2190
Ind. Eng. Chem. Res. 2000, 39, 2190-2198
An Adaptive Sorbent for the Combined Desulfurization/Denitration Process Using a Powder-Particle Fluidized Bed Guangwen Xu, Guohua Luo, Hisashi Akamatsu, and Kunio Kato* Department of Biological and Chemical Engineering, Gunma University, Gunma, Kiryu 376, Japan
This study is devoted to examining an adaptive sorbent for the combined desulfurization/ denitration (DeSOx/DeNOx) process reported in work by Xu et al. (Trans. Inst. Chem. Eng. 1999, 77, Part B, 77-87). The so-called powder-particle fluidized bed (PPFB) is used by the process so that SOx in flue gas is absorbed by a continuously supplied fine DeSOx sorbent and NOx is reduced to N2 by ammonia under the catalysis of a coarse DeNOx catalyst, the fluidization medium particles (FMP). Experiments were conducted in a laboratory-scale PPFB reactor by using model flue gases containing SO2 and NO. It was found that the sodium carbonate supported by fine alumina particles (Na2CO3/Al2O3) had not only a high efficiency in absorbing SO2 but also little negative effect upon the simultaneous NOx reduction of the process. Using a DeNOx catalyst, such as V2O5‚WO3/TiO2 or WO3/TiO2, to catalyze the DeNOx reactions, high SO2 absorption and NO reduction in excess of 95% were achieved for model gases free of CO2 at stoichiometric ratios of SO2 to sorbent (Na/S ) 2.0) and NH3 to NO (NH3/NO ) 1). As for the simulated flue gas SO2-NO-H2O-N2-air with an oxygen fraction of 2 vol % and a water vapor fraction of 5 vol %, such high removals of SO2 and NO were found available even in a shallow PPFB with a static catalyst packing height of 0.1 m. By using silica sand as FMP, the pure SO2 absorption by the sorbent Na2CO3/Al2O3 was also examined with respect to various influential factors, such as reaction temperature (523-673 K), flue gas composition (oxygen fraction and water vapor fraction), sorbent diameter, and operating conditions (gas velocity and FMP bed height). 1. Introduction Emission of SOx and NOx from combusting fossil fuels such as coals and heavy oils has usually been tackled by separately installing the processes of flue gas desulfurization (FGD) and selective catalytic reduction (SCR).1 In comparison with these kinds of processes, the so-called combined desulfurization/denitration (DeSOx/ DeNOx) processes removing SOx and NOx in a single set of units have shown potentials in both costs (especially the capital cost) and removal efficiency.1,2 Accordingly, extensive efforts have been made in the latest decades by academic scholars and industrial engineers to develop these kinds of techniques.2-4 With the use of the so-called powder-particle fluidized bed (PPFB) originally developed by Kato et al.,5,6 a new combined DeSOx/DeNOx process has recently been proposed by some of the authors.7-10 As conceptualized in refs 8 and 10, a fine DeSOx sorbent, several to tens of micrometers in diameter, is continuously fed to a PPFB reactor in which a coarse DeNOx catalyst, several hundred micrometers in size, is fluidized with flue gas. Ammonia for NOx reduction is usually fed to the bottom of the reactor. As a consequence, both SOx and NOx in the flue gas can be simultaneously removed with the single reactor. The spent sorbent entrained with the cleaned gas is collected at the reactor exit by using a solid separator, such as a bagfilter, which then is either regenerated or disposed of as waste. As has been analyzed by Xu et al.,10,11 definite potentials are afforded by this combined DeSOx/DeNOx process for use in flue gas cleaning, with thus the work * To whom correspondence should be addressed. Tel: 0081-277-30-1459. Fax: 0081-277-30-1457. E-mail: kato@ bce.gunma-u.ac.jp.
on selection of DeSOx sorbents and DeNOx catalysts ongoing since 1995.7,12,13 Recently, the removals of SOx and NOx have been further evaluated by Xu et al.,11 for showing whether they are really caused by the DeSOx reactions between SOx and sorbent and the NOx reduction by ammonia, respectively. A latest work of Xu et al.14 on the process has also looked into the influences of flue gas composition on the removal efficiency. All of these previous examinations have shown that the titanium-supported DeNOx catalysts (in granular state), such as V2O5‚WO3/TiO2 and WO3/TiO2, are adaptive for the NOx reduction of the process. At NH3/NO ) 1.0, removal of NO can approach 100% if any of such catalysts is used to catalyze the DeNOx reactions between NOx and NH3.9,10 Various metallic salts and oxides have been tested for sieving highly efficient DeSOx sorbent.7,12,13 It was found that the copper oxide sorbent supported by alumina particles, CuO‚V2O5‚K2SO4/Al2O3, enables high SO2 removal in excess of 95% at the stoichimetric ratio of Cu/S ) 1.0.9,13 Among the other sorbents tested, sodium bicarbonate (NaHCO3) has an efficiency next to that of copper oxide, which can afford an SO2 removal around 70% at Na/S ) 3.0.9-12 However, the sorbent CuO‚V2O5‚K2SO4/Al2O3 gives serious poisoning to the simultaneous NOx reduction taking place in the same reactor, while NaHCO3 allows NOx to be completely reduced by ammonia. The SO2 and NO removals for a simulated flue gas, SO2-NO-H2O-N2air, obtained by using these two sorbents were reported in ref 10 (its Figure 11). Because of the poisoning of NOx reduction, the NO removal using CuO‚V2O5‚K2SO4/ Al2O3 hardly reached 90%, and it also decreased with an increase in the water vapor fraction (Figure 11a,b). Compared to this, it was shown that the NO removal using NaHCO3 as the DeSOx sorbent was close to 100%
10.1021/ie9908027 CCC: $19.00 © 2000 American Chemical Society Published on Web 04/26/2000
Ind. Eng. Chem. Res., Vol. 39, No. 7, 2000 2191
Figure 1. Schematic diagram of the experimental apparatus.
under all conditions tested (Figure 11c). The poisoning of NOx reduction by CuO‚V2O5‚K2SO4/Al2O3 has been considered to be due to the presence of potassium sulfate (K2SO4) in this sorbent,9,10,13 as was similarly observed by other researchers.15-17 Therefore, one can see that there has not yet been an adaptive sorbent for the combined DeSOx/DeNOx process using PPFB such that the sorbent not only enables high SO2 removal but also gives few side effects, such as poisoning, to the simultaneous NOx reduction. The result summarized above implies that the sorbent composed of element sodium (Na) does not poison the titanium-supported DeNOx catalysts (the used ones are V2O5‚WO3/TiO2 and WO3/TiO2). On the other hand, sodium salts, especially NaHCO3 and Na2CO3, have long been recognized as kinds of applicable sorbents for DeSOx and have been extensively tested in both laboratory and application scales.18-22 In addition to a good efficiency in SOx adsorption, the sorbent is also easily manufacturing because NaHCO3 is rich in natural ores such as Nahcolite and Trona.18,23 With these issues, one thus may suggest that enhancing the absorption capacity of sodium carbonate should be a good solution to the adaptive DeSOx sorbent for the combined DeSOx/DeNOx process using PPFB. This is just the objective of this study, that is, to prepare a Na2CO3-based sorbent and to test its performance. 2. Experimental Section Shown in Figure 1 is a schematic diagram of the laboratory-scale facility used in this study. While the major sizes of the reactor are mentioned in Figure 1, the details about the apparatus, measurement method, and operation procedure can be found in Xu et al.10 In addition, the spent sorbent collected by a bagfilter at the reactor exit was measured using an electron probe microanalyzer (EPMA), for exploring the profiles of element sulfur and sodium across a sorbent particle. Most experiments were performed using the simplest model flue gas SO2-NO-air. Similar to refs 10, 11, and 14, a model gas, SO2-NO-H2O-N2-air (CO2 was not considered), was employed to simulate actual flue gas, of which nitrogen (N2) was used to adjust the oxygen fraction of the model gas. Removals of SO2 and NO were
Table 1. Properties of Fine Sorbent Powders and Coarse Medium Particles Used in This Study particles NaHCO3 Na2CO3 Na2CO3/Al2O3b silica sand WO3/TiO2 V2O5‚WO3/TiO2c
dpf, dpc [µm]
Fp [kg/m3]
Fb [kg/m3]
45 45 39 ( 673 K for the experimental conditions mentioned in Figure 3. This implies that the SOx absorption efficiency of Na2CO3 (similarly NaHCO3) is greatly enhanced by supporting it onto particle alumina. At T > 673 K, the sorbent even enables an SOx absorption in accordance with the stoichiometric ratio of Na/S ) 2.0. Therefore, one could deduce that the new sorbent made from Figure 2 is promising for the use in the combined DeSOx/DeNOx process using PPFB, noting that the preferred operating temperatures of the process are in 573-773 K.10 With an increase in the reaction temperature T in 523-873 K (the tested range), Figure 3 demonstrates that the SO2 abatement for pure Na2CO3 and Na2CO3/ Al2O3 gradually increases, while that for NaHCO3 first decreases and then increases. This evidences again the previous analysis that little Na2CO3 supported by alumina particles has been converted into NaHCO3 in experiencing the production procedure specified in Figure 2. The distinct variation of YSO2 with reaction temperature T for NaHCO3 was suggested to be caused by the thermal decomposition of NaHCO3 into Na2CO3, CO2, and H2O(vapor).6,9,10,11,28 The high-efficiency absorbing SO2 of the fine Na2CO3/ Al2O3 should be attributed not only to the support of Na2CO3 by alumina but also to the small size of sorbent particles. Figure 4 indicates that the larger the sorbent size, the smaller the removal YSO2 under the same operating conditions. Sodium carbonate impregnated on coarse alumina with an average size of 1600 µm was used by the so-called NOXSO process in which SOx and NOx of flue gas are simultaneously absorbed by the sorbent at a temperature of around 393 K.2,22,29,30 According to the examination results in the so-called life-cycle-test unit (LCTU),29 the molar ratio of Na to 2SOx + NOx was raised even up to 2.5, which is much larger than the stoichiometric value of 1.0 (the stoichiometric ratios are Na/SOx ) 2.0 and Na/NOx ) 1.0, respectively). The reason is that the reactions between the alumina-supported Na2CO3 and gas SOx (as well as NOx in the NOXSO process) still mainly take place on the surface of the sorbent particles. Element profiles across a spent sorbent particle, 53-88 µm in size, are shown in Figure 5, with Figure 5a for sodium (Na) and Figure 5b for sulfur (S). In each of the figures, the
Figure 5. Element distribution across a spent particle for Na2CO3/Al2O3 (dpf ) 53-88 µm, FMP ) silica sand, SO2-air, T ) 623 K, Ls ) 0.3 m, Na/S ) 2.0, YSO2 ) 88%).
brighter the color, the more the measured element implies. Figure 5a shows that, even for the fine alumina particle used in this study, the impregnated Na2CO3 still mainly remains on the surface of the particle, although more or less Na can be traced inside the particle. Corresponding to this, the adsorption of sulfur in desulfurization also mainly occurs on the sorbent surface, as demonstrated by the sulfur profile shown in Figure 5b. Therefore, the smaller the sorbent size, the higher the absorption ability should be, as benefited from the increased specific surfaces of the sorbent. In response, the removal of SO2 for the 39 µm sorbent can reach 100% in Figure 4 (at Na/S ) 3.0), whereas YSO2 for the other larger sorbents hardly approaches such a high value, showing the importance using fine alumina particles. 3.1.2. Parametric Dependencies. The SO2 removal for a model gas simulating an actual flue gas, that is, with a water vapor fraction of 5-10 vol % and an oxygen content of 2-4 vol %,27 is given in Figure 6 (here, CO2 was not considered). It shows that the SO2 abatement by the fine Na2CO3/Al2O3 can be greatly enhanced with water vapor, regardless of the oxygen fraction CO2 in the model gas. With an increase in the water vapor fraction CH2O, the removal YSO2 evidently increases and reaches 100% at CH2O ) 5 vol %. Using other DeSOx sorbents, our previous examinations have also found
2194
Ind. Eng. Chem. Res., Vol. 39, No. 7, 2000
Figure 6. SO2 removal with respect to the water vapor fraction under two different oxygen fractions for sorbent Na2CO3/Al2O3.
Figure 7. SO2 removal with respect to the static bed height, gas velocity, and apparent gas residence time for sorbent Na2CO3/ Al2O3.
that the DeSOx reactions between sorbent and SOx can be more or less promoted by water vapor.10,11,14 Nonetheless, the sorbent Na2CO3/Al2O3 used here appears more sensitive to the change in the water vapor fraction, which should be attributed to the high porosity of the sorbent. From the bulk density and particle density specified in Table 1, one can figure out that this fine sorbent has a very high packing voidage, implying a strong absorption for water. Thus, at a given percentage of water vapor in a model gas, the fine Na2CO3/Al2O3 can catch more water vapor than other low-porosity sorbents such as pure Na2CO3 and NaHCO3. As a consequence, the SO2 removal by the fine Na2CO3/Al2O3 shows greater variation with the change in the water vapor fraction, as is demonstrated in Figure 6. When the oxygen fraction in the model gas is reduced from 20 vol % (SO2-H2O-air) to the lowest value of CO2 ) 2 vol % for the actual flue gas,27 Figure 6 shows that the removal YSO2 takes a slight decrease at CH2O ) 0. However, with water vapor presented in the gas, such a reduction in CO2 causes little effect on the removal efficiency. This is consist with our observations upon the sorbent CuO‚V2O5‚K2SO4/Al2O3 measured in the same apparatus11,14 and is considered to result from two causes: first, the high SO2 absorption efficiency of these sorbents, which makes YSO2 insensitive to the variation in the DeSOx reaction rate raised by reducing CO211,14 and, second, the promotion of water vapor to the removal efficiency, which actually compensates the decrease in SOx capture caused by lowering the oxygen content. With an assumption of a plug gas flow, the influences of the static height Ls of coarse particle bed and the superficial gas velocity Ug are plotted in Figure 7 in terms of the relationship between YSO2 and the apparent
Figure 8. SO2 removal for pure Al2O3 powders under different water vapor fractions.
gas residence time in the coarse particle bed, Ls/Ug. In accordance with Figure 6, the model gas SO2-H2Oair with a water vapor fraction of CH2O ) 5 vol % was used in this measurement. One can see that SO2 removal increases with increasing Ls under a specified gas velocity Ug and decreases with increasing Ug at a given static coarse particle bed height Ls, just consistent with the dependence of YSO2 upon Ls/Ug demonstrated in the figure. Furthermore, Figure 7 indicates that the removal for fine Na2CO3/Al2O3 is subject to a unified relation of YSO2 to the apparent gas residence time Ls/ Ug, and there is little difference among operations at different Ls. According to Xu et al.,11 this is valid only for high efficient sorbent, and the SO2 removal for lowefficiency sorbent, such as pure Na2CO3 and NaHCO3, actually has different curves of YSO2 versus Ls/Ug distinguished by different Ls. The latter phenomenon is considered to be related to the de-efficiency in SO2 absorption caused by gas bubbles prevailing in deeper fluidized beds. For high-efficiency sorbent, however, this de-efficiency can be alleviated by the fast reactions between SOx and sorbent, thus enabling YSO2 to be little affected by the flow structure, such as gas bubbling.11 3.1.3. Removal Evaluation. While Figures 6 and 7 indicate that the sorbent Na2CO3/Al2O3 in a mean diameter of 39 µm can stoichiometrically absorb SO2, if CH2O is greater than 5 vol % in flue gas, the experiment has also observed that the SO2 removal at Na/S < 2.0 is usually a little bit higher than the value of Na/S. For example, YSO2 can be 95% at a Na/S of around 1.80, although it should not be in excess of 90% in case gas SO2 is absorbed only by Na2CO3. It is considered that this result is related to the physical adsorption of SO2 onto the porous alumina particles. Accordingly, a measurement was extended to detect the SO2 adsorption by a pure alumina sorbent manufactured by similarly following the procedure given in Figure 2 but without adding Na2CO3. The SO2 adsorption at the same sorbent feed rate as that for Na/S ) 2.0 is shown in Figure 8, indicating that this adsorption hardly reaches 20% and also varies little with water vapor fraction CH2O in the range of 0-10 vol %. Therefore, one can suggest that the SO2 removal for the sorbent Na2CO3/Al2O3 at Na/S < 2.0 should mainly stem from the SO2 absorption by Na2CO3. In this case (Na/S < 2.0), the adsorption on alumina possibly contributes to the total SO2 removal, but it should surrender to the SO2 absorption by the active component Na2CO3. That is, SO2 should first react with Na2CO3, and only if the supplied Na2CO3 is insufficient to capture all SO2 fed to the system (Na/S < 2.0) can unreacted SO2 hold in the reactor and can a part of it be allowed to adsorb onto alumina particles.
Ind. Eng. Chem. Res., Vol. 39, No. 7, 2000 2195
Figure 9. Removals of SO2 and NO at different Na/S for sorbent Na2CO3/Al2O3 and catalyst V2O5‚WO3/TiO2.
Figure 10. Removals of SO2 and NO under different operating conditions using two different catalysts and sorbent Na2CO3/Al2O3.
3.2. Simultaneous SOx and NOx Removals. From the examinations shown above, it can be seen that the fine alumina-supported Na2CO3 is of high efficiency in absorbing SO2 within the PPFB reactor operated at 573-773 K. With DeNOx catalyst, either V2O5‚WO3/TiO2 or WO3/TiO2, as FMP, the performance of the sorbent in the combined DeSOx/DeNOx process using PPFB is further demonstrated in Figures 9 and 10. 3.2.1. Overall Conception. Figure 9 shows the SO2 and NO removals versus Na/S for a model gas SO2NO-air, indicating that a high NO abatement of YNO > 95% is allowed by the fine Na2CO3/Al2O3 at NH3/NO ) 1, the stoichiometric ratio of ammonia to NO for NO reduction. This implies that the alumina-supported sodium does not have negative action, such as poisoning, onto the NOx reduction simultaneously taking place in the same reactor. On the other hand, the sorbent also absorbs SO2 with high efficiency; that is, a nearly complete SO2 abatement is available at the stoichiometric value of Na/S ) 2.0. Therefore, one can conclude
that the fine Na2CO3/Al2O3 is an adaptive sorbent for the combined DeSOx/DeNOx process using PPFB, with high efficiency in absorbing SO2 and low negative effects on NOx reduction. At the same time, Figure 9 also shows that the SO2 removal for every specified Na/S at Na/S < 2.0 is higher than the available value calculated from Na/S, i.e., (Na/ S)/2.0. For example, a sorbent feed rate at Na/S ) 1.0 can reach an SO2 removal near 85% in Figure 9, which is much higher than 50% calculated from Na/S ) 1.0 under the assumption that SO2 is completely absorbed by Na2CO3 supported by alumina. In addition, the removal at the same Na/S is much larger in Figure 9 using the DeNOx catalyst as FMP compared to that in Figure 3 where silica sand is used as FMP. These greater removals of SO2 are considered to be due to the SOx adsorption onto the catalyst particles. As revealed by Xu et al.,11,14 SOx, including SO3 formed in SO2 oxidation, may be adsorbed onto DeNOx catalyst, especially when oxygen presents in flue gas. However, the SOx adsorption surrenders to the DeSOx reactions for highly efficient sorbent, that is, SOx tends to first react with sorbent, and the adsorption can come into effect only if the DeSOx sorbent is insufficient to capture all SOx fed to the system. The sorbent feed rate in Figure 9 is lower than the one required by the stoichiometric Na/S of 2.0, thus causing the insufficiency of the sorbent supply and making the overall removal of SO2 become larger than the value calculated from Na/S. Nonetheless, the overall YSO2 is still less than 100%, because only a part of unreacted SOx can be adsorbed onto the surfaces of the catalyst. As Na/S increases, this SOx adsorption reduces gradually because of the gradually decreased amount of unreacted SO2, resulting in the gradual shrink of the difference between the apparent YSO2 and (Na/S)/2.0 during the increase of Na/S to its stoichiometric value of 2.0. Once Na/S reaches 2.0, the sorbent enables all SO2 fed to the system to be absorbed, leading to a zero SOx adsorption. In this case, the removal of YSO2 ) 100% is actually afforded by the DeSOx reactions of SOx with sorbent, as identified by Xu et al.11 through measuring the sulfur content in the spent sorbent. 3.2.2. Removal for Simulated Flue Gas. Plotted in Figure 10 are the removals of SO2 and NO under different operating conditions using the sorbent Na2CO3/Al2O3 (dpf ) 39 µm) and two titanium-supported catalysts. The model gas, SO2-NO-H2O-N2-air, was used to simulate the actual flue gas, with thus an oxygen fraction of 4 vol %.27 While gas velocity Ug was maintained at 0.5 m/s, the static height of catalyst bed, Ls, was set at lower values of Ls ) 0.15 and 0.1 m, aimed at examining the performances at lower gas residence time within the catalyst bed (refer to Figure 7 and ref 11). From Figure 10a one can see that the NO removal in excess of 95% can be ensured with the process for the simulated flue gas (CO2 ) 4 vol %, CH2O ) 0-10 vol %), even in a shallow catalyst bed with Ls ) 0.1 m. Also, both the titanium-supported catalysts, V2O5‚WO3/TiO2 and WO3/TiO2, appear adaptive for such high removals of NO. Water vapor and oxygen fractions in the flue gas slightly affect the NO removal, but their effects are so weak that YNO actually remains at similar values larger than 95% when changing CH2O in 0-10 vol % and CO2 in 4-21 vol %. While more details about the influences of CH2O, CO2, Ls and Ug upon NO removal can be found
2196
Ind. Eng. Chem. Res., Vol. 39, No. 7, 2000
in Xu et al.10,11,14 over the same catalysts but the sorbent NaHCO3, one can see that Figure 10a falls in good consistence with those previous examinations. This shows further that the sorbent Na2CO3/Al2O3 does not negatively affect the NOx reduction, as has been observed for pure NaHCO3. Removal of SO2 in Figure 10b further evidences the high efficiency of the fine Na2CO3/Al2O3 in absorbing SO2. Because of this, there is little difference among the SO2 removals using different catalysts. At Na/S ) 2.0, the removal has relatively lower values at CH2O ) 0 (9095%) but rapidly increases to values near 100% when CH2O is raised to 5 vol %. This finding is consistent with the demonstration of Figure 6; that is, the SO2 absorption of fine Na2CO3/Al2O3 can be considerably facilitated by increasing the water vapor fraction in the range of CH2O < 5 vol % (particularly for the case of Na/S ) 2.0). Because YSO2 already reaches 100% at CH2O ) 5 vol %, the further increase of the water vapor fraction to 10 vol % results in little variation in SO2 removal (CH2O in the real flue gas is 5-10 vol %). At CH2O ) 0, a little bit lower value of YSO2 shows for CO2) 4 vol %, in comparison with that for CO2 ) 20 vol % (CO2 for model gas SO2NO-H2O-air is about 20 vol %). This is due to the reduced desulfurization reaction rate at lower oxygen fraction. Nonetheless, such an influence shrinks to zero when CH2O increases to 5-10 vol %, implying that the decreased removal by lowering CO2 is completely compensated for by the removal increase which stemmed from raising CH2O. In addition, the result in Figure 10b suggests that a catalyst bed with a Ls between 0.1 and 0.2 m should be enough for realizing high SO2 removal by the fine Na2CO3/Al2O3 (lower Ls is expected for practical application). Comparing with Figure 6, one can also see that YSO2 is a little bit higher in Figure 10a than in Figure 6 at CH2O ) 0. It should be ascribed to the catalysis of the DeNOx catalyst for the DeSOx reactions as well as the SOx adsorption onto DeNOx catalyst.11,14 Nonetheless, the SO2 removals using different FMP (silica sand and catalysts) actually have the same value near 100% at CH2O > 5 vol %. As a summary, Figure 10 shows that high SO2 and NO removals can be simultaneously achieved with the combined DeSOx/DeNOx process using PPFB, if using the fine Na2CO3/Al2O3 as DeSOx sorbent and the titanium-supported DeNOx catalyst as FMP. Also, removals of SO2 and NO in excess of 95% appear available for the model gas SO2-NO-H2O-N2-air even in a shallow PPFB reactor with a catalyst packing height between 0.1 and 0.2 m (this value may vary with scales of reactors). Nonetheless, a further study is necessary for the influence of carbon dioxide on the DeSOx efficiency of the sorbent, because the real flue gas contains 10-15 vol % CO2.27 3.3. Prospect and Problems. The foregoing examinations indicate that the fine Na2CO3/Al2O3 is an adaptive sorbent for the combined DeSOx/DeNOx process using PPFB (the adaptation of titanum-supported catalyst to the NO removal of the process has been shown in refs 10 and 11). The main component of the sorbent, Na2CO3, is rich in natural ores, such as Nahcolite and Trona. In addition, one may suggest that many other alkali wastes generated in various industrial processes could be used for the production of the sorbent. While all of these show the prospects of the sorbent for use in practical flue gas cleaning, there are also a few points calling for further researches.
Figure 11. Conception of an application-oriented process using PPSB for producing fine Na2CO3/Al2O3.
3.3.1. Regeneration of Spent Sorbent. Sodium sulfates are easily dissolvable in water; thus, the spent sorbent should not be disposed of as natural wastes, while on the other hand, the direct disposal might also greatly increase the cost of the sorbent. Fortunately, the NOXSO process reported that the sulfate in the spent sorbent can be reduced into SO2, H2S, and sulfide using a reducing gas, such as natural gas. These gas products then can be further derived from the sorbent with stream and can be used to produce sulfur.22,29,30 As for the case of this study, more consideration should be made to look for a suitable technique for treating the fine sorbent. It is considered that the regeneration of small-size particles may be performed in a bagfilter, that is, using a reducing gas to directly purge the bagfilter collecting the spent sorbent. Of course, the use of PPFB could also be a technique suitable for this treatment. 3.3.2. Production of Sorbent. In laboratory tests, the fine Na2CO3/Al2O3 can be made in accordance with the procedure showing Figure 2, which obviously is unsuitable for the utilization in industrial production. Essentially, the preparation of the sorbent comprises two major steps, coating Na2CO3 onto fine alumina particles and drying the Na2CO3-coated sorbent. The socalled powder-particle spouted bed (PPSB) has been used by Guo et al.31 and Xu et al.32 for continuously drying microparticles. It is considered that this kind of spouted bed is also promising for the commercial production of the fine Na2CO3/Al2O3. As conceptualized in Figure 11, fine alumina is carried by hot air and continuously fed into a PPSB in which coarse silica sand is fluidized as FMP. At the bed entrance, a solution of Na2CO3 is sprayed into the gas flow, allowing it soon to mix with the fine alumina particles scattered in the gas flow. Ahead of this, both alumina particles and Na2CO3 solution droplets further advance into PPSB where they turbulently mix with the FMP. Fine alumina powders tend to adhere onto the surfaces of FMP and thus can reside in the bed for a long time.6 Benefiting from this, the alumina particles are allowed to be evenly wetted by the Na2CO3 solution and then dried in the same vessel through contacting hot air, the fluidizing gas. When FMP is peeled, the dried alumina powders are again entrained with the fluidizing gas and finally collected at the bed exit. Evidently, the fine particles obtained via these steps must be the same sorbent as the one prepared according to Figure 2, because a layer of Na2CO3 has already been coated on the surfaces of the collected alumina powders. That is, the production
Ind. Eng. Chem. Res., Vol. 39, No. 7, 2000 2197
of the fine Na2CO3/Al2O3 is likely to be implemented in a single PPSB. As a matter of fact, a similar method has been used by Xu et al.33 for manufacturing salt powders (NaCl) from drying a salt solution, demonstrating that PPSB is capable of producing fine powers in a few micrometers. In addition, one can see that the technique illustrated in Figure 11 is also appropriate for commercialization. 3.3.3. Optimization of Sorbent. As for the sorbent itself, further researches are needed for optimizing its performance. In this study, the weight ratio of Na2CO3 to alumina is set at 20%, an arbitrarily selected value in fact. From the viewpoint of practical application, the larger the ratio, the lower the costs should be, in terms of both materials and operation. On the other hand, the efficiency of the sorbent declines with an increase in this ratio because of the decreased supporting surfaces of the alumina powders. Therefore, what should be the optimal value of Na2CO3 versus Al2O3 calls for further researches. The selection of alumina materials should be another important subject related to the sorbent. The alumina particles with the least cost are expected, but the lower the cost, the less the specific areas usually are. The third point significant to the sorbent is the choice of the main component of the sorbent. Pure Na2CO3 is used in this study. How about the use of natural ores? The direct use of ores containing Na2CO3 and NaHCO3 could greatly reduce the material cost. Of course, the more economic way might be the use of alkali wastes generated in various industrial processes. Nonetheless, to put the process into practical applications, several other problems must also be considered and further studied, such as the scale-up of the PPFB reactor, the total pressure loss across the process, and the overall economy of the process in comparison with other techniques. 4. Conclusions An adaptive sorbent for the combined DeSOx/DeNOx process using the so-called PPFB was examined in a laboratory-scale PPFB reactor. It was found that sodium carbonate supported by fine alumina particles (Na2CO3/ Al2O3) has not only a high efficiency in absorbing SO2 but also little negative action on the NOx reduction simultaneously taking place in the process. By using the sorbent Na2CO3/Al2O3 and a titanium-supported DeNOx catalyst, either V2O5‚WO3/TiO2 or WO3/TiO2, removals of SO2 and NO in excess of 95% were achieved for the simulated flue gas SO2-NO-H2O-N2-air at the stoichiometric ratios of SO2 to sorbent (Na/S ) 2.0) and NH3 to NO (NH3/NO ) 1). For flue gas containing oxygen greater than 2 vol % and water vapor between 5 and 10 vol % (CO2 was not considered), such high removals were found available even in a shallow PPFB with a static catalyst packing height of 0.1 m. By using silica sand as the FMP, similarly high SO2 removal was also achieved. On the basis of the laboratory examinations, discussions indicated that the fine Na2CO3/Al2O3 possibly be produced with a technique suitable for commercialization, which uses the so-called PPSB to simultaneously implement the coating of Na2CO3 onto fine alumina particles and the drying of the Na2CO3coated sorbent. Further studies were also suggested for the regeneration of the used sorbent, the optimizion of the sorbent performance in terms of cost and removal efficiency, and a few issues related to the practical applications of the process.
Acknowledgment The authors are grateful to the Steel Industry Foundation for the Advancement of Environmental Protection for its financial support. Nomenclature Ci ) volume fraction of component gas i ()SO2, NO, O2, H2O) [vol % or ppm] Ci,0 ) Ci ()SO2, NO) at the inlet before reactor [ppm] Ci,1 ) Ci ()SO2, NO) at the reactor exit [ppm] dpc ) surface to volume average particle diameter of coarse medium particles [m] dpf ) surface to volume average particle diameter of fine sorbent particles [m] Ls ) static height of the coarse particle bed [m] T ) reaction temperature [K] ut ) terminal velocity [m/s] Ug ) superficial gas velocity [m/s] Yi ) removal of gas i ()SO2, NO) Fp ) particle density [kg/m3] Fb ) bulk density of particles [kg/m3]
Literature Cited (1) Soud, H. Suppliers of FGD and NOx Control Systems; IEA Coal Research: London, 1995; pp 18-27, 66. (2) Soud, H. N.; Fukasawa, K. Developments in NOx Abatement and Control; IEA Coal Research: London, 1996; pp 62-70, 8291. (3) Boer, F. P.; Hegedus, L. L.; Gouker, T. R.; Zak, K. P. Controlling Power Plant NOx Emissions. CHEMTECH 1990, 20, 312-319. (4) Rhoads, T. W.; Marks, J. R.; Siebert, P. C. Overview of Industrial Source Control for Nitrogen Oxides. Environ. Prog. 1990, 9, 126-130. (5) Kato, K.; Takarada, T.; Koshinuma, A.; Kanazawa, I.; Sugihara, T. Decarbonization of Silicon Carbide-Carbon Fine Particles Mixture in a Fluidized Bed. in Fluidization VI; Grace, J. R., Shemilt, L. W., Bergougnou, M. A., Eds.; Engineering Foundation: New York, 1989; pp 351-358. (6) Kato, K.; Takarada, T.; Matsuo, N.; Suto, T.; Nakagawa, N. Residence Time Distribution of Fine Particles in a PowerParticle Fluidized Bed. Kagaku Kogaku Ronbunshu 1991, 17, 970975. (7) Gao, S.; Suzuki, H.; Nakagawa, N.; Bai, D.; Kato, K. Simultaneous Removal of SO2 and NO in a Powder-Particle Fluidized Bed by Using Iron Oxide Dust as Sorbent. Sekiyu Gakkaishi 1995, 38, 399-406. (8) Gao, S.; Nakagawa, N.; Kato, K.; Inomata, M.; Tsuchiya, F. Simultaneous SO2/NOx Removal by a Powder-Particle Fluidized Bed. Catal. Today 1996, 29, 165-169. (9) Kato, K.; Gao, S. A New Process for Simultaneous Removal of SO2 and NO from Flue Gas by Powder-Particle Fluidized Bed. Sekiyu Gakkaishi 1997, 40, 443-453. (10) Xu, G.; Gao, S.; Suzuki, H.; Wang, B.; Ma, X.; Nakagawa, N.; Kato, K. An Innovative Combined Desulfurization/Denitration Process Using a Powder-Particle Fluidized Bed. Trans. Inst. Chem.Eng. 1999, 77, 77-87. (11) Xu, G.; Wang, B.; Suzuki, H.; Gao, S.; Ma, X.; Nakagawa, N.; Kato, K. Removal Efficiency of the Combined Desulfurization/ Denitration Process Using Powder-Particle Fluidized Bed. J. Chem. Eng. Jpn. 1999, 32, 82-90. (12) Gao, S.; Suzuki, H.; Nakagawa, N.; Bai, D.; Kato, K. Selection of SO2 Sorbents for Simultaneous Removal of SO2 and NO in the Powder-Particle Fluidized Bed. Sekiyu Gakkaishi 1996, 39, 59-64. (13) Gao, S.; Suzuki, H.; Nakagawa, N.; Kato, K. Performance of a New Copper-based Sorbent with the Addition of Potassium Sulfate for SO2/NO Removal in a Powder-Particle Fluidized Bed. Sekiyu Gakkaishi 1997, 40, 227-231. (14) Xu, G.; Wang, B.; Suzuki, H.; Kato, K. Performance with Respect to Flue Gas Composition of a Combined Desulfurization/ Denitration Process Using Powder-Particle Fluidized Bed. Chin. J. Chem. Eng. 1999, 7, 295-306.
2198
Ind. Eng. Chem. Res., Vol. 39, No. 7, 2000
(15) Nishijima, A.; Kurita, M.; Sato, T.; Kiyozumi, Y.; Hagiwara, H.; Ueno, A.; Todo, T. Effect of the Dust Components on the Life of Catalysts Used for the Reduction of NOx in the Sintering Furnace Gas. Nippon Kagaku Kaishi 1978, 6, 893-898. (16) Shikada, T.; Fujimoto, K. Effect of Added Alkali Salts on the Activities of Supported Vanadium Oxide Catalysts for Nitric Oxide Reduction. Chem. Lett. 1983, 77-80. (17) Fujimoto, K.; Shikada, T. Regeneration of V2O5-TiO2 Catalysts for Nitrogen Monoxide Reduction Poisoned by Potassium Salts. Chem. Lett. 1983, 515-518. (18) Furlong, D. A.; Ostop, R. L.; Drehmel, D. C. Selection, Preparation, and Disposal of Sodium Compounds for Dry SOx Scrubbers. Environ. Inter. 1981, 6, 99-101. (19) Jorgensen, C.; Chang, J. C. S. Evaluation of Sorbents and Additives for Dry SO2 Removal. Environ. Prog. 1987, 6, 26-32. (20) Muzio, L. J.; Offen, G. R. Dry Sorbent Emission Control Technologies, Part I Fundamental Processes. JAPCA 1987, 37, 642-654. (21) Erdo¨s, E.; Mocek, K.; Lippert, E.; Uchytilova´, V.; Neuzˇil, L.; Bejcek, V. Application of the Active Soda Process for Removing Sulfur Dioxide from Flue Gas. JAPCA 1989, 39, 1206-1209. (22) Ma, W. T.; Haslbeck, J. L. NOXSO SO2/NOx Flue Gas Treatment Process Proof-Concept Test. Environ. Prog. 1993, 12, 163-168. (23) Genco, J. M.; Rosenberg, H. S.; Anastas, M. Y.; Rosar, E. C.; Dulin, J. M. The Use of Nahcolite Ore and Bag Filters for Sulfur Dioxide Emission Control. JAPCA 1975, 25, 1244-1253. (24) Xu, G.; Li, J. Analytical Solution of the Energy-Minimization Multi-Scale Model for Gas-Solid Two-Phase Flow. Chem. Eng. Sci. 1998, 53, 1349-1366. (25) Keener, T. C.; Davis, W. T. Study of the Reaction of SO2 with NaHCO3 and Na2CO3. JAPCA 1984, 34, 651-654. (26) Keener, T. C.; Frazier, G. C.; Davis, W. T. Thermal
Decomposition of Sodium Bicarbonate. Chem. Eng. Commun. 1985, 33, 93-105. (27) Yeh, J. T.; Drummond, C. J.; Joubert, J. I. Process Simulation of the Fluidized Bed Copper Oxide Process Sulfation Reaction. Environ. Prog. 1987, 6, 44-50. (28) Howatson, J.; Dewald, H. D.; Outka, D. A.; Diller, D. J.; Cain, M. B.; Smith, J. W. Reaction of Nahcolite with Sulfur Dioxide. JAPCA 1980, 30, 1229-1230. (29) Ma, W. T.; Halsbeck, J. L.; Neal, L. G.; Yeh, J. T. Life Cycle Test of the NOXSO SO2 and NOx Flue Gas Treatment Process: Process Modeling. Sep. Technol. 1991, 1, 195-202. (30) Yeh, J. T.; Ma, W. T.; Pennline, H. W.; Halsbeck, J. L.; Joubert, J. I.; Gromicko, F. N. Integrated Testing of the NOXSO Process: Simultaneous Removal of SO2 and NOx from Flue Gas. Chem. Eng. Commun. 1992, 114, 65-88. (31) Guo, Q.; Hikida, S.; Takahashi, Y.; Nakagawa, N.; Kato, K. Drying of Micro-Particle Slurry and Salt-Water Solution by a Powder-Particle Spouted Bed. J. Chem. Eng. Jpn. 1996, 29, 152158. (32) Xu, J.; Osada, S.; Kato, K. Limiting Efficiency for Continuous Drying of Microparticles in a Powder-Particle Spouted Bed. J. Chem. Eng. Jpn. 1998, 31, 35-40. (33) Xu, J.; Osada, S.; Kato, K. Production of Very Fine Particles from Salt-Water Solution by a Powder-Particle Spouted Bed. In Circulating Fluidized Bed Technology VI; Werther, J., Ed.; Dechema e.V.: Wu¨rzburg, Germany, 1999; pp 561-566.
Received for review November 8, 1999 Revised manuscript received March 14, 2000 Accepted March 21, 2000 IE9908027