Measurement and Modeling of Elemental Mercury Sorption on Various Activated Carbons in a Fixed- Bed Adsorber

Transcription

1 J. Chin. Inst. Chem. Engrs., Vol. 34, No. 1, 17-23, 2003 Measurement and Modeling of Elemental Mercury Sorption on Various Activated Carbons in a Fixed- Bed Adsorber T. C. Ho [1], Y. Lee, N. Kobayashi and J. R. Hopper Department of Chemical Engineering, Lamar University Beaumont, TX 77710, U.S.A. J. Lin Department of Civil Engineering, Lamar University Beaumont, TX 77710, U.S.A. Abstract The characteristics of mercury sorption by activated carbon were experimentally measured and theoretically modeled. Experiments were carried out in a fixed-bed mercury sorption facility composed of a mercury permeation tube embedded in an isothermal water bath, a fix-bed adsorber enclosed in a furnace, and an on-line GC for mercury measurements. The proposed sorption model involved the coupling of a kinetic model based on the mechanisms of surface equilibrium and external mass transfer, and a material balance model based on the tankin-series approach. Three different equilibrium expressions were used in the model, i.e., the Henry s Law, the Langmuir isotherm and the Freundlich isotherm. In additional to kinetic simulations using the developed model, an equilibrium model was also used to simulate the thermodynamically preferred mercury species under the experimental conditions. The experimental results have indicated that the factors affecting the adsorption efficiency include the type of activated carbon, the adsorption temperature, the inlet mercury concentration, and the gas flow rate. The developed kinetic model has been found to describe well both the current experimental results and those reported in the literature. Key Words : Mercury, Adsorption, Activated carbon, Fixed-bed, Kinetic modeling INTRODUCTION Mercury emissions from coal combustion and waste incineration have been a great environmental concern and are targeted under the Clean Air Act Amendments of 1990 (Keating et al., 1997). Unlike most other trace elements, mercury is highly volatile and exists almost exclusively in the vapor phase of combustion flue gases, either in the form of elemental mercury or mercury salts such as HgCl 2, HgO, HgS and HgSO 4 (Keating et al., 1997). To protect public health, mercury emission standards of as low as 30 µg/dscm (dry standard cubic meter) have been imposed and are expected to be even stricter in the future (Keating et al., 1997). A promising method for effective mercury emission control is to employ suitable sorbents to absorb/adsorb mercury from the combustion flue gas (Krishnan et al., 1994, 1997; Korpiel and Vidic, 1997; Liu and Vidic, 2000). Activated carbon with or without chemical impregnation has been reported to be effective for mercury sorption (Keating et al., 1997). It is generally observed that mercuric chloride is more easily adsorbed by non-chemicallyimpregnated activated carbons than elemental mercury. However, sulfur-impregnated activated carbons have been found to dramatically enhance elemental mercury sorption (Keating et al., 1997; Krishnan et al., 1994, 1997; Korpiel and Vidic, 1997). The other flue gas components, e.g., SO 2, NO x, CO 2, and moisture, have negligible effect on the sorption process for elemental mercury (Liu and Vidic, 2000). In addition to experimental investigations, efforts have also been devoted to the modeling of mercury sorption processes (Carey et al., 1998; Meserole et al., 1999). Carey et al. (1998) developed a mass transfer model involving surface equilibrium to simulate mercury removal in a fixed bed. They concluded that the Freundlich isotherm provides a slightly better fit than the Langmuir isotherm during the simulations. Meserole et al. (1999) presented a theoretical model that combines the adsorption characteristics measured in the laboratory with mass transfer considerations to predict mercury removal by the duct injection process in actual flue gas streams. In their study, the Freundlich isotherm was [1] To whom all correspondence should be addressed

2 18 J. Chin. Inst. Chem. Engrs., Vol. 34, No. 1, 2003 used to describe the surface equilibrium. Their simulations were to predict when mercury removal is limited by mass transfer and when it is limited by sorbent capacity when injecting a powdered sorbent upstream of either an electric precipitator or fabric filter. In this study, both the experimental and theoretical investigations were carried out to characterize the mercury sorption process on activated carbons. Experiments were conducted in a fixed-bed mercury sorption facility and the proposed model was based on a kinetic model considering the mechanisms of surface equilibrium and external mass transfer, coupled with a material balance model based on the tank-in-series approach. Three different surface equilibrium expressions were used in the model, i.e., the Henry s Law, the Langmuir isotherm and the Freundlich isotherm. In additional to kinetic simulations using the developed model, an equilibrium model was also used to simulate the thermodynamically preferred mercury species under the experimental conditions. THEORETICAL The proposed process model involved the coupling of a kinetic model and a material balance model based on the tank-in-series approach. The kinetic model was based on the mechanisms of surface equilibrium and external mass transfer described below. Model derivation The transport rate for mercury between the bulk of the fluid phase and the outer surfaces of the sorbent granules is given by: (Perry and Chilton, 1973) dn/dt = K g S ex (C C i ). (1) For a segment of an adsorption system, the following material balance principle must be satisfied: dn/dt = (FC i FC e ). (2) Equations (1) and (2) can be combined to yield: K g S ex (C C i ) = (FC i FC e ). (3) With the assumption that the segment of the bed acts like a complete stirred tank reactor (i.e., C = C e ), Eq. (3) may be rearranged to yield: C = (K g S ex C i + FC i )/ (K g S ex + F). (4) Equation (4) describes the concentration of mercury in a segment of the bed at any bed location. The C i appearing in Eqs. (1), (3) and (4) is estimated by one of the three surface equilibrium expressions described below. Surface equilibrium expressions Three surface equilibrium laws were proposed Fig. 1. Conceptual sketch of the proposed adsorption model. to estimate C i in the model. They were: the Henry s Law, the Langmuir isotherm, and the Freundlich isotherm. The corresponding equations are expressed below: (1) Henry s Law In this approach, the C i in the model was estimated using the ideal gas law expressed as: C i = P i /RT, (5) where P i in the above equation was estimated by the following Henry s Law expression: P i = x H. (6) (2) Langmuir isotherm In this approach, the C i in the model was estimated using the following Langmuir expression: q = nc ii /(1 + kc ii ). (7) After rearrangement, the above equation becomes: C ii = q/(n kq). (8) (3) Freundlich isotherm In the Freundlich isotherm approach, the C i in the model was estimated using the following expression: q = [1/k ][C ii ] 1/m. (9) After rearrangement, the above equation becomes C ii = [k ] m [q] m. (10) Model simulation A tank-in-series model involving fifty tanks was employed for model simulation. The fifty tank was selected based on preliminary simulations that any additional tanks would not improve simulation results for the current process. Figure 1 describes the

3 T.C. Ho, Y. Lee, N. Kobayashi, J. R. Hopper and J. Lin : Measurement and Modeling of Elemental Mercury 19 Sorption on Various Activated Carbons in a Fixed-Bed Adsorber Fig. 2. Sherwood number for fixed and fluidized bed (Resnick and White, 1949). Fig. 3. Schematic diagram of the experimental set-up. schematic diagram of the model. The mass transfer coefficients (K g ) for the simulations were estimated by the experimental measurements reported by Resnick and White (1949) shown in Fig. 2. Note that the Modified Henry s Law Constant (H) and the values of parameters for Langmuir (n and k) and Freundlich (m and k ) isotherms under different mercury-sorbent pairs were evaluated by fitting the model to the experimental. In each model simulation, the initial and boundary conditions were set up as follows: Initial conditions: t = 0, C i =C o all tanks, x (or q) = 0 all tanks; Boundary conditions: t > 0, C i = C o tank 1. The simulation was to determine C e and x (or q) at each tank at any time t. The t used in the simulation was 1 minute. EXPERIMENTAL Experiments were carried out in a fixed-bed mercury sorption facility. A schematic diagram of the experimental set-up is shown in Fig. 3. As indicated in the figure, the facility was composed of a mercury permeation tube embedded in an isothermal water bath, a fix-bed adsorber enclosed in a furnace, and an on-line GC for mercury measurements. Ultrapure nitrogen gas from a gas cylinder was used as the carrier gas and elemental mercury was used as the mercury source. In an experimental run, a predetermined amount of a specific type of activated carbon was placed in the adsorber enclosed in the furnace set at a specific temperature. The nitrogen gas at a controlled flow rate was set to flow through Fig. 4. Equilibrium mercury speciation in a Hg o -Air- Sulfur system (Hg o : 1.5 wt%; Air: 94.5 wt%; Sulfur: 4.0 wt%). the permeation tube to pick up mercury at a predesigned concentration. The mercury containing gas was then flowing through the adsorber and the mercury in the gas stream was completely or partially removed by the activated carbon. The exit gas was then flowing through an on-line GC for mercury measurements. The differences between the inlet and outlet mercury concentrations represent the amount adsorbed. The activated carbons used and the experimental parameters tested are summarized in Table 1. RESULTS AND DISCUSSION Equilibrium simulation results Equilibrium simulations were performed to

4 20 J. Chin. Inst. Chem. Engrs., Vol. 34, No. 1, 2003 Table 1. Experimental parameters for the current experiments. A.C. Type Inlet Mercury Concentration (µg/m 3 ) Flow Rate (l/min) Temp. ( C) Column Diameter (cm) Bed Height (cm) Particle Diameter (mm) Mass Transfer Coefficient (m 3 /s) Reg. A.C. a HGR b a Without sulfur impregnation b With 9.7% sulfur impregnation Fig. 5. Equilibrium mercury speciation in a HgCl 2 -Air- System (Hg o : 15 wt%; Air: 94.5 wt%; Sulfur: 4.0 wt%; Chlorine: 1.5 wt%). identify the predominant mercury species during coal combustion and waste incineration. Figure 4 shows a typical set of simulation results involving coalmercury-air-sulfur. The results clearly indicate that elemental mercury is the predominant species during coal combustion at high temperatures. However, when chlorine is present as in waste incineration, mercuric chloride is seen to become the predominant species as shown in Fig. 5. Note that mercuric chloride is water-soluble and can be effectively controlled by wet scrubbers. Elemental mercury, however, is not water-soluble and requires alternative control strategies such as activated carbon sorption described in this study. Experimental and modeling results Two typical sets of experimental as well as modeling results for mercury sorption by two activated carbons, a regular activated carbon (AC) and a sulfur-impregnated AC (HGR), are shown in Figs. 6 and 7. As indicated, the regular AC appears to be a better adsorbent than HGR under the experimental conditions given on the figures. Because the sulfurmercury reactions are not expected to occur at 25 C, the observed low sorption performance associated Fig. 6. Mercury adsorption by Regular A.C. Fig. 7. Mercury adsorption by HGR. with HGR appears to be due to two main factors, large particle size and lower available carbon surface due to the sulfur impregnation. Additional experiments are currently being carried out to investigate the sorption process at higher sorption temperatures. As also indicated in the figures, the model is seen to describe the experimental results reasonably well. It is worth pointing out that all three equilibrium expressions describe the experimental data equally well and the best-fit values of the parameters for all the

5 T.C. Ho, Y. Lee, N. Kobayashi, J. R. Hopper and J. Lin : Measurement and Modeling of Elemental Mercury 21 Sorption on Various Activated Carbons in a Fixed-Bed Adsorber Table 2. Best-fit parameters for the current experiments. A.C.Type Temp. ( C) Hennry s Law Eq. H (mmhg) Langmuir Eq. n (m 3 /g) Langmuir Eq. k (m 3 /µg) Freundich Eq. k (g/m 3 ) Freundich Eq. m Reg. A.C HGR Fig. 8. Comparison between the Henry s Law and the Langmuir adsorption isotherm. Fig. 9. Adsorption profiles(hg o -PC-100 & Hg o -FGD by Krishnan et al., 1994). three expressions are summarized in Table 2. The primary reason for this observation is that the mercury sorption process occurs in an extremely low concentration range where the Henry s Law and the Langmuir isotherm behaves identically (see Fig. 8). In addition, the Freundlich isotherm is identical to the Henry s Law when m = 1.0. Comparison with literature data Additional experimental data for mercury sorp- Fig.10.Adsorption profiles(hg o -HGR & Hg o -BPL-S by Korpiel and Vidic, 1997). tion from two different research groups (Krishnan et al., 1994; Korpiel and Vidic, 1997) were also used in model simulations. It was again observed that the model is capable of simulating their data well (see Figs. 9 and 10) and that all three equilibrium expressions describe equally well these literature data (Ho et al., 2002). The best-fit values of the parameters for these simulations are summarized in Table 3. It is worth noting that the values shown in Table 3 appear to be in the same order of magnitude as those shown in Table 2 from the current study. It is also worth pointing out that a smaller H value is associated with a more efficient sorbent. Among the sorbents, the HGR used in the study of Korpiel and Vidic (1997) apparently is the most effective sorbent for elemental mercury sorption at 25 C. CONCLUSION An experimental and modeling study of mercury sorption by activated carbon in a fixed bed adsorber has been carried out. The results have indicated that many factors affect the adsorption efficiency including the type of activated carbon, the adsorption temperature, the inlet mercury concentration, and the gas flow rate. The developed model has been found to describe well both the current experimental results and those reported in the literature. All the

6 22 J. Chin. Inst. Chem. Engrs., Vol. 34, No. 1, 2003 Table 3. Best-fit parameters for the literature available data. Sorbent Type Temp( C) Hennry s Law Eq. Langmuir Eq. Langmuir Eq. Freundich Eq. Freundich Eq. H (mmhg) n (m 3 /g) k (m 3 /µg) k (g/m 3 ) m PC PC PC FGD FGD HGR HGR BPL-S BPL-S three surface equilibrium laws, i.e., the Henry s Law, the Langmuir isotherm, and the Freundlich isotherm, appear to describe the sorption profiles equally well. ACKNOWLEDGEMENT The authors are grateful for the financial support of this work by U.S. EPA through the Gulf Coast Hazardous Substance Research Center (Project No. 099LUB3697). The contents, however, do not necessarily reflect the views and policies of U.S. EPA. NOMENCLATURE C mercury concentration in gas phase, kgmol/m 3 C e mercury concentration in exit stream of a bed segment, kg-mol/m 3 C i mercury concentration in inlet stream of a bed segment, kg-mol/m 3 C o mercury concentration in incoming flue gas, kg-mol/m 3 C i mercury concentration at gas-particle interface, kg-mol/m 3 C ii mercury concentration at gas-particle interface, µg/m 3 d p particle size, m F volumetric flow rate, m 3 /min F e volumetric flow rate at the outlet (F e = F i = F), m 3 /min F i volumetric flow rate at the inlet (F i = F e = F), m 3 /min H Modified Henry's Law Constant, mmhg K g mass transfer coefficient, m/min k constant in Langmuir isotherm expressed in Eqs. (7) and (8), m 3 /µg k constant in Freundlich isotherm expressed in Eqs. (9) and (10), g/m 3 m constant in Freundlich isotherm expressed in Eqs.(9) and (10) N moles of solute (mercury), kg-mol n constant in Langmuir isotherm expressed in Eqs.(7) and (8), m 3 /g P i partial pressure of solute at gas-particle interface, mmhg q mercury concentration in sorbent, mg/g R universal gas constant, m 3 mmhg/kgmol K Re Reynolds number S ex total external surfaces of sorbent particles, m 2 Sc Schmidt number Sh Sherwood number T temperature, C or K T e temperature at the outlet (T e = T i = T), C or K T i temperature at the inlet (T i = T e = T), o C or K t time, min x mercury mole fraction in sorbent REFERENCES Carey, T. R., O. W. Hargrove, Jr., C. F. Richardson, R. Chang and F. Meserole, Factors Affecting Mercury Control in Utility Flue Gas Using Activated Carbon, J. Air Waste Manage. Assoc., 48, 1166 (1998). Ho, T. C., N. Kobayashi, Y. K. Lee, C. J. Lin and J. R. Hopper, Modeling of Mercury Sorption by Activated Carbon in a Confined, a Semi-Fluidized and a Fluidized Bed, Waste Manag., 22, 391 (2002). Keating, M. H., W. H. Maxwell, L. Driver and R. Rodriguez, Mercury Study Report to Congress, USEPA Office of Air Quality Planning and Standards, EPA-452/R , (1997). Korpiel, J. A. and R. D. Vidic, Effect of Sulfur Impregnation Method on Activated Carbon Uptake of Gas-Phase Mercury, Environ. Sci. Technol., 31, 2319 (1997). Krishnan, S. V., B. K. Gullett and W. Jozewicz, Sorption of Elemental Mercury by Activated Carbons, Environ. Sci. Technol., 28, 1506 (1994).

7 T.C. Ho, Y. Lee, N. Kobayashi, J. R. Hopper and J. Lin : Measurement and Modeling of Elemental Mercury 23 Sorption on Various Activated Carbons in a Fixed-Bed Adsorber Krishnan, S. V., B. K. Gullett and W. Jozewicz, Mercury Control in Municipal Waste Combustors and Coal-Fired Utilities, Environ. Prog., 16, 47 (1997). Liu, W. and R. D. Vidic, Impact of Flue Gas Conditions on Mercury Uptake by Sulfur-Impregnated Activated Carbon, Environ. Sci. Technol., 34, 154 (2000). Meserole, F., R. Chang, T. R. Carey, J. Machac and C. F. Richardson, Modeling Mercury Removal by Sorbent Injection, J. Air Waste Manage. Assoc., 49, 694 (1999). Perry, R. H. and C. H. Chilton, Chemical Engineers Handbook, McGraw-Hill Book Company, New York, U.S.A. (1973). Resnick, W. E. and R. R. White, Mass Transfer in Systems of Gas and Fluidized Solids, Chem. Eng. Prog., 45, 377 (1949). (Manuscript Received October 18, 2002)