Experimental investigation of a pilot-scale jet bubbling reactor for wet flue gas desulphurisation

Chemical Engineering Science58(2003)4695–4703/locate/cesExperimental investigation of a pilot-scale jet bubbling reactorfor wet ue gas desulphurisat ionYuanjing Zheng,S ren Kiil∗,Jan E.JohnssonDepartment of Chemical Engineering,Technical University of Denmark,Building229,DK-2800,Kgs.Lyngby,DenmarkReceived22August2002;received in revised form17February2003;accept ed21July2003AbstractIn the present work,an experimental parameter study was conducted in a pilot-scale jet bubbling reactor for wet ue gas desulphurisation (FGD).The pilotplantis downscaled from a limest one-based,gypsum producing full-scale wetFGD plant.Import antprocess paramet ers, such as slurry pH,inlet ue gas concentration of SO2,reactor temperature,and slurry concentration of Cl−have been varied.The degree of desulphurisation,residual limestone content of the gypsum,liquid phase concentrations,and solids content of the slurry were measured during the experimental series.The SO2removal e ciency increased from66.1%to71.5%when the reactor slurry pH was changed from3.5to5.5.Addition of Cl−(in the form of CaCl2·2H2O)to the slurry(25g Cl−=l)increased the degree of desulphurisation to above99%,due to the onset of extensive foaming,which substantially increased the gas–liquid contact area.An increase in the inlet ue gas SO2concentration from502 to991ppmv led to a decrease in the SO2removal e ciency from80.1%to69.4%.A temperature increase from296to323K caused a reduction in the degree of desulphurisation from69.4%to68.1%,but this result is almost within the experimental uncertainty.The residual limestone level in the gypsum formed increased with increasing values of reactor slurry pH,inlet ue gas SO2concentration,and slurry concentration of Cl−.?2003Elsevier Ltd.All rights reserved.Keywords:Wet FGD;Jet bubbling reactor;Absorption;Dissolution;Slurries;SO21.IntroductionCombustion of sulphur-containing fossil fuels,such as coal and oil,results in sulphur dioxide(SO2)emissions. The SO2is known to have detrimental e ects on human health and the environment,and as a consequence,many countries have imposed stringent regulations on coal-ÿred power plants over the past two decades(Soud,2000).In Denmark,for instance,an environmental levy of20Danish Kroner(equivalent to US$2.5)/kg sulphur emitted to the atmosphere was introduced on January1st2000(Frandsen, 2000).Consequently,reductions in SO2emissions are es-sential for the economy of power plants and research in desulphurisation technologies,in order to further optimise the processes,is needed.In the past10years, ue gas desulphurisation(FGD) technology has made considerable progress in terms of∗Corresponding author.Tel.:+45-4525-2827;fax:+45-4588-2258.E-mail address:sk@kt.dtu.dk(S.Kiil).e ciency,reliability(deÿned here as FGD operating hours divided by the boiler operating hours in a given period) and costs(Soud,2000).Presently,there are several ways of reducing SO2emissions from coal utilisation.Wet scrub-bers,especially the limestone-gypsum process,are the front-running FGD technologies.There are four main types of wet scrubbers,the spray scrubber,the packed tower, the jet bubbling reactor(JBR),and the double-loop reactor (Takeshita&Soud,1993).Fundamental studies have focused on the spray scrub-ber(e.g.Agrawal&Rochelle,1993;Brogren&Karlsson, 1997;Eden&Luckas,1998)and the packed tower(Kiil, Michelsen,&Dam-Johansen,1998;Frandsen,Kiil,& Johnsson,2001;Kiil,Nygaard,&Johnsson,2002;Zheng, Kiil,Johnsson,&Zhong,2002).However,few fundamen-tal studies on the JBR system have been reported.The in-vestigations(e.g.Gilbert,Kogawa,Sawahata,&Kumagai, 1987;Someshwar,Bhagat,&Hopke,1989;Yanagioka &Uchida,1993;Burford,Pearl,&Stevens,1994;Burford, 1997)consist of full-scale measurements where the0009-2509/$-see front matter?2003Elsevier Ltd.All rights reserved. doi:10.1016/j.ces.2003.07.0024696Y.Zheng et al./Chemical Engineering Science58(2003)4695–4703Fig.1.Schematic illustration of a full-scale jet bubbling reactor.After Burford(1997).performance of the commercial Chiyoda Thoroughbred121 (CT-121)process has been tested.These studies involved, for instance,the e ects of JBR slurry pH and pressure drop on the SO2removal and limestone utilization.Krasnopoler, Shields,and Shoji(1991)have explored the advantages of the CT-121process as a throwaway FGD system.These conditions,however,are only relevant for American wet FGD plants.In most European and Japanese plants the by-productmustbe high-qualit y gypsum,useful for wall-board manufacture.Kovacs and Fulop(1998)conducted SO2absorption experiments in a JBR using water and lime (as opposed to limestone)suspensions.Recently,Yoo and Kim(1998)presented some experimental results obtained using fairly low SO2concentrations(¡400ppmv).Their results showed the e ect of HCl in the ue gas on the degree of desulphurisation.Due to the increased importance of the JBR for wet FGD in Europe there is a need to increase the JBR process knowledge.A schematic drawing of a full-scale JBR is shown in Fig.1.The central feature of the JBR process is the absorber (reactor),in which the ue gas is blown into a limestone slurry,thereby forming aÿne bubble bed where SO2can be absorbed.The dissolved SO2dissociates to HSO−3,which is subsequently oxidised to SO2−4by injected air.Finally,SO2−4 ions combine with Ca2+ions,originating from dissolved limestone,and crystallise as gypsum.The overall reaction in the plant can be written asCaCO3(s)+SO2(g)+1=2O2(g)+2H2O(l)→CaSO4·2H2O(s)+CO2(g):(1) The JBR process eliminates the use of large recycle pumps and spray heads,making it less complex than conventional spray tower systems.However,the pressure drop in the JBR is signiÿcantly higher than that of the spray tower(Takeshita &Soud,1993).One major di erence between the JBR and other FGD absorption vessels is the gas–liquid contacting pattern.The ue gas enters the JBR through the gas-cooling duct where it is cooled and saturated with water by injection of water and slurry.The ue gas is then distributed through a large number of gas sparging tubes,which are immersed 12–18cm below the surface of the gypsum slurry(Burford, 1997).This injection of gas into the slurry creates a froth layer,which promotes absorption of SO2from the ue gas. The froth zone provides a very high gas–liquid interfacial area for SO2mass transfer to the slurry,as well as capture of y ash residuals(Yanagioka&Uchida,1993).The bub-bles in the froth zone are continuously collapsing and re-forming thereby generating new interfacial area and trans-port of reaction products away from the froth zone to the bulk slurry solution(Chiyoda Corporation,2000).Another major di erence between the JBR system and conventional FGD units is the operating slurry pH.The JBR system typ-ically operates between a pH of3.5and4.5in the slurry tank,whereas most conventional limestone systems operate ata holding t ank pH of5.0and above(Behrens&Delleney, 1981).In the JBR process the ue gas passes through a con-tinuous liquid phase.Conventional wet FGD plants,using spray towers,have ue gas as the continuous phase.The former enhances the mass transfer of SO2by providing a large gas–liquid contact area.Hence,a high SO2removal e ciency can be gained ata lower pH,leading t o a very low residual limestone content of the gypsum produced. To gain a deeper understanding of the JBR process,a limest one-based wetFGD pilotplant,downscaled from a full-scale JBR plant,has been constructed.A number of experiments,at di erent operating conditions,have been conducted.2.Experimental equipment and procedure2.1.JBR wet FGD pilot plantThe JBR pilotplantis downscaled from a full-scale wet FGD plantcleaning t he ue gas from a285MW e coal-ÿred power plant.A typical ue gas ow rate from the full-scale power plantis285N m3=s and the number of sparging tubes in the full-scale JBR is about2000.The diameter of the JBR is17m and the volume of the gypsum slurry is1300m3.A schematic illustration of the pilot plant is shown in Fig.2.The main components are the JBR,the feed and product tanks,and the natural gas burner.All pipes and vessels are made of PVC or stainless steel to avoid cor-rosion.The absorber is a stirred vessel with continuous feeding of both gas and liquid phases.The reactor tem-perature is controlled by means of a heat exchanger.The reactor vessel is a cylinder with an inner diameter of 52cm,and a height of101cm,ÿtted with four vertical ba es,a pH probe and a oating contact system for slurry level control.To avoid plugging,the level control was ushed with N2for30s every10min.An axial four-blade stirrer(diameter of20cm),with adjustable speed from 0to1420rpm,creates a high degree of mixing in the liquid phase.Y.Zheng et al./Chemical Engineering Science 58(2003)4695–47034697Gas from burnerTo stackGas sample boardDistilled water Pure SO 2Product tankFeed tankCompressed airSlurry pumpHeating water bathJet bubbling reactorWaterHeat exchanger Fig.2.Schematic illustration of the experimental pilot-scale JBR appa-ratus.The feed gas is supplied from either a natural gas burner or as compressed air.The gas mixture enters the heat ex-changer where it is cooled from burner temperatures to 74◦C.The gas enters the JBR through the gas-cooling duct where the temperature is reduced further to 50◦C and the gas is saturated by injection of distilled water.It was not possible to have a slurry quench zone upstream the reactor like in the full-scale absorber (Fig.1),as there was no water cleaning system in the gas distribution box.Keeping the gas distribu-tion box and sparger tubes free of solids is critical to ensure continuous performance of the system.In the experiments at room temperature,compressed and saturated air at 23◦C was used instead of burner ue gas.The low temperature (as opposed to the pertinent temperature of 50◦C)was used due to practical limitations of the temperature control at the high ue gas ow rates employed.However,as shown later in Section 3the degree of desulphurisation is practically in-dependent of reactor temperature in the range studied.In order to simulate a coal combustion ue gas,SO 2is added to the saturated gas just before it enters the absorber.The ue gas is then distributed through nine gas sparger tubes (Fig.3)with an internal diameter of 27:2mm and a length of 436mm.The tubes are immersed 5cm below the stag-nant surface of the gypsum/limestone slurry,and the injec-tion of gas into the slurry creates a froth layer,which pro-motes absorption of SO 2from the ue gas.The treated ue gas ÿnally ows to the stack.Air is introduced into the reactor through a probe to ox-idize HSO −3o SO 2−4.The SO 2−4ions combine with Ca 2+,originating from the dissolved limestone,and crystallize as gypsum.The addition of fresh limestone slurry is controlled by maintaining the holding tank pH at a constant value (typically 4.5).The changing weight of the feed tank over time is recorded and allows the feed consumption rate to be calculated.The feed slurry is prepared by mixing distilled water and a required amount of commercial limestone.JBR orificeGasdistribution boxFlue gasFig.3.Photo of gas distribution box with nine sparger tubes.A product stream of JBR slurry is withdrawn intermittently to the product tank.To study the e ect on the performance of the pilot plant from Cl,which is released during coal combustion as HCl and subsequently absorbed in the slurry in the full-scale plant,Cl was added to the feed tank in the form of CaCl 2·2H 2O(s)in one experiment.A steady-state concentration level of 25g Cl −=l,typical for a Danish wet FGD plant,was used (Frandsen etal.,2001).2.2.AnalysesThe amountof residual limest one in gypsum samples was determined by acid titration.Solids content (gypsum with residual limestone)of the slurry samples were determined by evaporation (9days at 40◦C)and subsequentweighing.Particle size distributions (PSDs)of limestone and gypsum samples were measured on a Malvern analyser,employ-ing laser di raction (Rawle,1997),and using ethanol as dispersant.2.3.Dimensional analysisA dimensional analysis for the volumetric mass transfer coe cientwas carried outin t he process of downscaling from full-to pilot-scale (Riber,2001).The reason for fo-cusing on this particular variable in the downscaling is that it contains the mass transfer and chemical reaction charac-teristics of the wet FGD plant and as such all the essential process parameters.It was chosen to have nine gas sparging tubes in the pilot plant to include possible interaction among the gas outlet streams of the various tubes.The dimensional analysis showed that in order to establish complete similarity between the pilot-and full-scale plants (at the same tempera-ture and pressure),the values of the following dimensionless numbers should be identical:Re =ud g =Ág ,Fr =u 2=(gd ),4698Y.Zheng et al./Chemical Engineering Science58(2003)4695–4703Table1Numerical values of the dimensionless numbers in three cases(each one ensuring that one particular dimensionless number is identical for the pilot-and full-scale plant)Scale of plant Re Fr l=(du2 g) Full35,89414460.0048 Pilot(case1)35,8941:2×1080.0001 Pilot(case2)10,8461:7×1050.0048 Pilot(case3)416214460.084The values of the dimensionless numbers for the full-scale plant are determined by the process parameters,which areÿxed in the analysis.and l=(du2 g),for nomenclature,see list of symbols.Here, it should be noted that similarity for the dimensionless num-ber(p2−p1)=(u2 g)=5,also appearing in the analysis, could be established by an adequate choice of immersion depth of gas sparging tubes(the latter being proportional to the total pressure drop in the pilot plant,p2−p1).Itwas not possible,due to the appearance of the same parameters in more t han one dimensionless number and pilotplant ow rate limitations,to obtain complete similarity for Re,Fr,and l=(du2 g)at the same time.In Table1the numerical val-ues for three cases(each having one of the three dimension-less numbers identical for pilot-and full-scale)are shown. The Froude number,Fr,physically represents the ratio of the initial jet momentum ow to the buoyant force(Bird, Stewart,&Lightfoot,1960)and is very important for the transition between jet stream and bubble formation,and so for the performance of the JBR.It was chosen to have the same Froude number in t he pilotand full-scale plantand relax on the other dimensionless parameters(i.e.case3in Table1).An additional argument,supporting this choice, was that the Re number was of the same order of magni-tude when keeping Fr constant,while a constant Re num-ber would cause a deviation of a factor of as much as105 between the Froude number in the pilot and full-scale plant (see Table1).The reason for this di erence is to be found in the way the various physical parameters appear in the di-mensionless numbers.Also,the ratio of ue gas ow rate to reactor slurry volume of the pilot plant is identical to that of the full-scale plant(=0:265s−1).In summary,itis not possible to obtain complete similarity between the pilot-and full-scale plantand one mustrely on physical knowledge of the process to make an adequate choice of downscaling(i.e. the design of a representative pilot plant).3.Results and discussionTo investigate the e ects of important process parame-ters,four experimental series were conducted in the pilot plant.Experimental parameters for a selected Base Case are presented in Table2.The other experiments involved per-turbations from the Base Case with respect to reactor pH, inlet ue gas concentration of SO2,reactor temperature,and Table2Experimental parameters of the Base CaseParameter ValueFlue gas ow rate(m3=s STP)0.0275Inlet ue gas concentration of SO2(ppmv)991 Limestone content of feed stream(wt%)10 Temperature(K)296Slurry level in the reactor(m)0.585 Oxidation air rate to reactor(m3=s STP)1:07×10−4 Pressure in plant(Pa)1:013×105 Reactor pH 4.5 Limestone type Faxe limestone Slurry concentration of Cl−0.0 Chemical analysis of Faxe limestone is provided in Kiil etal.(1998).slurry concentration of Cl−.Other parameters in Table2 such as ue gas ow rate,and oxidation air rate,deviated a few percentage points among experiments.The most essen-tial parameter for a wet FGD plant is the degree of desul-phurisation.However,for plants employing the forced oxi-dation mode(most European and Japanese plants),another essential parameter is the amount of residual limestone in the gypsum.There are two reasons for this:(1)good utilisation of limestone and(2)a saleable gypsum product(i.e.less than about3wt%residual limestone,(Ib k,1996)).More recently several wet FGD plants in the USA have converted their wet lime/limestone/waste plants to gypsum producing facilities,and the trend is continuing(Soud,2000).How-ever,the majority of wet FGD plants in the USA produce a mixture of calcium sulphite and gypsum as opposed to commercial quality gypsum.In Europe the wet FGD plants are designed and built to produce gypsum suitable for wall-board utilisation rather than disposal(Clarke,1993).Con-sequently,the degree of desulphurisation and the amount of residual limestone in the gypsum were analyzed for each experiment.3.1.E ects of reactor slurry pHFig.4shows the SO2removal e ciency atdi erentpH values with an SO2inlet concentration of991ppmv.The SO2removal e ciency appears to increase almost linearly from66.1%to71.5%when increasing the reactor pH in a range from3.5to5.5.At a higher pH,the dissociation reaction of SO2:SO2+H2O HSO−3+H+(2) is shifted to the right,leading to an increase in the enhance-ment factor and thereby in the degree of desulphurisation. Zhong(1998)did some wet FGD experiments in the falling ÿlm pilotplantdescribed in Kiil etal.(1998)at50◦C,with an inletSO2concentration of about1000ppmv.In this case, the SO2removal e ciency also increased approximately lin-early when increasing the holding tank pH in a range of pH of4.5–6.1(Fig.4).The results also agree well with YooY.Zheng et al./Chemical Engineering Science 58(2003)4695–4703469936Slurry pH value020406080100S O 2 r e m o v a l e f f i c i e n c y %5parison of pH e ects on SO 2removal e ciency between JBRs and a falling ÿlm pilot plant.The JBR of this work was run at 23◦C,the ue gas inletSO 2concentration was 991ppmv,and other parameters are given in Table 2.The falling ÿlm pilotplantwas run at50◦C,with a ue gas inlet SO 2concentration of 1000ppmv.The inlet SO 2concentration of Yoo and Kim’s JBR was about 300ppmv with a reactor pressure drop of about 1100Pa.All experimental results were obtained at steady state operation of the pilot plants.Error bars are shown on the ÿgure.and Kim’s (1998)JBR experiment s.Itis notpossible t o extrapolate directly the SO 2removal e ciency for the ÿrst two processes to full-scale plants using the results in Fig.4,because in both cases the full-scale plant has additional SO 2removal.In a packed tower there is a spray section above the falling ÿlm section and a gas–liquid contact zone below.In the JBR process there is injection of slurry into the ue gas before it enters the bubbling reactor (Fig.1).Only the trend in the dependence of removal e ciency with pH is provided in Fig.4.The total e ciency of a full-scale plant cannotbe seen.However,in bot h cases itcan be seen t hat a substantial desulphurisation takes place.The reason for a higher SO 2removal e ciency in Yoo and Kim’s exper-iments,which also did not include a desulphurisation zone prior to the JBR,may be due to a di erent setup and ex-perimental conditions.In their case the reactor has a diam-eter of 220mm and a height of 500mm.The diameter of the sparger tube is about 6mm and the oriÿce diameter is 2mm.The inletSO 2concentration is only about 300ppmv and there is also a deeper sparger immersing length corre-sponding to a reactor pressure drop of about 1100Pa.In our case the reactor pressure drop is only about 600Pa.Fig.5shows the residual limestone level in the gypsum atdi erentpH values wit h an SO 2inlet concentration of 991ppmv.The residual limestone level of the gypsum in-creases with increasing pH values.These results are in good agreement with full-scale results of Gilbertetal.(1987)and Someshwar etal.(1989).In the wet FGD process,the pH value of the slurry is very important because pH a ects the dissolution rate of limestone,the relative concentrationdistribution of HSO −3and SO −3,and the dissolution equi-346Slurry pH value00.40.81.21.62R e s i d u a l l i m e s t o n e i n g y p s u m w t %Fig.5.The residual limestone in the gypsum at di erent pH values.Parameters are given in Table 2.The experimental results were obtained at steady state operation of the pilot plant.Error bars are shown on the ÿgure.librium of SO 2(Eq.(2)).Low pH favors the dissolutionrate of limestone.A decrease in pH of 2corresponds to a 100-fold increase in hydrogen ion concentration and almost the same in limestone dissolution rate (Kiil,Johnsson,&Dam-Johansen,1999).The operation at lower pH ranges has a smaller risk of scaling and plugging (Burford,1997).The JBR process is designed to operate in a pH range of 3–5where the driving force for limestone dissolution is high,resulting in nearly complete reagent utilisation.At low pH,the prevailing sulfurous species is bisulphite (HSO −3)as op-posed to sulÿte ion (SO 2−3).HSO −3is more easily oxidizedo SO 2−4than SO 2−3(Takeshita &Soud,1993).Thus,itis advant ageous t o run a wetFGD plantatlow values of pH from a limestone utilization and gypsum production point of view.However,the degree of desulphurisation decreases atlower pH,as seen in Fig.4.A transient mass balance calculation for total S or Ca over the JBR,assuming mixed ow for the slurry phase,provides an equation for the time,t ,it takes to obtain a given concentration (total S or Ca)in the slurry (Fogler,1986)t =− ln 1−CC o;(3)where is the space time (i.e.the slurry volume divided by the volumetric feed rate)and C o is the steady-state concen-tration in the JBR.The time it takes the e uent concentration to reach 99%of its steady-state value is found by inserting C =0:99C o in Eq.(3).This gives t =4:6 =4:6×35:5=163h,i.e.about7days.Itcan be seen in Fig.6that experimental data suggests that the time to reach the steady-state value is about 9days (216h)rather than 7.However,taking into account the uncertainties of the measurements and the as-sumption of perfect mixing in the above calculation the agreement seems reasonable.The solids concentration in the Base Case experimentwas used as an indicat ion of when4700Y.Zheng et al./Chemical Engineering Science 58(2003)4695–4703246810Running time (days)S o l i d s c o n c e n t r a t i o n w /w %Fig.6.Solids concentration levels of the Base Case experiment as a function of time.Parameters are given in Table 2.Error bars are shown on the ÿgure.Table 3Degree of desulphurisation,residual limestone in the gypsum,and solids concent rat ion atdi erent ue gas inletSO 2concentrations InletSO 2concentration (ppmv)991502Degree of desulphurisation (%)69.480.1Residual limestone in the gypsum (wt%)0.660.30Time to reach steady state (h)163260Solids concentration (wt%)27.232.3Parameters are given in Table 2.the system had reached a steady state.In the pilot plant the steady-state concentration is in uenced by the limestone concentration in the feed tank.In full-scale systems,on the other hand,the level of solids concentration at steady state is determined by the water balance.However,it should be mentioned that the advantages of operating the scrubber at low slurry concentrations include reduced plugging and reduced wear of JBR internals.There is also more risk for the sparger tubes to be plugged when the solids concentra-tions become high (¿24%)(Burford,1997).In the experi-mentatpH 5.5,aboutone-t hird of t he cross sect ion of each sparger tube was plugged after a running time of 5days.3.2.E ects of inlet SO 2concentrationTable 3shows the degree of desulphurisation,residual limestone in the gypsum,and solids concentrations at dif-ferentinletSO 2concentrations at a pH of 4.5.The e ect of inletSO 2concentration on SO 2removal e ciency is quite signiÿcant.An increase in inlet SO 2concentration leads to a decrease in SO 2removal e ciency ata given setof scrubber operating conditions.This drop in SO 2removal e ciency is a result of the lower enhancement factor for absorption of SO 2,which may be evidenced by comparing with the sim-ulations of Kiil etal.(1998).Table 4Cl −e ects on SO 2removal e ciency,and residual limestone in the gypsumInletSO 2concentration (ppmv)502605Cl −concentration (g/l)025Degree of desulphurisation (%)80.199¡Residual limestone in the gypsum (wt%)0.3 3.6Parameters are given in Table 2.The limestone utilisation increased with a decrease in in-letSO 2concentration.At a lower SO 2concentration,less limestone is needed to maintain the same pH of the holding tank,and the limestone slurry feed rate is decreased,result-ing in an increase in JBR solids residence time.It took a longer time to reach a steady state in the case of a lower inletSO 2concentration because the feed rate of limestone slurry was lower.To make sure a steady state was reached,the mass balances of Ca 2+and S were calculated.In the Base Case,the balance deviation of Ca 2+and S was −5:5%and 2.2%,respectively.Most of the deviations were within ±7%.The time needed for the system to reach a steady state at di erent conditions was also estimated.At an inletSO 2concentration of 991ppmv it should take 163h for the system to reach a steady state and at an inlet SO 2concentration of 501ppmv the value was 260h (these esti-mated values may,due to the assumption of perfect mixing made in the calculations,be somewhat lower than the actual values as discussed in the previous paragraph).The mean residence time of limestone slurry increased from 35.5to 56:5h.The reason for this increase in residence time is that the limestone slurry feed rate dropped from 3.5to 2:2kg =h because the limestone concentration in the feed tank was the same in all experiments.3.3.E ects of Cl −Table 4illustrates the e ects of Cl −on SO 2removal ef-ÿciency and residual limestone level of the gypsum.The addition of Cl −enhances the SO 2removal e ciency from about80%t o above 99%.This is in good agreementwit h Yoo and Kim’s results (1998).They also found that the SO 2removal e ciency showed a maximum value (about99%)in the range of 10–30g Cl −=l when Cl −was added to the slurry in the form of CaCl 2.The main reason for this e ect is assumed to be an increase in the gas–liquid surface area due to foaming.The addition of Cl −caused the slurry level to rise 8cm higher in the reactor and extensive foaming was observed.The gas/liquid interfacial area is expected to in-crease accordingly as the slurry level above the sparger tube is increased (Riber,2001).Other reasons for the very low outlet concentration of SO 2measured could be a sampling error because of fouling of the sample line or because of an increased level of residual limestone in the gypsum.As the free space above the liquid surface is 41cm and the slurry level only rose 8cm,it is unlikely that foam could penetrateY.Zheng et al./Chemical Engineering Science58(2003)4695–47034701 Table5Degree of desulphurisation at di erent reactor temperaturesReactor temperature(K)Degree of desulphurisation(%)29669.432368.1Parameters are given in Table2.the sampling system and impair the SO2measurement.Theaddition of Cl−to the slurry elevates the content of residuallimestone from about0.3to3:6wt%.This is because of alower limestone dissolution rate due to the increased con-centration of Ca2+(Frandsen etal.,2001).One could argue that the large excess of CaCO3content may contribute to the high SO2removal rate rather than the foaming.However,in our previous work on a fallingÿlm column the removal ef-ÿciency was only83%with a residual limestone content of 4:6wt%,using the same Cl−level.In these earlier exper-iments,an SO2removal increase with the addition of Cl−was also observed(Frandsen,2000;Frandsen etal.,2001). The mechanism through which Cl−modiÿes the froth ten-dency of the slurry is unknown.Probably,the increase in ionic strength of the slurry(as a result of an elevated con-centration of Cl−)modiÿes the surface tension and thereby the foaming capacities of the slurry,but this could not be conÿrmed.Detailed measurements of the slurry surface ten-sion as a function of Cl−concentration is needed.3.4.E ects of reactor temperatureTable5shows the e ect of reactor(i.e.absorber and ue gas)temperature on the degree of desulphurisation.The SO2 removal e ciency decreased from69.4%to68.1%when the reactor temperature was increased from296to323K. These experimental results are so close(¡1:5%di erence) that the e ect of experimental errors cannot be excluded. However,a lower degree of desulphurisation with increas-ing temperature is indeed expected due to the lower water solubility of SO2at higher temperatures(Kiil etal.,1998). Itshould be ment ioned t hatt he experimentat323K did not reach a steady-state solids concentration level due to a poor long-term performance of the burner at high temper-ature.However,previous experiments have shown that the degree of desulphurisation is practically insensitive to the solids content of the slurry and the results seem credible. Obviously,the residual limestone level of the gypsum could not be estimated for this particular experiment.3.5.Particle size distributions of gypsumFig.7shows that the PSD of gypsum varied only slightly with changing values of slurry pH and inlet SO2con-centration.For all the experiments,the volume moment mean(i.e.equivalentvolume mean diamet er,deÿned in Rawle,1997)of the gypsum particles are in the range of0.010.11101001000Particle diameter, µm20406080100Cumulativevolumefraction,%limestonepH=4.5pH=5.5pH=3.5502 ppmvFig.7.Particle size distributions of gypsum and Faxe limestone.Process parameters are given in Table2.The502ppmv in the legend refers to the inlet ue gas SO2concentration.The experimental results were obtained at steady state operation of the pilot plant.42–50 m.In the Base Case experiment,about50vol% is smaller than44 m and about90vol%are smaller t han 94 m.According to results obtained at Georiga Yates plant,a particle size distribution displaced towards smaller part icles was found athigher inletSO2concentrations and slurry pH values(Burford,1997).The author stated that this was probably because of lower solids residence time at higher inletSO2concentrations and slurry pH values.How-ever,this tendency was not found in our experiments.The secondary nucleation of gypsum in a JBR system may be less than that in a common spray tower.The latter employs large centrifugal pumps to move reacted slurry,probably causing crystal attrition and secondary nucleation.It was also found that the gypsum particle size was reduced when y ash was present during an elevated ash-loading test at Georiga Yates plant(Burford,1997).However,there was no y ash addition in our experiments,only limestone im-purities.Notice,that the mean limestone particle size(about 30 m)is signiÿcantly smaller than that of the gypsum particles(42–50 m).4.ConclusionsAn experimental parameter study with a pilot-scale jet bubbling reactor for wet ue gas desulphurisation was con-ducted.Parameters varied were ue gas SO2inletconcen-tration,slurry pH,reactor temperature,and concentration of Cl−.They all a ect,to various extents,the degree of SO2 removal,limestone utilisation,and gypsum quality.The presence of Cl−can decrease the purity of the gypsum sig-niÿcantly.Running the system at a lower pH value or remov-ing Cl−before the reactor can reduce the residual limestone level in the gypsum.There is,however,a trade-o between the limestone utilization,gypsum production and the de-gree of desulphurisation.In future work,experiments with。