Article pubs.acs.org/EF
Effects of Reduced Bed Temperature in Laboratory- and Full-Scale Fluidized-Bed Boilers: Particle, Deposit, and Ash Chemistry Frida Jones,*,†,‡ Fredrik Niklasson,† Daniel Lindberg,‡ and Mikko Hupa‡ †
SP Technical Research Institute of Sweden, Box 857, SE-501 15 Borås, Sweden Åbo Akademi Process Chemistry Centre, Biskopsgatan 8, FI-20500 Åbo, Finland
‡
ABSTRACT: This study focuses on how the temperature of the bed in a waste-fired fluidized-bed boiler affects the chemical composition of ash and deposits. The aim was to reduce the concentration of corrosive elements in the convection pass, which can lead to both less frequent soot blowing intervals and extended superheater lifetimes. Complementary laboratory-scale tests were carried out in a single-pellet reactor to study online alkali and Zn emissions during temperature changes, using an ICP-MS instrument. The full-scale study was based on full-scale experiments at a plant consisting of two 20 MWth fluidized-bed boilers firing a mixture of municipal solid waste and industrial waste. The boilers are normally operated at bed temperatures of about 870 °C. This normal operation was compared in this work to a test case in which the bed temperature was reduced below 720 °C by altering the air staging and flue gas recycling. The experimental work included collecting samples of fuel, ash, and particles under the two different sets of operating conditions. Furthermore, deposits on temperature-controlled probes were sampled upstream of the superheaters. By reducing the bed temperature, the sand consumption of the plant could be reduced by roughly 25%. The measurements showed that the amount of submicrometer particles decreased and the fouling rate on deposit probes was reduced by about 20%. The measured concentration of HCl in the flue gas increased as the bed temperature was reduced. This might be a consequence of the reduced formation of alkali chlorides. In addition, results from the laboratory-scale tests indicated a trend of reduced alkali emissions from the fluidized bed with reduced temperature, and thermodynamic equilibrium calculations confirmed the trends.
■
volatilized at higher temperatures (>600 °C) or remain in a solid or liquid state under combustion conditions.8 The trace element zinc has gained attention in the past decade. During combustion, zinc is released and readily available to form gaseous products in the flue gas, and previous studies indicate that zinc readily forms zinc chloride, which melts at 318 °C.9 Findings of ZnCl2 in fireside deposits in municipal solid waste (MSW) boilers have led to the general belief that ZnCl2 might play a role in the corrosion of heattransfer tubes in these boilers owing to its low melting temperature and high corrosivity when molten.10−17 In the combustion of municipal solid waste (MSW) and industrial waste (IW), the chemical reactions that remove alkali chlorides from flue gas are hard to predict and control owing to the fuel heterogeneity. A “normal” bed temperature in a fluidized-bed boiler could be on the order of 850 °C. Reducing this bed temperature is expected to change the behavior of the alkali metals, hence reducing the formation of particles that might cause corrosive deposits on heat-exchanging surfaces. However, while doing this, the boilers still must obey the European Waste directive and keep the flue gas at 850 °C for at least 2 s at the top of the boiler. The purpose of this work was to study this hypothesis with full-scale experiments in an industrial boiler and laboratory-scale experiments in a singlepellet fluidized-bed reactor. Results from the full-scale project were also reported by Pettersson et al.18
INTRODUCTION
Waste combustion is becoming more common in Europe as increasingly strict regulations regarding landfills are being introduced. Landfills are known to emit, among other species, methane, which is a much more powerful greenhouse gas than carbon dioxide.1−3 Instead, combustible waste could be used for energy recovery in combined heat and power plants. Nevertheless, because of the heterogeneity of waste fuels and the high concentrations of inorganic elements in combination with high concentrations of chlorine, there is an issue concerning fouling and deposition on heat-transfer surfaces in the boilers. Especially at high temperatures, Cl-containing deposits are known to cause severe corrosion on superheaters. To minimize such problems, waste boilers are generally operated at a significantly lower steam temperature than, for example, fossil fuel boilers.4 During combustion, alkali metals present in the fuel are partly released to the gas and partly bound in the ash.5 Once released to the gas, the elements are available to take part in chemical reactions. The boiling temperatures for pure potassium and sodium are 774 and 811 °C, respectively,6 but as these metals are rarely found in pure form, the volatilization temperatures depend on the way they are bound in the fuel. As described by Zevenhoven et al.,7 ash-forming matter can be present in the fuel in four different classes: organically bound material, dissolved salts, included mineral matter, and excluded mineral matter. The K and Na that is organically bound can be volatilized already at lower temperatures (approximately 300− 400 °C), whereas the K and Na in the other classes are © 2013 American Chemical Society
Received: May 6, 2013 Revised: July 10, 2013 Published: July 10, 2013 4999
dx.doi.org/10.1021/ef400836e | Energy Fuels 2013, 27, 4999−5007
Energy & Fuels
■
Article
Fly ash particles were sampled at position A, port P3, in Figure 1, with a quench/dilution probe in combination with a Dekati Low Pressure Impactor (DLPI), which is a cascade impactor separating particles according to their size. The method was further described by Johansson et al.19 Inside the impactor, the particles were collected on preweighed polycarbonate substrates mounted on jet plates. After sampling, the polycarbonate substrates were weighed before the chemical analysis. As described by Pettersson et al.,18 deposit probes were used to study deposit formation. The temperatures of the two probes used for deposit sampling were controlled by air cooling. The probes were 2 m long and had a diameter of 38 mm. On each probe, two high alloyed steel rings (253 Ma) were placed approximately 10 cm from the tip. At the superheaters (position A in Figure 1), two different material temperatures were used: 435 °C, which corresponds to the material temperature during present operation, and 500 °C, which is a goal for future waste combustion. Deposits were also sampled after the economizer (position B in Figure 1) using a material temperature of 230 °C, roughly corresponding to the material temperatures of the economizer. In addition to the particle and deposit samples, other types of ash samples were collected from different locations in the boiler during the tests: bottom ash, cyclone ash, ash from the empty gas pass, ash from the textile filter, and recycled sand. After the experiments, the deposit rings, particle samples, and ash samples were dissolved in an acid solution before being analyzed by inductively coupled plasma optical emission spectroscopy (ICP-OES) or ICP mass spectrometry (ICP-MS) for ash-forming matter according to Swedish Standards SS028311 and SS028150-2, and trace elements were analyzed according to standard method ASTM D 3638. This method allows detection of small amounts of the elements analyzed except for Si, which has a relatively poor solubility in the acid used to dissolve the samples. The following elements were analyzed: Cl, S, Al, Ba, Ca, Cu, K, Mn, Na, P, Pb, Ti, Zn, and Si. Fe was not determined on the deposit probes because the analysis method included deposit ring material. Gaseous HCl and SO2 were measured in the flue gas after the economizer. An extracted flow of the flue gas was bubbled through a solution of 0.1 M NaOH (aq) and 0.3% H2O2 (aq). The solution was then analyzed to provide average concentrations over the 2-h sampling period. Continuous measurements of CO2, CO, and NOx were made by Fourier Transform infrared (FTIR) spectrometry. Laboratory Experiments. Small-scale experiments were carried out in a laboratory fluidized-bed reactor; see the schematic overview in Figure 2. The reactor was made of quartz glass with an inner diameter of 60 mm and a length of 1.2 m. The reactor was placed in an electrically heated oven with three individual temperature zones (up to 1100 °C). The fluidized bed rested on a porous plate in the middle of the glass reactor. The bed consisted of 180 g of cleaned quartz sand, where 90% of the particles had diameters between 0.1 and 0.3 mm. The inlet and outlet of the reactor were made of water-cooled metal flanges attached to the glass reactor at the bottom (inlet) and top (outlet). The fluidizing gas, a mixture of nitrogen and air, was controlled by mass flow regulators. The flue gas extracted at the top of the reactor was split between an online ICP-MS instrument and other gas analyzing equipment (O2, CO, CO2, and total hydrocarbons). Before entering the gas analyzing instruments, the flue gas passed through a heated ceramic filter (175 °C) and a condenser (3.5−4 °C). The flue gas diverted to the ICP-MS instrument did not pass through the heated filter; instead, the gas was cooled (5 °C), and a sample was drawn through a capillary to a nebulizer, driven by a constant flow of Ar. The nebulizer controlled the dosage of aerosol sample into the reactor. A controlled flow of Kr was used as an internal standard in the ICP-MS instrument. Single fuel pellets were dropped into the reactor from the top, after which the concentrations of alkali metals and Zn in the flue gas were monitored online. The waste pellets used were made out of MSW collected at the full-scale boiler, using the methods of sampling and preparation for chemical analysis for BFB boilers presented by Jones et al.20 The sample preparation provided a homogenized fuel-particle (1-
METHODS
The full-scale tests were carried out at a heat and power plant in the Swedish town of Borås. The plant is equipped with two parallel bubbling fluidized-bed (BFB) boilers of 20 MWth each, delivering superheated steam at 49 bar and 405 °C. The boilers are hereafter referred to as B1 and B2. The fuel consists of a mixture of 30% municipal solid waste (MSW) and 70% industrial waste (IW). The IW is made up of combustible fractions from industry sites and contains paper, wood, fabrics, plastics, building and demolition waste, old rubber tires, and so on, in different, undetermined mixtures. The IW does contains no hazardous waste and no, or almost no, food waste. A schematic of one boiler is presented in Figure 1.
Figure 1. Sketch of one of the two 20 MWth waste combustors used for the experiments. Dimensions = 12 × 4.05 × 4.95 m3. Flue gas velocity = 40000 m3n/h dry at 8−9% O2. Components: (1) combustion chamber, (2) fuel feed chute, (3) primary air, (4) secondary and tertiary air, (5) empty gas pass, (6) superheaters followed by an evaporator bundle, (7) cyclone (average particle size = 70 μm, efficiency = 60%), (8) economizer.
The bed temperature is normally kept at around 870 °C. If this temperature is exceeded, the bed will be cooled by means of flue gas recirculation mixed with primary air (labeled 3 in Figure 1). However, flue gas recirculation is mainly used to increase the flue gas residence time and reduce NOx emissions. Secondary and tertiary air are introduced over the bed at a constriction of the combustion chamber (4 in Figure 1). At the top of the combustion chamber, the flue gas temperature is normally between 900 and 950 °C. After the combustion chamber, the flue gas passes an empty gas pass (5) before reaching the superheaters (6), which are followed by an evaporator bundle. The flue gas temperature is around 350 °C prior to the economizer (8), in which the gas is cooled to around 150 °C. Finally, the flue gas passes a cleaning system before leaving the plant or being partially recycled into the boiler to regulate the bed temperature. The bottom ash and other noncombustible debris are continually drained from the bed through a chute in the center of the bottom air distributor. After being sieved, the fine fraction of the bed material is recycled to the boiler, whereas the coarse fraction leaves the plant as bottom ash. Fresh sand is regularly added to maintain the bed height and renew the bed material. Fly ashes are collected at the bottom of the empty gas pass and at the cyclone, as well as from the fabric filter in the flue gas cleaning system. Full-Scale Tests. In the full-scale experiments, a test case with a reduced bed temperature (RBT) was compared to a reference case (ref). For the RBT case, the bed temperature was reduced by the increased flow of recirculated flue gas through the bed, and some of the primary air was diverted to secondary and tertiary air. The measurements were carried out over two days, one day per bed temperature. The two boilers were operated under the experimental conditions for approximately three days before the measurements were started. 5000
dx.doi.org/10.1021/ef400836e | Energy Fuels 2013, 27, 4999−5007
Energy & Fuels
Article
merged data set were not evaluated, but the combined set of data for the liquid phase allowed a larger set of liquid-phase components to be considered at the same time. The merging of the different data sets for the molten salts was done in cooperation with the developers of the software Factsage. The fuel compositions determined in the study for all three cases (RBT, reference, and laboratory) were used as inputs for the calculations in addition to air to achieve an air-to-fuel ratio of λ = 0.7 for reducing conditions and λ = 1.4 for oxidizing conditions. In addition, a calculation excluding the main silicate-forming elements Mg, Fe, Al, and Si was performed to compare to the predictions when nonvolatile silicates were included. When the silicate-forming elements were excluded, the calculations were based on simple alkali salts and Pb, Zn, and Ca compounds.
■
RESULTS AND DISCUSSION The measurements of the two full-scale tests were made on separate days, one week apart, to minimize any memory effects in the bed. Fuel samples were extracted from the fuel feeder a few times daily during the measurements. The analysis results for the fuels are reported in Table 1, which also includes the Table 1. Fuel Analysis case component
Figure 2. Schematic of the laboratory setup.
moisture ash S Cl C H N Ob Al2O3 SiO2 Fe2O3 TiO2 MgO CaO Na2O K2O P2O5 PbO ZnO sum
mm) mix that was used for pellet production. The results from the ICP-MS are presented as relative mass-weighted concentrations, Cx,rel, calculated as
Cx ,rel =
1 Sx FKr m SKr Frg
(1)
where m is the mass of the fuel sample burned; Sx and SKr are the mass-weighted signals from the instrument for elements x and Kr, respectively; and FKr and Frg are the gas flows of Kr and flue gas, respectively, to the instrument. To provide quantitative results, Cx,rel should be multiplied by a constant that depends on the element being studied, among other factors. The values of such constants were not determined in this study, so that the results presented are qualitative values, proportional to the actual concentrations. Thermodynamic Equilibrium Calculations. Thermodynamic equilibrium calculations were performed to predict the speciation of the ash-forming elements as a function of temperature for the three different cases (RBT, reference, and laboratory) under both oxidizing and reducing conditions. Advanced thermodynamic modeling was performed using the software package Factsage, version 6.3.21 A tailormade thermodynamic database was compiled for the calculations. The data for the gas phase and the stoichiometric solid phases of elements C, H, O, N, S, Cl, Na, K, Zn, Pb, Ca, Mg, Fe, Al, Si, P, and Ti were taken from the FACT53 database in the Factsage software. It was assumed that N2 was the only stable nitrogen compound, as the formation of NOx compounds in biomass combustion is strongly dependent on kinetics and N speciation in the fuel. The thermodynamic data for the multicomponent liquid phase containing Na+, K+, Ca2+, Mg2+, Zn2+, Pb2+//SO42−, CO32−, Cl−, S2−, and O2− were developed recently, and more detailed descriptions of the liquid phase can be found in the works by Lindberg et al.22−25 and Robelin and Chartrand.26 A review of the present status of thermodynamic databases for ash components in biomass and waste combustion was published by Lindberg et al.23 Certain solid solutions and subsystems of the liquid phase also appear in the FTSalt and FTPulp databases in the Factsage software. The tailor-made database had the additional benefit compared to the standard data in the Factsage databases of merging several liquid-phase subsystems, such as Na+, K+, Ca2+, Mg2+, Zn2+, Pb2+//Cl− and Na+, K+//SO42−, CO32−, Cl−, S2−, O2−. Some of the binary and ternary subsystems in the
a
unitsa wt wt wt wt wt wt wt wt wt wt wt wt wt wt wt wt wt wt wt
% % % % % % % % % % % % % % % % % % %
(ar) (df) (df) (df) (df) (df) (df) (df) ash ash ash ash ash ash ash ash ash ash ash
reference
RBT
laboratory
33.1 17.9 0.2 0.47 46.2 6.1 1.1 28 11.61 41.83 3.75 2.14 2.69 20.32 4.82 2.96 2.69 0.07 0.37 93.26
35.7 19 0.35 0.49 46 6 1.1 27 10.94 50.67 7.37 2.20 2.71 19.88 5.75 2.28 1.69 0.07 0.39 103.94
29.2 16.2 0.26 0.53 45.2 5.8 1 31 9.33 38.30 3.62 2.37 2.56 20.73 4.74 2.31 1.27 0.05 0.37 85.64
ar, as received; df, dry fuel. bBy difference.
analysis of the MSW used to produce pellets for the laboratory tests. These compositions were used as inputs for the thermodynamic equilibrium calculations. Average operating data of the boiler during the measurements are provided in Table 2 and in Figures 3−8. From Figure 3, it is clear that the demand of keeping the flue gas at 850 °C for at least 2 s stated by the European Waste directive is fulfilled. The recirculated gas flow through the bed controlling the bed temperature was manually adjusted during the RBT case, while it was automatically controlled during the reference case. This explains the somewhat varying temperature of the bed during the RBT case. The results from the measurements of HCl and SO2 in the flue gas after the economizer are also included in Table 2. A higher concentration of HCl was detected in the flue gas during the RBT case. The increase in gaseous HCl might be related to 5001
dx.doi.org/10.1021/ef400836e | Energy Fuels 2013, 27, 4999−5007
Energy & Fuels
Article
Table 2. Operating Parameters during the Measurements, Including Measured SO2 and HCl Contentsa case RBT
reference
parameter
average
min
max
average
min
max
bed temperature (°C) O2 content (vol %, wet gas) CO content (mg/Nm3, dry gas, 11% O2) total air flow (Nm3/s) total flow of recycled flue gas (Nm3/s) HCl content (mg/Nm3, dry gas, 11% O2) SO2 content (mg/Nm3, dry gas, 11% O2)
724 6.1
691 5.9
761 6.3
876 7.3
865 7.0
888 7.7
12
2
60
6
4
9
7.1 4.2
6.8 4.0
7.4 4.5
8.3 4.4
7.8 3.6
8.8 5.0
660b
530b
111b
114b
Figure 5. Total gas flow and bed gas flow of the boiler for the RBT case.
a
Operating data are further presented in Figures 3−8. bHCl and SO2 analyses were performed by a wet chemical method and provide only average values for the samples.
Figure 6. Total gas flow and bed gas flow of the boiler for the reference case. Figure 3. Load and temperature of the boiler for the RBT case.
Figure 7. O2, NOx, and CO contents for the RBT case.
Figure 4. Load and temperature of the boiler for the reference case.
The plant was equipped with a semidry flue-gas cleaning system into which the feed rate of hydrated lime was regulated to keep certain emission limits on HCl and SO2. At most times, the hydrated lime feed rate was determined by the HCl concentration, because the HCl emission limit is stricter. Thus, a negative consequence of the reduced bed temperature was increased hydrated lime consumption.
less alkali being released from the bed, but it could possibly also be an effect of temperature-dependent chemical reactions involving S capture in the bed by Ca, which is abundant in the ash. If less S is captured by Ca in the bed, more sulfur will be available for other reactions, for example, with alkali metals, resulting in reduced formation of alkali chlorides. 5002
dx.doi.org/10.1021/ef400836e | Energy Fuels 2013, 27, 4999−5007
Energy & Fuels
Article
the RBT case was not detemined, but it could be related to lower agglomeration tendencies of the bed material at a reduced bed temperature. In Figure 9, two size fractions are indicated: “fine” and “coarse”. These samples were chemically analyzed and are presented in Figure 10 in mole percentages. Because O and C were not determined in the chemical analysis and it is plausible that they were present in the particle samples due to oxide and carbonate formation, a perfect balance between cations and anions was not obtained. As expected from the chemical analysis, higher concentrations of alkali metals and Zn were detected in the fine particles, predominantly formed by condensation in the gas phase. The coarser particle fraction contained more Al, Ca, and Mg. The same increase was seen for the Pb content. Furthermore, the fine fraction in the RBT case contained a slightly higher concentration of Cl than that in the reference case. However, the total concentration of fine particles in the flue gas decreased, which means that the total concentration of particle-bound Cl in the flue gas decreased. Using an average for the two reference cases, the total decrease in particle-bound Cl was approximately 20% when the boiler was run at RBT. Deposit Probe Measurements. The two deposit probes with different material temperatures were simultaneously inserted approximately 1 m from the inside wall at location A in Figure 1, through two separate ports in the same manhole door of boiler B1. The ports were separated horizontally by about 0.5 m (P1 and P2, respectively, in Figure 1). Rather unexpectedly, the results showed a significant difference between the ports: many more deposits formed on the probe at point P2. This is probably a consequence of an uneven flow pattern right after the flue gas turned upward through the third pass (see Figure 1). All deposit probe exposures were made for 2 h, and no soot blowing was carried out during the tests. Photographs of the windward side of the deposit probes and the amounts of mass gained by the rings are presented in Table 3. The amounts of mass gained were recalculated into deposit growth rates, as shown in Figure 11 and Table 4, in which the chemical compositions are also presented. The bars are balanced to the total weight with “others”, which probably mostly consist of oxygen bound as oxides. It is notable that Cl is lower in all RBT cases for comparable temperatures and
Figure 8. O2, NOx, and CO contents for the reference case.
Particle Samples. Particles were sampled with a watercooled probe in port P3 in Figure 1. This sampling was performed simultaneously with the collection of the deposits. A comparison of the particle size distributions showed that the concentration of submicrometer particles (1 μm) increased (see Figure 9). The cause of the increase in coarser particles in
Figure 9. Particle size distributions measured using a DLPI upstream of the superheaters.
Figure 10. Molar compositions of the particle samples, where others corresponds to small amounts of P, Al, Mn, Ti, Ba, Cu, Pb, Mg, and Fe. 5003
dx.doi.org/10.1021/ef400836e | Energy Fuels 2013, 27, 4999−5007
Energy & Fuels
Article
Table 3. Photographs and Amounts of Weight Gained for Sample Rings after Exposure at the Superheaters at Points P1 and P2 in Figure 1
positions. On average, the deposit growth decreased 20% when the bed temperature was reduced. The growth rates and corresponding chemical analyses of the deposits sampled at the economizer are shown in Figure 12 and listed in Table 4. These samples were collected in B2 because of some operational problems with B1 at the time for this part of the experiment. This should not affect the results because the two boilers were operated similarly. In the deposits, the alkali content was reduced by approximately 25%, whereas the S and the Ca contents increased. In addition, the Zn content also decreased by approximately 25% compared to the reference case. Generally, the S/Cl-ratios of the deposits increased when the boiler was operating with lower bed temperature. An increased ratio indicates improved deposit quality, although the deposits can still be considered corrosive because of the presence of Cl. Bottom Ash, Cyclone Ash, Filter Ash, and Return Sand. A visual inspection of the bottom ashes (Figure 13) shows that the reference-case bottom ash contained many small agglomerates and pieces of glass that had melted on the surface,
Figure 11. Deposit growth at superheaters in terms of the elements analyzed, with the different modes of boiler operation compared in pairs by position and material temperature. Others correspond mainly to the difference between the analysis result and the weight of the collected deposit and partly to P, Al, Mn, Ti, Ba, Cu, and Pb. Numberical values are presented in Table 4.
Table 4. Deposit Growth (g m−2 h−1) for the Elements Analyzed on Probes for the Superheater Area and Economizer Cl S P Zn K Na Ca Si Al Mn Ti Ba Cu Pb othersa sum a
RBT 435 A
ref 435 A
RBT 500 A
ref 500 A
RBT 435 B
ref 435 B
RBT 500 B
ref 500 B
RBT 230 eco
ref 230 eco
0.42 0.46 0.01 0.04 0.25 0.31 0.56 0.13 0.07 0.01 0.05 0.01 0.01 0.09 2.56 4.97
0.98 0.84 0.04 0.09 0.53 0.50 1.34 0.42 0.31 0.02 0.10 0.02 0.08 0.08 5.39 10.72
0.56 0.73 0.02 0.05 0.42 0.39 0.73 0.20 0.10 0.02 0.05 0.01 0.07 0.02 3.70 7.06
1.42 0.68 0.02 0.10 0.39 0.36 1.14 0.27 0.17 0.04 0.06 0.02 0.06 0.04 5.71 10.50
2.79 1.02 0.04 0.14 0.95 0.70 1.12 0.42 0.19 0.06 0.09 0.02 0.45 0.22 8.57 16.78
3.63 0.73 0.04 0.19 1.03 0.73 1.37 0.45 0.26 0.07 0.10 0.02 0.28 0.45 9.22 18.57
1.51 0.84 0.03 0.12 0.59 0.47 1.51 0.39 0.21 0.04 0.10 0.02 0.12 0.04 7.30 13.29
2.37 0.76 0.02 0.12 0.67 0.50 1.34 0.36 0.20 0.07 0.07 0.02 0.20 0.07 9.53 16.31
1.06 0.15 0.02 0.05 0.27 0.23 0.56 0.20 0.12 0.01 0.06 0.01 0.06 0.08 1.70 4.58
1.54 0.15 0.02 0.08 0.50 0.36 0.56 0.17 0.11 0.01 0.06 0.01 0.09 0.11 2.69 6.45
Others corresponds to unidentified material (difference between analysis results and weighed deposit). 5004
dx.doi.org/10.1021/ef400836e | Energy Fuels 2013, 27, 4999−5007
Energy & Fuels
Article
bed temperature decreases the sintering tendencies of the bed. Another observation considering a longer perspective is that the sand consumption of the plant could be reduced by roughly 25% when running the boiler under RBT conditions. Laboratory-Scale Experiments. Experiments with single waste pellets in the laboratory reactor of Figure 2 were performed to study the effect of varied reactor temperature on the release of alkali metals to the flue gas. During these tests, the reactor was operated with an atmosphere of 5% O2 in nitrogen. As an example of results from these tests, massweighted signals of Na + K and Zn measured by the ICP-MS instrument are shown in Figure 14. The horizontal axis shows the time from the start of each experiment, when a single waste pellet was dropped into the reactor. The vertical axis shows a mass-weighted value of the signal from the ICP-MS instrument, scaled to the signal for the internal standard (here, Kr) and to the weight of each pellet (on the order of 0.8 g) as given by eq 1. The lines illustrate the detected concentrations for three different test runs at different reactor temperatures. The initial peak, lasting about 1 min, corresponds to the devolatilization of the fuel, after which the char combustion occurs over a longer period. Alkali metals are released during both these stages of fuel conversion. A comparison of the tests at 850 and 650 °C clearly illustrates that the amount of alkali metals released to the flue gas depends on the reactor temperature. In the case of Zn, there is a clear difference between the 850 and 650 °C, with a major decrease in the release of Zn. In contrast, for 750 °C, the release of Zn continued over a longer time period and with more mass than in either of the two other cases. It should be mentioned, however, that the laboratory setup is still under development to improve the repeatability and to provide quantitative results. Nevertheless, the results of these experiments support the full-scale results, proposing that a lower bed temperature reduces emissions of alkali metals. Thermodynamic Calculations. Calculations were performed for each of the three fuel compositions presented in Table 1. No major differences between the three different fuels were noted, as the behavioral trends were similar, and therefore, only one fuel is presented, namely, the fuel sampled for the RBT case. For oxidizing conditions (Figure 15a), the calculations predicted that the volatilization of alkali chlorides increases with increasing temperature, confirming the hypothesis that a lower bed temperature would result in less release of potentially corrosive elements. At the same time, as the alkali chlorides
Figure 12. Deposit growth at the economizer in terms of the elements analyzed. Others correspond mainly to the difference between the analysis results and the weight of the collected deposit and partly to P, Al, Mn, Ti, Ba, Cu, and Pb. Numerical values are presented in Table 4.
Figure 13. Photographs of bed ash for the two different modes of operation of the boiler: (left) RBT case, (right) reference case with agglomeration.
allowing smaller sand particles to stick to it. The bottom ash from the RBT case did not contain agglomerates, and the glass fragments had no sand particles stuck to them. Obviously, the glass did not become sticky under RBT conditions. Chemical analyses of the ashes showed that the Cl concentration increased in the bottom ash and the recycled sand during the RBT case, whereas it was unchanged in the ash from the empty gas pass and decreased in the ash collected both in the cyclone and in the textile filter. Scanning electron microscopy energy-dispersive X-ray (SEMEDX) fluorescence spectrometry evaluation by Pettersson et al.18 confirmed an increase in Cl coating on the outside of the particles in both the bottom ash and the return sand of the RBT case. An enrichment of alkali chlorides in the bed could increase the risk of bed agglomeration, but on the other hand, the low
Figure 14. Cumulative measured masses of (left) alkali metals (Na + K) and (right) Zn released. Both graphs correspond to waste combustion at 5% O2 and different temperatures. 5005
dx.doi.org/10.1021/ef400836e | Energy Fuels 2013, 27, 4999−5007
Energy & Fuels
Article
Figure 15. Predicted concentrations of selected gaseous compounds under (a) oxidizing and (b) reducing conditions with all elements included in the calculation. All compounds correspond to the primary y axis on the left, except for the Zn compounds, which correspond to the secondary y axis on the right. Under reducing conditions (panel b), Zn reaches a stable level of just over 40 ppm at 800 °C (above top axis).
Figure 16. Predicted concentrations of selected gaseous compounds under (a) oxidizing and (b) reducing conditions for the simplified calculation. All compounds correspond to the primary y axis on the left, except for the Zn and Pb compounds, which correspond to the y axis on the right. Under reducing conditions (panel b), Zn reaches a stable level of just over 40 ppm at 780 °C (above top axis).
decreased with reduced temperatures, the HCl formation increased, suggesting that more chlorine formed HCl when less alkali was available from the fuel. The formation of gaseous ZnCl2 took place at much lower temperatures and increased, with a maximum release peaking around 800 °C and then decreasing at higher temperatures. At temperatures above 1000 °C, gaseous Zn was the main volatile Zn species. Above 800 °C, the formation of SO2 increased significantly, and SO2 became the dominant gaseous species close to 1000 °C whereas HCl decreased as a result of the formation of other chlorides, such as alkali chlorides. For the same fuel mixture under reducing conditions, an even higher concentration of alkali chlorides formed around 850 °C (see Figure 15b). Under these reducing conditions, no SO2 formed, and the ZnCl2 behavior was similar to that observed under oxidizing conditions (concentration too low to be seen in the figure). Significant formation of gaseous Zn occurred at 700 °C, with a sharp increase up to 800 °C and then stabilization, showing a steady level at higher temperatures. The trends from the calculations suggest that reducing conditions give rise to the release of corrosive elements in higher concentrations than oxidizing conditions. They also show that a temperature drop from 850 to 650 °C significantly reduces the formation of alkali chlorides in favor of the formation of HCl.
The simplified calculations excluded Si, which thus prevented formation of alkali silicates. This affected the calculations in that alkali was available for reaction with chlorine, inhibiting the domination of gaseous HCl. This is visible in Figure 16a (note the difference in scale of the secondary y axis compared to that in Figure 15), where the HCl concentration stays rather low throughout the temperature interval. However, at lower temperatures, the alkali chlorides were in the solid state, and gaseous alkali chlorides increased with increasing temperature, as in the calculations with all elements, but formed at lower temperatures and in higher concentrations. In addition to the alkali chlorides, alkali hydroxides formed in noticeable amounts just under 900 °C and increased with temperature, which was also the behavior of gaseous Zn. The calculations also indicated formation of PbO at 500 °C and stabilization from approximately 770 °C. In a reducing environment (Figure16b), b), the increased formation of gaseous Zn and Na was significant above 900 °C. Around 1100 °C, the formation of alkali hydroxides sharply decreased, and SO2 formed. Pb and PbS formed in low concentrations and fluctuated with respect to each other from 500 °C and on. As in the calculations with all elements, the simplified calculations showed that a reducing environment gives rise to a higher concentration of potentially corrosive substances. 5006
dx.doi.org/10.1021/ef400836e | Energy Fuels 2013, 27, 4999−5007
Energy & Fuels
■
Article
(13) Åmand, L.-E.; Leckner, B.; Eskilsson, D.; Tullin, C. Energy Fuels 2006, 20, 1001−1007. (14) Backman, R.; Hupa, M.; Hiltunen, M.; Peltola, K. Proceeding of the 18th International American Society of Mechanical Engineers (ASME) Conference on Fluidized Bed Combustion; Toronto, Ontario, Canada, May 22−25, 2005. (15) Enestam, S. Ph.D. Thesis, Åbo Akademi University, Turku, Finland, 2011. (16) Bankiewicz, D.; Enestam, S.; Yrjas, P.; Hupa, M. Fuel Process. Technol. 2013, 105, 89−97. (17) Elled, A. L.; Åmand, L.-E.; Eskilsson, D. Energy Fuels 2008, 22, 1519−1526. (18) Pettersson, A.; Niklasson, F.; Moradian, F. Fuel Process. Technol. 2013, 105, 28−36. (19) Johansson, L.; Leckner, B.; Tullin, C.; Åmand, L.-E.; Davidsson, K. Energy Fuels 2008, 22, 3005−3015. (20) Jones, F. C.; Blomqvist, E. W.; Bisaillon, D.; Lindberg, D. K.; Hupa, M. Waste Manage. Res., published online Jun 9, 2013, 10.1177/ 0734242X13490985. (21) Bale, C. W.; Bélisle, E.; Chartrand, P.; Decterov, S. A.; Eriksson, G.; Hack, K.; Jung, I.-H.; Kang, Y.-B.; Melançon, J.; Pelton, A. D.; Robelin, C.; Petersen, S. CALPHAD: Comput. Coupling Phase Diagrams Thermochem. 2009, 33, 295−311. (22) Lindberg, D.; Backman, R.; Chartrand, P.; Hupa, M. Fuel Process. Technol. 2013, 105, 129−141. (23) Lindberg, D.; Backman, R.; Chartrand, P. J. Chem. Thermodyn. 2007, 39, 1001−1021. (24) Lindberg, D.; Backman, R.; Chartrand, P. J. Chem. Thermodyn. 2007, 39, 942−960. (25) Lindberg, D. Ph.D. Thesis, Åbo Akademi University, Turku, Finland, 2007. (26) Robelin, C.; Chartrand, P. J. Chem. Thermodyn. 2011, 43, 377− 391.
CONCLUSIONS In this work, full-scale experiments showed a reduction in the rate of deposit growth of 20%, on average, when the boiler was operated at reduced bed temperature. There was also a decrease of Cl in the deposits during this mode of operation. In addition, the total concentration of fine particles in the flue gas decreased, which means that the total concentration of particle-bound Cl in the flue gas decreased. The bottom ash and the recycled sand from the reduced-bedtemperature (RBT) case contained fewer agglomerates and increased concentrations of Cl. Meanwhile, the Cl content was unchanged in the ash from the empty gas pass and reduced in the textile filter ash. For the RBT case, a higher concentration of HCl was detected in the flue gas, whereas the SO2 concentration remained almost unchanged. These findings can be explained by the lowered vapor pressure of alkali metals, which decreased the total amount of alkali available for reaction in the gas phase, resulting in the formation of HCl. Additional laboratory-scale tests with bed temperatures of 850 and 650 °C clearly indicated that the amount of alkali metals released to the flue gas depends on the reactor temperature. Furthermore, thermodynamic equilibrium calculations supported the findings of the experiments.
■
AUTHOR INFORMATION
Corresponding Author
*E-mail:
[email protected]. Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS The authors express gratitude toward Borås Energi och Miljö, Dalkia, and Metso Power for their assistance during the experimental work and kindly acknowledge financial support from Waste Refinery and Värmeforsk. This project was performed in cooperation with the University of Borås.
■
REFERENCES
(1) Bogner, J.; Spokas, K. Chemosphere 1993, 26, 369−386. (2) Zhang, H.; He, P.; Shao, L. Atmos. Environ. 2008, 42, 5579−558. (3) IPCC Fourth Assessment Report: Climate Change 2007; Intergovernmental Panel on Climate Change (IPCC): Geneva, Switzerland, 2007. (4) Nielsen, H. P.; Frandsen, F. J.; Dam-Johansen, K.; Baxter, L. L. Prog. Energy Combust. Sci. 2000, 26, 283−298. (5) Knudsen, J. N. Ph.D. Thesis, Technical University of Denmark, Lyngby, Denmark, 2004. (6) Lewis, R. J., Sr. Sax’s Dangerous Properties of Industrial Materials, 9th ed.; Van Nostrand-Reinhold: New York, 1996; Vol. III. (7) Zevenhoven, M.; Yrjas, P.; Skrifvars, B.-J.; Hupa, M. Energy Fuels 2012, 26, 6366−6386. (8) Davidsson, K. Ph.D. Thesis, Gothenburg University, Gothenburg, Sweden, 2002. (9) Robelin, C.; Chartrand, P. J. Chem. Thermodyn. 2011, 43, 377− 391. (10) Bøjer, M.; Arendt Jensen, P.; Frandsen, F. J.; Dam-Johansen, K.; Hedegaard Madsen, O.; Lundtorp, K. Fuel Process. Technol. 2008, 89, 528−539. (11) Niemi, J.; Enestam, S.; Mäkelä, K. In Proceedings of the 19th International Conference on Fluidized Bed Combustion; Winter, F., Ed.; Springer: Dordrecht, The Netherlands, 2006; Part II. (12) Verhulst, D.; Buekens, A.; Spencer, P. J.; Eriksson, G. Environ. Sci. Technol. 1996, 30, 50−56. 5007
dx.doi.org/10.1021/ef400836e | Energy Fuels 2013, 27, 4999−5007