Scaling Trace Organic Contaminant Adsorption Capacity by Granular

Jun 18, 2010 - ... with GAC adsorption. Christopher J. Corwin , R. Scott Summers. Journal - American Water Works Association 2012 104 (1), E36-E47 ...
0 downloads 0 Views 357KB Size
Environ. Sci. Technol. 2010, 44, 5403–5408

Scaling Trace Organic Contaminant Adsorption Capacity by Granular Activated Carbon CHRISTOPHER J. CORWIN* AND R. SCOTT SUMMERS Department of Civil, Environmental and Architectural Engineering, University of Colorado, 1111 Engineering Drive, 428 UCB, Boulder, Colorado 80309-0428

Received December 11, 2009. Revised manuscript received June 3, 2010. Accepted June 10, 2010.

The role of particle size on the reduction of granular activated carbon (GAC) adsorption capacity for trace organic contaminants by dissolved organic matter (DOM) is examined and applied to performance scale-up. The adsorption capacity reduction, termed fouling, must be scalable in order to use bench scale tests, such as the rapid small-scale column test (RSSCT) to predict full-scale breakthrough. Equilibrium adsorption capacity tests with GAC preloaded with DOM and RSSCT breakthrough curves at three different GAC particle sizes indicate that GAC adsorption capacity is dependent on GAC particle size when DOM is present. Thus, the RSSCT cannot be expected to match full-scale results regardless of which RSSCT design approach is used (constant or proportional diffusivity), unless a scaling factor is applied to the results. Proportional diffusivity RSSCT breakthrough curves demonstrate that surface concentration of DOM is not a good measure of fouling. It is hypothesized that pore blockage is the mechanism responsible for the dependence on particle size. As GAC particle size increases, the microporous surface area behind a constricted pore also increases. The result is lower adsorption capacity per mass of adsorbent in the larger GAC particles. A scaling methodology for equilibrium and breakthrough data is presented that accounts for the dependence of NOM preloading effects on GAC particle diameter.

model (1-3). By crushing the GAC to a smaller size the RSSCT can predict breakthrough behavior of a full-scale adsorber in a fraction of the time (e.g., 1-10%) required in the fullscale process. Two key assumptions are (1) GAC properties must not change when the GAC is crushed from full-size to small-size (e.g., adsorption capacity, bulk density, porosity), and (2) the substances which compete with the target compound for adsorption sites must depend on particle size in the same manner as the target compound. Two RSSCT design approaches are commonly used; the constant diffusivity (CD-RSSCT) test assumes no dependence of intraparticle diffusivity on GAC particle size, and the proportional diffusivity (PD-RSSCT) test assumes a linearly proportional dependence of intraparticle kinetics on GAC particle size (2, 3). The PD-RSSCT has been shown to yield good results for the prediction of DOM breakthrough (4, 5), as measured by dissolved organic carbon (DOC). Crittenden et al. (4) showed the CD-RSSCT adequately predicted breakthrough curves of organic contaminants when DOM was not present. While several studies have evaluated these two RSSCT approaches for their match to full-scale data for trace organic contaminants when DOM is present, a conclusion about which test is most appropriate has not evolved (4, 6-10). Typical full-scale and RSSCT organic contaminant breakthrough curves in the presence of DOM from the literature (6) are shown in Figure 1. Time is expressed as throughput, the ratio of operation time to EBCT, in bed volumes. Fullscale data yielded an earlier TCE breakthrough than data from either of the RSSCTs, indicating a lower TCE adsorption capacity. Data from the two RSSCTs exhibit about the same TCE adsorption capacity, determined by integrating the area above the breakthrough curves, or comparing the throughput at 50% breakthrough. The primary difference observed between the CD- and PD-RSSCT results is the slope of the breakthrough curves which is a result of the different approaches for scaling intraparticle diffusion (adsorption kinetics). The slope of the full-scale breakthrough curve is often better matched by the PD-RSSCT, while the onset of the breakthrough curve is better predicted by the CD-RSSCT. The results shown in Figure 1 suggest the difference in adsorption capacity between full- and small-scale GAC is

Introduction Granular activated carbon (GAC) adsorption is an effective technology for removing a wide array of dissolved organic contaminants from drinking water. However, full-scale GAC performance assessment can be time-consuming and expensive, as adsorption is a complex, nonsteady state, mass transfer limited process that is difficult to scale down. The rapid small-scale column test (RSSCT) has been the primary bench-scale tool for GAC evaluation to date; however, RSSCT results do not adequately simulate full-scale data for trace organic contaminants when dissolved organic matter (DOM) is present. Thus RSSCT utility is typically limited to the selection of GAC type and optimum empty bed contact time (EBCT). The RSSCT uses the concept of similitude to scale the adsorption process using dimensionless parameters developed from the dispersed-flow pore and surface diffusion * Corresponding author phone: +1 303 735 4147; fax: +1 303 492 7317; e-mail: [email protected]. 10.1021/es9037462

 2010 American Chemical Society

Published on Web 06/18/2010

FIGURE 1. Comparison of full-scale TCE breakthrough curves with CD-RSSCT and PD-RSSCT predictions (6). Neither RSSCT result matches the adsorption capacity of the full-scale column, but the PD-RSSCT appears to better match the kinetics (EBCT ) 5 min, C0 ∼ 1 mg/L, DOC ∼ 4 mg/L). The dashed line shows the PD-RSSCT data normalized for particle size using Y ) 0.15. VOL. 44, NO. 14, 2010 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

5403

caused by competitive interactions between the target organics and the background DOM. Because the DOM is largely irreversibly adsorbed (11) the permanent reduction in adsorption capacity is termed fouling. The scalability of fouling has been assessed by preloading different GAC particle sizes with DOM, and then performing adsorption isotherm tests (7). Results indicated that fouling was not scalable. The authors hypothesized that fouling was not mass transfer limited, but that the rate limiting step was a surface reaction that was not scalable. Constant diffusivity scaling equations were used to determine equivalent times for DOM preloading (7). Later work established that the PD-RSSCT should be used to simulate removal of DOM (4), which is primarily responsible for fouling. Classically, competitive adsorption has been described by the ideal adsorbed solution theory (IAST) (12-14). However, IAST only considers the direct site competition mechanism and has been shown to under-predict removal of target compounds at trace levels in the presence of DOM at much higher levels (15, 16). The discrepancy has been attributed to the IAST assumption that all adsorption sites are equally available to all adsorbates. High molecular weight molecules are subject to size exclusion in small GAC pores, a violation of the IAST assumption. Li et al. (17) found that a high molecular weight model compound blocked pores which resulted in slower adsorption kinetics of a target compound than without the competing substance, but that the pore blockage had little effect on adsorption capacity. Adsorption capacity reduction was mostly attributed to direct site competition from competing materials in the same molecular weight range as the target compound. Schideman et al. (18, 19) detail these mechanisms of competition, as well as a third mechanism where accumulated DOM increases external mass transfer resistance in film diffusion. The objective of this study was to evaluate the scalability of fouling related to the use of the RSSCT. The approach was to evaluate (i) the equilibrium adsorption capacity of GAC preloaded with DOM to determine the relationship between adsorption capacity reduction and particle size, and (ii) RSSCT breakthrough curves to determine the relationship between trace organic contaminant breakthrough and particle size. Finally, a fouling relationship is developed and applied to the collected RSSCT and equilibrium adsorption capacity data, pilot data, and data from the literature.

Experimental Section Adsorbates. Bisphenol A and erythromycin were selected as target compounds for bench-scale testing because of their concern in the environment. Bisphenol A and erythromycin have molecular weights of 228 and 734 Da, pKa values of 10.5 and 8.88, and log Kow values of 3.32 and 3.06, all respectively. Radiolabeled compounds were used to allow efficient quantification of target organics at trace levels. 14C bisphenol A and 3H erythromycin were obtained from American Radiolabeled Chemical, Inc. (St. Louis, MO). The pilot scale system was limited to compounds detected in the source water at high enough concentrations for quantification. Adsorbents. Fresh bituminous based GACs (Calgon F300, dp ) 1.3 mm; and Norit N1240, dp ) 0.92 mm) were carefully crushed with a mortar and pestle and separated with U.S. Standard sieves on a sieve shaker. The fractions between the #20 and #40 sieves (dp ) 0.61 mm), the #60 and #100 sieves (dp ) 0.20 mm), and the #100 and #200 sieves (dp ) 0.11 mm) were collected for bench-scale experiments. The crushed GAC fractions were washed, dried, and stored in a desiccator until use (20). Bituminous based GAC (Calgon F820, dp ) 1.47 mm) was used in the pilot adsorber. Waters. Two surface waters were used for bench-scale testing. Water A was collected from a reservoir in the front range of Colorado, coagulated with 40 mg/L alum, and filtered 5404

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 44, NO. 14, 2010

TABLE 1. Influent Water Quality Parameters for Bench-Scale and Pilot-Scale Testing water

source

A reservoir B river pilot river

DOC (mg/L) pH UVA254 (cm-1) SUVA (L/mg/m) 2.5 2.3 2.4

7.7 7.6 8.0

0.035 0.062 0.036

1.4 2.7 1.9

through a 0.45 µm cartridge filter. Water B was a southwest Colorado river water pretreated with pressure driven microfilter membranes. Measured properties of the two laboratory waters and the pilot adsorber influent water are presented in Table 1. Analytical Methods. Bisphenol A and erythromycin samples were prepared by placing 4 mL of sample and 16 mL of Ultima Gold scintillation cocktail into a 20 mL polyethylene vial. Samples were analyzed on a Packard Tri Carb 2300 liquid scintillation analyzer for a 40 min counting time. A series of dual labeled quenched standards was prepared and analyzed to determine scintillation counting efficiencies. The method resulted in reporting levels of 5 ng/L for bisphenol A and 15 ng/L for erythromycin. Pilot samples were analyzed for atrazine, DEET, simazine, and prometon using liquid chromatography followed by tandem mass spectrometry at the Center for Environmental Mass Spectrometry at the University of Colorado with a reporting level of 1 ng/L each. DOC samples were analyzed with a Sievers 800 TOC analyzer with an inorganic carbon removal unit using the ultraviolet irradiation/persulfate oxidation method in accordance with Standard Method 5310C (21). Ultraviolet absorbance (UVA) was analyzed at a wavelength of 253.7 nm using a HACH DR4000 Spectrophotometer in accordance with Standard Method 5910 (21). Equilibrium Tests. Water A was spiked with about 500 ng/L bisphenol A. For each GAC particle size, five amber bottles with preweighed GAC doses were filled headspace free. Identical control bottles with no GAC were used to detect losses other than adsorption by the GAC. Quarter liter samples were placed on a tumbler and liter samples on an orbital shaker. The 0.11 mm fraction was equilibrated for 21 days, the 0.20 mm fraction for 38 days, and the 0.61 mm fraction for 118 days. The equilibration times were proportional to the GAC particle diameter, and confirmation that equilibrium was reached was ensured by continuing equilibration for at least 10% beyond the listed times and checking that the liquid phase concentration was stable. RSSCT. RSSCTs were designed according to the proportional diffusivity approach, because the PD-RSSCT results in a better prediction of DOM breakthrough (4). The 0.11 mm and 0.20 mm GAC fractions were packed into a 4.76 mm inside diameter Teflon column. The 0.61 mm GAC fraction was packed into a 7.0 mm inside diameter glass chromatography column to maintain an aspect ratio of 11, which is above the 8-10 needed to avoid wall effects (22). The system flow rates were calibrated to 2.0 mL/min and checked daily. Effluent samples were collected directly from the column approximately every 24-48 h. The volume of water in the discharge tank was measured and recorded to determine the throughput at each sampling time. Occasionally, the composite water collected in the discharge tank was analyzed for target compound concentration, DOC concentration, and UVA as a quality assurance measure. Influent samples were periodically taken from immediately above the column. Preloading of GAC with DOM for equilibrium tests was accomplished with PD-RSSCTs performed with Water A in the absence of any spiked target compounds. At scaled sampling times of 1, 2, 4, 8, and 16 weeks GAC was removed from the column and mixed and a representative composite

FIGURE 2. DOC breakthrough results for three GAC particle sizes show no dependence of DOC adsorption capacity on GAC particle size (F300 GAC, EBCT ) 10 min, Water A). sample was obtained. The remaining GAC was then repacked into the column, and the flow rate was adjusted to achieve a constant 10 min EBCT throughout the preloading run. Pilot System. A one million gallon-per-day demonstration unit was operated for one year at a surface water utility in the southeastern United States. Coagulated and settled water was applied to a GAC filter adsorber with an EBCT of 7.1 min. Filter influent and effluent samples were analyzed monthly. A PD-RSSCT designed to simulate plant conditions with the same GAC, EBCT, and influent water was also performed.

Results and Discussion Preloaded GAC Testing. DOC breakthrough curves from the PD-RSSCTs conducted with three different GAC particle size fractions are presented in Figure 2. The DOC results, and the UVA results (available in the Supporting Information (SI)), indicate the PD-RSSCTs result in a single breakthrough curve and hence, equal surface loadings of DOM on the three GAC size fractions at any scaled time. Thus, the adsorption capacity of the GAC for DOM is not dependent on particle size when the PD-RSSCT is used. These results are consistent with previous findings that the proportional diffusivity approach is appropriate for simulating DOM removal (4, 5). Also, the data in Figure 2 displays a continuous breakthrough curve which indicates that mixing and repacking of the GAC media during sampling did not disrupt the mass transfer zone in a manner that affected the breakthrough curve. Adsorption isotherm tests were performed with preloaded GAC to assess the adsorption capacity reduction caused by the adsorbed DOM. Adsorption isotherm results for bisphenol A with the 4 week preloaded GAC are shown in Figure 3 and are representative of the other preloading times. Figure 3 depicts the fraction remaining at equilibrium as a function of the GAC dose. Little difference is observed between the two smaller GAC particle sizes, but the largest GAC particle size exhibits a lower adsorption capacity. These data indicate a dependence of adsorption capacity of bisphenol A on GAC particle size when DOM is present. The largest particle size exhibiting the lowest adsorption capacity is consistent with the full-scale GAC exhibiting earlier breakthrough than either CD- or PD-RSSCT as illustrated by Figure 1. Figure 2 shows that the preloaded GAC had approximately the same surface concentration of DOC for the three particle sizes at all sampling points, which indicate that DOC breakthrough, or DOC surface concentration, is not a good indicator of fouling. The results also confirm that direct site competition cannot be the only mechanism of fouling.

FIGURE 3. Equilibrium uptake results for bisphenol A with three GAC particle sizes preloaded to the equivalent time of 4 weeks indicate a dependence of adsorption capacity on GAC particle size (F300 GAC, C0 ) 500 ng/L, Water A). It is hypothesized that pore blockage is responsible for the observed dependence of fouling on particle size. When pore blockage occurs within the internal structure of the GAC, the surface area behind the blockage becomes either unavailable or kinetically less available to the target organics. Crushed GAC has been shown to have the same total surface area, cumulative pore volume, and pore size volume fractions per gram of adsorbent (23). However, smaller particle sizes have more of the meso- and micropores, that are subject to blockage, open to the bulk flow because the number of particles is increased per gram of adsorbent. The impact of pore blockage on fouling is thus reduced in the small GAC particles. When pore blockage occurs in the larger GAC particles there is relatively more surface area behind the blockage becoming unavailable which explains the reduced adsorption capacity compared to small GAC particles (4, 6, 7, 24). Blockage of a pore may not be total and previous research indicates adsorption kinetics are more impacted by pore blockage than adsorption capacity (17). The previous work was performed on powdered activated carbon, which has particle sizes 20-50 times smaller than GAC. Thus, the impact of pore blockage on adsorption capacity is expected to be small. Also, mass transfer to some adsorption sites can become so slow that the site is not kinetically available during the time frame of normal column operation, which reduces the apparent adsorption capacity. Results shown in Figure 3 violate two underlying assumptions of the RSSCT method, namely that adsorption capacity of the GAC is not dependent on particle size and that fouling scales in the same manner as the target compound. Since there is no mechanism by which adsorption capacity can vary with particle size in the RSSCT methodology, neither CD- or PD-RSSCT can be expected to be a good predictor of adsorption capacity when fouling is dependent on particle size. Therefore, the dependence of adsorption capacity on GAC particle size when DOM is present must be taken into account separately from the RSSCT scaling equations. Coloaded GAC Testing. Coloaded RSSCTs were performed on three size fractions of GAC with bisphenol A and erythromycin in Water B as corroborating evidence of the dependence of adsorption capacity on GAC particle size. Breakthrough curves are shown in Figure 4. For each particle size the breakthrough curves are systematically different and show that fouling increases as VOL. 44, NO. 14, 2010 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

5405

FIGURE 4. Breakthrough curves of bisphenol A and erythromycin for three GAC particle sizes show a dependence of adsorption capacity on GAC particle size (N1240 GAC, EBCT)10 min, C0 ∼ 100 ng/L, Water B). the GAC particle size increases. Again, the DOC breakthrough curves indicated equal surface concentrations of DOM for each GAC particle size fraction (shown in SI), suggesting that the degree of fouling was not just related to DOM surface concentration (as shown in Figure 3). Scalability of Fouling. Results indicate fouling is a function of particle size, so a function of the ratio of particle sizes should normalize the data. The scale factor, SF, is designated as the ratio of the diameters of the full size GAC, dp,LC, to the small-scale GAC, dp,SC: SF )

dp,LC dp,SC

(1)

The scale factor is raised to an exponent, Y, to allow variability in the relationship that accounts for the reduction in adsorption capacity attributed to fouling. Thus, it is termed the fouling index: FoulingIndex ) SFY

(2)

The fouling index is defined in this way so the normalized data is always projecting the performance of the full-scale GAC. Adsorption isotherm data, like those shown in Figure 3, are then normalized by multiplying the dimensionless fouling index by the dose. A manual search for the value of Y that visually collapsed the data to a single curve was then perfomed. Results of this normalization are shown in Figure 5, which are the same data used in Figure 3. The equilibrium data collapse onto a single curve which confirms that fouling is dependent on GAC particle diameter. Normalization is applied to the dose because the GAC dose of a batch test is analogous to the use rate in column test, both with units of mass of GAC per volume of water treated. The use rate is the primary figure of merit when determining the economics of GAC adsorbers because it determines bed life and is defined by eq 3 which relates use rate, UR, to bed density, FGAC, and throughput measured in bed volumes, BV: UR )

FGAC BV

(3)

Thus, normalization of the throughput axis can also be applied to the breakthrough curves from Figure 4 by dividing by the fouling index, as presented in Figure 6. Results indicate that the fouling index normalizes the breakthrough curves for adsorption capacity of bisphenol A and erythromycin at trace levels with the same value of Y ) 0.8. After application of the fouling index, both adsorption isotherm and breakthrough data obtained with three GAC particle sizes collapse 5406

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 44, NO. 14, 2010

FIGURE 5. Equilibrium data with the dose axis normalized to particle size for a full-scale particle size of dp ) 1.3 mm show a single curve (Y ) 0.8). onto a single curve, indicating that the effect of particle size is captured by the method. The fouling index was applied to pilot data shown in Figure 7. No spiking of compounds was performed at the demonstration unit so influent concentration of the detected compounds varied with season. Therefore, breakthrough curves were calculated on a mass basis for the pilot and the PD-RSSCT. The compounds detected in the influent at high enough concentrations for analysis were; atrazine (∼100 ng/ L), DEET (∼40 ng/L), simazine (∼500 ng/L), and prometon (∼10 ng/L). All four compounds resulted in a good match between pilot and PD-RSSCT data with a Y value of 0.6. The PDRSSCT was terminated when the water ran out; therefore, data was only available for the beginning of the breakthrough curve as these are strongly adsorbing compounds. The beginning of the breakthrough curve is typically all that is needed to determine performance because high levels of removal are usually desired. A lower background DOC in the pilot system led to higher adsorption capacities for the target compounds than in the RSSCT. The slightly lower DOC concentration in the pilot tests resulted in breakthrough curves that were closer together than would be expected if they were run at the same DOC concentration, which resulted in a smaller Y value. Similarly, the fouling index was applied to PD-RSSCT breakthrough data from the literature (6) and a constant value of Y ) 0.15 was determined as the best fit. A good match of full-scale adsorption capacity was observed for dibromo-

FIGURE 6. A single breakthrough curve is obtained with the data shown in Figure 4 when the throughput axis is normalized to a full-scale particle size of dp ) 0.92 mm (Y ) 0.8).

FIGURE 7. Comparison of pilot and PD-RSSCT breakthrough curves. Closed symbols show the PD-RSSCT breakthrough data shown at a Y value of 0.6 (F820 GAC, dp ) 1.47 mm; EBCT ) 7.1 min). Note the breakthrough curves are presented on a mass basis. chloromethane, 1,2-dibromoethane, bromoform (available in SI), and TCE (shown in Figure 1). The normalization does not perform as well for chloroform (available in SI) which breaks through before the DOC. The early chloroform breakthrough changes the nature of fouling and its susceptibility to pore blockage. Although adsorption capacity was well predicted with Y ) 0.15 the slope of the breakthrough curve departed from the pilot data suggesting that adsorption kinetics may not be matched perfectly by the PD-RSSCT. The significantly smaller value of Y is probably due to the higher initial concentration of the target organics. While an initial concentration of 100 ng/L was used in this study for bisphenol A and erythromycin the reproduced data was from a study that used influent concentrations of about 1 mg/L for each of six compounds (6). The smaller value of Y for the reproduced data means less adsorption capacity reduction due to fouling. A Y value of 0 means no adjustment for GAC

particle size as would be expected in distilled/deionized water. Najm et al. (25) determined that the relative concentrations of the target compound to DOM (C0/DOC) impacted the competition between the target compound and the DOM. The target compounds with higher C0/DOC ratios were able to better compete against the background, providing evidence that Y may be concentration dependent. Further, in each set of data; laboratory, pilot, and from the literature, a single value of Y normalizes data for multiple compounds, thus the value of Y does not appear to be compound specific. At this time, data at two particle sizes needs to be obtained to determine the value of Y.

Acknowledgments The lead author was funded by the Malcolm Pirnie Inc. Fellowship during a portion of the research. Dr. E. Michael Thurman and Dr. Imma Ferrer of the Center for EnvironVOL. 44, NO. 14, 2010 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

5407

mental Mass Spectrometry performed the sample analysis of the pilot waters. The pilot plant data was collected under Water Research Foundation Project 4155 in collaboration with Malcolm Pirnie, Inc.

Supporting Information Available Additional information including three additional figures. This material is available free of charge via the Internet at http://pubs.acs.org.

Literature Cited (1) Frick, B. R. Theoretische betrachtungen zu den problemen des scale-up von aktivkohlefestbettadsorbern. In Aktuelle Probleme der Wasserchemie und der Wasseraufbereitung; Engler-Bunte Institut, Universitat Karlsruhe: Germany, 1982. (2) Crittenden, J. C.; Berrigan, J. K.; Hand, D. W. Design of rapid small-scale adsorption tests for a constant diffusivity. J.-Water Pollut. Control Fed. 1986, 58 (4), 312–319. (3) Crittenden, J. C.; Berrigan, J. K.; Hand, D. W.; Lykins, B. Design of rapid fixed-bed adsorption tests for nonconstant diffusivities. J. Environ. Eng.-ASCE 1987, 113 (2), 243–259. (4) Crittenden, J. C.; Reddy, P. S.; Arora, H.; Trynoski, J.; Hand, D. W.; Perram, D. L.; Summers, R. S. Predicting GAC performance with rapid small-scale column tests. J. Am. Water Works Assoc. 1991, 83 (1), 77–87. (5) Summers, R. S.; Hooper, S.; Hong, S. The use of RSSCTs to predict GAC field-scale performance Proc. Am Water Works Assoc. Annu. Conf., 1994. (6) Crittenden, J. C.; Reddy, P. S.; Hand, D. W.; Arora, H. Prediction of GAC Performance Using Rapid Small-Scale Column Tests, AWWARF report #90549; American Water Works Association: Denver, CO, 1989. (7) Summers, R. S.; Haist, B.; Koehler, J.; Ritz, J.; Zimmer, G.; Sontheimer, H. The influence of background organic-matter on GAC adsorption. J. Am. Water Works Assoc. 1989, 81 (5), 66–74. (8) Vidic, R. D.; Sorial, G. A.; Papadimas, S. P.; Suidan, M. T.; Speth, T. F. Effect of molecular-oxygen on the scaleup of GAC adsorbers. J. Am. Water Works Assoc. 1992, 84 (8), 98–105. (9) Cerminara, P. J.; Sorial, G. A.; Papadimas, S. P.; Suidan, M. T.; Moteleb, M. A.; Speth, T. F. Effect of influent oxygen concentration on the GAC adsorption of VOCs in the presence of BOM. Water Res. 1995, 29 (2), 409–419. (10) Hineline, D.; Crittenden, J. C.; Hand, D. W. Use of the rapid small-scale column tests to predict full-scale adsorption capacity and performance Proc. Am. Water Works Assoc. Annu. Conf. 1987.

5408

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 44, NO. 14, 2010

(11) Summers, R. S.; Roberts, P. V. Activated carbon adsorption of humic substances 0.1. heterodisperse mixtures and desorption. J. Colloid Interface Sci. 1988, 122 (2), 367–381. (12) Myers, A. L.; Prausnitz, J. M. Thermodynamics of mixed-gas adsorption. AIChE J. 1965, 11 (1), 121–127. (13) Radke, C. J.; Prausnitz, J. M. Thermodynamics of multi-solute adsorption from dilute liquid solutions. AIChE J. 1972, 18 (4), 761–768. (14) Crittenden, J. C.; Luft, P.; Hand, D. W.; Oravitz, J. L.; Loper, S. W.; Arl, M. Prediction of multicomponent adsorption equilibria using ideal adsorbed solution theory. Environ. Sci. Technol. 1985, 19 (11), 1037–1043. (15) Knappe, D. R. U.; Matsui, Y.; Snoeyink, V. L.; Roche, P.; Prados, M. J.; Bourbigot, M. M. Predicting the capacity of powdered activated carbon for trace organic compounds in natural waters. Environ. Sci. Technol. 1998, 32 (11), 1694–1698. (16) Graham, M. R.; Summers, R. S.; Simpson, M. R.; MacLeod, B. W. Modeling equilibrium adsorption of 2-methylisoborneol and geosmin in natural waters. Water Res. 2000, 34 (8), 2291–2300. (17) Li, Q. L.; Snoeyink, V. L.; Marinas, B. J.; Campos, C. Elucidating competitive adsorption mechanisms of atrazine and NOM using model compounds. Water Res. 2003, 37 (4), 773–784. (18) Schideman, L. C.; Marinas, B. J.; Snoeyink, V. L.; Campos, C. Three-component competitive adsorption model for fixed-bed and moving-bed granular activated carbon adsorbers. Part I. Model development. Environ. Sci. Technol. 2006, 40 (21), 6805– 6811. (19) Schideman, L. C.; Marinas, B. J.; Snoeyink, V. L.; Campos, C. Three-component competitive adsorption model for fixed-bed and moving-bed granular activated carbon adsorbers. Part II. Model parameterization and verification. Environ. Sci. Technol. 2006, 40 (21), 6812–6817. (20) ICR Manual for Bench- And Pilot-Scale Treatment Studies, EPA 814-B-96-003; U.S. EPA: Washington, DC, 1996. (21) Clesceri, L. S.; Greenberg, A. E.; Eaton, A. D. Stand. Methods Exam. Water Wastewater 1998. (22) Knappe, D. R. U.; Snoeyink, V. L.; Roche, P.; Prados, M. J.; Bourbigot, M. M. Atrazine removal by preloaded GAC. J. Am. Water Works Assoc. 1999, 91 (10), 97–109. (23) Patni, A. G.; Ludlow, D. K.; Adams, C. D. Characteristics of ground granular activated carbon for rapid small-scale column tests. J. Environ. Eng.-ASCE 2008, 134 (3), 216–221. (24) Speth, T. F.; Miltner, R. J. Effect of preloading on the scale-up of GAC microcolumns. J. Am. Water Works Assoc. 1989, 81 (4), 141–148. (25) Najm, I. N.; Snoeyink, V. L.; Richard, Y. Effect of initial concentration of a SOC in natural-water on its adsorption by activated carbon. J. Am. Water Works Assoc. 1991, 83 (8), 57–63.

ES9037462