Pollutant emissions released during sewage sludge combustion in a bubbling fluidized bed reactor

Waste Management 105 (2020) 27–38

Contents lists available at ScienceDirect

Waste Management journal homepage: www.elsevier.com/locate/wasman

Pollutant emissions released during sewage sludge combustion in a bubbling fluidized bed reactor Antonio Soria-Verdugo a,⇑, Juho Kauppinen b, Teemu Soini b, Luis Miguel García-Gutiérrez a, Toni Pikkarainen b a b

Carlos III University of Madrid (Spain), Department of Thermal and Fluids Engineering, Avda. de la Universidad 30, 28911 Leganés (Madrid), Spain VTT Technical Research Centre of Finland Ltd, Renewable Energy Processes. Ruukinmestarintie 2, Fl-02330 Espoo, Finland

a r t i c l e

i n f o

Article history: Received 3 October 2019 Revised 27 January 2020 Accepted 27 January 2020

Keywords: Ash accumulation Bubbling fluidized bed (BFB) Cynara cardunculus L. Combustion Pollutant emission Sewage sludge

a b s t r a c t The combustion of dry sewage sludge particles in a bubbling fluidized bed was studied in detail, analyzing the composition of the exhaust gases by means of a FTIR equipment. The operating conditions of the fluidized bed, i.e., the bed temperature and mass flow rate of fluidizing air, were varied to quantify their effect on the exhaust gas composition. The bed material was also varied, using sepiolite, silica sand and braunite particles, to evaluate the effect of different bed densities on the pollutant emissions. The results obtained for the combustion of sewage sludge particles in the different fluidized beds tested were compared to combustion tests run for the same operating conditions and bed materials using Cynara cardunculus L. as a fuel. Pollutant emissions derived from sludge combustion are much higher than those obtained from combustion of Cynara. The operating conditions also affect the emissions, e.g., the concentration of CO in the exhaust fumes decreased substantially when increasing bed temperature and air flow rate. The bed density has an effect on the combustion efficiency of sludge, obtaining higher efficiencies in low-density beds for high temperature and air flow rates, while the efficiency was increased in highdensity beds for low bed temperature and fluidizing air flow rate. The effects of ash accumulation and agglomerates formation were also analyzed. Ó 2020 Elsevier Ltd. All rights reserved.

1. Introduction The volume of sewage sludge generated as a residue has been increasing substantially during the last decades due to rapid growing of population and fast rate of urbanization (Syed-Hassan et al., 2017), causing a problem of waste management. Landfilling, agricultural use, and thermochemical conversion are the main ways of sewage sludge disposal (Fonts et al., 2012). However, landfilling is restricted by European regulations due to environmental problems, while the agricultural use of sewage sludge is also limited because of harmful compounds contained in the sludge, e.g., heavy metals, polychlorinated biphenyls, and polyaromatic hydrocarbons (Kaminsky and Kummer, 1989). In contrast, thermochemical conversion of sewage sludge may permit energy recovery, reduces the volume of the residue by up to 70%, and destructs pathogens (Fytili and Zabaniotou, 2008). Combustion and co-combustion are considered as promising thermochemical processes to dispose sewage sludge, considering the reduction of the net CO2 emissions ⇑ Corresponding author. E-mail address: [email protected] (A. Soria-Verdugo). https://doi.org/10.1016/j.wasman.2020.01.036 0956-053X/Ó 2020 Elsevier Ltd. All rights reserved.

associated with fuel conversion due to the biogenic nature of sludge (Werther and Ogada, 1999). Nevertheless, the emissions of other pollutants produced during sewage sludge combustion should be also considered. Fluidized beds are widely used as industrial chemical reactors due to their ability to convert low quality solid fuels with a high efficiency and with an associated low emission of pollutants. The high thermal inertia and heat transfer coefficients characteristic of fluidized beds permit the conversion of even low-grade solid fuels (Kunii and Levenspiel, 1991), such as sewage sludge. The homogeneous and low operating temperature in fluidized beds limit emissions of NOx, and sorbent bed materials can be used for in-bed capture of SOx emissions. However, the performance and emission level of fluidized beds are influenced by fuel mixing (Leckner, 1998). Mass and heat transfer between the solid fuel particles and the bed material are influenced by fuel axial mixing (Lundberg et al., 2016) and high concentration of volatile matter and char close to the fuel feeding ports may occur when fuel mixing is poor (Gómez-Barea and Leckner, 2010), resulting in undesired temperature profiles (Winaya et al., 2007) and increased pollutant emissions (Leckner, 1998).

28

A. Soria-Verdugo et al. / Waste Management 105 (2020) 27–38

Nomenclature CC CN CS DT FT HT mac mb q ST t Tb

carbon conversion [%] nitrogen conversion [%] sulfur conversion [%] deformation temperature [°C] fluid temperature [°C] hemispheric temperature [°C] mass of ash accumulated in the bed [g] mass of bed material [g] air flow rate [lpm] softening temperature [°C] time [min] bed temperature [°C]

Axial or vertical fuel mixing in bubbling fluidized beds is enhanced by the ascension of bubbles. Several authors have studied in detailed the sinking and rising processes of solid fuel particles in a bubbling fluidized bed, relating this motion to that of bubbles (Rios et al., 1986; Kunii and Levenspiel, 1991; Rees et al., 2005;Pallarès and Johnsson, 2006; Soria-Verdugo et al., 2011a). Most of these studies are experimental works based on tracking fuel particles in pseudo-2D beds and concluded that fuel particles sink continuously following the bed material motion, whereas the rising fuel motion is composed by various jumps induced by various passing bubbles (Rios et al., 1986). Buoyancy forces play an important role in axial dispersion of fuels. Soria-Verdugo et al. (2011b) demonstrated that flotsam particles, i.e., fuel particles with lower density than the dense bed, are more prone to be found close to the bed surface, resulting in a poor axial mixing. However, these buoyancy effects can be avoided by increasing the gas velocity to impose a more vigorous fluidization of the bed, improving the fuel mixing (Soria-Verdugo et al., 2011b; Köhler et al., 2017). This improvement of fuel axial mixing for increasing gas velocities was confirmed by simulations run by Garcia-Gutierrez et al. (2017) using a fully couple hybrid-model. Regarding fuel lateral mixing, mixing cells, defined as regions of the fluidized bed containing a bubble preferential path at the center (Pallarès and Johnsson, 2006), affect the dispersion of solid fuel particles inside the bed (Sette et al., 2014), while their motion in the freeboard can be modeled as a ballistic trajectory (Garcia-Gutierrez et al., 2014). The effect of fuel lateral mixing is relevant in industrial scale reactors (Olsson et al., 2011), where the number and location of fuel feeding ports should be optimized to enhance a homogeneous distribution of fuel in the bed (Garcia-Gutierrez et al., 2015). Biomass combustion in bubbling fluidized beds may also present operational problems such as formation of agglomerates. In fact, severe agglomeration induces the defluidization of the bed, and thus, the shutdown of the combustor operation, which may reduce the profitability of the plant due to the cost associated with outages (Morris et al., 2018). The main affecting parameters for agglomeration are the alkali form and content of the solid fuel ash, bed temperature, bed material properties, and fluidizing agent velocity (Rizeq and Shadman, 1989). Nevertheless, early detection techniques and countermeasures to palliate possible problems derived from formation of agglomerates are available, as reviewed by Scala (2018). In this work, the composition of the exhaust fumes from sewage sludge combustion in a bubbling fluidized bed is studied. The effect of bed temperature and fluidizing air flow rate on the concentration of pollutant emissions was evaluated. The composition of the exhaust fumes obtained from the combustion of sewage sludge in each case was compared to the concentration of pollutant emis-

top X

Dp

gC

operation time until agglomerates formation [min] percentage of mass remaining in the pan during TGA tests [wt.%] bed pressure drop increase during operation [Pa] combustion efficiency [%]

Abbreviations: BFB Bubbling Fluidized Bed HHV Higher Heating Value TGA Thermogravimetric Analyzer

sions in the exhaust gases released from the combustion of Cynara cardunculus L. for the same operating conditions. As a main novelty, various bed materials, namely silica sand, sepiolite and braunite, were also tested to check the effect of the dense bed density on the sludge combustion process and the pollutant emissions. The effect of the bed material on the combustion efficiency and the conversion of carbon, nitrogen and sulfur to the exhaust fumes was also studied. Ash accumulation and agglomeration problems were also analyzed and discussed for the combustion of both sludge and Cynara. 2. Materials and methods 2.1. Biomass characterization The sewage sludge studied was obtained from the municipal sewage treatment plant of Loeches (Madrid, Spain) in January 2019, after being pre-dried at 80 °C in a fluidized bed in the plant. The potential of sewage sludge collected from this plant in 2016 as a fuel for pyrolysis in a fluidized bed was already analyzed by Soria-Verdugo et al. (2017a). Samples of Cynara cardunculus L. pellets were also tested for comparison of the pollutant emissions to those obtained from sewage sludge combustion, since cardoon is a widely used lignocellulosic energy crop. The samples of Cynara used were previously subjected to pyrolysis (Morato-Godino et al., 2018) and gasification (Gómez-Hernández et al., 2016) studies in bubbling fluidized bed reactors. Both Cynara and sludge were grinded and sieved to a particle size around 3 mm. A basic characterization of both biomass types was performed, consisting in proximate and ultimate analyses, conducted in a TGA Q500 from TA Instruments and a Leco TruSpec CHN Macro and TruSpec S analyzer, respectively, and a heating value test carried out in an isoperibolic calorimeter Parr 6300. The elemental analysis was performed following the standard ISO 18125-2017 and the heating value was measured according to the standard ISO 16948-2015. The proximate analysis consisted in two thermogravimetric tests, one conducted under an inert atmosphere and the other under an oxidative atmosphere. The temperature profile of the inert atmosphere test had a temperature increase from room temperature up to 105 °C, followed by an isothermal process until no mass variation was detected. Then, temperature was further increased up to 900 °C, where another isothermal process started. The moisture content of the samples was determined as the percentage of mass released at the end of the first isothermal process, i.e., at 105 °C, while the volatile mater content was measured as the mass percentage released between the two isothermal processes, i.e., the percentage of mass released at 900 °C discounting

A. Soria-Verdugo et al. / Waste Management 105 (2020) 27–38

the moisture content. A temperature increase from room temperature up to 300 °C was programmed for the oxidative atmosphere test, followed by an isothermal process to induce a smooth releasement of volatiles. After that, temperature was further increased up to 550 °C and an isothermal process occurred until no mass variation was detected in the sample. The ash percentage of the sample was obtained from the oxidative atmosphere test as the percentage of mass remaining after the test, i.e., after a complete combustion of the sample at 550 °C. Finally, the fixed carbon content was determined by difference. The elemental composition of the samples was determined conducting the exhaust fumes from a complete oxy-combustion of the samples through infrared absorption cells to measure the content of carbon, hydrogen and sulfur, and through a thermal conductivity cell to determine the nitrogen content. The content of oxygen of the sample was then calculated by difference. The higher heating value of the feedstocks was determined by measuring the temperature increase due to the complete combustion of the samples in an atmosphere of pressurized pure oxygen. Further details of the equipment and methods employed for the basic characterization of the feedstocks can be found in Soria-Verdugo et al. (2018). Table 1 summarizes the results obtained from the basic characterization of the feedstocks. Cynara is characterized by high contents of volatile matter and carbon and low contents of ash, nitrogen, and sulfur. These characteristics are favorable for their use as a fuel in combustion applications, where the emissions of NOx and SOx expected from the combustion of this lignocellulosic biomass are low. In contrast, sewage sludge is characterized by much higher contents of ash, nitrogen, and sulfur, thus, a detailed study of the pollutant emissions associated with its combustion should be accomplished to evaluate the potential of sewage sludge as a fuel in combustion applications. The composition of Cynara cardunculus L. reported in Table 1 is very similar to that measured by Fernández Llorente et al. (2006), Aho et al. (2008), and Pallarès et al. (2009). In the case of sewage sludge, despite the inherent heterogeneous composition of sludge, the composition shown in Table 1 is within the range of variability reported by SoriaVerdugo et al. (2017b), obtained from a detailed review of reported values measured by several authors. The ash melting temperature of each biomass was also determined by means of an ash fusibility test performed in an Automatic Ash Fusion Temperature Analyzer, AF700, from LECO, following the standard ISO 540-2008. Cones made of a mixture of ash from the feedstocks and a dextrin solution were built and subjected to a heating process under an oxidative atmosphere from 400 to 1500 °C. The cones were video recorded during the heating process to determine the relevant ash fusibility temperatures: (i) Deformation Temperature (DT) for which deformation of the cone tip

Table 1 Results of the characterization of sewage sludge and Cynara cardunculus L. (wb: wet basis, daf: dried ash free basis, * calculated by difference).

Moisture [% wb] Volatile matter [% wb] Fixed carbon* [% wb] Ash [% wb] C [% daf] H [% daf] N [% daf] S [% daf] O* [% daf] HHV [MJ/kg wb] DT [°C] ST [°C] HT [°C] FT [°C]

Sewage sludge

Cynara cardunculus L.

8.2 54.1 15.3 22.4 44.8 7.7 6.5 2.9 38.1 13.8 1117 1203 1241 1259

5.7 74.5 12.3 7.5 45.4 6.6 2.6 0.2 45.2 14.9 1103 1137 1149 1189

29

occurs, (ii) Softening Temperature (ST) for which the cone height equals the cone base width, (iii) Hemispheric temperature (HT) for which the cone height equals half the cone base width, and (iv) Fluid Temperature (FT) for which the cone is spread over a flat layer. The values of the ash fusibility temperatures obtained for sewage sludge and Cynara cardunculus L. are also reported in Table 1. The ash melting temperatures obtained for Cynara are similar to those measured by Fernández Llorente and Carrasco García (2005), while the melting temperatures obtained for sludge ash are in good agreement with those reported by Folgueras et al. (2015), Magdziarz et al. (2016), and Li et al. (2019). The characteristic ash fusibility temperatures for the sludge are roughly 5% higher than those of Cynara. Further details of the equipment employed, and the procedure followed to determine the four relevant ash fusibility temperatures, can be found in de Palma et al. (2019). 2.2. Experimental setup A bench scale bubbling fluidized bed (BFB) combustor with continuous fuel feeding was used in the study. A schematic and a photography of the facility can be observed in Fig. 1. In these experiments, the apparatus was operated with continuous fuel feeding. Fuel was fed by a screw located 155 mm above the distributor of the fluidized bed. The height of the riser tube (heated part) was 669 mm, and the inner diameter of the lower part (height 225 mm) was 36.0 mm. The inner diameter of the upper part of the riser tube was 53.1 mm. Pre-heated primary gas was fed into the reactor through a perforated plate. Different types of gas mixtures, including O2, CO2, SO2, CO, H2O, air, and nitrogen, can be used as fluidizing agent. However, only air was employed for the combustion experiments conducted in this work. The temperature and pressure profiles inside the combustor was measured at six levels, and the temperature was controlled by two surrounding electric heaters. Flue gas composition after the filter was measured with on-line gas analyzers and a FTIR spectrometer Gasmet DX4000 with 2.5 m cuvette, operated with a cell temperature of 180 °C and a pressure of 1006 mbar. 2.3. Experimental procedure Combustion tests were conducted in the BFB combustor for sewage sludge and Cynara cardunculus L. During the sewage sludge experiments, bed temperatures of 750 and 800 °C and air flow rates of 6 and 10 lpm were tested. These bed temperatures are within the range of temperatures typically used to analyze pollutant emissions derived from solid fuels combustion (Chirone et al., 2008; Guo and Zhong, 2017; Li et al., 2018; Li and Chyang, 2020). In contrast, the combustion tests of Cynara were only carried out for the limiting cases, i.e., for the lower bed temperature and air flow rate, Tb = 750 °C and q = 6 lpm, and for the higher values of both parameters, Tb = 800 °C and q = 10 lpm. For each fuel and operating condition tested, three different bed materials, namely sepiolite, silica sand, and braunite, were used to evaluate the effect of the different density and characteristics of these three materials on the combustion process. All combustion tests started with the bed preheated by the furnace surrounding it at a preset temperature depending on the final temperature desired, the type of fuel used, and its flow rate, which is a function of the air flow rate. The sewage sludge flow rate was adjusted for each air flow rate to obtain an average oxygen concentration at the exhaust fumes of around 7 vol%, in agreement with the study carried out by Guo and Zhong (2017), whereas for the combustion of Cynara cardunculus L., due to the easier combustion of this lignocellulosic biomass, the concentration of oxygen of the exhaust fumes was reduced to 6 vol%. The temperature at which the bed is to be preheated and the fuel flow rate required for each

30

A. Soria-Verdugo et al. / Waste Management 105 (2020) 27–38

Fig. 1. Schematic and photography of the experimental facility.

experimental condition were determined previously to the combustion tests. 3. Results and discussion 3.1. Time evolution of pollutant emissions The time evolution of the pollutant emissions generated was analyzed in detail for the combustion of sewage sludge in a silica sand bed operated at an average bed temperature of 750 °C and an air flow rate of 6 lpm. After that, the effect of varying the bed temperature, air flow rate, bed material, and type of fuel will be discussed in the following subsections. As an example, Fig. S1 (supplementary material) shows the temperature measured by a type K thermocouple immersed in the silica sand dense bed during the combustion of sewage sludge at 750 °C with an air flow rate of 6 lpm. For this case, the bed was preheated at 690 °C and the average fuel flow rate was 1.6 g/min. At the beginning of the test, the bed temperature is stable at the preset value, 690 °C in this case, and it starts to increase as soon as sewage sludge is supplied to the bed due to its exothermic combustion. After approximately 3 min from the beginning of the sludge feeding, the bed temperature reaches the desired value of 750 °C and it is uniform for the whole test, i.e., for 30 min. Slight fluctuations of the temperature of around 10 °C in amplitude are observed in the figure, which are caused by the operation of the screw feeder and the variable particle size of the sewage sludge. In this case, the average bed temperature for the last 30 min is 751.8 °C, which proves the reliability of the control system and the procedure followed to obtain the desired bed temperature. The composition of the exhaust fumes was monitored by the FTIR for the whole combustion tests with a frequency of 1 Hz. The time evolution of the concentration of different compounds in the exhaust fumes produced during the combustion of sewage sludge in the silica sand bed operated at 750 °C and using an air flow rate of 6 lpm are depicted in Fig. 2. The concentration of oxygen, water vapor and all the compounds containing carbon, nitrogen, and sulfur detected by the FTIR with concentrations above 15 ppm are shown in the figure. The evolution of the O2, CO2 and H2O concentration, measured in vol.%, can be observed in Fig. 2a. The initial concentration of O2 is around 21 vol%, corresponding to the concentration of oxygen in air. Then, as sewage sludge starts

to be supplied and burned in the bed, the concentration of O2 decreases to an average value of approximately 7 vol%, stable for the whole test, but with fluctuations around the average value due to the effect of the feeding system and the chemistry associated with sewage sludge combustion. Both the concentration of carbon dioxide and water vapor in the exhaust fumes increase at the beginning of the test, reaching stable values after around 3 min from the beginning of the sludge feeding. In this case, the fluctuation of the concentration of CO2 and H2O is much slower than that of O2. Fig. 2b shows the concentration of CO and SO2 in the exhaust fumes of sewage sludge combustion, measured in ppm. The concentration of carbon monoxide is highly variable during the test, with several peaks of concentration above 3000 ppm. In contrast, the fluctuation of concentration of sulfur dioxide is much lower, increasing from zero to a stable average value, attained after 10– 15 min from the beginning of the sludge feeding. The evolution of the concentration of compounds containing nitrogen in the sludge combustion exhaust fumes, measured in ppm, is plotted in Fig. 2c, including nitrogen monoxide NO, nitrous oxide N2O, hydrogen cyanide HCN, and ammonia NH3. The concentration of nitrogen dioxide NO2 was also measured, however, its average concentration for this sewage sludge combustion test is only 5 ppm, thus, it is not represented in the figure. The concentrations of NO, N2O and HCN increase when the sludge is fed to the bed and reach stable values in around 3–5 min. The highest concentration of the compounds containing nitrogen corresponds to NO, with concentration varying between 400 and 600 ppm. A peak in the NO concentration was reached when the concentration of O2 and CO2 were equal, i.e., after approximately 3 min from the beginning of the fuel feeding due to the consumption of oxygen, as previously stated by Guo and Zhong (2017). In the case of the concentration of hydrogen cyanide, although its average value is stable during the test, it is characterized by a high variability, with peaks of amplitudes of approximately 150–200 ppm. In contrast, the concentration measured for N2O is more stable, obtaining only slight fluctuations around the average value. In the case of ammonia, the concentration detected in the exhaust fumes is low during the first 5–10 min of the combustion test, increasing then to a stable value after around 15 min from the beginning of the sludge feeding. The fluctuation of the concentration of NH3 is negligible compared to the rest of compounds containing nitrogen.

A. Soria-Verdugo et al. / Waste Management 105 (2020) 27–38

31

Fig. 2. Time evolution of exhaust fumes composition during combustion of sewage sludge in a silica sand BFB at Tb = 750 °C and q = 6 lpm.

The time evolution of the rest of compounds of the exhaust gas containing carbon with an average concentration above 15 ppm, i.e., methane CH4, ethylene C2H4, and benzene C6H6, can be observed in Fig. 2d. Acetylene C2H2, formaldehyde CHOH, propane C3H8, and ethane C2H6 were also monitored, however, their average concentrations measured for this sludge combustion test are 13, 10, 3, and <3 ppm, respectively, and thus they are not included in the figure. The time evolution of the concentrations of CH4, C2H4 and C6H6 in the exhaust fumes of the sludge combustion follow a similar trend, starting to increase around 3 min after the beginning of the sludge feeding and showing peaks of increasing concentration of each compound. In fact, the concentration peaks of CH4, C2H4 and C6H6 are correlated, occurring at the same time, which may be attributed to an instantaneous decrease of the local oxygen concentration around the fuel particle. The peaks of concentration of hydrogen cyanide detected in Fig. 2c are also correlated with the peaks of the compounds containing carbon. Therefore, the variability of the concentration of HCN may also be attributed to an instantaneous reduction of the local oxygen concentration around the sludge particles. 3.2. Variation of operating conditions Both the bed temperature and the air flow rate were varied to determine their effect on the pollutant emissions generated during sewage sludge combustion in a silica sand fluidized bed. Increasing the air flow rate has an effect on the bed fluid-dynamics and on the dispersion of fuel particles in the bed. Higher gas flow rates induce a more homogeneous distribution of fuel particles throughout the bed height, even for fuels that differ in density with the bed material (Soria-Verdugo et al., 2011c; Lundberg et al., 2017). This improvement of the axial mixing of fuel particles in the bed is especially important for char combustion, since char particles

experience an important reduction of density compare to the raw feedstock (Soria-Verdugo et al., 2019), being flotsam particles in a silica sand fluidized bed and, thus, more prone to be moving close to the bed surface. However, increasing the air flow rate reduces also the residence time in the reactor, limiting the reaction time for all the compounds generated from the solid fuel combustion. In view of the time evolution of the pollutant emissions generated, shown in Fig. 2, average values were calculated for each compound considering the last 20 min of the tests, when the emission of all the compounds are stable. Table 2 shows the average values of the concentration of each compound generated during sewage sludge combustion in a silica sand bed for temperatures of 750 and 800 °C and air flow rates of 6 and 10 lpm. In all cases, the oxygen content in the exhaust gases was fixed at around 7 vol% varying the fuel feeding rate. Similar concentrations of around 8 vol% and 9 vol% were obtained for H2O and CO2, respectively, for all the bed temperatures and air flow rates tested. The concentration of CO decreases with the bed temperature due to a higher thermal energy available in the reactor. The effect of the increase of gas flow rate on the concentration of CO in the exhaust gases is different for the two bed temperatures tested. For Tb = 750 °C, the increase of the air flow rate from 6 to 10 lpm induces a slight increase of the CO concentration, which may be attributed to the shorter residence time of the exhaust fumes in the reactor. However, for a bed temperature of 800 °C, a significant decrease of the CO emission was detected when the air flow rate was increased, which might be due to a better axial mixing of the fuel particles in the bed, resulting in the releasement of CO inside the dense bed where higher temperatures enhance the oxidation of CO molecules to CO2. The effects detected for the carbon monoxide emissions, i.e., slight reduction of its concentration with temperature and substantial reduction for higher gas flow rates at elevated temperature, were also observed for other hydrocarbons and com-

32

A. Soria-Verdugo et al. / Waste Management 105 (2020) 27–38

Table 2 Average composition of exhaust fumes of sludge combustion in a silica sand BFB. Tb = 750 °C

O2 [vol.%] H2O [vol.%] CO2 [vol.%] CO [ppm] SO2 [ppm] N2O [ppm] NO [ppm] NO2 [ppm] NH3 [ppm] HCN [ppm] CH4 [ppm] C2H2 [ppm] C2H4 [ppm] C2H6 [ppm] C3H8 [ppm] C6H6 [ppm] CHOH [ppm] HCl [ppm] HF [ppm]

Table 3 Average composition of exhaust fumes of Cynara combustion in a silica sand BFB.

Tb = 800 °C

q = 6 lpm

q = 10 lpm

q = 6 lpm

q = 10 lpm

6.4 8.1 9.6 1282 1265 225 512 5 61 165 86 13 59 <3 3 20 10 83 20

7.7 7.6 8.6 1316 1241 305 498 7 19 178 88 18 78 <3 3 23 13 82 28

6.7 8.1 9.5 1169 1428 167 501 3 19 121 75 18 45 <3 3 20 10 81 42

7.3 7.7 9.3 336 1400 248 419 5 5 66 13 4 7 <3 <3 6 4 97 36

pounds containing carbon, e.g., CH4, C2H4, C6H6, CHOH, and HCN, and even for some compounds containing nitrogen, such as NO and NH3. In contrast, the concentration of SO2, HCl and HF in the exhaust fumes of sewage sludge combustion in a silica sand fluidized bed reactor increase slightly with the bed temperature. Regarding the emission of N2O, its concentration increases with the air flow rate for the two bed temperatures tested, obtaining lower values for the higher bed temperature in both cases. An inverse relation is also observed for the concentration of N2O and NO, when the concentration of the former increases, the concentration of the latter decreases and vice versa. In view of the pollutant emissions generated during combustion of sewage sludge in a silica sand fluidized bed, high temperature and air flow rates are recommended. Furthermore, the use of inbed sorbing materials to limit the emission of SOx is also advisable. 3.3. Comparison of emissions caused by sludge and Cynara combustion The pollutant emissions obtained from the combustion of sewage sludge in a silica sand fluidized bed were compared to those produced during the combustion of a widely used lignocellulosic energy crop, namely Cynara cardunculus L. For the comparison, combustion tests of Cynara were run for the two limiting cases, i.e., low bed temperature and gas flow rate (Tb = 750 °C and q = 6 lpm), and high bed temperature and gas flow rate (Tb = 800 °C and q = 10 lpm). The average concentration of the different compounds detected in the exhaust fumes of Cynara cardunculus L. combustion in a silica sand fluidized bed can be found in Table 3. During the combustion of Cynara, the targeted concentration of oxygen in the exhaust fumes was around 6 vol%. As a result of combustion, average concentrations of H2O and CO2 of approximately 8.5 vol% and 12.0 vol%, respectively, were measured for both operating conditions tested. In general, a significant reduction of the concentration of all pollutant emissions can be observed in the results of the analysis of exhaust fumes generated during Cynara combustion compared to combustion of sludge, confirming the critical pollution problem associated with sewage sludge incineration. The only exception is the emission of HCl, which is higher for the combustion of Cynara (Gominho et al., 2018) than for sludge combustion due to the higher content of chlorine in the lignocellulosic biomass. In fact, the chlorine content reported in the literature for Cynara cardunculus L. varies between 0.78 and 1.76 wt% db (González et al., 2004; Fernández Llorente et al., 2006; Aho

O2 [vol.%] H2O [vol.%] CO2 [vol.%] CO [ppm] SO2 [ppm] N2O [ppm] NO [ppm] NO2 [ppm] NH3 [ppm] HCN [ppm] CH4 [ppm] C2H2 [ppm] C2H4 [ppm] C2H6 [ppm] C3H8 [ppm] C6H6 [ppm] CHOH [ppm] HCl [ppm] HF [ppm]

Tb = 750 °C q = 6 lpm

Tb = 800 °C q = 10 lpm

6.5 8.7 12.5 587 4 16 249 <3 5 <3 <3 <3 <3 <3 <3 4 <3 214 7

6.6 8.3 11.6 425 8 8 250 <3 <3 <3 <3 <3 <3 <3 <3 4 <3 133 <3

et al., 2008; Bartolomé et al., 2010), whereas for sewage sludge the content of Cl ranges from 0.04 to 0.29 wt% db (Arena and di Gregorio, 2014; Folgueras et al., 2015; Ronda et al., 2019). Specifically, the concentration of carbon monoxide produced during the combustion of Cynara is around three times lower than that generated by sludge combustion at 750 °C and 6 lpm, due to the higher volatility of Cynara. However, the CO concentration is similar, even higher for the combustion of Cynara, when the combustion occurs at 800 °C and 10 lpm. Furthermore, the emission of SO2 derived from the combustion of Cynara is remarkably lower than that associated with sludge combustion, which was expected considering the much lower sulfur content of Cynara cardunculus L. than sewage sludge obtained in the basic characterization of the feedstocks (Table 1). The emission of compounds containing nitrogen was also lower for the combustion of Cynara, due also to its lower nitrogen content compared to sewage sludge (Table 1), being the most relevant emission that of NO. 3.4. Variation of bed material The bed material used for the combustion tests was also varied to evaluate its effect on the process and the pollutant emissions released. The main difference between the three bed materials used, namely sepiolite, silica sand, and braunite, is their particle density, which affects the density of the dense fluidized bed. The characteristics of each bed material are reported in Table 4, including their particle and bulk density and their average particle size, which was selected higher for lower density particles, resulting in type B particles according to Geldart’s classification (Geldart, 1973) in all cases. Considering the density of each bed material, Cynara and sludge particles, and especially the char produced from them, will be flotsam particles in the braunite bed and, thus, more probably located close to the bed surface, whereas they will be jetsam particles in a bed of sepiolite, where they will be located close

Table 4 Characteristics of the bed materials employed. Bed material

Particle density [kg/m3]

Bulk density [kg/m3]

Average particle size [lm]

Sepiolite Silica Sand Braunite

1550 2600 4760

790 1730 2270

400 275 100

A. Soria-Verdugo et al. / Waste Management 105 (2020) 27–38

to the distributor with a higher probability. However, the axial mixing of even flotsam and jetsam particles can be improved by increasing the gas velocity in the fluidized bed (Soria-Verdugo et al., 2011b; Köhler et al., 2017). In addition to the difference of particle density, braunite particles are oxygen carriers (Pikkarainen and Hiltunen, 2017) and sepiolite has some sorption/catalytic effect during biomass thermochemical conversion (SoriaVerdugo et al., 2019), whereas silica sand particles are known to be inert during biomass conversion (Koppatz et al., 2011). As a first step to evaluate the effect of the different bed materials, the response of sepiolite, silica sand, and braunite particles to high temperature and oxidative atmosphere was tested in the reactor prior to the combustion tests. This analysis consisted in heating the reactor to the desired temperature, e.g., 800 °C, and supplying a bed of 15 ml of the different materials, corresponding to a mass of 11.8 g of sepiolite, 26.0 g of silica sand, and 34.0 g of braunite. These beds were fluidized using an air volumetric flow rate of 10 lpm to test the different bed materials under the more extreme conditions, i.e., the highest temperature and air flow rate used for the combustion studies. The composition of the exhaust gases from the reactor was analyzed by the FTIR to determine if any interaction of the bed material occurred. The results of the time evolution of the main compounds found in the exhaust gases after the bed material supply, i.e., O2, H2O, CO2, and CO, can be observed in Fig. 3. Fig. 3a shows the time evolution of the O2, H2O, CO2, and CO concentration in the exhaust gases of the reactor after supplying the bed of sepiolite at a temperature of 800 °C with an air flow rate of 10 lpm. A sharp peak in the concentration of H2O up to 16.3 vol% can be observed in the figure, whereas a smoother increase of the CO2 concentration is also detected. The concentration of H2O in the exhaust gases could be due to moisture contained in the fresh sepiolite, however, the sharp peak of H2O concentration is correlated with a reduction of the concentration of O2 to 13.0 wt%, informing of a reaction of the sepiolite particles inside the reactor instead of merely moisture releasement. A peak was also detected for the concentration of CO in the exhaust gases, attaining a maximum value of 69 ppm (notice the factor 103 used in the figure to improve visualization). This maximum concentration of CO in the exhaust gases coincides with the minimum concentration measured for O2, so it is caused by a defect of oxygen in the reactor to oxidize completely the carbon monoxide molecules to carbon dioxide. In the case of silica sand and braunite particles, the results of the time evolution of the main compounds detected in the exhaust gas of the reactor after supplying the beds are depicted in Fig. 3b. Similar results were found for the silica sand and braunite bed, with low concentrations of CO2 and CO (the latter not represented

33

in the figure due to the negligible concentration) in the exhaust gases and with a wide and smooth peak for the concentration of H2O, which is attributed to the moisture content of the bed particles in both cases. In these cases, concentrations of H2O and CO2 were detected for a shorter time than in the sepiolite bed (notice the different scale in x-axis for Fig. 3a and b). The concentration of H2O in the exhaust gases of the braunite bed is higher than in the silica sand bed, however, this is due to the higher mass of braunite, 34.0 g, compared to silica sand, 26.0 g, supplied to obtained beds with the same volume, i.e., same bed height. The main difference detected for the braunite and silica sand particles is the concentration of O2 in the exhaust gases (notice the cut in the y-axis in Fig. 3b), which is slightly lower for the braunite particles, showing a reduction to a value of 19.8 vol% after supplying the braunite bed, compared to the silica sand bed. This result confirms the oxygen carrier capability of the braunite particles and the inert character of silica sand particles at high temperature under an oxidative atmosphere. A mass balance was carried out for the analysis of the effect of high temperature and oxidative atmosphere on the different bed materials tested. For the mass balance, the concentration of nitrogen in the exhaust gases was determined by difference and compared to the inlet nitrogen mass flow rate contained in the 10 lpm of air supplied to the reactor as fluidizing agent. Then, the instantaneous mass flow rate of each compound detected in the exhaust gases can be determined and the mass loss of the bed material can be calculated by integration of the time evolution curves. Following this procedure, the resulting mass loss for the sepiolite particles during their calcination in the reactor at 800 °C with an air flow rate of 10 lpm was 20.2 wt%, whereas a negligible mass loss of only 0.3 wt% was obtained for both silica sand and braunite beds. The reaction of sepiolite particles during calcination was confirmed by a basic characterization of the fresh sepiolite consisting in an ultimate and a proximate analysis. The ultimate analysis confirmed a content of 5.3 wt% of carbon and 1.5 wt% of hydrogen in the sepiolite, which, combined to the moisture content of sepiolite of 6.4 wt%, is the cause of the high H2O and lower CO2 concentration detected in the exhaust gases after supplying the sepiolite bed to the reactor. In addition, a thermogravimetric analysis (TGA) of the fresh sepiolite particles confirmed the mass loss detected in the reactor. The mass loss of fresh sepiolite during the TGA test was 19.3 wt%, a close value to the 20.2 wt% measured in the reactor, confirming the mass loss of sepiolite during calcination. TGA tests conducted under the same conditions for silica sand and braunite particles also confirmed a negligible mass loss for these bed materials at high temperature under an oxidative atmosphere.

Fig. 3. Effect of temperature and oxidative atmosphere on the bed materials. (a) Sepiolite, (b) silica sand and braunite (notice the y-axis in figure b) is cut).

34

A. Soria-Verdugo et al. / Waste Management 105 (2020) 27–38

Once the effect of temperature and oxidative atmosphere on the bed material was evaluated, combustion tests using different bed materials were also performed. The concentration of each compound detected by the FTIR spectrometer in the exhaust fumes of combustion for each test are included in the supplementary material, in Tables S1 and S2. In the following sections, the main trends observed for the composition of the exhaust fumes of combustion of sewage sludge and Cynara cardunculus L. in each bed and for each operating condition are discussed. The combustion tests using sewage sludge as a fuel in bubbling fluidized beds of sepiolite and braunite particles were run for all the experimental conditions used for the silica sand bed, i.e., bed temperatures of 750 and 800 °C and air flow rates of 6 and 10 lpm. Some similarities can be observed for the sewage sludge combustion tests conducted in the sepiolite and braunite beds and those performed in a bed of silica sand particles. For instance, a significant reduction of CO concentration in the exhaust fumes produced during combustion of sewage sludge in sepiolite and braunite beds was also detected when increasing the air flow rate for a bed temperature of 800 °C. This reduction in concentration is also observable for other compounds containing carbon, such as CH4, C2H4, C6H6, CHOH, and HCN, although the reduction is not as remarkable as for CO. In addition, the slight increase of the concentration of SO2, HCl and HF in the exhaust fumes of sewage sludge combustion with bed temperature and the increase of N2O concentration with the gas flow rate detected in the experiments run in a silica sand fluidized bed were also observed for the beds of sepiolite and braunite particles. However, some effects of the bed material used for the sewage sludge combustion tests can be observed in the composition of the exhaust fumes. The emission of hydrogen cyanide HCN is much lower when the sewage sludge combustion occurs in a sepiolite fluidized bed, especially for the lowest bed temperature tested Tb = 750 °C. In contrast, the concentration of HF in the exhaust fumes produced during sludge combustion in a fluidized bed of sepiolite particles at 800 °C is much higher than that of a silica sand or braunite fluidized bed. This higher concentration might be enhanced by a faster and more homogeneous heating of low-density char and ash particles generated during sewage sludge combustion, due to a better mixing and more homogeneous distribution in a fluidized bed conformed by lower density particles like sepiolite. Moreover, NH3 emission is lower in both sepiolite and braunite beds compared to sewage sludge combustion in a fluidized bed of silica sand particles. This is especially evident for the sludge combustion tests in a fluidized bed of braunite operated at a bed temperature of 800 °C. In the case of sewage sludge combustion in a braunite fluidized bed, the concentration of CH4 and C2H4 in the exhaust fumes are higher for a bed temperature of 750 °C and an air flow rate of 6 lpm, compared to the tests run in sepiolite and silica sand beds. This higher concentration may be caused by the lower residence time of the exhaust gases in the hot zone of the reactor due to a poor mixing of the fuel particles in this high-density bed. For the low air flow rate, the sludge particles would be more probably moving close to the surface of the bed of braunite particles due to their flotsam character in this bed. Therefore, the time required to oxidize the gases released at the top of the bed for a temperature of 750 °C is higher than the residence time of these gases in the bed, enhancing the emission of low molecular weight hydrocarbons for these operating conditions. This effect is also confirmed by the high concentration of CO in the exhaust fumes of sludge combustion in the braunite bed compared to the tests run in sepiolite and silica sand beds. The effect of the bed material employed was also quantified for the combustion of Cynara cardunculus L. Very similar results were obtained for the cardoon combustion in the sepiolite and braunite beds and the bed of silica sand particles. The only significant differ-

ences are the much lower emission of CO in the sepiolite bed, which can be attributed to the better mixing of biomass and larger residence time of gases in the low-density bed of sepiolite particles, where CO is released by fuel particles inside the dense bed. A lower emission of HCl in the braunite bed was also detected, which might be enhanced by the poor mixing and inefficient heating of Cynara particles in the high-density bed of braunite particles. In view of the results obtained from the combustion tests of sewage sludge and Cynara cardunculus L. conducted in fluidized beds of different particles, it can be concluded that the bed material employed influences the composition of the exhaust fumes and, thus, the pollutant emissions associated with the solid fuel combustion. This effect is mainly caused by the distribution of fuel particles in the dense bed, which affects both the heating rate of solid fuel particles and the residence time of the gases released during combustion in the hot zone of the reactor. 3.5. Carbon, sulfur and nitrogen conversion and combustion efficiency The conversion of the carbon, sulfur and nitrogen contained in the fuel to gaseous compounds was calculated for each combustion test. For the calculation, the mass flow rate of exhaust gases was determined by a mass balance to the nitrogen flow rate. The nitrogen content in the exhaust gases was calculated by difference, assuming that the whole content of non-detected compounds by the FTIR corresponds to nitrogen. Then, equaling the exhaust mass flow rate of nitrogen to the mass flow rate supplied as fluidizing air, the mass flow rate of exhaust gases can be calculated. Once the mass flow rate of the exhaust gases was known, the mass flow rate of each compound can be calculated considering its molecular weight. Therefore, the carbon, sulfur and nitrogen conversion was determined as the ratio of mass flow rate of each element in the exhaust gases, considering all the compounds containing this element, to the mass flow rate of the element supplied in the fuel, determined as the average fuel feeding rate times the content of the element obtained from the elemental analysis (Table 1). The combustion efficiency was also calculated as the mass flow rate of CO2 in the exhaust gases over the mass flow rate of carbon supplied in the fuel, as proposed by Ren et al. (2017). The values of the carbon, sulfur and nitrogen conversion obtained for each test of both sewage sludge and Cynara cardunculus L. combustion are reported in Table 5. High carbon conversions above 65% and low nitrogen conversion below 10% were obtained for both sludge and Cynara in all cases. Also, the values of combustion efficiency were higher than 61% for sludge and 76% for Cynara in all cases, obtaining higher values of the efficiency for the combustion of cardoon. In contrast, the sulfur conversion obtained differs significantly for the two solid fuels tested, ranging from 33 to 45% for the combustion of sewage sludge, whereas the maximum sulfur conversion obtained for the combustion of Cynara cardunculus L. was only 3.8%. For the combustion of sludge, the lower values of the carbon conversion, and thus the combustion efficiency, were obtained for the sepiolite bed at 750 °C and for the braunite bed at 800 °C, however, the sulfur conversion is also low in these cases. The cause of the low carbon and sulfur conversion in these cases may be a deficient heating of the sludge particles, which contributes to decrease the combustion efficiency. In the case of Cynara cardunculus L. combustion, the bed of sepiolite particles contributed to increase the carbon conversion, i.e., the efficiency of combustion of the solid fuel, which might be due to a more homogeneous distribution of the low-density cardoon particles in the low-density sepiolite bed. The low values of nitrogen conversion obtained for both fuels is in good agreement with the results reported by Cammarota et al. (2013) and Li and Chyang (2020), who obtained an exponential decrease of the nitrogen conversion with the nitrogen content in the solid fuel, during its combustion

35

A. Soria-Verdugo et al. / Waste Management 105 (2020) 27–38 Table 5 Carbon CC, sulfur CS and nitrogen CN conversion and combustion efficiency gC for each combustion test. Fuel

Bed material

Tb [°C]

q [lpm]

CC [%]

CS [%]

CN [%]

gC [%]

Sludge Sludge Sludge Sludge Sludge Sludge Sludge Sludge Sludge Sludge Sludge Sludge Cynara Cynara Cynara Cynara Cynara Cynara

Silica sand Silica sand Silica sand Silica sand Sepiolite Sepiolite Sepiolite Sepiolite Braunite Braunite Braunite Braunite Silica sand Silica sand Sepiolite Sepiolite Braunite Braunite

750 750 800 800 750 750 800 800 750 750 800 800 750 800 750 800 750 800

6 10 6 10 6 10 6 10 6 10 6 10 6 10 6 10 6 10

76.2 73.4 73.2 74.2 65.4 69.3 72.2 74.4 76.8 73.3 68.3 66.6 76.3 85.6 82.4 87.7 81.8 81.0

40.6 42.8 44.7 46.0 31.0 40.0 37.8 43.6 36.6 40.4 33.4 39.2 1.5 3.8 1.9 3.6 1.3 2.5

7.4 8.8 5.9 6.3 6.9 8.1 5.9 7.1 8.0 9.1 5.7 5.8 3.5 3.9 3.8 4.4 3.8 3.8

74.8 71.7 71.9 73.8 64.1 67.9 71.0 73.9 74.9 72.0 67.4 66.3 76.0 85.3 82.2 87.5 81.5 80.7

at 750 and 800 °C. In fact, they stated that there are limited values of the nitrogen conversion available in the literature for fuel nitrogen contents higher than 2.5%, and the value of the corresponding nitrogen conversion in these cases are low. In our case, both solid fuels are characterized by a nitrogen content above 2.5%, being 2.6% for Cynara and 6.5% for sludge (Table 1). The values reported in Table 5 for the conversion of carbon, sulfur and nitrogen and the combustion efficiencies are within the range of variation obtained by Ren et al. (2017), who studied the combustion of several raw and torrefied biomass types. Considering the combustion efficiency obtained for sewage sludge, high-density beds are more efficient if the combustion process occurs at low temperature and air flow rates, i.e., 750 °C and 6 lpm. In contrast, for a higher bed temperature and air flow rate, 800 °C and 10 lpm, a higher sludge combustion efficiency is attained in a low-density bed of sepiolite. Therefore, taking into account also the significant reduction of carbon monoxide emissions for high temperature and air flow rate, using sepiolite particles to conform the bed is recommended. Furthermore, the lower density of sepiolite implies a lower bed pressure drop, which permits to design lower pressure drop gas distributors for the reactor (Karri and Werther, 2003), saving air pumping cost during the operation of the combustor. 3.6. Ash accumulation Since the experimental facility employed to conduct the solid fuel combustion experiments has no ash removal system, some

ash accumulation in the bed occurs. This ash accumulation during the short combustion experiments carried out, ranging from 30 min to 1 h, should not be a problem for the Cynara combustion tests, considering the low ash content of only 7.5 wt% wb characteristic of this lignocellulosic biomass and the very low density and fragility of the ash produced in this case, which due to attrition inside the bed would generate small ashes directly blown out of the reactor by the gas flow rate. In fact, no significant bed pressure drop increase was detected as the combustion of cardoon progressed for any of the operating conditions and bed materials tested. In contrast, sewage sludge is characterized by a high ash content, 22.4 wt% wb in this case. Furthermore, the ash generated from sludge combustion is much denser than that generated from burning cardoon, and sludge ash is also characterized by a higher hardness. Therefore, in the case of sludge combustion, ash would be accumulated in the bed. In fact, ash accumulation was detected by a linear increase of the bed pressure drop with time during the sludge combustion tests. Fig. 4a shows the increase of bed pressure drop as a function of time during sludge combustion in a fluidized bed of silica sand particles. Similar results were obtained for the combustion tests run in the sepiolite and braunite beds and for bed temperatures of 750 °C, since the sludge feeding rate is an exclusive function of the air flow rate. A linear increase of the bed pressure drop with time can be observed for the tests performed for both an air flow rate of 6 and 10 lpm. Obviously, the slope of the bed pressure drop increase is higher for the air flow rate of 10 lpm, due to the higher sludge feeding rate required to obtain the same concentration of oxygen

Fig. 4. Variation of (a) increase of bed pressure drop and (b) mass of ash accumulated in the bed.

36

A. Soria-Verdugo et al. / Waste Management 105 (2020) 27–38

in the exhaust fumes when increasing the air flow rate. Fluctuations in the bed pressure drop signal were detected for both air flow rates, as a result of the bubbles’ motion and eruption in the freeboard (Davidson and Fluidization, 1971; Kunii and Levenspiel, 1991). These fluctuations are higher when increasing the air flow rate due to a more vigorous fluidization of the bed caused by the presence of bigger bubbles (Puncochár et al., 1985). Considering the bed cross section, the mass of ash accumulated in the bed can be calculated from the measured bed pressure drop increase. The time evolution of the mass of ash accumulated during the sludge combustion tests in the silica sand bed operated at 800 °C is plotted in Fig. 4b for each air flow rate, together with the calculated value. The estimation of the mass of ash accumulated in the bed was made using the sludge feeding rate in each case, i.e., 1.6 g/min of sludge for 6 lpm of air and 2.3 g/min for 10 lpm, and the ash content of sludge, and considering that all the ash contained in sludge is accumulated in the bed. The mass of ash accumulated in the bed follows a linear trend, due to the constant feeding rate of sludge, and shows a very good agreement with the calculated value. The total mass accumulated after 30 min of sludge combustion test is around 11 g and 17 g for air flow rates of 6 and 10 lpm, respectively. These masses of ash accumulated are comparable to the bed material initial mass, which is 11.8 g for sepiolite, 26.0 g for silica sand, and 34.0 g for braunite. These increases of the bed mass due to sludge ash accumulation increase the bed height and affect its fluid-dynamics, since the size of bubbles in the bed depends on both the gas velocity and the bed height (Darton et al., 1977). Therefore, the duration of the sludge combustion tests was limited to 30 min of stable operation, whereas for

the combustion of Cynara, experiments of even 1 h of stable operation were performed because of the negligible effect of ash accumulation during the combustion of this lignocellulosic biomass. Even though ash is accumulated in the bed during sewage sludge combustion, the concentration of the main compounds detected by the FTIR in the exhaust fumes was unaffected, as can be seen from the stable average values obtained for the time evolution of the main compounds in Fig. 2. However, the emission of hydrogen fluoride HF seems to be influenced by the presence of ash in the bed. The time evolution of the concentration of HF in the exhaust fumes of sludge combustion in each fluidized bed is depicted in Fig. 5, for each operating condition. Both for the combustion in the silica sand and the braunite beds, the emission of HF increases as time progresses due to a higher presence of ash in the bed. In addition, the concentration of HF is higher for the cases of a bed temperature of 800 °C, for which the thermal energy, and then the heating capability, of the bed is higher. In the case of sludge combustion in a sepiolite fluidized bed, the results obtained for the air flow rate of 10 lpm are different. In these cases, the concentration of HF in the exhaust fumes increases faster and to higher values compared to the tests with silica sand and braunite beds, and then the value is almost stable or even decreases slightly. This different evolution of the HF emission in the sepiolite bed might be caused by a faster heating of the ash particles produced, due to a better distribution throughout the whole sepiolite bed because of its lower density, which is enhanced by a higher air flow rate. Considering the high ash content characteristic of sewage sludge, the use of ash removing systems in fluidized bed sludge

Fig. 5. Time evolution of the HF concentration in the exhaust fumes produced by combustion of sewage sludge in a BFB of (a) silica sand, (b) braunite, (c) sepiolite particles, for various operating conditions and bed materials.

A. Soria-Verdugo et al. / Waste Management 105 (2020) 27–38

combustors is recommended to avoid changes in the bed fluiddynamics caused by variations of the bed height due to ash accumulation and to limit the emissions of hydrogen fluoride. The proper design of these ash removing systems are specially relevant for low-density beds, such as sepiolite beds, due to a higher effect of ash accumulation in the bed fluid-dynamics, caused by the lower mass of the bed material, and the higher emission of hydrogen fluoride when ash is accumulated in a sepiolite bed. 3.7. Agglomeration problems Agglomeration problems were detected during some of the combustion tests using Cynara cardunculus L. as a fuel. Table S3 (supplementary material) shows the operation time until agglomeration detection as a function of the operating conditions and the mass of bed material employed. Originally, the mass of bed material employed was that required to have the same bed height in all cases, i.e., 11.8 g of sepiolite, 26.0 g of silica sand, and 34.0 g of braunite. However, for those combustion tests in which agglomeration led to bed defluidization, the mass of bed material was increased, and the test was repeated with a higher dense phase mass to evaluate its effect. The maximum measuring time was 1 h, therefore, an operation time >60 min was reported in Table S3 for those cases for which no agglomeration was detected during this time period. Agglomeration problems leading to the bed defluidization were detected during the combustion of cardoon only for some tests conducted in sepiolite and silica sand beds. In contrast, no relevant agglomerates were detected for the combustion experiments run in braunite beds. This might be attributed to the higher mass of bed material during the tests in the braunite bed due to its higher density. In fact, increasing the mass of silica sand and sepiolite, the agglomeration problems were solved. For a bed temperature of 750 °C and an air flow rate of 6 lpm, using 26.0 g of either sepiolite or silica sand was enough to prevent agglomeration problems. However, a mass of sepiolite particles of 11.8 g resulted in agglomeration problems after only 16 min from the beginning of the Cynara feeding to the bed. For a higher bed temperature of 800 °C and air flow rate of 10 lpm, a mass of 34.0 g of either sepiolite or silica sand, equal to that of the braunite bed, was found to be able to avoid any bed defluidization during the measuring time. Nonetheless, for the mass of bed material required to have the same bed height with all the bed materials, agglomeration problems were detected for both the sepiolite and the silica sand beds. Specifically, using a mass of 11.8 g of sepiolite resulted in bed defluidization after around 11 min of the cardoon feeding to the bed operated at 800 °C and 10 lpm. For the same operating conditions, defluidization occurred after 13 min in a bed of 26.0 g of silica sand. Therefore, defluidization due to agglomerates formation during combustion tests was found to be enhanced by a low mass of bed material. In contrast to the results obtained for the Cynara cardunculus L. combustion, no defluidization caused by agglomeration formation was detected during any of the tests conducted for sewage sludge combustion. This may be attributed to the slightly higher ash melting temperatures measured for sewage sludge (see Table 1) and mainly to the higher content on alkalis in the Cynara ashes, which contributes to the formation of agglomerates (Rizeq and Shadman, 1989). 4. Conclusions The combustion of sewage sludge in a bubbling fluidized bed was studied varying the bed temperature and the fluidizing air mass flow rate. The dense bed density was also varied by changing

37

the bed material particles, employing sepiolite, silica sand and braunite. The pollutant problems derived from sewage sludge incineration were evaluated by analyzing the composition of the exhaust fumes obtained from sludge combustion by means of a FTIR equipment. The main compounds of the exhaust fumes, apart from O2, CO2 and H2O, were CO and SO2, with average concentrations around 1000 ppm, and NO with an average concentration of approximately 500 ppm. Furthermore, lower concentrations of a wide variety of low-weigh hydrocarbons and compounds containing nitrogen were also detected in the exhaust fumes. In contrast, the concentration of pollutant emissions in the exhaust fumes of the combustion of Cynara cardunculus L. is much lower, being around 50% lower for CO and NO, and negligible for the rest of compounds. The efficiency of combustion of sludge is also low, around 70% depending on the operating conditions, compared to that obtained for the combustion of Cynara, approximately 80%. Considering the pollutant emissions and the efficiency of sludge combustion, a low-density fluidized bed of sepiolite particles operated at high temperature and air flow rate are recommended. Furthermore, to limit the high emissions of sulfur oxides characteristic of sludge combustion, the use of in-bed sorbing materials is advisable. In the case of sludge combustion, an ash removing system should be used to avoid an increase of the bed mass and pressure drop during operation and a higher emission of HF. Agglomeration problems in the bed were also detected, but only for the combustion of cardoon due to its higher content of alkali compared to sewage sludge. Declaration of Competing Interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Acknowledgments This project has received funding from European Union’s Horizon 2020 Research and Innovation Program under grant agreement number 731101 (BRISK II). Appendix A. Supplementary material Supplementary data to this article can be found online at https://doi.org/10.1016/j.wasman.2020.01.036. References Aho, M., Gil, A., Taipale, R., Vainikka, P., Vesala, H., 2008. A pilot-scale fireside deposit study of co-firing Cynara with two coals in a fluidized bed. Fuel 87, 58– 69. Arena, U., di Gregorio, F., 2014. Gasification of a solid recovered fuel in a pilot scale fluidized bed reactor. Fuel 117, 528–536. Bartolomé, C., Gil, A., Ramos, I., 2010. Ash deposition behavior of cynara–coal blends in a PF pilot furnace. Fuel Process. Technol. 91, 1576–1584. Cammarota, A., Chirone, R., Salatino, P., Solimene, R., Urciuolo, M., 2013. Particulate and gaseous emissions during fluidized bed combustion of semi-dried sewage sludge: effect of bed ash accumulation on NOx formation. Waste Manage. 33, 1397–1402. Chirone, R., Salatino, P., Scala, F., Solimene, R., Urciuolo, M., 2008. Fluidized bed combustion of pelletized biomass and waste-derived fuels. Combust. Flame 155, 21–36. Darton, R.C., LaNauze, R.D., Davidson, J.F., Harrison, D., 1977. Bubble growth due to coalescence in fluidized beds. Trans. Inst. Chem. Eng. 55, 274–280. Davidson J.F., Harrison, D., 1971. Fluidization. Academic Press Inc. de Palma, K.R., Garcia-Hernando, N., Silva, M.A., Tomaz, E., Soria-Verdugo, A., 2019. Pyrolysis and combustion kinetic study and complementary study of ash fusibility behavior of sugarcane bagasse, sugarcane straw, and their pellets – case study of agro-industrial residues. Energy Fuels 33 (4), 3227–3238. Fernández Llorente, M.J., Carrasco García, J.E., 2005. Comparing methods for predicting the sintering of biomass ash in combustion. Fuel 14–15, 1893–1900.

38

A. Soria-Verdugo et al. / Waste Management 105 (2020) 27–38

Fernández Llorente, M.J., Escalada, Cuadrado R., Murillo Laplaza, J.M., Carrasco García, J.E., 2006. Combustion in bubbling fluidised bed with bed material of limestone to reduce the biomass ash agglomeration and sintering. Fuel 85, 2081–2092. Folgueras, M.B., Alonso, M., Folgueras, J.R., 2015. Effect of sludge addition to coal on Na, K and S volatilization in ashing process and ash fusibility. Fuel Process. Technol. 138, 714–723. Fonts, I., Gea, G., Azuara, M., Ábrego, J., Arauzo, J., 2012. Sewage sludge pyrolysis for liquid production: a review. Renew. Sustain. Energy Rev. 16, 2781–2805. Fytili, D., Zabaniotou, A., 2008. Utilization of sewage sludge in EU application of old and new methods – a review. Renew. Sustain. Energy Rev. 12, 116–140. Garcia-Gutierrez, L.M., Soria-Verdugo, A., Marugán-Cruz, C., Ruiz-Rivas, U., 2014. Simulation and experimental study on the motion of non-reacting objects in the freeboard of a fluidized bed. Powder Technol. 263, 112–120. Garcia-Gutierrez, L.M., Soria-Verdugo, A., Ruiz-Rivas, U., 2015. Optimization of the feeding ports location in a fluidized bed combustor based on Monte Carlo simulations of fuel particles motion. Fuel 141, 82–92. Garcia-Gutierrez, L.M., Hernandez-Jimenez, F., Cano-Pleite, E., Soria-Verdugo, A., 2017. Improvement of the simulation of fuel particles motion in a fluidized bed by considering wall friction. Chem. Eng. J. 321, 175–183. Geldart, D., 1973. Types of gas fluidization. Powder Technol. 7, 285–292. Gómez-Barea, A., Leckner, B., 2010. Modeling of biomass gasification in fluidized beds. Prog. Energy Combust. Sci. 36, 444–509. Gómez-Hernández, J., Serrano, D., Soria-Verdugo, A., Sánchez-Delgado, S., 2016. Agglomeration detection by pressure fluctuation analysis during Cynara cardunculus L. gasification in a fluidized bed. Chem. Eng. J. 284, 640–649. Gominho, J., Curt, M.D., Lourenço, A., Fernández, J., Pereira, H., 2018. Cynara cardunculus L. as a biomass and multi-purpose crop: a review of 30years of research. Biomass Bioenergy 109, 257–275. González, J.F., González-García, C.M., Ramiro, A., González, J., Sabio, E., Gañán, J., Rodríguez, M.A., 2004. Combustion optimisation of biomass residue pellets for domestic heating with a mural boiler. Biomass Bioenergy 27, 145–154. Guo, F., Zhong, Z., 2017. Experimental studies on combustion of composite biomass pellets in fluidized bed. Sci. Total Environ. 599–600, 926–933. Kaminsky, W., Kummer, A.B., 1989. Fluidized bed pyrolysis of digested sewage sludge. J. Anal. Appl. Pyrol. 16, 27–35. Karri, S.B.R., Werther, J., 2003. Gas distributor and plenum design in fluidized beds. In: Yang, W.C. (Ed.), Handbook of Fluidization and Fluid-particle Systems. Marcel Dekker Inc., New York, pp. 164–179. Köhler, A., Rasch, A., Pallarès, D., Johnsson, F., 2017. Experimental characterization of axial fuel mixing in fluidized beds by magnetic particle tracking. Powder Technol. 316, 492–499. Koppatz, M., Pfeifer, C., Hofbauer, H., 2011. Comparison of the performance behavior of silica sand and olivine in a dual fluidized bed reactor system for steam gasification of biomass as a pilot plant scale. Chem. Eng. J. 175, 468–483. Kunii, D., Levenspiel, O., 1991. Fluidization Engineering. Butterworth-Heinemann, Boston. Leckner, B., 1998. Fluidized bed combustion: Mixing and pollutant limitation. Progr. Energy Combust. Sci. 24, 31–61. Li, P.W., Chyang, C.S., 2020. A comprehensive study on NOx emission and fuel nitrogen conversion of solid biomass in bubbling fluidized beds under staged combustion. J. Energy Inst. 93, 324–334. Li, P.W., Chyang, C.S., Ni, H.W., 2018. An experimental study of the effect of nitrogen origin on the formation and reduction of NOx in fluidized-bed combustion. Energy 154, 319–327. Li, M., Li, F., Liu, Q., Fang, Y., Xiao, H., 2019. Regulation of ash fusibility for high ashfusion-temperature (AFT) coal by industrial sludge addition. Fuel 244, 91–103. Lundberg, L., Soria-Verdugo, A., Pallarès, D., Johansson, R., Thunman, H., 2017. The role of fuel mixing on char conversion in a fluidized bed. Powder Technol. 316, 677–686. Lundberg, L., Tchoffor, P.A., Pallarès, D., Johansson, R., Thunman, H., Davidsson, K., 2016. Influence of surrounding conditions and fuel size on the gasification rate of biomass char in a fluidized bed. Fuel Process. Technol. 144, 323–333. Magdziarz, A., Wilk, M., Gajek, M., Nowak-Wozny, D., Kopia, A., Kalemba-Rec, I., Kozinski, J.A., 2016. Properties of ash generated during sewage sludge combustion: a multifaceted analysis. Energy 113, 85–94. Morato-Godino, A., Sanchez-Delgado, S., Garcia-Hernando, N., Soria-Verdugo, A., 2018. Pyrolysis of Cynara cardunculus L. samples – effect of operating

conditions and bed stage on the evolution of the conversion. Chem. Eng. J. 351, 371–381. Morris, J.D., Daood, S.S., Chilton, S., Nimmo, W., 2018. Mechanisms and mitigation of agglomeration during fluidized bed combustion of biomass: a review. Fuel 230, 452–473. Olsson, J., Pallarès, D., Johnsson, F., 2011. Lateral fuel dispersion in a large-scale bubbling fluidized bed. Chem. Eng. Sci. 74, 148–159. Pallarès, D., Johnsson, F., 2006. A novel technique for particle tracking in cold 2dimensional fluidized beds simulating fuel dispersion. Chem. Eng. Sci. 61, 2710– 2720. Pallarès, J., Gil, A., Cortés, C., Herce, C., 2009. Numerical study of co-firing coal and Cynara cardunculus in a 350 MWe utility boiler. Fuel Process. Technol. 90, 1207–1213. Pikkarainen, T., Hiltunen, I., 2017. Chemical looping combustion of solid biomass – performance of ilmenite and braunite as oxygen carrier materials. Proc. from the 25th European Biomass Conference and Exhibition. Puncochár, M., Drahos, J., Cermák, J., Selucky, K., 1985. Evaluation of minimum fluidization velocity in gas fluidized beds from pressure fluctuations. Chem. Eng. Commun. 35, 81–87. Rees, A.C., Davidson, J.F., Hayhurts, A.N., 2005. The rise of a buoyant sphere in a gasfluidized bed. Chem. Eng. Sci. 60, 1143–1153. Ren, X., Dun, R., Meng, X., Vorobiev, N., Schiemann, M., Levendis, Y.A., 2017. Carbon, sulfur and nitrogen oxide emissions from combustion of pulverized raw and torrefied biomass. Fuel 188, 310–333. Rios, G.M., Dang, Tran K., Masson, H., 1986. Free object motion in a gas fluidized bed. Chem. Eng. Commun. 47, 247–272. Rizeq, R.G., Shadman, F., 1989. Alkali-induced agglomeration of solid particles in coal combustors and gasifiers. Chem. Eng. Commun. 81, 83–96. Ronda, A., Gómez-Barea, A., Haro, P., de Almeida, V.F., Salinero, J., 2019. Elements partitioning during thermal conversion of sewage sludge. Fuel Process. Technol. 186, 156–166. Scala, F., 2018. Particle agglomeration during fluidized bed combustion: mechanisms, early detection and possible countermeasures. Fuel Process. Technol. 171, 31–38. Sette, E., Pallarès, D., Johnsson, F., 2014. Experimental quantification of lateral mixing of fuels in fluid-dynamically down-scaled bubbling fluidized beds. Appl. Energy 136, 671–681. Soria-Verdugo, A., García-Gutiérrez, L.M., Sánchez-Delgado, S., Ruiz-Rivas, U., 2011a. Circulation of an object immersed in a bubbling fluidized bed. Chem. Eng. Sci. 66, 78–87. Soria-Verdugo, A., García-Gutiérrez, L.M., García-Hernando, N., Ruiz-Rivas, U., 2011b. Buoyancy effects on objects moving in a bubbling fluidized bed. Chem. Eng. Sci. 66, 2833–2841. Soria-Verdugo, A., García-Hernando, N., Almendros-Ibáñez, J.A., García-Hernando, U., 2011c. Motion of a large object in a bubbling fluidized bed with a rotating distributor. Chem. Eng. Process. 50, 859–868. Soria-Verdugo, A., Morato-Godino, A., Garcia-Gutierrez, L.M., Garcia-Hernando, N., 2017a. Pyrolysis of sewage sludge in a fixed and a bubbling fluidized bed – estimation and experimental validation of the pyrolysis time. Energy Convers. Manage. 144, 235–242. Soria-Verdugo, A., Goos, E., Morato-Godino, A., Garcia-Hernando, N., Riedel, U., 2017b. Pyrolysis of biofuels of the future: sewage sludge and microalgae – thermogravimetric and modelling of the pyrolysis under different temperature conditions. Energy Convers. Manage. 138, 261–272. Soria-Verdugo, A., Goos, E., Garcia-Hernando, N., Riedel, U., 2018. Analyzing the pyrolysis kinetics of several microalgae species by various differential and integral isoconversional kinetic methods and the Distributed Activation Energy Model. Algal Res. 32, 11–29. Soria-Verdugo, A., von Berg, L., Serrano, D., Hochenauer, C., Scharler, R., Anca-Couce, A., 2019. Effect of bed material density on the performance of steam gasification of biomass in bubbling fluidized beds. Fuel 257, 116118. Syed-Hassan, S.S.A., Wang, Y., Hu, S., Su, S., Xiang, J., 2017. Thermochemical processing of sewage sludge to energy and fuel: fundamentals, challenges and considerations. Renew. Sustain. Energy Rev. 80, 888–913. Werther, J., Ogada, T., 1999. Sewage sludge combustion. Progr. Energy Combust. Sci. 25, 55–116. Winaya, I.N.S., Shimizu, T., Yamada, D., 2007. A new method to evaluate horizontal solid dispersion in a bubbling fluidized bed. Powder Technol. 178, 173–178.