Environ. Sci. Technol. 2006, 40, 584-589
Electrostatic Precipitator Performance and Trace Element Emissions from Two Kraft Recovery Boilers TERTTALIISA LIND,† JOUNI HOKKINEN,† J O R M A K . J O K I N I E M I , * ,†,# RISTO HILLAMO,‡ ULLA MAKKONEN,‡ ANTTI RAUKOLA,§ JAAKKO RINTANEN,⊥ AND KARI SAVIHARJU¶ VTT Processes, P.O. Box 1602, FIN-02044 VTT, Finland, Finnish Meteorological Institute, P.O. Box 503, FIN-00101 Helsinki, Finland, Kvaerner Power Oy, P.O. Box 109 FIN-33101 Tampere, Finland, Alstom Finland, P.O. Box 92, FIN-01721 Vantaa, Finland, and Andritz, P.O. Box 500, FIN-48601 Kotka, Finland
Fine particle emissions from combustion sources have gained attention recently due to their adverse effects on human health. The emission depends on the combustion process, fuel, and particulate removal technology. Particle concentrations at Kraft recovery boiler exits are very high, and the boilers are typically equipped with electrostatic precipitators (ESP). However, little data are available on the ESP performance in recovery boilers. Particle concentrations and size distributions were determined at two modern, operating recovery boilers. In addition, we determined the fractional collection efficiency of the ESPs by simultaneous measurements at the ESP inlet and outlet and the particulate emissions of trace metals. The particle mass concentration at the ESP inlet was 11-24 g/Nm3 at the two boilers. Particle emissions were 30-40 mg/ Nm3 at boiler A and 12-15 mg/Nm3 at boiler B. The particle size distributions had a major particle mode at around 1 µm. These fume particles contained most of the particle mass. The main components in the particles were sodium and sulfate with minor amounts of chloride, potassium, and presumably some carbonate. The ESP collection efficiency was 99.6-99.8% at boiler A and 99.9% at boiler B. The particle penetration through the ESP was below 0.6% in the entire fume particle size range of 0.3-3 µm. Trace element emissions from both boilers were well below the limit values set by EU directive for waste incineration.
Introduction Fine particle formation and emissions from combustion processes have gained increasing attention lately due to several investigations that have addressed adverse health * Corresponding author phone: +358 20 722 6158; fax: +358 20 722 7021; e-mail:
[email protected]. † VTT Processes. ‡ Finnish Meteorological Institute. § Kvaerner Power Oy. ⊥ Alstom Finland. ¶ Andritz. # University of Kuopio, Kuopio, Finland. 584
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 40, NO. 2, 2006
effects of the fine particles (1). The particle concentrations in the flue gases of Kraft recovery boilers are very high, and average particle size is typically one micrometer (1 µm, aerodynamic diameter). Therefore, potentially, fine particle emissions may be significant. In Kraft recovery boilers, black liquor, which is a waste sludge from paper pulping process, is burned to recycle the pulping chemicals, and the heat from combustion of the organic matter is used for steam and electricity production. The main pulping chemicals are NaOH and Na2S. Consequently, the particles that are formed during combustion are typically rich in sodium and sulfur in the form of sodium sulfates, unlike, for example, ash particles from coal combustion which contain large amounts of, e.g., silicates and metal oxides. In the year 1997, 72 million tons of oil equivalent (mtoe) of black liquor was used for energy production globally (2). Fume particles make up most of the particle mass formed during black liquor combustion. Fume particles are approximately 1 µm in diameter. They are formed by vapor condensation processes from gas-phase species that are released during char burning and smelt oxidation (3-6). Fume particle size, concentrations, and other characteristics have been determined in both laboratory and field measurements (4, 7). The rest of the particle mass concentration is made up by coarse particles that can be divided into two particle modes: intermediate and carry-over particles. Intermediatesized particles have been observed in the size range of 10100 µm (7), and they are formed by ejection from liquor droplets during devolatilization and char burning, as well as by agglomeration. The coarse carry-over particles are liquor droplets and smelt, or their fragments, in various stages of burn-out. Carry-over particle formation and deposition mechanisms have an extensive literature (e.g., 8). Particle mass concentrations at the recovery boiler are very high, typically 30-60 g/Nm3 at economizer exits (9). Because of high particulate loading, all recovery boilers are equipped with particulate removal devices, most commonly electrostatic precipitators (ESP). However, few investigations have addressed electrostatic precipitator performance in recovery boilers (10). In coal and biomass combustion, ESP collection efficiencies are very high, many times 99.9% and more. However, they typically have lower collection efficiency in the particle size range of 0.1-1 µm (particle diameter). In coal combustion, even 10% of the particles in this size range may penetrate the ESP (11, 12), and in fluidized bed combustion of biomass values as high as 60% have been reported for penetration in this size range (13). In coal and biomass combustion, the fly ash particle size distributions are typically bimodal, with a fine particle fraction in the size range of 0.1-1 µm and a coarse particle mode in the size range of 1-100 µm. Most of the particle mass is in the coarse particle mode, mostly in particles larger than 10 µm. Since the particle size distributions from coal and biomass combustion are distinctly different from those from recovery boilers where most of the particle mass is in fume particles, it can be expected that ESP performance may also differ. In this investigation, we determined particle characteristics and concentrations at two modern, operating recovery boilers. We determined the fractional collection efficiency of the ESPs and the particle emissions of those trace metals that are regulated by EU directive for waste incineration. Mercury was not included in this investigation because of its occurrence partly in the gas-phase compounds at ESP outlet 10.1021/es0503027 CCC: $33.50
2006 American Chemical Society Published on Web 12/13/2005
TABLE 1. Gas Compositions and Process Values during Different Measurement Days at Boilers A and B boiler A
furnace temperature [°C] black liquor flow [m3/h] O2 [%] NOx [mg/Nm3] SO2 [mg/Nm3]
TABLE 2. As-Fired Black Liquor Analysis Results from Boilers A and B
boiler B
day 1 70% load
day 2 70% load, NCGs off
day 3 100% load
day 1 100% load
day 2 70% load
970
1010
960
1040
1050
59.3 4.1 200 2.0
65.0 4.2 200 2.0
74.8 3.7 230 2.5
73.2 3.5 170 3.5
55.0 4.2 150 3.0
conditions. Other trace metals are typically not found as gasphase compounds at the temperatures of the ESPs.
Methods Fine particle measurements were carried out at two Kraft recovery boilers to determine fine particle mass concentrations, size distributions, ESP collection efficiency, and trace element emissions. The measurements were conducted at the ESP inlet and outlet. Both boilers were modern, relatively new boilers. Boiler A had a capacity of 2000 tons of dry solids/day. Noncondensable gases (NCG) are combusted in the boiler along with black liquor. The measurements were carried out during normal operation, which was approximately 70% load at the time of our measurements (day 1), during 70% load when NCGs were not fed into the boiler (day 2), and during full 100% load (day 3). The furnace temperature as determined by optical pyrometers in the lower furnace, black liquor feed, and gas composition values for different days at both boilers are given in Table 1. CO concentration varied at both boilers between 0 and 500 mg/Nm3, but it was generally higher at boiler A than at boiler B. Boiler B had a capacity of 1800 tons of dry solids/day. NCGs are not combusted in the recovery boiler. The measurements were carried out during normal full capacity load (day 1) and during 70% load (day 2). As the ESP inlet was not accessible for our measurement devices, particle measurements were carried out at economizer 1 at the flue gas temperature of 190 °C. ESP outlet measurements were carried out at stack. The specific collection area of the ESP at boiler A was 10% smaller than that at boiler B. As-fired black liquor samples and ash collected by the ESP were collected every 2 h during measurements. Samples were then combined into one sample per day. Black liquor samples were analyzed for dry solids, inorganic ash compounds, and calorimetric value according to standard analysis methods at KCL, Finland. Ash samples from the ESP were analyzed for water-soluble compounds with ion chromatography (IC). The particle mass size distributions and mass concentrations were determined with a Berner-type low-pressure impactor (BLPI) from the electrostatic precipitator (ESP) inlet and outlet (14). A cyclone was used before the BLPI to collect particles larger than 4 µm (aerodynamic diameter). The precutter cyclone was placed inside the duct, and the sample coming out of the cyclone was diluted with a porous tube diluter. The low-pressure impactor sampling method and uncertainties are described in detail in ref 14. In this investigation, mass concentration uncertainty due to the sampling method (14) and sample extraction (15) was approximately 5% at the ESP inlet measurements for the fume particles. At the ESP outlet measurements, the uncertainty was 20% at boiler A and 30% at boiler B. In addition to the impactor measurements, total particle emissions were
boiler A
dry solids [%] in dry matter: ash [%] Na [%] K [%] S [%] Cl [%] SO4 [%] CO3 [%] NH4 [mg/kg] Cd [mg/kg] Pb [mg/kg] calorimetric value [MJ/kg] nNa/nSO4 nK/nNa nCl/nK n(2 × SO4)/n(K + Na)
boiler B
day 1
day 2
day 3
day 1
day 2
74.9
75.0
75.3
77.4
77.3
53 19 3.3 4.5 0.4 6.6 5.5 16 0.5 1.5 12.7 12.0 0.10 0.13 0.15
50 19 3.2 4.2 0.5 6.5 5.6 6.0 0.6 1.7 12.4 12.2 0.10 0.17 0.15
50 19 3.4 4.6 0.5 5.6 5.0 6.0 0.5 1.6 13.1 14.2 0.11 0.16 0.13
52 20 2.6 5.5 0.6 4.8 5.9 5.0 0.4 1.2 12.7 17.4 0.08 0.25 0.11
53 20 2.7 5.5 0.6 5.4 6.0 6.0 0.8 2.2 13.0 15.5 0.08 0.25 0.12
measured in-stack using filter sampling according to standard method SFS3866. No gas-phase compounds were measured. The major uncertainty in the recovery boiler measurements typically arises from the unstable combustion conditions in the furnace. This is reflected in uneven flow patterns and temperatures in the furnace and flue gas channels. This causes changes in the pressure and velocity of the flue gas as well as particle concentrations. In this investigation, the measurement times and sites were chosen to minimize uncertainties caused by unstable combustion. Soot-blowing was switched off in the last sections of the boiler during the ESP inlet measurements to avoid uncertainties caused by collection of re-entrained coarse particles. Elemental and water-soluble ion size distributions were determined by analyzing the BLPI and cyclone-collected samples with ion chromatography (IC) and inductively coupled plasma mass spectroscopy (ICP-MS). The collection efficiency of the ESP was studied by simultaneous BLPI measurements from the ESP inlet and outlet. Particle samples for morphology analysis with scanning electron microscopy (SEM) were collected with a thermophoretic sampler at the ESP inlet (4). The samples were collected on copper grids that were coated with a thin carbon film. No sample preparation was used prior to SEM analysis.
Results Black Liquor and ESP Ash Analysis. The dry solids content in black liquors from the two boilers A and B were 75% and 77%, respectively, Table 2. The liquor composition was very stable and did not vary much from day to day. Sodium content in the as-fired liquor was approximately 20%, and chlorine content was 0.5% in the dry solids in both boilers. Potassium and sulfate contents were slightly higher, and total sulfur content slightly lower, at boiler A than at boiler B. The sulfateto-alkali ratio (2 × SO4)/(Na + K) was 0.15 at boiler A and 0.12 at boiler B. This ratio indicates how large a fraction of alkalis sodium and potassium may react with sulfur to form sulfates during combustion. If this ratio is 1, all alkalis and sulfur in the black liquor may form alkali sulfates, and if it is
