be observed when comparing A4 x A7 and B4 x B7 conditions. For the same type of coal, the number of moles formed per SO 2 and CaO moles consumed were higher for the lower U s . Here, it can be observed that reactants are consumed better at a higher residence time, as expected. When comparing conditions A4 x B4 and A7 x B7, i.e., for the same fluid dynamic conditions but different coals, there are some few points that should be considered. Different values are observed for the average SO 2 conversion rate. Even with lower SO 2 conversion values for Coal B under conditions A4 x B4 and a similar conversion under conditions A7 x B7, Coal B presented higher conversion rate values; SO 2 conversion can be observed in Figure 3. This behavior is a result of the S wt.%, which is higher for Coal B and responsible for a higher reaction rate.
Moreover, for the same coal, the fluid dynamic conditions indicated an influence. For U s = 0.38 m/s, the reaction extent was 10.35 moles SO 2 /h under condition B4 and -8.79 moles SO 2 /h under condition B7. This decrease in the conversion rate with Coal B, which was not observed for Coal A, could explain the behavior observed in Figure 3 for this coal with U s = 0.46 m/s.

Ash Characterization
During coal combustion, a part of the total sulphur in the coal will be retained as solid compounds in the ash. This signifies that the coal ash has a sulphur-retaining property, which makes a significant contribution to alleviating the problem of SO 2 emissions (Sheng et al., 2000). According to the same authors, most of the calcium in coal is active and capable of participating in sulphur-retention reactions during coal combustion. Additionally, the content of calcium in most coals is generally significantly higher compared to that of other basic elements. Therefore, variations in the Ca/S molar ratio in coal can markedly affect the percentage of sulphur retained.
The X Ray Diffraction (XRD) patterns for the two coal ashes studied are illustrated in Concerning SO 4 2based compounds, most organic sulphur and pyrite in coal are oxidized and converted to SO 2 gas during the combustion in furnaces. A small part of the sulphur may be retained as solid compounds due to the contribution of alkaline components, such as CaO, MgO, Al 2 O 3 , K 2 O and Na 2 O in the coal ash. The desulphurization property of the coal ash during combustion is primarily affected by the boiler shape, flame temperature, residence time in the furnace, initial molar ratio of Ca/S and reaction activity of the alkaline components. It is difficult for sulphates of other minor elements, such as MgSO 4 , Al 2 (SO 4 ) 3 , Fe 2 (SO 4 ) 3 , K 2 SO 4 and Na 2 SO 4 , which are less thermally stable than CaSO 4 , to act as sulphation products during the coal combustion at high temperatures (Cheng et al., 2003).
Muscovite (KAl 2 (AlSi 3 O 10 )(OH) 2 ) was the primary silicate identified in the Coal A and Coal B ashes. A strong presence of hematite was observed in the Coal A ashes. By analyzing Figure 4, it is clear that differences exist between the ashes generated with and without desulphurization. The anhydrite peaks appear to be more intense after dolomite (CaMg(CO 3 ) 2 ) and lime (CaO) additions due to the desulphurization mechanism, as previously discussed.
Considering the hematite formation, according to Brady et al. (1994), the dominant reaction for FeS contained in coals can be written as follows: The presence of calcium sulphite in conditions A4, A7 and B4 indicates that the desulphurization reactions did not have time to be completed. This may have occurred because, as soon as the gases exited the reactor, they were captured by Cyclone #1, and the reactor's temperature rapidly decreased.
The rate of formation of sulphur-containing solid components (sulphide, sulphite and sulphate) appears to be nearly independent of the presence of oxygen at temperatures below approximately 735 °C. At 830 °C, where calcium sulphite is unstable in the prevailing atmosphere, the rate of formation of sulphurcontaining components is increased by the presence of oxygen (Dam-Johansen and Ostergaard, 1991). The same authors note that the formation of calcium sulphate from calcium oxide, sulphur dioxide and oxygen may occur through the formation of calcium sulphite and the subsequent disproportionation to calcium sulphate and calcium sulphide, oxidation to calcium sulphate, or through catalyzed or uncatalyzed oxidation of sulphur dioxide to sulphur trioxide and the subsequent reaction with calcium oxide. Calcium sulphide was not observed in all of the ashes, which suggested that the disproportionation of calcium sulphite may not have occurred. The XRD patterns suggest that the formation of CaSO 3 occurs first and is then followed by oxidation to CaSO 4 , as indicated by Reactions 1 and 2.

Elutriation
The elutriation behavior of the ashes and the desulphurization products was analyzed. Consequently, conditions A4 and A7 were compared, because the Ca/S relation was the same for both conditions but the superficial velocities applied were different. The elutriation analysis had the primary objective of assessing the desulphurization products' partitioning behavior within the bed and Cyclone #1.
The XRF analysis of the ashes from conditions A4 and A7 are presented in Table 9. The goal was to consider the quantitative oxide analysis to serve as a basis for determining the CaO partitioning. Thus, Table 10 presents the mass collected and the ash mass flow within Cyclone #1 producing the ashes in Silo #1. Consequently, from the total ash generation, the wt. % ashes obtained in Silo #1 from the operation of Cyclone #1 can be determined. Furthermore, by obtaining the CaO feed from the dolomite and the non-desulphurized ashes, the wt. % CaO collected in Cyclone #1 was also determined in Table 10.
From Table 10, it can be seen that 60.5 wt.% of the ashes in condition A4 were obtained in Silo #1 and 55.4 wt.% from condition A7 were obtained in the same silo. The values for the wt. % CaO fed into the reactor (dolomite + non-desulphurized ashes), elutriated and collected by Cyclone #1 and stored in Silo #1 were not within the same range; for condition A4, 43.07 wt.% of the fed CaO was collected by cyclone #1 whereas 37.32 wt.% was collected by the same cyclone under condition A7.
From the first analysis, it can be seen that the elutriation and the collection of ashes and Ca-containing products do not present the same behavior.  The typical tendency is that the elutriation of solids increases as U s increases. However, it can be seen that the wt. % of CaO in Silo #1 was lower for condition A7 when a higher U s value was applied. This result was distinctly different than expected. Additionally, Huda et al. (2006) observed white and yellow bed materials larger than 2.0 mm after the in situ desulphurization. They noted that the yellow particles were CaCO 3 coated with CaSO 4 and Ca-aluminosilicate while the white particles were CaCO 3 or CaCO 3 coated with CaSO 4 . Consequently, the authors defined the Ca/S molar ratio of the fine sorbent formed by attrition (< 200 µm), which was significantly lower than the total Ca/S molar ratio.
From the elutriation analysis, it can be noted that, for long dolomite desulphurization runs on the bubbling fluidized bed, there is a tendency of the bed properties to change for coal types similar to the Brazilian ones. It is important that this be controlled in industrial plants so that the fluidization quality is not affected.

CONCLUSIONS
After analyzing different aspects that involve in situ desulphurization in a bubbling fluidized pilot plant, some conclusions can be developed for the SO 2 conversion, ash characteristics and elutriation mechanisms.
Based on the stoichiometric analysis performed on the SO 2 conversion, i.e., a comparison of one coal under different fluid dynamics conditions, Coal B presented higher SO 2 conversion rates. This was the expected result for a higher S wt.% for this fuel. Additionally, better conversion was observed for lower U s values; however, Coal B presented lower conversion rate values for SO 2 consumed per hour for U s = 0.46 m/s. The fluidization slugging behavior when applying a higher superficial velocity, U s = 0.46 m/s, and a bed height of 0.4 m for 0.262 m 2 , indicated a considerable decrease in the SO 2 conversion, which was most likely due to changes in the mass transfer mechanism. Together, the changes in the residence time must be considered to some extent. However, the residence time was not able to explain the SO 2 conversion because, for example, in the Coal B conditions, it showed the same value for both superficial velocities applied. Consequently, the fluidization condition with a mass transfer mechanism can be a better path for achieving the SO 2 conversion relations.
Mineralogical analyses for the ash characterization indicated the presence of CaSO 4 and CaSO 3 as desulphurization products. Because no CaS was observed in the ashes, it can be suggested that the mechanism for CaSO 4 formation occurs by the formation and oxidation of CaSO 3 . This issue is important to be controlled when considering the use of ash. The formation of CaSO 4 instead of CaSO 3 is often preferred when considering the different structural uses of coal ashes.
The different fluid dynamic conditions applied clearly indicated a lower percentage of CaO in Cyclone #1 than the quantity fed compared with the same relation for the ashes fed into the reactor. This behavior was explained by the tendency of desulphurization products to accumulate in the bed during the use of dolomite. It was confirmed by visually detecting the desulphurization products as larger white particles in the bed. Again, it is an important issue to be considered in the continuous run of bubbling fluidized bed plants working with in situ desulphurization due to the strong tendency of changes in the bed properties. Consequently, some continuous bed control must be developed by an appropriate procedure. Moreover, property changes in long desulphurization runs are still an issue to be analyzed in future studies.