ADSORPTION PROCESSES W ALTER J. WEBER JR The University of Michigan, College of Engineering, Ann Arbor, Michigan 48104, USA ABSTRACT Adsorption is a fundamental process in the physicochemical treatment of municipal wastewatcrs, a treatment which can economically meet today's higher effluent standards and water reuse rcquirements. Activated carbon is the most effective adsorbent for this application. Expandcd-bed contact systems permit most efficient use of granular carbon for waste treatmcnt. The adsorption proccss is enhanced by in-situ partial regeneration effected by biological growth on thc surfaces of the carbon. Physicochemical systems using adsorption with activated carbon consistently produce high Ievels of treatment and have a high degree of stability and reliability. Advantages over biological trcatment systems include: lower land area requircmcnts; lower sensitivity to diurnal flow and concentration variations and to toxic substances; potential for significant heavy metal removal; grcatcr flexibility in design and operation; and, superior removal of organic wastes. INTRODUCTION Adsorption is integral to a broad spectrum of physical, biologicat and chemical processes and operations in the environmental field. Purification of gases by adsorption has played a major role in air pollution controL and adsorption of dissolved impurities from solution has been widely employed for water purification. Adsorption is now viewed as a superior method for wastewater treatment and water reclamation. Applications of adsorption for chemical processing. air pollution control. and water treatment are well known; applications in wastewater treatmcnt and water pollution control are generallynot as weil recognized, nor as weil understood. The process has been demonstrated to be widely effective for removing dissolved organic substances from wastewaters, but it should not be viewed as a catholicon for waste treatment, nor should its application be made in an empirical fashion. The purpose of this paper is to develop the details of this application, highlighting advantages over other wastewater purification processes, and defining major factors and considerations involved in its design and use. ADSORPTION PROCESSES Adsorption occurs at least partly as a result of-and likewise influences and alters-~~forces active within phase boundaries, or surface boundaries: 375
WALTER J. WEBER JR these forces result in characteristic boundary energies. Classical chemistry defines a system by the properties of its mass: for surface phcnomena the significant properties are those of the surface or boundary. A pure liquid reduces its free surface energy through the action of surface tension, which is quantitatively equal to the amount of work necessary to compensate the natural reduction in free surface energy. A large nurober of soluble materials can effectively alter the surface tension of a liquid. Deter-gents, for example, lower surface tension dramatically. lf a material which is active at surfaces is present in a liquid system, a decrease in the tension at the surface will occur upon movement ofthe solute to the surface. Migration of the substance to the surface or boundary results in a reduction of the work required to enlarge the surface area, the reduction being proportional to the concentration of adsorbate at the surface. The energy balance of the system thus favours adsorptive concentration of such surface-active sub-stances at the phase interface. The tendency of an impurity to lower the surface tension of water is referred to as hydrophobicity: that is, the impurity 'dislikes' water. Adsorption of an impurity from water on to activated carbon may result from solute hydrophobicity, or it may be caused by a high affinity of the solute for the carbon. For most systems encountered in waste treatment, adsorption results from a combination of these factors. The solubility of a substance in water is significant: solubility in the sense of the chemical compatibility between the water and the solute. The more hydrophilic a substance the less likely it is to be adsorbed. Conversely, a hydrophobic substance will more likely be adsorbed. In the context of solute affinity for the solid, it is common to distinguish between three types of adsorption. The affinity may be predominantly due to: ( 1) electrical attraction of the solute to the adsorbcnt (exchange adsorp-tion); (2) van der Waals attraction (physical or ideal adsorption): or, (3) chemical reaction ( chemisorption or chemical adsorption). Many adsorptions of organic substances by activated carbon result from specific interactions between functional groups on the sorbate and on the surface of the sorbent. These interactions may be designated a.'i 'specific adsorptions'. lt is possible for specific adsorptions to exhibit a large range of binding energies, from values commonly associated with 'physicar adsorp-tion to higher energies associated with 'chemisorption '. The adsorptive interactions of aromatic hydroxyl and nitro-substituted compounds with active carbon, for example, are specific adsorption processes resulting from formation of donor-acceptor complexes with surface carbonyl oxygen groups, with adsorption continuing after these sites are exhausted by com-plexation with the rings of the basal planes of the carbon microcrystallite 1• Adsorption results in the removal of solutes from solution and their concentration at a surface, until the amount of solute remaining in solution is in equilibrium with that at the surface. This equilibrium is described by expressing the amount of solute absorbed per unit weight of adsorbent qe, as a function of C, the concentration of so]ute remaining in so]ution. An expression of this type is termed an adsorption isotherm. Two equations, the Langmuir equation and the Freundlich equation, find common use for describing adsorption isotherms for water and wastewater treatment appli-376
ADSORPTION PROCESSES cations. The Langmuir isotherm is qe = QbC/(1 + bC) (1) in which b is a constant related to the energy or net enthalpy of adsorption, and Q is the ultimate adsorption capacity (maximum value of qe). The Freundlich equation has the general form (2) where KF and n are constants and n > 1. Data are usually fitted to the log-arithmic form of the equation, which gives a straight line with a slope of 1/n and an intercept equal to the value of log KF for C = 1 (log C = 0). The intercept is roughly an indicator of sorption capacity and the slope, 1/n, of adsorption intensity. The Freundlich equation generally agrees weil with the Langmuir equation and experimental data over moderate ranges of concentration, C. Unlike the Langmuir equation however, it does not reduce to a linear adsorption expression at very low concentrations, nor does it agree weil with the Langmuir equation at very high concentrations, since n must reach some Iimit when the surface is fully covered. The adsorption isotherm is useful for representing the capacity of an activated carbon for adsorbing organics from a waste. and in providing description of the functional dependence of capacity on the concentration of pollutant. The steeper the isotherm, the more effective is the activated carbon; that is, the sharper the rise of the isotherm to a given ultimate capacity as concentration increases, the higher will be the effective Capacity at the concentration Ievel desired for the treated water. Experimental determination of the isotherm is routine practice in evaluating the feasibility of adsorption for treatment, in selecting a carbon, and in cstimating carbon dosage requirements. The Langmuir and Freundlich equations provide means for mathematical description of the experimentally observed depend-ence of capacity on concentration. The adsorption isotherm relates to an equilibrium condition, however, and practical detention times used in most treatment applications do not provide sufficient time for true equilibrium to obtain. Rates of adsorption are thus significant, for the morerapid the approach to equilibrium, the greater is the fraction of equilibrium capacity utilized in a given contact time. There are three primary rate steps in the adsorption of materials from solution by granular activated carbon. First is the transport of the adsorbate through a surface film to the exterior of the adsorbent ('film diffusion'); second is the diffusion of the adsorbate within the pores of the adsorbent ('pore diffusion'); third is adsorption of the solute on the interior surfaces bounding pore and capillary spaces. For most operating conditions. transport of adsorbate through the 'surface film· or boundary layer is rate-limiting. If sufficient turbulence is provided, transport of the adsorbate within the porous carbon may control the rate of uptake. The method by which the carbon is contacted with the water determines in large part which of the transport or reaction steps is rate-limiting. For a completely and vigorously mixed batch reactor, pore diffusion may be rate-limiting. For continuous flow systems (e.g. beds of granular carbon) 377
WALTER J. WEBER JR film diffusion is usually rate-limiting for normal flowrates of 2--10 gpmjft2 (81-408 1/min · m2). COMPONENTS AND CONDITIONS Surface area Extent of adsorption is generally proportional to specific surface area, specific surface area being that portion of the total surface available for adsorption. If the mechanism of uptake is one of adsorption on external sites of a non-porous adsorbent, the rate should vary reciprocally with the first power of the diameter. This holds also for porous adsorbents when the rate is controlled by an external resistance, i.e. 'film transporf. Conversely, for cases in which intrapartide transport controls, the variation should be with the reciprocal of a higher power of the particle diameter 1. Solute properties In general, an inverse relationship between extent of adsorption and water solubility can be anticipated. The water solubility of organic compounds within a particular chemical dass decreases with increasing chain length, because the coropound becomes roore hydrocarbon-like as the nurober of carbon atoms becomes greater. Thus, adsorption from aqueous solution increases as an homologaus series is ascended, largely hecause the expulsion of increasingly large hydrophobic molecules from water pcrmits an increasing nurober of water-water bonds to reform. Molecular size is of significance if the adsorption rate is controlled by intraparticle transport. in which case the reaction generally proceeds more rapidly the smaller the adsorbate molecule. It must be emphasized however. that rate dependence on molecular size can be generalized only within a particular chemical dass or series of molecules. As shown by W cber and Morris2 large molecules of one chemical dass may adsorb more rapidly than smaller ones of another if higher energies (driving forces) arc involved. Many organic compounds exist, or havc the potential of existing. as ionic species. Fatty acids, phenolic species. amines. and many pesticides are a few materials which ionize under appropriate conditions of pH. Activated carbon commonly has a net negative surface charge · further many of the physical and chemical properties of certain compounds undergo changes upon ionization. Most observations point to the generalization that as long as compounds are structurally simple, adsorption is at a minimum for neutral species. As compounds become more complex. thc cffect of ionization decreases. Studies of amphoteric compounds indicate an adsorp-tion maximum at the isoelectric point. consistent with other observations that adsorption is at a maximum for neutral species. Apolar solute will be strongly adsorbed from a non-polar solvent by a polar adsorbent. but will prefer a polar solvent to a non-polar adsorbent. Polarity of organic compounds is a function of charge separation within the molecule. Almost any asymmetric compound will be more or less polar, but several types of functional groups tend to produce fairly high polarities in compounds. Examples of Lhese are hydroxyL carboxyl, nitro, nitrile, carbonyl, sulphonate. and amine. Thus ethanol, C2H50H, is polar, having an incremental negative charge on the 378
ADSORPTION PROCESSES hydroxyl group and a corresponding positive charge on the cthyl group. Because solvation by water involves formation of a hydrogen bond from one of the positively charged hydrogens of the water to a group bearing morc or less of a negative charge, along with some bonding in the reverse direction to the water oxygen, water solubility is expected to increase with increasing polarity. It therefore follows that adsorption decreases as polarity increases. even though active carbon is a polar adsorbent. Because hydrogen and hydroxide ions are adsorbed quite strongly. the adsorption of other ions is influenced by the pH of the solution. Further. to the extent to which ionization of an acidic or basic compound affects its adsorption, pH affects adsorption in that it governs t hc degree of ionization. In general, adsorption of typical organic pollutants from water is increased with decreasing pH. The organic components of a waste mixture may mutually enhance adsorption, may act relatively independently, or may interfere with one another. Mutualinhibition can be expected if the adsorption affinitics of the solutes do not differ by several orders of magnitude and there is no specific interaction between solutes enhancing adsorption. Similarly. because the adsorption of one substance will tend to reduce the number of open sites and, hence, the 'concentration' of adsorbent available. mutually depressing effects on rates of adsorption may be predicted. lt should be apparcnt from the foregoing discussion of the effects of solute character on adsorption that an analytical characterization of the impuritics present in a waste is helpful to a thoughtful prediction of thc effectivencss of carbon in water purification. Temperature Adsorption reactions are normally exotherrnie: thus the extcnt of adsorp-tion generally increases with decreasing temperature. Changes in enthalpy for adsorption are usually ofthe order ofthose for condensation or crystalliza-tion reactions, thus small variations in temperature tend not to alter the adsorption process in waste treatment to a significant extent. Adsorbent properties The properties of different carbons can have profound effects on both rate and capacity for adsorption. The surface chcmistry of active carbon has been a subject of much intcrest for more than a century. yet surprisingly little is known about the nature of the surface functional groups of this material. Recent work has provided an examination of the character of functional groups formed on active carbon under different conditions of activation, using the technique of multiple internal reflectance spectroscopy (MIRS), as a mcans for characterizing surface functional groups 1• Commercial carbons can be prepared from a variety of raw materials, including wood, Iignite, coal, hone, petroleuro residues, and nut shells. The raw material is generally activated in an atmosphere of carbon dioxide. carbon monoxide. oxygen, water vapour, air or other selected gases, at a temperature between 300" and 1 OOO.'C, often followed by quenching in air or water. Because of the 'impure' nature of the raw materials used in the production of commercial carbons, and because of the concentration and 379
WALTER J. WEBER JR temperature gradients that develop within the beds of carbon during activa-tion, very heterogeneaus or, at best, difficult to characterize surfaces result. Oxygen is known to react to a significant extent with activated carbons. Tt has been shown that carbons activated in an atmosphere of pure carhon dioxide, or in a vacuum, react with molecular oxygen at room temperature and below. This affinity for irreversibly 'chemisorbed · oxygen strongly suggests the formation of organic oxygen functional groups on the carbon surface. Several types of oxygen surface groups have been postulated to explain these phenomena. lt is generally thought that two principal types of oxygen functional groups are present on an active carbon surface. those which desorb as CO, and those which desorb as C02. Several investigators have shown experimentally that carbons activated at higher temperatures are 'basic carbons'. 'Acidic' carbons are defined a~ carbons which are capable of lowering the pH of neutral or alkahne distilled water, and which are relatively hydrophilic. 'Basic' carbons are not really basic in the acid--base sense, as they interact with acidic solutions in a specific anion adsorption manner. but they are characterized by the ability to raise the pH of a neutral or acidic solution, and by relative hydrophobicity. We have determined with MIRS techniques the presence of significant amounts of carbonyl and carboxyl groups on activated carbon surfaces. directly substantiating what had long been speculated. The behaviour of active carbon as an adsorbent has to be related to its surface chemistry: the evidence for chemical interaction at the surface between carbonyl and carboxyl groups and organic adsorbates is convincing. Enhancement of the adsorption capacity of active carbon may weil be accomplished by increasing the con-centration of appropriate surface functional groups 1. The most characteristic physical property of activated carbon is its extremely large surface area, which is comprised mainly of surfaces bordering inner pore spaces. The surface area of active carbon is approximately 1 000 m2 jg. Relative to the small geometric area of the granules or particles of this material, the large total area requires the existence of a considerable internal surface which can be provided only by small capillaries. In explaining many observed relationships associated with adsorption of materials from solution by carbon, it is essential to considcr the physical structure of the adsorbent because the size and arrangement of the capillaries (micropores: I 0 30 Ä) and channels or interstices (macropores: 30-1 OO<XlO Ä) appear to play a significant role in adsorption processes. In general, the system of macropores contributes little to the total surface area and adsorptive capacity of active carbons. A high pcrcentagc of macro-pore volume in an active carbon is often a distinct disadvantage because of loss of density to extraneous pore volume. On the other hand, for those instances in which intraparticle transpoft processes control the kinetics of uptake by porous carbon, an extensive system of larger macropores may be of considerable benefit in that full utilization of the capacitywill be attained more rapidly because of the relative ease and specd of transport in larger pores. The total pore volume of an active carbon may be measured by dis-placement of an inertgassuch as helium to account for the micropore volume and by displacement of mercury to account for the macropore volume. At 380
ADSORPTION PROCESSES atmospheric pressure, however, mercury will not penetrate pores less than about 10 A in diameter, hence simple mercury and helium displacement measurements will not yield sufficiently accurate information regarding pore size distribution. For characterization of the distribution of micropores in active carbons, the water desorption method of Jtihola and Wiig is often employed1. Regeneration is an important consideration in the use of aetive earbon for wastewater treatment. Detailed diseussion of this aspect of aetivated carbon lies beyond the seope of this paper. lt should be noted, however, that it is eurrently feasible to regenerate granular earbon by eonventional thermal teehniques for at least 15 eycles of sueeessive saturation and regenera-tion. The results of attempts to develop effieient and eeonomieal means for ehemieal regeneration of earbon have thus far been disappointing. CONTACT SYSTEMS AND REACTORS The manner in whieh to eontaet earbon most effeetively with the solution to be treated is of partieular signifieanee for large-seale treatment of water. Rates of adsorption from solution on granular adsorbents have becn found to be dependent upon the particle size of the adsorbent. lt is therefore desir-able to employ particles of as small a diameter as eonditions of effieient operation allow, so that high rates of adsorption may be obtained. In bateh-type eontact proeesses a quantity of earbon is mixed eontinuously with a speeifie volume of water until the eontaminants have been deereased to a desired Ievel. The earbon is then removed and either disearded or regenerated for use with another volume of solution. If finely powdered earbon is used in this type of system, separation of the spent adsorbent from the water may be diffieult. Conversely, the use of large particles of earbon. whieh are removed more rapidly when exhausted, requires Ionger periods of eontact between solution and adsorbent, neeessitating larger basins or tanks in whieh to retain the water during treatment. Column-type eontinuous-flow operations have an advantage over bateh-type operations beeause rates of adsorption depend on the eoneentration of solute in the solution being treated. For eolumn operation the earbon is eontinuously in eontaet with a fresh solution. Consequently, the eoneentra-tion in the solution in eontaet with a given layer of earbon in a eolumn ehanges very slowly. For bateh treatment, the concentration of solute in eontaet with a speeifie quantity of earbon deereases mueh morc rapidly as adsorption proeeeds, thereby deereasing the effeetiveness of the adsorbent for removing the solute. The rate of exhaustion of earbon in most waste treatment applieations is usually not high enough to justify moving-bed adsorbers for eolumn or bed type systems. Thus a fixed-bed adsorber is generally preferred. Upflow expanded operation of fixed beds of aetivated earbon permits the use of small particles for faster adsorption rates without the assoeiated problems of exeessive head-loss, air-binding, and fouling with partieulate matter eommon to paeked-bed operation with fine media. In expanded-bed opera-tion, the water flows upward through a eolumn of relatively fine granular earbon at a veloeity suffieient to suspend the carbon. Paeked-bed adsorption teehniques have eonventionally been used for water treatment. Expanded-bed 381
WAL TER J. WEBER JR technology is relatively new. The advantages of expanded-bed adsorbers over packed-bed adsorbers have been dcmonstrated and discussed by Weber1• For fixed-bed (either packed or expanded) adsorption operations with activated carbon the water or wastewater to be treated is passed through a stationary bed. Nonsteady-state conditions prevail in that the carbon continues to remove increasing amounts of impurities from solution over the entire period of useful operation. Figure 1 is a plot of the adsorption pattem which normally obtains for a fixed-bed nonsteady-state adsorber. The impurity is adsorbed most rapidly and effectively by the first few layers of fresh carbon during the initial stages c ~<;<>cpt,oc no no zone -~-_i --~T -~-l_ . Adsorption Ad:~l1ont;J L __ j zon~ --1 ° :~~~~~:--~ ~1------~--------~3 -~----~4 1 I I I I I I I : I I I I I I I rc,I~~Z'~l ____ ----l_ ______ _ !C21coV-=. --~-------"'"'-'--o Breakpoint Time or volume of water treated Figure 1. Schematic represcntation of the movement of thc adsorption 1.one and thc rcsulting breakthrough curvc (after Wcber1 ). of operation. These first layers are in contact with the solution at its highest concentration IeveL C0. The small amounts of solute which escape adsorp-tion in the first few layers of adsorbent are then removed from solution in subsequent strata. and essentially no solute escapes from the adsorber initially ( C = 0). The primary adsorption zone is concentrated near the influent end of the column. As the polluted feedwater continues to flow into the column. the first few layers of carbon become practically saturated with solute and less effective for further adsorption. Thus. the primary adsorption zone moves through the column to regions offresher adsorbent. The wavelike movement of this zone, accompanied by a movement of the C0 concentration front, occurs at a rate much slower than the linear velocity of the water or wastewater. As the primary adsorption zone moves through the column, more and more solute tends to escape in the effiuent. as indicated in the sequence of schematic drawings in Fiqure /. The plot of C/C0 versus time (for a constant flowrate) or volume of water treated depicts the increasc 382
ADSORPTION PROCESSES in the ratio of effiuent to influent concentrations as the zone moves through the column. The breakpoint on this curve represents that point in operation where-for all practical purposes-the column is in equilibrium with the influent water, and beyond which little additional removal of solute will occur. At this point it is desirable to reactivate or replace the carbon. The method chosen for operation of a fixed.:.bed adsorber depends to a large cxtent on thc shape of the curve given by plotting C/C0 versus time or volume. As noted previously, this curve is referred to as a breakthrough curve. For most adsorption operations in water and wastewater treatment, breakthrough curves exhibit a characteristic S shape. but with varying dcgrees of steepncss and position of breakpoint. Factars which affect the actual shape of the curve include all of the parameters discussed earlier (shape of the adsorption isotherm, solute concentration, pH. rate-limiting mechanism for adsorption and nature of the equilibrium conditions, particle size, etc.) and, in addition, the depth of the column of carbon and the velocity offlow. As a general rule, the time to breakpoint is decreased by: (1) increased particle size of the carbon; (2) increased concentration of solutc in the influent; (3) increased pH of the water; (4) increased flowrate; and. (5) decreased bed depth. lf the total bed depth is smaller than the length of the primary adsorption zone required for effective removal of solute from solution, then the concentration of solute in the effiuent will rise sharply from the time the effiuent is first discharged from the adsorber. Thus. for each type of adsorption operation there exists a critical minimum carbon depth. PREDICTION AND DESIGN Quantitative prediction of the performance of fixed-bed adsorbers involves prediction of the shape and position of the breakthrough curve. representing the movcment of the adsorption front through an adsorber. lf the breakthrough curves for adsorber systems can be reliably predicted on the basis of some easily determined Iabaratory measurements, extensive pilot plant scale studies can bc obviated, and considerable savings in time and money can bc realized. This 'requirc~ application of appropriate mathe-matical modeling techniques for operation on information developed from component and condition analysis. Modcling procedures can take several Ievels of sophistication. A simple mass transfer model and a second-order kinetics model have been described by Weber1. Usinowicz and Weber3 have examined the use of the second-order model for mixtures of contaminants exhibiting markedly different adsorption characteristics. Figure 2 is repre-sentative of the results obtained. The use of second-order kinetics facilitates analytical solution of the resulting partial differential modeling equations. but for multiple solute systems does not provide a true prediction of the entire breakthrough curve. As may bc noted from Figure 2, however, the prediction is generally good for the first 5~60 per cent of the operational period of the adsorber; this suffices for most practical purposes, since adsorbers are normally not operated to more than 15-30 per cent break-through in wastewater treatment. Nonetheless, models amenable to analytical solution must employ relatively simple types of rate expressions describing the uptake of adsorbates (e.g. second-order kinetics or film transport). The 383
cc .. ... :l ::J 0.8. ~ ~ 0.2 0 10 20 WALTER J. WEBER JR 0 0 0 0 30 Time, h 0 0 0 0 0 0 o Experimental data -Kinetic modet prediction Initial COD = 275 mg/1 L.O 50 60 70 Figure 2. Experimental and predicted COD breakthrough curves for a complex mixture of organic substances (after Usinowicz and Weber~). models are similarly limited to linear or Langmuir type adsorption isotherms: both of these analyses are inadequate for describing adsorption behaviour at the high concentrations found in many wastewaters. Usinowicz and Weber3 applied the mathematical technique ofThomas (1944) to mixtures of compounds exhibiting a broad spectrum of adsorption behaviour: since this method was developed for a homogeneaus single sorbate system. the systems studied were necessarily treated in terms of one lumped-parameter quantity, COD(Chemical Oxygen Demand). Thedatapresented by Usinowicz and Weber could be predicted for only the first half of the effiuent break-through profile by the analytical solution of Thomas. Models of this type predict symmetrical breakthrough curves, hence, although they are often adequate for dilute single-solute or dominant-solote solutions. or similarly simple systems, their use for complex heterogeneaus systems is limited. Other shortcomings are the Iack of flexibility to permit inclusion of multi-component interactions, biological activity, solids mixing, dispersion, and multicomponent equilibrium hysteresis. To eliminate these handicaps Weberand Crittenden5 have developed several numerical techniques which can include these factors. A material balance on an adsorption results in two partial differential equations (PDEs), orte for the liquid phase and one for the solid phase. Extending this to a multisolute system, a pair of PDEs is obtained for each solute. A suitable expression for the rate ofuptake, an equation of equilibrium, and information on the mixing of the solids is needed before the solution of the PDEs can be efTected. Weber and Crittenden have used film transport, and Thomas's second order expression to describe the equilibrium behaviour of the solute and have treated the solid phase as both completely mixed and fixed. For multicomponent systems, the only difTerence is the means in which 384
ADSORPTION PROCESSES the equilibrium positions is described. Extension of the Langmuir expression to multisolute systems as reported by Snoeyink and Jain6 and by ideal solution and ideal adsorption theory developed by Radke and Prausnitz 7 were used to characterize the equilibrium behaviour in multisolute systems. Four mathematical techniques were compared to each other by Weber and Crittenden5 and to the data of Keinath and Weber8, Usinowicz and Weber3 and Valencia and Gloyna9. A Iist ofthe four mathematical techniques evaluated, including abrief description of the numerical techniques, follows. ( 1) Thomas's analytical solution to the PD Es, the solid phase is fixed. (2) Numerical techniques developed by the authors. An Implicit Backward Finite Difference Method (IBFDM) is utilized to solve the liquid phase PDE and a second-order Runge-Kutta is employed to solve the solid phase PDE. This technique is applied for both extremes of the solids mixing range, i.e. fixed and completely mixed. The authors demonstrated that the IBFDM used on the liquid phase PDE converged to the identical breakthrough curve that the Implicit Forward Finite Difference Method (IFFDM) used on the liquid phase PDE would predict for small time and distance steps, i.e. about 400 time steps and 400 distance steps. However, the IBFDM converges more rapidly than the IFFDM, hence requires fewer time steps and fewer distance steps. (3) Michael's mass transfer model, in which the solid phase is taken tobe fixed and the rate law used for the uptake of the adsorbate is film transport. (4) The distributed parameter techniquc, in which the adsorber is divided into a series of completely mixed flow reactors (CMFRs). Two Ordinary Differential Equations (ODEs) per CMFR is the outcome of this analysis. The ODEs are solved by a 1 6th-order predictor-corrector method known as DVDQ. This method was chosen over the Continuous Systems Modeling Program (CSMP) since it is more economical and much more accurate. It will meet a local truncation error specified by the user. For a multicomponent system methods 2-A can be extendcd from the single solute case to incorporate multicomponent interactions. but not the analytical solution of Thomas. The reasons these can account for multi-component behaviour in carbon column is that methods 2 and 4 are just marehing time and distance problems for which the addition of more com-ponents produces additional algebraic equations at any given time step, while for method 3, additional equations in the numerical quadrature technique pose no difficulty. Figure 3 is a plot of predicted breakthrough profiles for the numerical technique (number 2 above) of Weber and Crittenden. Film transport controls rate of uptake in this case. The Langmuir expression is used to describe the equilibrium position. Plots are given for fixed and completely mixed solid phase with several values for the film transfer coefficient. DiGiano and Weber 10• 11 have examined and discussed the kinetics of adsorption in, and performance characteristics of. completely mixed batch (CMB) and CMF reactors. The batch or CMB reactor requires consideration of nonsteady reaction conditions for both solution and solid (activated carbon) phases, while the CMF reactor is normally operated in a region of steady-state (constant concentration) for the solution phase. The types of modefing equations discussed by DiGiano and Weber for these systems 385
1.0 4J 0.5 \.:1" 0 t..J 0 W ALTER J. WEBER JR 1 0 Throughput porameter Legend 1. q/qe, CMSP, k10= 9. 00 min-1 2 qlqe, FSP , kta=" 9. 00 min-1 3. qiqe, CMSP, kt0= I. 33 min-1 1.. qlqe, FSP , kt0= 4 33 min-1 5C/Co,FSP ,kta= 4 33 min~1 6 C/C0, CMSP, kto. = 4. 33 min-1 7. C!Co, FSP , kta= 9 00 min-1 8. C!Co, CMSP, k1n = 9 00 min-1 Fiqure 3 Effiuent concentration influent conccntration (C1C0) and uptakc ofsolute cquilihrium uptake of solute at the influent conccntration (q!q,.) versus the throughput paramctcr for a completely mixed solid phase (CMSP) and a fixed solid phase (FSP). Overall film cocfficicnt 14) given for each curve. apply both for the distributed parameter simulation technique for fixed bed adsorbers and for single and/or multiple stage application of carbon in tank reactors. As noted earlier. granular carbon is normally applied in fixed-bed (PF or PFD) reactors while powdered carbon is gcncrally applied in tank-type (CMB or CMF) reactors. APPLICATIONS As noted earlier_ the most common type of adsorption system is one in which wastewater is passed through fixed bcds of carbon. In such systems the waste is applied to the beds at rates generally ranging from 2 gpm/ft 2 to 10 gpm/ft2 (81-408 l/min.m2). In this flow range esscntially equivalcnt adsorption efficiency is obtained for equivalent contaci times. At flowrates below 2 gpmjft2 (81 1/min m2) adsorption efficiency is reduccd. whilc at flowrates above 10 gpm/ftl (408 l/min.m2) excessive pressure drop takes place in packed beds. Contact times employed are in the range of 30 minutes to 60 minutes. In generat increases in contact time up to 10 minutcs yicld proportionale increases in organic removaL Beyond 30 minutes the rate of increase falls ofT with increases in contact time, and at about 60 minutes the effects of additional contact time become negligible. Carbon beds opcrated at the lower end of the flow range are generally designed for gravity flow. Systems designed for higher flowrates must employ pressure vessels if packed beds are used. A pressure vesscl is rnore expensive to construct than a gravity flow vesseL but commonly requires less land area. and pro-vides greater ability to handle fluctuations in flow. Provision must be made regularly to backwash packed-hcd carbon systems because they collect suspended solids and attached biological growths 386
ADSORPTION PROCESSES which tend to develop in this application. Backwashing alone generally relieves clogging due to suspended solids, but does not completely remove attached biological growth. lt is advisable to include a surface wash and air scour to be assured of removal of gelatinous biological growth. Attached growth can Iead to development of anaerobic conditions in packed beds. Aeration of the feed is partially effective in preventing anaerobic conditions, but this also accelerates growth to the extent that excessive backwash is required; air-binding can also result. Packed beds of granular carbon areweil suited for treatment of solutions containing little or no suspended solids, and under such circumstances normally operate effectively for extended periods without clogging or excessive pressure loss. However, the suspended solids invariably present in municipal and industrial wastewaters and the potential for biological growth on the surfaces of the carbon present some problems for the use of packed beds as noted above. By passing wastewater upward through a bed of carbon at velocities sufficient to expand the bed, problems of fouling, plugging and increasing pressure drop are minimized. Effective operation over Ionger periods of time results, as has been demonstrated in comparative laboratory sturlies and in pilot field investigations in both 'tertiary' and direct physicochemical applications (Weber12, Hopkins et a/.u, Weber et al. 14· 1 5). Another advantage of the expanded bed is the relatively small dependence of pressure drop on particle size. lt is possible to use carbon of smaller particle size in an expanded bed than is practicable in a packed bed. thus taking advantage of higher adsorption rates. Perhaps the most significant potential benefit provided by expanded-bed adsorption systems is the extension of the operational capacity of activated carbon observed by Weber et a/.14• 15. These researchers found that apparent sorption capacities in excess of 1 00 wt ~~ as organic matter and 150 wt ~~~ as chemical oxygen demand (COD) could be obtained in expanded-beds of activated carbon in which biological growth was allowed to develop fully. Because expanded beds rcquire little maintenance, extended periods of undisturbed operation facilitate the development and continuous growth of bacteria on the carbon surfaces. These bacteria utilize organic matter adsorbed on the carbon as a food source, functioning to provide in-situ regeneration by renewing the carbon surface for continued adsorption. Figure 4 after Weber et al.15 is a schematic representation oftbis adsorption-biological oxidation process. To provide a good effiuent and to utilize the sorption capacity most effectively, an approach to countercurrent contact is commonly required. This can be achieved by causing the wastewater to flow through a nurober of contactors or stages in series in one direction while the carbon moves in the opposite direction. In powdered carbon contacting systems this is the procedure used. With granular carbon the procedure is generally modified somewhat to avoid undesirable attrition Iosses due to cxcessive handling of the carbon. F or most granular carbon contact systems the Iead contactor in a series of adsorption columns is removed from service when the carbon it contains is cxhausted and, after being refilled with fresh carbon, is placed at the end of the series. Each contactor is thus advanced one position in the series by piping and valving arrangements which permit shifting of the inflow 387
WALTER J. WEBER JR ' " " -.... Aerobic \ '-"boundary \ Anaerob1~ film \ boundary\ \ film \ \ \ \ I I 3 I I -~; Bulk solution I -----!+_ 1 I ---.....co2 + / / / I _...... I ../ / / _......./ / / / Fiyure 4. Schcmatic interpretation of in-situ biological regeneration. Evcnt sequcncc ( ll Diffu-sion of !arge adsorbing organic molecule~!, to surface of carbon. (2) Anaerobic degradation of !arge molecule .5 tosmall molecule B . (3) Diffusion of small non-adsorbing organic molecule H' away from surface of carbon. (4) Aerobicdegradation of small molecule B to C02 and H20. and outflow points ofthe series accordingly. As the number of stages increases, the piping and valving arrangement becomes more complex and costly. A compromise between the advantage of employing multiple stages to utilize carbon capacity more effectively and the cost of each additional stage must be achieved. Principal factors which must be considered in the design of a carbon adsorption system may be summarized as: ( 1) type of carbon, granular or powdered: (2) contact times: (3) flowrate: (4) configuration, series or parallel: ( 5) number of stages: (6) mode of operation, packed-bed or expanded-bed, pumped or gravity flow: and, (7) adsorption capacity. Table 1 gives carbon capacities obtained in field operations at several physicochemical pilot plants. In that the wastes, effiuent criteria numbcr of contact stages, etc., varied from plant to plant, it is not surprising that some spread in the results is observed. For generat planning purposes a COD capacity of 50 wt ~<~ is reasonable if no biological extension of carbon capacity is taken into account. This is approximately cquivalent to a rcquircmcnt of Tahle I. Carhon capacities obtainccl in physicochemical pilot plants Plant Capacitics. Wt";, T<K: COD Bluc Plains (Washington) 15 41 Ewing Lawrcncc (Ncw Jersey)* 50 ISO Ncw Rochelle (New York) 20 24 60 Lebanon (Ohio) 22 50 • Blolo,e.cally-extcndcd expantled-nerl operatwn tWelwr 1'/ o/1 '1 388
W. J. WeberSearch for this author in:
