Emissions of SO 3 from a Coal-Fired Fluidized Bed under Normal and Staged Combustion

This paper reports the measurements of SO 3 emissions with and without limestone under unstaged and staged fluidized-bed combustion, carried out on a 0.3 × 0.3 m 2 and 2m high stainless-steel combustor at atmospheric pressure. The secondary air was injected 100cm above the distributor. SO 3 emissions were monitored for staging levels of 85 :15, 70:30, and 60:40, equivalent to a primary air/coal ratio (PACR) of ∼ 0.86, 0.75, and 0.67. Experiments were carried out at 0%–60% excess air level, 1-2m/s fluidizing velocity, 800–850 ∘ C bed temperature, and 20–30 cm bed height. During unstaged combustion runs, SO 3 emissions were monitored forawiderangeofCa/Sratiosfrom0.5to13.However,forthestagedcombustionruns,theCa/Sratiowasfixedat3.SO 3 wasretained to a lesser extent than SO 2 , suggesting that SO 2 reacts preferentially with CaO and that SO 3 is involved in the sulphation process to a lesser degree. The SO 3 emissions were found to be affected by excess air, whereas the fluidizing velocity and bed temperature had little effect. SO 3 was depressed on the addition of limestone during both the staged and unstaged operations, and the extent of the reduction was higher under staged combustion.


Introduction
The presence of SO 3 in flue gas corrodes the equipment and ducts of combustion system and therefore needs to be removed [1]. In order to control emissions of SO 3 , more studies on its formation and dissociation are required under air-fired and oxy-fired combustion conditions. The simulation study of Zheng and Furimsky [2] shows that SO 3 emissions would be unaffected during oxy-fuel combustion, being governed only by oxygen concentration. The kinetics of reactions occurring in the combustor were studied by Burdett et al. [3] using a TGA microbalance. They proposed the following mechanisms for the formation of SO 3 . 3 Homogeneous Gas Phase Reaction. SO 2 may be oxidized to SO 3 by two reactions:

SO 2 /SO
where M is a chaperon third body molecule. The large temperature dependence of reactions (1) and (2) ensures that the rate of production falls rapidly with decreasing gas temperature, and, in fact, 90%-95% of SO 3 is formed in the bed and freeboard and the remaining 5%-10% in the region between the freeboard and sampling point. SO 3 increases sharply with temperature, but the homogeneous reaction cannot account for all the SO 3 produced.

Heterogeneous Catalysis of SO 2 on Bed Particles and
Heat Transfer Surfaces. In a coal burning combustor, a more effective catalytic material, iron oxide, is present in fly ash. While the SO 3 formation in this process is important, the experimental data is insufficient to quantify the SO 3 formation. Dennis and Hayhurst [4] used an 80 mm diameter fluidized-bed combustor and mass spectrometer for measuring the concentration of SO 3 . They confirmed the amount of SO 3 formed at atmospheric pressure to be very low and much less than the equilibrium concentration. The rate measured 2 ISRN Environmental Chemistry was 100 times faster than expected for oxidation in the gas phase.
Willium and Gibbs [5] measured SO 2 and SO 3 emissions from a 0.3 m 2 fluidized-bed combustor. They reported that the higher the excess air level, the lower the SO 2 emissions (on removal of the dilution effect, SO 2 increased with an increase in excess air, reaching a limiting value of 1300 ppm above 30% excess air), and that SO 3 emissions increased slightly with the increase in excess air. Barnes [6] also observed the similar effect of excess air on SO 2 and SO 3 emissions. Willium and Gibbs [5] and Barnes [6] found that the higher the sulphur content of fuel, the higher the SO 2 emissions, while the SO 3 was unaffected by the fuel sulphur content [5]. Coal fed to the bed caused SO 2 and SO 3 emissions to increase than when fed to the surface [5]; SO 3 shows a weak dependence on bed temperature [5,6]. Barnes [6] studied the effect of sand particle size, fluidizing velocity, and bed depth on SO 2 and SO 3 emissions. Barnes' findings indicate that fine sand (0.300 mm) produces high SO 2 and SO 3 emissions. Increasing fluidizing velocity from 1 to 2 m/s caused reduced formation of SO 2 in the bed and freeboard. An increase in bed depth increased SO 3 emission; a deep bed (30 cm) and fine sand resulted in a slight decrease in SO 2 emission.
Oxygen availability and fluidizing characteristics within the bed also affect SO 3 formation. Ahn et al. [7] found that for pulverized coal, concentrations of SO 3 and SO 2 were significantly higher for oxy-fired conditions as compared to air-fired conditions. In circulating fluidized bed, SO 3 concentrations were notably higher for oxy-fired conditions too. For higher sulfur coal, SO 3 concentrations were 4-6 times greater on average. Their findings contradict the finding of Barnes [6].
Hindiyarti et al. [8] investigated the reaction of SO 3 with H, O, and OH radical. The revised rate constant calculated by them suggests that SO 3 and O reaction is found to be insignificant during most conditions. According to them, SO 3 + H is the major consumption reaction for SO 3 .
Stanger and Wall [9] reviewed published work on SO 3 concentrations and emissions under oxy-fuel firing. Their conclusion is that the conversion of SO 2 to SO 3 is considerably variable.
Willium and Gibbs [5] found that coal char had a very significant removal effect on SO 3 emissions because SO 3 above the bed was 50% greater than in the exit. Barnes [6] said that unburnt char does not have a major effect on SO 3 as carbon carryover increases under the given operating conditions. Her results showed that the quantity of inert particles (30 cm deep bed) resulted in an increase in heterogeneous catalytic reaction of SO 2 to form SO 3 .
Burdett et al. [3] carried out experiments in a microbalance to study the effect of limestone on SO 3 emissions. According to Burdett et al. [3], the reaction of CaO, O 2 , and SO 2 , in order to yield CaSO 4 (reaction (3)), must occur in two separate steps. Two possibilities exist, either SO 2 reacts with CaO and CaSO 3 is formed, which is then oxidized, or else the formation of SO 3 in gas phase or on a stone surface is followed by an attack on the CaO. Consider the following: Route 1 CaO + SO 2 → CaSO 3 (4) Route (2) It is not possible, due to the lack of available experimental data, to say which of these mechanisms is operative under given conditions, although both may be important. A most interesting comparison between the SO level detected with and without limestone (proposed by Burdett et al. [10]) is the SO 2 /SO 3 ratio. Without limestone, the ratio they found was 2200/33 (or 67 : 1), and with limestone it increased to 350/1.5 (or 230 : 1). It is clear that SO 3 is depressed to a greater extent than SO 2 on the addition of limestone, and they attribute this to the higher reactivity of SO 3 compared with that of SO 2 . Burdett [11] and Burdett et al. [10] have assumed that SO 2 oxidizes to SO 3 in the particles at a rate dependent on local SO 2 and O 2 concentrations, with SO 3 diffusing through the CaSO 4 shell and reacting with CaO. The rate of formation of CaSO 4 may be linked to different rates of production of SO 3 at different locations within the stone. At a high oxygen level, oxidation increases preferentially at the edge of a particle. High utilization is achieved when SO 2 diffusion in the interior of the stone is maximized, and this, in turn, implies a low SO 3 formation rate. Barnes [6] reported a decrease in SO 2 conversion to SO 3 with oxygen concentration and an increase with SO 2 concentration. Fieldes et al. [12] reported the achievement of a high fractional sulphation (0.36) when coal is burnt in the bed. It appears that the fraction of the sulphur gas phase, which is SO 3 , has a substantial effect on the fractional sulphation of the limestone. Ash also appears to remove SO 3 selectively. The mechanism of this hypothesis supposes that the direct reaction of CaO with SO 3 is faster than a reaction via the CaSO 3 intermediate. This paper examines the factors responsible for formation and reduction of SO 2 and SO 3 from a coal-fired fluidized bed under varying operating conditions.

Apparatus and Procedure
The main features of the fluidized-bed combustor and ancillaries are presented in Figure 1. The bed consisted of silica sand of mean size 0.700 mm. Fluidizing air was supplied by a fan and metered and introduced through a distributor plate. For staged combustion, the secondary air was introduced into the combustor through a stainless-steel pipe 100 cm above the bed surface. In staged combustion mode, the total combustion air is separated into a primary air stream supplied  to fluidize the bed and a secondary air stream injected above the bed to complete the combustion. For example, in 70 : 30 staging, 30% of the total air is injected as secondary air. The bed was preheated by a propane burner that was fixed above the bed, and the fluidizing airflow rate was adjusted to the lowest level to minimize heating time. Coal was fed in the combustor when the bed temperature reached 550 ∘ C. When the bed temperature reached 800 ∘ C, the desired coal feed rate was adjusted to a constant value, the propane burner was switched off, and the fluidizing air was adjusted to the required level. The bed temperature was maintained constant by using an adjustable cooling coil with circulating water. Concentrations of O 2 , CO, CO 2 , and SO 2 were recorded continuously by an ADC-RF infrared gas analyzer.
The experiments were carried out at bed temperatures of 800-850 ∘ C, fluidizing velocities of 1-2 m/s, and excess air levels of 0%-60%. Static bed height was 20-30 cm. Two types of coal bituminous Linby and Daw Mill of 3-16 mm (large) diameter in size and two types of limestone, Ballidone and Penrith of <3 mm mean diameter in size, were used. In both cases, the coal was premixed with the limestone and fed overbed at 42 cm above the distributor (see Table 1 for proximate and ultimate analyses of the coal and Table 2 for chemical composition of the limestone). The Ca/S ratio was 3 : 1 mole per mole or otherwise as indicated. Three levels of staging (15%-40% secondary air) were used to investigate the effect of fluidizing velocity, bed temperature, and excess air on SO 3 reduction during air staging.

Sampling of SO 3
An SSL/MEL SO 3 analyzer, developed by Severn Science Labs/Marchwood Eng. Labs, was used for continuous monitoring of the SO 3 . A detailed account of the principles and operating procedures of an SSL/MEL analyzer can be found in Jackson et al. [13] and Hotchkiss et al. [14].
The representative sample of flue gas was extracted from the sample point located near the exit of flue gas to the cyclone (200 cm above the distributor), where the gas temperature was around 550 ∘ C. Under these conditions, the use of a quartz sampling probe was found to be adequate. This enabled the extraction of the acid-containing gas directly into the filter-contactor of the SO 3 analyzer. The temperature of the sampled gas was maintained (by keeping the length of the tube as short as possible) at a value in excess of the acid dewpoint and below the temperature at which significant dissociation to SO 3 occurred; accurate determination of the acid gas content in the gas could then be made precisely over a significant period of time. This procedure effectively eliminates any interaction between the SO 3 in the gas sample and other species within the sampling probe itself.

SO 3 Emissions without Limestone under Unstaged Combustion.
The effect of some operating variables on the SO 3 emissions under unstaged combustion is presented in Figure 2. The results are corrected to 5% oxygen in the flue equivalent. The flue gas SO 3 emissions ranged from 5 to 10 for the Linby coal. SO 3 emissions decrease with excess air when corrected for dilution, increase slightly with excess air, reach limiting values, and then gradually decrease. The fluidizing velocity also affects emissions to some extent. The effect of bed height on SO 3 emissions was associated with the size of sand particles. As the fine sand produces more reducing environment, the oxygen stoichiometry influences the rates of oxidation of SO 2 , and as a result, the generation of SO 3 is reduced. The SO 3 emissions were also less sensitive to change in bed temperature. The rise in emissions was typically 0.5/10 ∘ C.
In another set of experiments, the Daw Mill Coal was tested for SO 3 emissions. The flue gas SO 3 emission ranged from 4 to 19.5 ppm for Daw mill coal. Changing the operating parameters resulted in a maximum change in SO 3 emission of 13 ppm. Increasing the bed depth resulted in higher SO 3 emission, the fluidizing velocity also affected emission depending on the size of bed material used. The SO 3 emission increased as the bed size was varied from coarse to fine with the bed depth. SO 3 emissions were weakly dependent on temperature typically rising 0.7 ppm/10 ∘ C. [4] have reported that the SO 3 formation under atmospheric pressure was very low (e.g., mole fraction of SO 3 in off gas = 7 ×10 −5 for mole fraction of SO 2 gas entering a bed = 2.3 ×10 −3 at 875 ∘ C). The oxidation rate in the gas phase was 100 times faster than expected. An Eley-Rideal mechanism was 4 ISRN Environmental Chemistry   proposed in which O 2 and SO 2 competitively chemisorb on the surface, and the rate of reaction is controlled by gas-phase molecule of SO 2 reacting with adsorbed O atom. Willium and Gibbs [5] have tested many coal types for SO 3 concentration without limestone in 750-900 ∘ C temperature range. Their findings suggest that ash (having traces of Ca, Mg, Na, K, etc.) is the principle removing species of SO 3 . In another experiment, when pure SO 2 was introduced, the SO 3 reacted with added char at 850 ∘ C in the absence of oxygen to give SO 3 of 7 vpm in the outlet, which suggests that char is important in the removal of SO 3 . He also observed a 50% reduction in SO 3 in the freeboard. According to Willium, the reduction was due to the reaction of SO 3 with unburnt char.

Comparison to Reported Work. Dennis and Hayhurst
SO 3 emissions are dependent on the oxygen and sulfur dioxide concentrations and were found to follow a similar trend. Willium and Gibbs [5] found that in contrast to the effect on SO 2 emissions, fine coal produced lower SO 3 emissions. In this study, SO 3 emissions were slightly higher when fine sand was used and tended to increase with bed height. This suggests that unburnt char does not have a significant effect on SO 3 emissions. The results of this study indicate that the amount of particles in the bed could have a significant effect on SO 3 emissions, resulting in an increase in the heterogeneous catalytic reaction of SO 2 to form SO 3 as the quantity of bed particles increases. Higher bed height, therefore, will also result in high SO 3 emissions. The oxygen concentration and fluidizing velocity will also affect SO 3 formation.

SO 3 Emissions with Limestone under Unstaged
Combustion. SO 3 emissions decrease in the presence of limestone and the reduction is temperature sensitive. The SO 3 reductions were less sensitive than the reductions achieved for SO 2 at similar conditions [15,16]. At a temperature around 850 ∘ C, the SO 3 reductions were only 28% of the SO 2 reductions, but  at 800 ∘ C, the reductions reached 70% of the SO 2 reduction level. The results corresponding to the operating conditions are shown graphically in Figures 3 and 4. It should be noted that the SO 3 reduction shown in Figure 3 was obtained when limestone was injected 12 cm above the distributor, and Figure 4 represents the results when limestone was injected 42 cm above the distributor. At a temperature of 850 ∘ C, some of the SO 2 will always be converted to SO 3 via reaction (6). The conversion decreases with oxygen concentration and increases with sulfur dioxide concentration. An increase in temperature enhances the rate of SO 3 formation. SO 3 can react with CaO to form CaSO 4 via reaction (7). The rate of this reaction is temperature dependent. Yilmaz et al. [17] studied the thermal dissociation of SO 3 in the range of 800-1200 ∘ C under atmospheric pressure. At the location in the flame where the net SO 3 formation rate is zero, he determined a rate constant of 6.9 × 10 10 cm 3 mol −1 s −1   24  22  20  18  16  14  12  10  8  6  4  2  0  0  2  4  6  8  10  12  14  16 18 Ca/S molar ratio Limestone injection height 42 cm, excess air, 32% Limestone injection height 42 cm, excess air, 60% for SO 3 + N 2 → SO 3 + O + N 2 ; that was consistent with other flame results. A high temperature lowers the reaction rate. Therefore, at a high temperature, more SO 3 is produced but less will be consumed in sulphation. As a result, a larger decrease in SO 3 emissions is observed at lower temperature. The effect of excess air can also be seen in these graphs. SO 3 emissions have been found to increase with excess air, but upon removing the dilution effect, the increase is within a narrow range, indicating that there could be an optimum reduction at a particular excess air beyond which the SO 3 reduction decreases. An increase in the fluidizing velocity has little effect on the overall reduction of SO 3 emissions.
During another set of experiments, Penrith limestone was added to the Daw Mill coal. It was observed that SO 3 emissions were decreased in the presence of limestone, and the reduction was temperature dependent. At the higher temperature of 850 ∘ C, the SO 3 reductions were 18%-20% of the SO 2 reduction, but at 800 ∘ C the SO 3 reductions reached 55% of the SO 2 reduction level. Figure 5 shows the results of this set.

Comparison to Reported Work (Conducted on Microbalance or Small Bed of 36-78 mm ID).
Burdett et al. [3] have reported that the reaction between limestone and sulfur oxides is highly sensitive to changes in O 2 , SO, and SO 3 concentrations. Absorption of SO 3 by the coal ash cannot be quantified on the microbalance, and the microbalance results are not applicable to fluidized combustor.
Fieldes et al. [12] have reported that extent of SO 2 oxidation to SO 3 varied with SO 2 and O 2 concentration. The coal combustion test showed that the lower SO 3 concentrations are due to its selective removal by ash. They had tested a variety of limestone, and in all the cases the mole fraction of CaO converted to CaSO 4 was affected by inlet oxygen in the same way as Penrith limestone.
Thibault et al. [18] have conducted experiments on a small (6 mm) fixed bed packed with CaO particle. They have tested two grain size of the sorbent and reported that for efficient capture of SO 3 a small grain size and open macropore structure are essential.

Comparison to Reported Work (Conducted on Pilot Scale).
Burdett et al. [19] have reported fractional conversion of SO 2 to SO 3 decreased from about 1.5% in the limestone-free case to around 0.35% when the limestone and alkaline ash were present, which was due to the greater reaction of SO 3 with limestone compared with ash, the absorption occurring both in the bed itself and in the freeboard.
Burdett et al. [10] have reported that combustion of a 3% sulfur coal in a bed burning at 900 ∘ C generated 33 vpm of SO 3 and proposed that the effect of O 2 on sulphation capacity results from the formation of SO 3 within the pores of the stone. [20] have conducted staged experiments on a quartz tube methane burner producing stable methane-H 2 S flame within desired fuel-air ratio without a sorbent presence under staged combustion conditions. They have reported that when the remaining excess air was injected into these gases, the maximum amount of SO 3 formed was greater than formed when this additional air was included with the initial combustion air, the overall excess of air being the same in both cases. The experimental conditions of Merryman and Levy do not match with our fluidized bed; therefore, their results are not comparable with this study.

SO 3 Emissions without Limestone under Staged Combustion. Merryman and Levy
During this study, the SO 3 emissions under staged combustion without limestone could not be monitored extensively due to malfunctioning of SO 3 analyzer.

SO 3 Emissions with Limestone under Staged Combustion.
The concentration of SO 3 emissions at 1.5 m/s and 20% excess air was 17.0 ppm, which decreased to 5.5 ppm in the presence of limestone. The SO 3 emissions at 70/30 staged (1.5 m/s, 850 ∘ C) combustion (without limestone) were similar to those of unstaged combustion (without limestone). However, it was observed that in the presence of limestone, staged combustion results in a higher reduction of SO 3 than unstaged. Figure 6 gives the SO 3 emissions as a function of PACR. The emissions at 15% secondary air were 1.5 ppm and increased to 7.2 ppm at 45% secondary air. It is clear that SO 3 is depressed on the addition of limestone during both the unstaged and staged operations, and the extent of reduction was higher under staged combustion. Figure 6 shows that the maximum removal of SO 3 occurs at a lower staging levels of 85/15, and, as the bed becomes more substoichiometric, the rate of SO 3 removal decreases. This trend indicates the formation of SO 3 in the freeboard which bypasses the limestone and appears in the flue. This increase in SO 3 reduction with the in-bed air ratio is in agreement with Barnes [6] findings.
The results of the staged combustion test with Daw Mill coal in the presence of Penrith limestone indicate that SO 3 emissions varied little with changes in excess air. However, if excess air is coupled with fluidizing velocity, then it had some effects on the emissions. At higher velocity of 2 m/s, the change was up to 4 ppm. The concentration of SO 3 emissions at 1.5 m/s and 30% excess air was 20 ppm which decreased to  It should be noted that there is no published work on SO 3 emissions under staged combustion conditions with or without limestone on any scale. Therefore, the results of this study could not be compared.

Conclusion
The experimental data shows that during unstaged combustion without limestone, SO 3 emissions are dependent on oxygen and SO 2 concentration. SO 3 emissions increase slightly with excess air, reaching a limiting value, and then slowly decrease. SO 3 emissions are less sensitive to change in bed temperature. However, the fluidizing velocity and bed height affect the emissions.
In the presence of limestone, SO 3 emissions are reduced during both staged and unstaged operations, and the reduction is temperature sensitive. However, during staged combustion, the reduction is enhanced. As staged fluidizedbed combustion is a proven technique to reduce NO and SO 2 emissions, therefore, it should be possible to operate a fluidized-bed combustor under a staged mode with limestone to keep SO 2 , SO 3 , and NO emissions to a minimum.