338
Energy & Fuels 2008, 22, 338–342
Pilot Plant Testing of Elemental Mercury Reemission from a Wet Scrubber John C. S. Chang* U.S. EnVironmental Protection Agency, National Risk Management Research Laboratory, Air Pollution PreVention and Control DiVision (MD-E305-03), Research Triangle Park, North Carolina 27711
Yongxin Zhao ARCADIS G & M, Inc., 4915 Prospectus Dr., Suite F, Durham, North Carolina 27713 ReceiVed June 22, 2007. ReVised Manuscript ReceiVed September 7, 2007
This paper is to discuss the recent observations of elemental mercury (Hg0) reemissions from a pilot-scale limestone wet scrubber. Simulated flue gas was generated by burning natural gas in a down-fired furnace and doped with 2000 ppm of sulfur dioxide (SO2). Mercuric chloride (HgCl2) solution was delivered to the scrubber at a controlled rate to simulate the absorption of ionized mercury (Hg2+). Testing results have shown that, after Hg2+ was injected, elevated Hg0 concentrations were soon detected both in the scrubber effluent flue gas and the hold tank air, which reflected the occurrence of Hg0 reemissions in both places. When the HgCl2 feed was stopped, the Hg0 reemission continued for more than 2 h. In addition, a significant Hg0 reemission was also detected outside the scrubber loop. In an attempt to understand the Hg0 reemission increase across the wet scrubber system under transient and steady states and to understand the underlying relationship with the mercury complexes retained in the wet scrubber system, a mercury reemission model was developed. With this model, it was found that the Hg0 reemission rate under the current testing conditions can be simulated by a first-order reaction, and only a portion of Hg · S(IV) complexes retained in the slurry were participating in the reemission reaction.
Introduction Wet flue gas desulfurization (FGD) scrubbers can provide the co-benefit of mercury removal by absorbing ionized mercury (Hg2+) in the flue gas generated from coal-fired power plants.1 This co-benefit was included in the newly promulgated Clean Air Mercury Rule (CAMR) by the U.S. Environmental Protection Agency (EPA) to reduce mercury emissions from new and existing coal-fired electric-utility steam-generating units.2 CAMR will be implemented in two phases with a cap-and-trade program. Initially, a national annual cap of 38 tons is to be reached by 2010, with emission reductions coming primarily as a co-benefit of technologies that control other air pollutants (e.g., wet-FGD scrubbers). The second phase sets a national annual cap of 15 tons by 2018. The co-benefit will also play an important role in achieving the secondphase mercury reduction goal. However, it is known that a portion of the Hg2+ absorbed in the wet scrubbers can be converted back to elemental mercury (Hg0) and reemitted.1,3 The Hg0 reemission results in an increase of flue gas Hg0 concentration across the scrubber by as much as 40% and significantly reduces the co-benefit * Corresponding author: e-mail
[email protected]; Tel (919) 5413747; Fax (919) 541-2157. (1) Kilgroe, J. D.; Sedman, C. B.; Srivastava, R. K.; Ryan, J. V.; Lee, C. W.; Thorneloe, S. A. Control of Mercury Emissions from Coal-Fired Electric Utility Boilers: Interim Report, EPA-600/R-01-109, 2001. (2) Standards of Performance for New and Existing Stationary Sources: Electric Utility Steam Generating Units; Final Rule, 40 CFR Parts 60, 63, 72 and 75 (OAR-2002-0056; FRL), http://www.epa.gov/air/mercuryrule/ pdfs/camr_final_preamble.pdf. Accessed on March 15, 2005. (3) Brown, T. D.; Smith, D. N.; Hargis, R. A., Jr.; O’Dowd, J. J. Air Waste Manage. Assoc. 1999, 49, 1–97.
of wet scrubber mercury removal.4 The Hg0 reemission can also have adverse financial impact on power plant operation under the CAMR cap-and-trade program. Chang and Ghorishi5 investigated Hg0 reemissions using a bench-scale wet scrubber simulator and sodium sulfite and bisulfite solutions under the worst-case conditions (no oxygen in gas and no solid precipitation in solution). The experimental results indicated that the absorbed Hg2+ can be reduced by aqueous S(IV) (sulfite and/or bisulfite) species and results in Hg0 reemission under simulated wet scrubber conditions. The S(IV)-induced Hg2+ reduction and Hg0 reemission mechanism was described by a model that assumes that only a fraction of the Hg2+ can be reduced, and the rate-controlling step of the overall process is a first-order reaction involving Hg · S(IV) complexes formed by the absorbed Hg2+ and the S(IV) species. The model predicted that as much as 58% of the Hg2+ absorbed can be reemitted as Hg0 in a wet scrubber system. In addition, the model predicted that the slurry in a wet scrubber may contain a significant amount of Hg · S(IV) complexes, which can function as a precursor to Hg0 reemission outside of the scrubber (e.g., in hold tank, clarifier, and vacuum filter). Hg0 reemission outside of the scrubber further reduces the co-benefit of wet scrubber mercury removal and may cause an occupational health hazard in coal-fired power plants. As a continued effort of pervious research,5 the objective of this investigation was to conduct pilot plant tests to evaluate (4) Srivastava, R. K.; Hutson, N.; Martin, B.; Princiotta. Control of Mercury Emissions from Coal-Fired Electric Utility Boilers. EnViron. Sci. Technol. 2006, 40, 1385–1393. (5) Chang, J. C. S.; Ghorishi, S. B. Simulation and Evaluation of Elemental Mercury Concentration Increase in Flue Gas Across a Wet Scrubber. EnViron. Sci. Technol. 2003, 37, 5763–5766.
10.1021/ef700355q CCC: $40.75 2008 American Chemical Society Published on Web 11/29/2007
Mercury Reemission from a Wet Scrubber
Energy & Fuels, Vol. 22, No. 1, 2008 339
Figure 2. Schematic of the Hg0 sampling train. Table 1. Typical Test Conditions of the Wet Scrubber
Figure 1. Schematic of the pilot-scale wet scrubber system.
the wet scrubber Hg0 reemissions under more realistic conditions, to validate the previous laboratory results, to evaluate the potential Hg0 reemission outside a wet scrubber, and to understand the relationship of Hg0 reemissions with the retained Hg species in the slurry. Experimental Methodology The pilot-scale scrubber system (Figure 1) consists of a scrubbing tower (i.e., three-stage turbulent contact absorber), a hold tank, a makeup tank, and a slurry pump. Simulated flue gas is generated from a down-fired cylindrical furnace, known as an innovative furnace reactor (IFR). The IFR is usually fired with natural gas and has a combustion capacity of 150 000 Btu/ h. The fuel is introduced to the top of furnace and combusted with air from axial and tangential directions. Since the simulated flue gas generated from natural gas contains no SO2 or mercury, pure SO2 from a gas cylinder was injected into the IFR to achieve the desired flue gas SO2 concentration. HgCl2 solution (i.e., a dilute aqueous solution of HgCl2) was used as the source of Hg2+. A peristaltic pump was used to deliver the HgCl2 solution to the middle stage of the TCA at a constant rate (e.g., 5 mL/min). The three-stage scrubbing tower is 10 cm in diameter. Each stage is 92 cm in length but contains only 20 cm deep plastic hollow balls (2 cm in diameter), which are supported by a grid at the bottom of each stage. The hollow balls, fluidized by the upward flue gas, facilitate slurry turbulence and promote intimate contact between flue gas and scrubbing slurry so as to improve the SO2 and Hg2+ removal efficiencies. The scrubbing slurry is delivered by a recirculation pump from the slurry hold tank to the spray nozzle located at the top of the TCA and travels downward through the spray tower back to the slurry hold tank. Prior to discharging into the exhaust duct, the flue gas passes through a chevron-type demister at the top of scrubber to remove the carryover of mist. The slurry hold tank, located below the scrubbing tower, is ∼200 L in volume but only holds ∼140 L of slurry during testing. The nominal slurry concentration is 5%. A heating pad, wrapped around the outside wall of the slurry hold tank, is used to bring the slurry to the desired temperature before each test. The slurry in the hold tank is agitated by a motor to keep the solids uniformly suspended. The pH of the slurry in the hold
experimental condition
parameter
flue gas flow rate (G), L/min inlet flue gas SO2 concentration, ppm flue gas O2 concentration, % flue gas CO2 concentration, % hold tank slurry volume (V), L hold tank pH slurry recirculation rate (L), L/min hold tank slurry temperature, °C Hg2+ feed rate (F), µg/min
850 2000 8-9 6-7 140 6.0 11.4 55 25
tank is controlled by the addition of calcium carbonate slurry from the limestone makeup tank. A feedback control loop is used to regulate the calcium carbonate feed rate to maintain pH in the hold tank at a desired level. The concentrations of SO2, O2, and CO2, etc., were continuously monitored at both the scrubber inlet and outlet using a CEM system. Continuous measurement of Hg0 was performed at the scrubber outlet with an ultraviolet (UV) spectrometer. To eliminate the possible adverse effect of SO2 on Hg measurements, a 20% NaCO3 solution was used to remove SO2 from the sampling stream as shown in Figure 2. After the SO2 is removed, the gas flowed through an ice bath and a NAFION gas dryer (Perma Pure Inc.) where moisture was removed. An SO2 analyzer was used to monitor the SO2 concentration in the sampling stream after the mercury analyzer to confirm the adequate removal of SO2 by the 20% Na2CO3 solution. Results and Discussion Hg0 Reemissions across the Scrubber. The first aim in this research is to confirm the Hg0 reemissions and their occurrence locations under more realistic wet scrubber operating conditions. The first expected Hg0 reemission location was the scrubbing tower, where Hg0 reemissions are generally observed in coalfired power plants. In order for a reliable conclusion to be drawn, the reemission of Hg0 from the wet scrubber needs to be below 1 µg/m3, making it possible to inject HgCl2 into the scrubber. This eliminates the effect of residual Hg2+ from previous tests as flue gas Hg0 concentration at the scrubber inlet was less than 0.5 µg/m3 when natural gas was burned in the furnace. The HgCl2 feed to the middle stage of the scrubber was initiated (time 0) after the major operating conditions (see Table 1) were stabilized. As demonstrated in Figure 3, where the measured flue gas Hg0 concentration at the scrubber exit vs time is plotted, flue gas Hg0 concentration across the scrubber started to increase a few minutes after time 0. Although the scrubber exit Hg0 concentration continued to increase as time proceeded, the rate of increase gradually slowed down and began to level off at ∼180 min. The gradual increase of Hg0 across the scrubber reflected the reduction of Hg2+ added to the slurry and the reemission as Hg0. In Figure 3, it can be concluded that Hg0 reemissions occurred in the scrubbing tower and that the cumulative Hg2+ in the
340 Energy & Fuels, Vol. 22, No. 1, 2008
Figure 3. Comparison of measured scrubber exit flue gas Hg0 concentrations with model predictions.
Figure 4. Linear relationship of Hg0 concentrations between scrubberexit flue gas and hold-tank air.
slurry—rather than the incoming Hg2+—functioned as the primary source of Hg0 reemissions. However, as the absorbed Hg2+ was mainly retained in the hold tank slurry, the question is: are Hg0 reemissions also occurring in the hold tank? Therefore, the Hg0 concentration above the slurry in the hold tank was also measured during the test by diverting Hg sampling to the hold tank, and the Hg0 concentrations which match up with the scrubber-exit Hg concentrations are plotted in Figure 4. Results showed that Hg0 reemissions not only occurred across the scrubbing tower but also in the tank bulk slurry. The regression in Figure 4 shows that a linear correlation exists between the hold-tank air and the scrubber-exit flue gas Hg0 concentrations. In other words, the hold tank Hg0 reemission rate was proportional to that in the scrubbing tower. The average ratio between hold tank air and scrubber-exit flue gas Hg0 concentration was ∼9.1. The hold tank was covered, but not sealed. Using a tracer gas technique, the hold-tank air exchange rate with the ambient air was estimated to be 24 L/min. When the Hg0 reemission rate was compared according to the occurrence locations, the Hg0 reemission in the hold tank was ∼25.7% of that in the scrubbing tower. Therefore, it is concluded that a significant amount of Hg0 reemission also occurred in the hold tank.
Chang and Zhao
Figure 5. Comparison of measured and model predicted residue mercury reemissions.
The cumulative Hg2+ in the hold tank served as the primary source of Hg0 reemissions. This was further demonstrated by the data shown in Figure 5. When the HgCl2 feed to the scrubber was terminated at the end of 240 min of the test, a mercury reemission test was continued for an additional 160 min by maintaining the same SO2 scrubbing operation conditions—except HgCl2 feedings. The Hg0 reemission decreased rapidly from ∼7.8 to ∼2.2 µg/m3 for the first 60 min and appeared to tail off thereafter. Slurry samples were also taken from the hold tank to evaluate the possibility of Hg0 reemission outside of the scrubber recirculation loop. Preliminary tests were conducted with 1.1 L/min of air flowing over the surface of 150 g of slurry in a glass impinger without any mixing or shaking. Less than 1.0 µg/m3 of Hg0 was detected in the impinger effluent air, indicating an extremely low Hg0 reemission. However, when the air was bubbled through the slurry, the impinger-exit air Hg0 concentration increased rapidly. To further evaluate the Hg0 reemission outside the scrubber loop, a group of three air-stripping tests were performed. The first test was carried out by bubbling 1.1 L/min of air through a glass impinger containing 150 mL of slurry maintained at 50 °C. Figure 6 shows that the air Hg0 concentrations (represented by solid circles) at the impinger exit increased immediately after the air bubbling started. The Hg0 concentration increase continued for about 15 min and reached a peak as high as 75 µg/m3. After the peak, the exit-air Hg0 concentration gradually decreased. The residue Hg0 emission lasted for more than 1 h with an Hg0 concentration in the impinger-effluent air of about 7.5 µg/m3 at the end of the test. The total amount of Hg0 emitted from the impinger in the first 60 min was estimated to be 2.4 µg, which is equivalent to an Hg0 reemission capacity of 16 µg/L in the slurry. The second test was performed at room temperature with the same operating conditions as the first. Hg0 reemissions, represented by solid squares in Figure 6, still occurred. However, the reemission pattern is different from that of the first test at 50 °C. Instead of a peak, the impinger-air effluent Hg0 concentration increased to a plateau within 25 min and decreased very slowly for the remainder of the test. In addition, the reemission rate was also considerably lower than that of the first test. The total amount of Hg0 emitted in 60 min was estimated to be only 0.36 µg, equivalent to an Hg0 reemission capacity of 2.4 µg/L in the slurry. The third test was conducted with slurry reheated to 50 °C after cooling to
Mercury Reemission from a Wet Scrubber
Energy & Fuels, Vol. 22, No. 1, 2008 341
When the hold tank slurry is well-mixed, the following mass balance equation can be established to represent the mercury absorption and reduction in the pilot plant: V
d[Hg·S(IV)] ) rF - kV[Hg·S(IV)] dt
(2)
where V is the volume of the slurry in the hold tank, L; [Hg · S(IV)] the concentration of Hg · S(IV) complexes in the slurry, µg/L; r the fraction of the Hg2+ that forms the Hg · S(IV) complexes; F the Hg2+ feed rate to the scrubber, µg/min; and k the first-order reaction rate constant, min-1. The solution of eq 2 under the initial conditions of [Hg · S(IV)] ) 0 at t ) 0 is rF (1 - e-kt) (3) Vk On the basis of assumptions (5) and (6), the Hg0 reemission can be modeled as [Hg·S(IV)] )
V
Figure 6. Mercury reemissions from the wet-scrubber loop.
room temperature by sitting undisturbed for about 2 h. Figure 6 shows that the residue Hg0 reemission rate and pattern, represented by solid triangles, are very similar to that of the first test. The estimated total amount of 60 min Hg0 reemission for the third test was 2.3 µg, equivalent to an Hg0 reemission capacity of 15.3 µg/L in the slurry, which was also very close to that of the first test. The results indicate that the hold tank slurry contained a considerable amount of mercury species that can be converted to Hg0 and subsequently released. Hg0 Reemission Simulation. Hg2+ absorbed in the scrubbing slurry can form different mercury species. To help understand the underlying Hg0 reemission mechanism, a model to simulate the Hg0 reemission is necessary. Chang and Ghorishi5 suggested that only a portion of Hg2+ species absorbed in the scrubbing liquor functioned as the precursor of Hg0 reemissions, while the remaining portion was relatively stable and would not serve as the precursor of Hg0 reemissions. This model, represented by eq 1, has shown success in describing the Hg0 reemission in homogeneous scrubbing solution containing sulfite and bisulfite ions but not verified under heterogeneous conditions. To confirm its viability under heterogeneous conditions, this model is employed here with minor modifications. r
HgCl2 + S(IV)
k
98 Hg·S(IV) complexes 98 Hg0 v (1)
fother stable complexes and byproducts The modified model was established on the basis of the following assumptions: (1) Because of its high solubility, all the HgCl2 fed to the scrubber was absorbed by the scrubbing slurry. (2) In the presence of S(IV) species, the absorbed HgCl2 reacted with S(IV) instantly to form a number of complexes and byproducts. (3) Only a fraction, r, of the HgCl2 fed in to the scrubber formed Hg · S(IV) complexes, which further underwent a series of chain reactions to generate Hg0. (4) The rate-controlling step of the chain reactions can be represented by a first-order reaction with respect to Hg · S(IV) complexes and a global rate constant, k. (5) Because of the insoluble nature, all the Hg0 in the recirculation slurry was either released in the hold tank or stripped off and exited the scrubber with the effluent flue gas. (6) The hold tank Hg0 reemission rate can be simulated by a mass transfer across a phase-boundary model assuming the reemission rate is proportional to the liquid phase Hg0 concentration in the hold-tank slurry.
d[Hg0] ) kV[Hg·S(IV)] - L[Hg]0 - kHAH[Hg]0 dt
(4)
where [Hg0] is the Hg0 concentration in the hold-tank slurry, µg/L; L the slurry recirculation rate through the scrubber, L/min; kH the interphase mass transfer coefficient in the hold tank, dm/ min; and AH the hold tank liquid-gas interphase surface area, dm2. The solution of eq 4 with initial condition of [Hg0] ) 0 at t ) 0 is [Hg0] )
[
rF L*(e-kt) - Vk(e1L* L * -Vk
(L/⁄V)t
)
]
(5)
where L* is the sum of L and kHAH. The flue gas Hg0 concentration at the scrubber exit can be estimated by CHg0 )
[
L LrF L * (e-kt) - Vk(e[Hg0] ) 1G L*G L * -Vk
(L/⁄V)t
)
]
(6)
where CHg0 is the scrubber-exit flue gas Hg0 concentration, µg/ m3; and G the flue-gas flow rate through the scrubber, m3/min. As presented in previous section, the pilot plant hold tank Hg0 reemission rate was estimated to be 25.7% of that in the scrubber, thus; L* ) 1.257L. When the values of L, G, F, and V are known (see Table 1), there are only two adjustable parameters, r and k, in eq 6. Using a nonlinear regression curve routine, implemented on a microcomputer, values of r and k can be obtained by fitting eq 6 to the pilot plant scrubber-exit flue gas Hg0 concentration data, as shown in Figure 3. The estimated values of r and k are 0.368 and 0.0124 min-1, respectively. In Figure 5, the HgCl2 feed to the scrubber was terminated at the end of 240 min of the test. A mercury reemission test was conducted by continuing SO2 scrubbing operation at the same conditions without HgCl2 feed for an additional 160 min. The model calculation shows that, after initial 240 min operation, the Hg · S(IV) complexes in the slurry should be equivalent to 5.0 µg/L of Hg0. In addition, there should also be 0.60 µg/L of Hg0 in the slurry. On the basis of model assumptions, the residual mercury (i.e., Hg · S[IV] complexes) should continue to form Hg0 and result in a long-lasting Hg0 reemission even without any Hg2+ supply. The Hg0 reemission without HgCl2 feed can be simulated by eqs 2 and 4 with F set at 0. According to the model, the residual mercury reemission reflected as the measured flue gas Hg0 concentration at the scrubber exit, Cd, can be represented by the following equation: Lk[Hg·S(IV)]0(e- L/⁄V t - e-kt) L[Hg0]0e+ G L* G kV (
Cd )
(
)
)
(L/⁄V)t
(7)
342 Energy & Fuels, Vol. 22, No. 1, 2008
where Cd is the scrubber-exit flue gas Hg0 concentration, µg/ m3; [Hg · S(IV)]0 the slurry Hg · S (IV) complex concentration when the HgCl2 feed was terminated, µg/L; and [Hg0]0 the slurry Hg0 concentration when HgCl2 feed was terminated, µg/L. With the values of k ) 0.0124 min-1, [Hg · S(IV)]0 ) 5.0 µg/L, and [Hg0]0 ) 0.60 µg/L as obtained by fitting the first 240 min baseline data in Figure 3, the continued Hg0 reemissions can be calculated with eq 7. Figure 5 compares the measured and the calculated mercury reemissions. Through the model simulation, it is believed that not all, but only a portion, of Hg2+ species retained in the slurry served as the Hg0 reemission precursor. Stable Hg2+, as complexes or byproducts in the slurry, can be as high as 63% of the total retained Hg2+ under this baseline testing conditions. Conclusions Pilot plant tests confirmed the previous laboratory-scale experimental results that a significant amount of absorbed Hg2+ can be reduced by S(IV) species, thereby an increase of flue gas Hg0 concentration across a wet-FGD scrubber. Elevated Hg0 concentrations were detected not only in the scrubbing tower where Hg0 reemission are generally observed but also in the hold-tank air. The estimated pilot plant hold-tank Hg0 reemission rate was as high as 25.7% of that in the scrubber. The holdtank Hg0 emissions can significantly decrease the effective
Chang and Zhao
mercury removal efficiency in a wet FGD scrubber system. Pilot plant data have also shown evidence for mercury reemissions outside the scrubber loop. Temperature and disturbance (e.g., mixing, pumping, bubbling, and vacuuming) can significantly affect the mercury reemission rate. In addition to the mercury reemissions outside a scrubber loop in coal-fired power plants, the results imply that reemissions can also occur in the waste/ byproduct treatment facilities. The mercury reemissions not only reduce the co-benefit mercury removal of a scrubber but also cause health hazard concerns of occupational mercury exposure. As a further effort to understand the relationship of Hg0 reemissions with the retained mercury species in the slurry, the Hg0 reemission across the wet scrubber was simulated with the model proposed by Chang and Ghorishi5 with minor modifications and showed success. With the simulation results, it is found that only a portion of Hg · S(IV) complexes retained in the slurry were participating in the reduction process, and the Hg0 reemission rate can be described by a first-order reaction mechanism under the testing conditions. Acknowledgment. The design, construction, and start-up of the pilot-plant facility were partially funded by a grant from the Illinois Clean Coal Institute. EF700355Q
