Shdation and design models for

adsorption processes
Walter J. Weber, Jr.
Edward H. Smith
7be University of Michigan
Ann Arbor, Mich. 481G9

Adsorption is a fundamental process

for separating inorganic and organic
contaminants from waters and waste
waters. The most common environmental applications of this process are
embodied in ion-exchange and activatedsarbon systems. Although such
processes are generally capable of
highly effective separations, the design
and operation of specific systems is frequently complicated by the composition
of waters and wastes and by the intricate matrix of interactions commonly
associated with heterogeneous systems.
These complications are particularly
severe in applications involving the removal of specifically targeted organic
compounds from complex mixtures, a
circumstance encountered with increasing frequency as water decontamination
requirements become more stringent.
Adsorption system design traditionally has been based on information
gained from pilot-plant test programs.
The pilot-plant design approach acknowledges the system specificity of
particular applications, but it does not
lend itself easily to prediction of system
responses to variables other than those
specifically tested, nor to elucidation of
design principles that can be e x t r a p
lated to other applications.
Mathematical process models can facilitate the design of full-scale systems
simply by reducing the number of pilotscale tests required to evaluate various
operating conditions and design parameters. Such models can generally be
calibrated from simple and well-controlled bench-scale experiments, thus
1040 Environ. Sci. Technal.. Vai. 21. No. 11, 1987

Q
positioning pilot test programs principally for verification of design bases
rather than for primary data development.
Once verified, a process model can
be used to examine conditions other
than those directly measured. Such a
model can also project adsorber response and sensitivity to a variety of
circumstances that may be anticipated
but not easily reproduced or experimentally simulated in the pilot program. The logistics, time, and expense
normally associated with pilot-scale
test equipment and programs make
these important considerations.
This paper provides an overview and
analysis of certain contemporary ele-

ments and applications of adsorption

process modeling. Included is a discussion that expands upon considerations
initiated in a recent ES&T feature article that described processes for removing dissolved organic contaminants
from water ( I ) . In addition, this paper
lays a foundation for more detailed descriptions of research developments in
adsorber modeling (2-4).
The work is not exhaustive in that it
makes no attempt to present and critique all models and approaches available. Rather, it focuses on the Freundlich
and Ideal Adsorbed Solution Theory
equilibrium models and a two-resistance, homogeneous-surface-diffusion
dynamic model as examples that seem
particularly suited for simulating adsorption processes and predicting their
behavior in water and waste treatment
applications. A discussion of modeling
approaches and parameter estimation
techniques that address environmental
complexities is given to bridge the tbeoretical aspects of modeling with some
specific challenges encountered in
field-scaledesign.
The bridging of theory and practice
is perhaps one of the most valuable engineering functions that mathematical
models can serve, particularly with respect to processes as complex as adsorption. A modeling effort can be focused either on the interpretation of
basic data to facilitate an unequivocal
understanding of mechanism or on the
translation of empirical observations
into functional design relationships.
Models developed pursuant to the
first objective must by nature be
mechanistic, whereas those structured primarily to serve the latter purpose may be pbenomenological in
character. As theory merges with practice, the interpretation of observed phenomena is enlightened by a greater

0013936W8710921-1040$01.5010 0 1987 American Chemical Smiely

understanding of underlying mecnanism, and the two types of modeling

efforts logically converge.
Indeed, it is often the convergence of
the two different modeling approaches
that both facilitates and signifies the
merging of theory and practice. Research and development on adsorption
models stands squarely at the crossroads of theory and practice today. Specific efforts may lean right or left, yet
all contribute to the bridging process.
An important objective of this review is
to identify and document specific areas
that require effective interaction and
communication between differently focused modeling approaches.
Fixed-bed admrbers
The most common adsorber configuration used in water and waste treatment applications is the fixed-bed reactor (FBR). One critical aspect of the
design of an FBR adsorber involves
characterization of the effluentconcentration profile as a function of throughput (Le., the volume processed or the
time of operation). This profile, commonly termed the breakthrough curve,
represents the specific combination of
equilibrium and rate factors that control
process performance in a particular application.

The aynamic behavior of an FBR aosorber may be visualized in terms of an

active adsorption zone that moves
through the bed as a function of mass
loading and saturation of the adsorbent,
illustrated in Figure l(a). Figure 1@)
translates this behavior into the breakthrough curve, which is essentially the
trace generated by movement of the adsorption zone through and out of the
adsorber.
The rate of appearance and the shape
of a breakthrough profile depend on
several factors, including the physical
and chemical properties of both the sorbate and sorbent, the particular ratelimiting mechanisms involved, the
depth of the bed, and the velocity of
flow. The relative effects of these factors are highly specific to each particular application, which constitutes one of
the major obstacles in the design of
such processes.
The first step in developing a mathematical model that describes or predicts
adsorption dynamics is to provide a
representation of the equilibrium behavior. Figure 2 depicts several different types of equilibrium capacity-concentration distributions, or adsorption
isotherms (5).The second step is to
mathematically characterize the associated rate phenomena. In general, the

rate of uptake of a contaminant by a

microporous adsorbent, such as activated carbn, is controlled by a resistance to mass transport rather than by a
reaction velocity.
Figure 3 is a schematic representation of a porous adsorbent surrounded
by a hydrodynamic boundary layer or
film. The macropores of such an adsorbent are typically large enough that
associated diffusion phenomena are unhindered by the pore walls, whereas the
micropores have radii comparable to
the size of diffising species, so diffision may be hindered by the pore walls.
This conceptual view translates into the
resistance rate series shown in Figure
3; the overall rate of adsorption is controlled by the step($ providing the
greatest resistance to mass transport.
The third step in structuring an adsorption model is implementation of the
thermodynamic principle of mass continuity. In an FBR, this results in a material balance for each component of
interest in both the liquid (water) and
solid (adsorbent) phases; that is, for a
control volume of either phase:
MlS
Mass
MUS
rateof = flux of . flux of
component
component
component
change
in
out

Mass
rite of
component
adsorption

Environ. Sci. T&hnol.. Val. 21. NO. 11. 1987 1041

The flux of a component entering or

leaving the control volume can include
bulk flow, dispersive flow, and molecular diffusion.
The normal procedure for applying
the continuity principle is to consider a
small but finite control volume representative of the overall system. After
the appropriate relationships describing
the l o c a l i i mass balance are developed, the control volume is assumed to
become infinitesimally small so as to
generate the continuity relations on a
differential scale. Boundary and initial
conditions are also established to specify the constraints of integration. Because the partial differential equations
comprising such a system are often
highly nonlinear, a numerical method is
generally required for their solution.

F&liirium modeling
Breakthrough profiles generated
from predictive dynamic models are
particularly sensitive to equilibrium
parameters (610).This can be especially crucial in the very low, so-called
Henrys region, of concentration as
well as in very high equilibrium concentration ranges. It is thus important
in the formulation of a modeling approach to select an equilibrium model
that can accurately describe single-solute isotherm data over several decades
of concentration and that can be translated into a predictive multicomponent
isotherm equation. Also, the model selected must be integrable with dynamic
models in such a manner as to minimize
computational efforts.
There are numerous single- and multicomponent equilibrium formulations,
both theoretical and empirical in nature, that vary considerably in applicability and utility (11).Multicomponent
models that are more predictive in nature yield values of liquid- and solidphase concentrations for competing adsorbates based upon single-solute
isotherm data or sorbate-solution characteristics. Difficulties variously associated with such models are that:
they may not be theoretically sound
for the systems they purport to represent;
they may contain parameters that are
difficult to obtain either experimentally or analytically;
they may be restricted as to the types
and concentrations of compounds for
which they produce accurate results;
or
they may not yield an explicit calculation of adsorption variables for integration into dynamic models.
Interests in this research area remain
strons. oarticularlv among those attemp&to utilize purely th&retical information and first principles as a

1
1042 Environ. Sci. Technol., Vol. 21, No. 11. I987

basis for engineering projections.

The Freundlich Model. Semi-empirical models, such as the familiar
Freundlich equation, generally fit single-solute experimental data well and
acknowledge the surface heterogeneity
of activated carbon; therefore, these
semi-empirical models are frequently
used in adsorber design efforts. This
model has the form:
= KFC"

(1)

where q and C are the equilibrium

solid- and solution-phase concentrations of solute; KF and IZ are characteristic constants relating to adsorption capacity and intensity, respectively.
Extensive use of the Freundlich isotherm model has led to a significant
data base of associated model coefficients for different single-adsorbate-adsorbent systems (12). A potential disadvantage of this model is that it does not
approach Henry's law at low concentrations. To accommodate this shortcoming, the Freundlich equation has been
altered in a variety of ways to adequately describe data over wider concentration ranges (13, 14).
Expansions of the Freundlich model
have also been proposed for the description of multicomponent equilibria
(2, 15, 16). Due to their empirical nature, formulations such as these require
multicomponent isotherm data to calibrate model coefficients.

tems (14, 19). The feasibility of incorporating IAST into kinetic models to
predict adsorption dynamics for multicomponent mixtures has also been
demonstrated (3, 20, 21).
The IAST model assumes thermodynamic equivalence of the spreading
pressure of each solute at equilibrium,
where spreading pressure, T ,is the difference in interfacial tension between
the pure solvent-solid interface and the
solution-solid interface at the same
temperature. Single-solute isotherm
data are required to compute the
spreading pressure of each solute, i, according to:

where R is the ideal gas constant, T is

absolute temperature, and A is the specific surface area of the adsorbent.
When single-solute isotherm data are
described by an appropriate mathematical model, such as [ql =f(Ci)l, Equation 2 can be expressed and spreading
pressure can be calculated according to
the modified relationship (22):

The Ideal Adsorbed Solution Theory Model. The accuracy with which

predictions of multicomponent adsorp- The terms Ci*and qi* in Equations 2

tion equilibria can be obtained was en- and 3 are, respectively, the liquid- and
hanced significantly, especially for di- solid-phase concentrations of species i
lute solutions, by development of the in single-solute systems yielding the
ideal adsorbed solution theory (IAST) same spreading pressure as that of the
model (17, 18). This model has re- mixture. C;. and qi are the respective
ceived widespread use in multisolute liquid- and solid-phase concentrations
adsorption research for a variety of rea- of solute i in the mixture.
Other equations required for the
sons. Besides the fact that application
of IAST necessitates only single-solute IAST calculation are:
data, the model is flexible in that multicomponent calculations can be per7rj = 7r
(4)
formed using several different singlesolute isotherm relationships.
In addition, this model has a solid
theoretical foundation, providing a useful understanding of the thermodyN
namic approach to adsorption. In this
regard it is similar to the Gibbs adsorpzj=1
tion equation upon which it is based.
i= 1
This is in contrast to other popularized
multicomponent models such as the
Langmuir competitive model, which is
founded on the same limiting assumptions as the single-solute Langmuir
model (Le., monolayer adsorption and
a homogeneous adsorbent surface).
Comparative studies have indicated
that the IAST model has greater accu- where zi is the mole fraction of solute i
racy than the Langmuir competitive in the adsorbed phase; qT is the total
model for certain multicomponent sys- quantity of material adsorbed from the

mixture; and N is the total number of

species in solution.
To obtain a solution, Equation 3 must
be integrated over the concentration
range of interest (utilizing the singlesolute equilibrium model) to calculate
spreading pressures, The remaining unknowns must be evaluated by simultaneous solution of Equations 3-8. The
model thus becomes increasingly difficult to use for increasing numbers of
solutes. In addition, care must be exercised to ensure that simple single-solute
models provide a sufficient degree of
precision over the concentration ranges
of interest, whereas more sophisticated
models may make evaluation of the
spreading pressure (Equation 3) prohibitively difficult.
A number of attempts have been
made to modify the IAST model to improve its accuracy and to minimize
computational requirements to render it
more amenable to complex systems.
One approach to reducing the computational effort required to implement
IAST is based on the assumption that
the spreading pressure for multicomponent and single-solute systems can be
equal only for the special case of identical isotherms, or when the single-solute
Freundlich isotherms for the components have equal slopes (23-26).
Average isotherm constants, which
become representative of hypothetical
(or fictive) solutes of the multisolute
system, can be calculated to reduce the
IAST equations to a single expression.
The solid-phase concentration of a solute in the mixture can be calculated
directly from the liquid-phase concentration and the actual and average isotherm constants.
This simplified competitive adsorption model (SCAM) allows IAST concepts to be applied to complex mixtures
of many solutes. However, in this form
the model does not incorporate the
mass balance relationship, has limited
applications over broad concentration
ranges, and is not always coincident
with more rigorous IAST calculations
when Freundlich n values (isotherm
slopes) for competing solutes are markedly different.
For some applications it is difficult to
obtain accurate calculations for the
spreading pressure of each solute for
use in the IAST model; such calculations require accurate descriptions of
single-solute data. Although the
Freundlich model is often suitable for
highly heterogeneous surfaces (such as
activated carbon) and is simple in form,
it is not always appropriate for describing equilibrium data in low concentration ranges.
In one approach (29, the log-log
plot of equilibrium data is divided into
several straight-line segments, each deEnviron. Sci. Technol., Vol. 21, No. l l , 1987 1043

scribed by appropriate Freundlich constants. The method is well-suited for

incorporation in multicomponent dynamic models because it calculates liquid-phase concentrations explicitly for
given solid-phase concentrations. A
major limitation is that the model application becomes increasingly difficult
for more than two solutes. In addition,
the mass balance equation is not included, rendering batch-mode verification of multicomponent concentrations
less meaningful. In other words, the
model is not truly predictive.
In a similar attempt a three-parameter Freundlich single-solute isotherm
was proposed for which an analytical
solution could be obtained for Equations 3 and 4 (14, 28). The mass balance expression for each component in
the mixture is incorporated:

where Co,Lis the initial concentration of

species i in the mixture and M/V is the
carbon dose. As expected, computations are more complex for the threeparameter isotherm model than for the
simple Freundlich equation, and manipulation of single-solute data to establish isotherm parameters is somewhat
arbitrary.
The standard Freundlich equation
(Equation 1) has been applied successfully to IAST calculations for simple
mixtures of similar solutes such as lowmolecular-weight halocarbons (29).
Different modifications of IAST have
been employed for systems in which
competitive interactions are significant.
In one case, the simplified IAST model
was modified by adding a term to account for the effects of irreversibility
and unequal competition (19).
The irreversibility term was determined experimentally by sequential
sorbate loading of a shallow GAC bed
incorporated within a completely mixed
batch reactor (CMBR) system, sometimes referred to as a differential batch
reactor (30).The model showed significant improvement over its predecessor
for two-solute systems of mostly
phenolic compounds. An alternative
scheme involves empiricizing the IAST
model by addition of a term, R,, to
Equation 5 to more accurately describe
experimental equilibria (3, 31, 32);
specifically:

Approaches such as these illustrate

how the interplay between scientific
and engineering disciplines can be used
to develop models that both maintain
theoretical integrity and describe, simu1044 Environ. Sci. Technol., Vol. 21, No. 11, 1987

late, and ultimately predict, process

phenomena with acceptable accuracy.
IAST has several inherent shortcomings that are revealed when the model is
applied to nonideal organic solutions,
especially mixtures containing humic
substances. The concept of equivalent
spreading pressure does not adequately with the associated boundary condi
address specific adsorption phenomena tions:
or so-called nonideal competition, nor
does it account for the differential availability of sorption sites on heterogeneous adsorbent surfaces induced by
steric or surface energy effects (33-35).
Moreover, for many mixtures, par=O
(14)
ticularly for complex mixtures containing humic materials, reliable information on the spreading pressures of
solution components is not available to
provide accurate input for the model. In
such cases, the researcher or design engineer can either abandon the existing In the above expressions, E represents
model in favor of a new approach or the bed void fraction; ps is the apparent
modify a given model to account for particle density; r is the particle radius;
observed anomalies and preserve the and Dp is the pore diffusion coefficient.
A number of investigators have apability to extract phenomenological significance from the empiricized ver- plied the pore diffusion model in conjunction with film resistance and varision(s).
ous equilibrium expressions (38).It has
been found, however, that pore diffuKinetic modeling
Dual resistance models. Dynamic sion coefficients calculated according to
models that account for both film and this model are frequently larger than
intraparticle diffusion appear most suit- corresponding free-liquid diffusivities.
able for describing the adsorption of This suggests that surface (rather than
both target contaminants and back- pore) diffusion may be the primary inground humic substances. Variations of traparticle transport mechanism, thus
such models are distinguished accord- promoting increased interest in surface
ing to the means for expressing the in- diffusion models (39-43).
Surface diffusion supposes that an adtraparticle mass transport step. The
sorbed
molecule migrates to an adjasimplest formulation for describing diffusion with microporous adsorbents is cent adsorption site on a carbon pore
the linear driving force approximation wall surface, a transport motivated by
the surface concentration gradient and
(36,37j:
available sorption energy. Several different
solutions to the film-surface difhi
= - (C, - C,,) =kSi(qsi
- qi);
fusion model for liquid-solid separa6t
0
tions have been presented (11,38, 44,
i = l , 2 , 3, . . . ,N
(11) 45)*
Where p is the bulk density of the adOne example of a two-resistance
sorbent and C, and qSiare the solid and model that has evolved from efforts to
interphase concentrations. This model simulate and predict adsorption procassumes that equilibrium exists at the esses for both CMBR and FBR systems
fluid-particle interface and that the is the homogeneous surface diffusion
mass transfer coefficients for the liquid (HSD) version of the Michigan Adand solid phases, kLand k,, respectively, sorption Design and Applications
are constant and independent of the Model, MADAM (4652). Figure 4
depicts a schematic conceptualization
transfer rates of competing species.
A more descriptive and complex of this model for an FBR adsorber (5).
analysis of intraparticle diffusion is The results of wide-scale testing of the
given by the pore transport model. Ad- MADAM-HSD and several subsesorbate molecules are assumed to dif- quent refinements thereof suggest that
the model provides a reasonable strucfuse in the pore voids and adsorb on
pore surface sites. The migration of ad- ture for simulating adsorption procsorbed solutes can occur only by de- esses for water and waste treatment
systems.
sorption followed by pore transport to a
MADAM-HSD: development and
new site, The corresponding material
balance equation expresses intraparticle application. The MADAM-HSD
sorbate concentration as a function of equations derived from mass balance
the radial dimension of the adsorbent equations for the fluid and solid phases
in an FBR adsorber are given below in
particle; for spherical coordinates:

averaged over the particle radius and

included in the intraparticle diffusion
coeficient;
dilute solution conditions prevail,
hence no bulk flow term is required
in the solid-phase material balance
expression;
diffusion coefficients are not concentration-dependent;
only longitudinal concentration gradients exist in the FBR (Le., no channeling); and
longitudinal dispersion is negligible
over the length of the column (53).
In the earliest applications of this
model, surface diffusion coefficients
were determined by a statistical search
of single-solute CMBR rate data for
both single-component and multicomponent systems. Film diffusion coeficients were estimated by empirical correlation techniques (54).
The HSD model and its variations
have been applied to a rather broad
range of singlecomoonent and multicomponent solutions comprising priority pollutants and unspecified background dissolved organic matter
(DOM). The implementation and important features of the HSD model have
been summarized (55,56) with the goal
of compiling a simplified, user-oriented
guide to designing pilot-scale adsorbers
for either CMBR or FBR systems.
Successful application of the model
has been well established for adsorption
and desorption in single-solute systems
of target organics in backgrounds containing little or no DOM (57-62) and in
waters containing only DOM (63, 64).
These applications have involved various adsorbent types and particle sizes,
including cases of variable particle size
distributions (2, 65).
The HSD model also has been shown
to handle transient organic loadings for
several one- and two-solute systems
(31, 66). Adequate prediction of breakthrough profiles for competing adsorbates, including chromatographic displacement phenomena, has been
demonstrated for several applications
(2, 3, 9,31, 62). The model has been
less successful in predicting effluent
concentrationsfor two situations: when
a multicomponent solution contains
compounds that exhibit markedly different adsorptive behaviors, or when
significant interactions occur within a
complex background of unknown composition (3, 6, 62). Increasing the complexity of the intraparticle diffusion
term has resulted in improved model
performance in some cases (52, 67,
68),but such modifications also involve
additional parameter evaluations and
analytical requirements.
A variety of other diffusion-controlled mass transport models have
been developed to better describe the

dimensionalized spherical-coordinate
form for singlecomponent systems.
Liquid-phase continuity equation:

=O
r=o

(I8)

where C and q are fluid- and solidphase concentrations of adsorbate, respectively; v, is the interstitial fluid velocity in the direction, z, of flow; f is
time; Dh is the hydrodynamic dispersion coefficient; p s is the apparent particle density; and e is the void fraction of
the bed.
Solid-phase continuity equation:

where r is the radial distance from the

center of the carbon particle and D, is
the surface diffusion coefficient. Specification of boundary and initial conditions include introduction of a film
transfer coefficient, kp These conditions are then given by:

where C, is the fluid-phase concentration near the surface of a carbon partcle of radius, a,, and C, is the influent
concentration to a bed of depth L.
Assumptions associated with this
model include the following:
local equilibrium OCCUIS at the exterior carbon surface;
the labyrinth factor (a measure of
carbon particle pore configuration) is

Enviran. Sci. Technol.. Vol. 21, No. 11. 1987 1045

behavior of FBR adsorption systems in

environmental applications (11). These
include models that combine pore and
surface diffusion (21, 69) and models
that incorporate concentration-dependent diffusion coefficients (30, 70).

Modeling approaches for complex

mixtures
A number of different techniques
have been used for modeling the adsorption of organic contaminants in the
presence of background DOM. One
approach, which treats background
organic matter as an important but
uncharacterized local condition, experimentally evaluates model rate and equilibrium parameters for target compounds as system-specific coefficients
(57, 59).
The idea of implicitly incorporating
background effects in phenomenological coefficients has been applied to both
single-target compound systems and
more complex systems containing mixtures of several targeted organic contaminants (3, 57, 59). Although the
methodology can be generalized for adaptation to various waters and wastes,
the system-specific nature of the technique prohibits extrapolation of coefficient values from one case to another.
Several attempts at a more general
approach have involved lumping all target organics and DOM into a single parameter, such as TOC, DOC, or TOX
(total organic halogen), and modeling
the behavior of that parameter (64,71,
72). In a related method, which involves multicomponent modeling, each
significant target compound represents
a single solute; all other organics are
characterized and modeled in terms of
a lumped concentration parameter (57).
One rather unique approach to unspecified mixtures of adsorbing species
involves hypothesizing the nonspecific
matrix as a number of fictive or
pseudospecies. The individual isotherm
properties of these fictive species are
evaluated from lumped parameter data
and the IAST model (73, 74). The initial concentrations and single-solute
isotherm parameters for each pseudocomponent must be determined before
IAST can be implemented to describe
the overall isotherm and predict its
competitive impact on selected target
compounds.
In addition to experimental determination of the overall isotherm, a characteristic mixture isotherm of a tracer
substance (i.e., a known compound
added to the unknown mixture and detected by single-substance analysis)
helps to establish initial qualitative estimates of the unknown parameters for
each component, These parameters are
varied systematically within a nonlinear, statistical curve-fitting routine until
1046 Environ. Sci. Technol., Vol. 21, No. 11, 1987

deviations between experimental isotherm points and those computed by the

model are minimized.
The calculated parameters can then
be used for predicting breakthrough
profiles and competitive adsorption
equilibria for specific compounds in the
presence of the mixture, thus moving
toward the goal of formulating a methodology to accurately predict adsorption of target contaminants from an unknown solution containing many
organic species.
This work has served as a forerunner
to related research on the development
of species-grouping methodologies for
mixtures containing a large number of
known solutes. In one development,
species grouping is predicated upon
Freundlich isotherm parameters for individual solutes, and a simplified expression for spreading pressure is employed in IAST model calculations
(75). Multicomponent solutions are reduced by this grouping to a system of a
fewer number of pseudospecies characterized by average adsorption parameters, with adsorbing solute concentrations measured in terms of a surrogate
parameter such as TOC.
This work has been extended by development of a search procedure for
calculating the pseudospecies composition and adsorption equilibrium parameters from total adsorbate equilibrium
concentration data (measured as TOC)
obtained from CMBR experiments
(76). Attempts to apply species grouping to the simplification of FBR multicomponent adsorption calculations
have met with less than complete success (77),possibly because of the simplicity of the dynamic model employed
(Le., a linear driving force relationship
similar to that given by Equation 11);
the lack of inclusion of competitive
chromatographic displacement effects;
and the fact that the criteria used for
species grouping were largely based on
only a sorption capacity parameter
(i.e., the preexponential coefficient in
the Freundlich model, Equation 1).
In this latter regard, it must be recognized that sorption energy or intensity,
as reflected by the exponential coefficient of the Freundlich model, distinguishes the equilibrium adsorption
characteristics of a compound as significantly as does the capacity coefficient.
Moreover, compounds that exhibit similar equilibrium characteristics may
very well have vastly different mass
transport properties.
In subsequent attempts to address
these shortcomings, the species-grouping criteria have been modified to
include requirements that adsorbates
of the same pseudospecies have
Freundlich exponent values that did not
differ by more than 20% and that the

intraparticle diffusion coefficients must

be of the same order of magnitude ( 78).
Even with these tighter grouping criteria, the approach has yet to demonostrate a capability for describing the
competitive displacement of specific,
weakly adsorbing compounds in FBR
systems.
A further refinement of attempts to
describe the adsorption behavior of
complex mixtures in terms of
Freundlich isotherm relationships and
coefficients for pseudo- or theoretical
components involves calibration of
these parameters with respect to a
known weakly adsorbing tracer compound. In this technique the tracer compound is either added to the mixture or
is an existing component that is singled
out for particular analysis. The theoretical component parameters are computed by matching the displacement of
the isotherm for the tracer compound to
its corresponding single-solute isotherm. The ability of the method to
simulate composite isotherms for the
components of background TOC and
total organic chlorine (TOX) has been
demonstrated (79).
Theoretical evidence has also been
given that the tracer compound principle and computed equilibrium parameters for pseudocomponents may be applied to determination of effective
solid-phase diffusivities for the same
pseudospecies (80). However, no experimental verification of this latter aspect of the tracer compound modeling
approach has been presented. Moreover, work on this approach has for the
most part been restricted to mixtures
containing only low-molecular-weight
micropollutants exhibiting similar
chemical structures.
It is clear from the foregoing that our
state-of-the-art ability to predictively
model the adsorption characteristics of
the individual components of complex
mixtures reflects at least as much art as
it does science. Nonetheless, work such
as that described is gradually enhancing
our ability to phenomenologically characterize the overall behavior of complex systems.

Model synopsis
Recent developments in adsorption
models have led to more sophisticated
and mechanistically more correct
structures and codes. The fact remains,
however, that it is still unrealistic for
most water and waste applications to
use models for design of prototypes directly from first principles. This is especially the case for complex waters
containing mixtures of target compounds and unspecified background organic matter, such as humic substances.
Although small-scale refinements of
existing models will continue to im-

prove their computational efficienc:

and predictive capability, it appears tha.
the mathematical description and numerical solution of dynamic models has
essentially reached a sufficient level of
sophistication, at least temporarily (7,
82). That is to say, the mechanistic and
mathematical refinement of theoretically oriented model efforts has outstripped our ability to adequately quantify all of the input parameters required
by these models.
For the present, priority challenges
in dynamic modeling involve improving the reliability of parameter estimation techniques and evaluating and enhancing the capability of existing
models to accommodate the complexities encountered in field applications.
As concept and practice converge,
feedback from such efforts will ultimately justify further theoretical sophistication.

Parameter estimation
The majority of the models currently
used to simulate and predict adsorption
processes for water and waste treatment
systems are phenomenological in the
sense that many of the reaction mechanisms in such systems, and the impacts
of solid-solute and solute-solvent interactions on these mechanisms, are incorporated implicitly into the model coefficients. This approach provides a
workable balance between simplistic
pilot-scale testing and ideal predictive
model development, which involves independent characterization and experimental verification of each reaction and
associated mechanism. The latter a p
proach is generally impractical for
engineering applications that deal with
the complexities of field conditions,
whereas the former approach does not
provide a logistically efficient means
for formulating and testing hypotheses
to explain how the dynamics of contaminants are controlled on a macroscopic
scale.
The ability of a phenomenological
model to describe and/or predict contaminant behavior in a given system is
ultimately vested in the accuracy with
which model input parameters are determined and how well the parameters
relate to conditions anticipated at full
scale. For cases of negligible dispersion, two-resistance models such as the
MADAM-HSD (as given by Equations
1 6 2 2 ) require estimation of a film diffusion coefficient, a surface diffusion
coefficient, and appropriate isotherm
parameters for each contaminant to be
modeled.
The film mass transport coefficient,
kl, is often calculated from one of a
number of semi-empirical correlations.
Each of these correlations is distinguished by a functional relationship be-

Literature correlations for determination of external mass transfer

coefficients
c.rnlalID"

Williamwn et al. (54)

lor

Ranpe(s) of VI

2.4 c R e O ~ ~ W ~ ' 2

c Re < 125
: Sc < 1300

Expr-Ion

Wilson and Geankoplis 1.09 c-mR'RSc'n

950

(83)

Ohashi et al. (84j

2 + 1.58 Re"' Sc'"

1.21

.:

Wfl

2 + 0.59 Reo.eSc'"

Gnielinski (85).

+ (ShL2 + Sh '."I

[*[I + 1.5(1 - e?
Kataokaetal. (86)
1.85 [(I - c)/c]'" Re'" Sc'"
Dwivedi-Upadhyay(87) (Ik)[0.765(~Re)~.'~
+ 0.365
( C R ~ ) ~S.C~"' ~~ ]

< Sc < 70.000

< R e < 5.8(stokes

.u...masO.Wl
. region)
&#$;,$2
,__

< Re < 500 (transition

region)
Re > 500 (Newton
1 < Re < 10,000
0.0 < Sc < 10,OM
Relr/(l - 4 < 101
0.01 < Re < 15,O
5.8

*Where Sh, = 0 . W Re'n Sc'" and Sh, = (0.037 Reo8Sc) I [I 2.443 W' ( S P-1)l.
Land Tsubsnim refer lo laminar and lurbulenlflow, rebpeclively.

tween the dimensionless Reynolds

(Re), Schmidt (Sc), and Sherwood (Sh)
numbers, the latter being related to the
film diffusion coefficient by:
Sh = k@/D[
where d is the adsorbent particle diameter, DL is the bulk liquid diffusivity of
the adsorbate and Re and Sc are defined
as:
Re = v,d/v
Sc = v/DL

where v, is the interstitial flow through

the bed, and vis the kinematic viscosity
of the fluid. Values of DL can be estimated from the relationship (82):

where p is absolute viscosity, X and M

are associated solvent parameters, and
V, is the molar volume of the solute at
its normal boiling point.
Table 1 contains a sample listing of
hydrodynamic correlations from the literature (54, 83-87). It is important to
recognize that calculations of free liquid diffusivity and of liquid diffusion
from semi-empirical correlations such
as these do not take into account interactions between the target contaminants
and background matrices. Moreover,
the equations presented in Table 1 have
been developed using materials that are
substantially different in chemical and
physical character than typical microporous adsorbents such as activated
carbon.
A number of studies have demonstrated that the surface topography and
roughness of an adsorbent can have an

impact on film-controlled mass transfer

rates (58,85,88). Further, no standardized criteria have been established for
determining which correlation may be
best suited to a particular system, other
than restrictions stipulated according to
hydrodynamic conditions. Caution
must therefore be exercised in applying
generic correlation techniques to determination of specific mass transfer coefficients, particularly for the complex
mixtures typical of water and waste
treatment applications.
Traditional dynamic modeling a p
proaches have utilized klvalues that are
determined from correlation procedures in conjunction with intraparticle
(e.g., surface) diffusion coefficients obtained from model calibrations to
CMBR rate data. Such studies are relatively easy to perform and do not necessitate the large volumes of solution
required for column studies. If the ultimate objective is to predict the performance of FBR systems, however, it
must be recognized that CMBRs do not
approximate the hydrodynamic and
contaminant removal patterns of
column reactors.
In a CMBR, for instance, the concentration gradient decreases more rapidly
with respect to the carbon surface (at
least in the initial stages of operation)
than it does in a column, for which the
influent concentration of adsorhate@)is
normally maintained at a relatively
constant level. Effective intraparticle
diffusion coefficients may also differ
between the two types of reactor systems, particularly if an adsorbate displays a hysteresis pattern with respect
to the sorption-desorption process.
These considerations are particularly
significant for mnlticomponent systems
and systems in which interactions between target compounds and backEnviron. Sei. Technol., MI. 21.NO. 11. 1987 1047

ground matrices are operative, especially for cases of nonideal competition

among adsorbates. For these reasons,
certain approaches may be suspect if
they employ D, values obtained from
single-solute rate studies for application
in multicomponent modeling exercises,
The differential batch reactor has
been utilized by several investigators as
a prospective improvement on the
standard CMBR apparatus (30, 89).
Although this system more accurately
approximates the hydrodynamics of an
FBR, it suffers from the same deficiencies as the CMBR with respect to simulating the concentration and hysteresis
profiles of adsorbates in flow-through
column-type systems.
Conversely, the short-bed adsorber
(SBA) technique (60, 61) enables determination of values for both kf and 0,in
an experimental reactor that more
closely approximates the configuration
of practical adsorption systems. The
SBA is an adsorber column sufficiently
short in length to allow incipient breakthrough of the compounds being studied, yet still exhibiting hydrodynamic
properties comparable to those of a
deep-bed adsorber.
The design of an SBA is such that the
initial stage of breakthrough is dominated by film transfer, regardless of the
type of overall mass transport control
that will eventually predominate. This
enables independent determination of
the film transfer coefficient followed by
calculation of the surface difisivity using a search-regression routine over the
entire breakthrough profile. In this way
both kf and D, are calibrated from the
same experimental data set using the
HSD model, thus circumventing the
need for correlation procedures and the
potential for error compounding that is
involved when CMBR systems and correlation techniques are used to determine the two rate parameters.
Investigations using the SBA/
MADAM-HSD approach have demonstrated good to excellent agreement between predicted and experimental FBR
breakthrough profiles in deep beds for
single-solute and bisolute systems of
typical organic contaminants (9, 60,
61). Descriptions of dynamic multicomponent hysteresis and displacement
phenomena have been obtained for
nearly every case studied. The potential
of the SBA technique for mass transfer
parameter determination in systems
containing complex mixtures of organics has been demonstrated (3, 57, 64).
Model isotherm parameters typically
have been obtained from equilibrium
experiments conducted by the bottlepoint technique in CMBR systems. A
substantial volume of data can be generated with relative ease by this
method. However, just as CMBR data
1048 Environ. Sci. Technol., Vol. 21,

No. 11, 1987

may not provide the most accurate basis

for estimating rate parameters for FBR
modeling, so the bottle-point method
may not provide the best estimate of
equilibrium relationships; this again by
virtue of the potentially irreversible
hysteresis patterns exhibited by certain
adsorbates (34) and the different approaches to multicomponent equilibria
that occur in CMBR and FBR adsorbers.
In an FBR, for example, the highest
solution-phase concentration of adsorbate encountered by an adsorbent is
equivalent to the equilibrium concentration at exhaustion. To achieve the
same equilibrium concentration in a
CMBR, the initial concentration must
be higher than that applied to the
column. Thus different concentration
gradient patterns result for the two
types of reactor systems.
As for kinetic studies, the impact of
such variables on estimated values ,for
the effective equilibrium capacities and
model coefficients is no doubt proportional to the number of adsorbing species in solution. Several investigations
have revealed dissimilarities between
effective capacities determined by
CMBR and FBR techniques (2, 90).
Efforts are currently underway to develop a standardized methodology for
column determinations of adsorption
equilibria (2).

Conclusions and future directions

The ultimate engineering objective of
process modeling is to provide a vehicle for efficient and cost-effective design. Although significant progress toward this objective has been made in
the development of adsorption models
and parameter estimation procedures,
more must be accomplished before
specifications for prototype systems can
be generated directly by modeling efforts alone.
The role of experimental testing and
empirical observation in the design of
specific systems remains as important
today as it has ever been, although an
improved understanding of the process
now enables more focused test programs and more informed interpretation of experimental information. This
latter fact is attributable to advances
that have been made toward a parallel
objective of process modeling; to wit,
the development of more rigorous
mechanistic interpretations of process
dynamics through continuously refined
integration of experimental observations with theoretical concepts and
model assumptions. Theory and practice are thus converging to benefit both
objectives of process modeling.
In considering avenues for future
progress, several deficiencies in our
current Understanding of the process

and in our abilities to represent it in

precise mathematical form emerge
from the foregoing discussion. With
few exceptions, the approaches described have not sufficiently emphasized interactions and impacts effected
by natural background organic matter
on the adsorption of specifically targeted organic contaminants in complex
multicomponent systems. Investigations to discern potential relationships
between experimentally determined coefficients and critical system properties
could prove particularly useful in addressing this deficiency.
Variations in background solution
characteristics can alter the adsorption
equilibrium and mass transfer behavior
of target species. Modeling approaches
can be structured to reflect these alterations in isotherm and kinetic model
coefficients. As lumped-parameter
techniques and phenomenological approaches for characterizing the physical
and chemical properties of unspecified
backgrounds are developed and refined, it may be possible to correlate
these parameters with model coefficients to predict adsorber performance
relative to perturbations in background
water composition.
Measureable solution parameters to
be examined in this regard would logically include pH, ionic strength, the
presence of multivalent cations, and the
character and concentration of background dissolved organic matter. In the
latter regard, background organic species characterizations may have to be
done phenomenologically, in terms of
such measures as apparent molecular
weight distributions and polarityhydrophobicity.
Column methods for determination
of isotherm parameters to be used in
FBR adsorber modeling should be
more extensively tested and analyzed.
Small, bench-scale column methodologies have been developed for singleand dual-component systems of target
organics in relatively simple background waters. Evaluation of both
equilibrium and rate parameters from
the same experimental design may
eliminate potential problems created
when CMBRs are used to estimate coefficients for subsequent column predictions.
Bench-scale column techniques could
enable such complex features of fullscale FBR systems as reactor hydrodynamics, influent conditions, time- variant concentration patterns, and
hysteresis phenomena to be more
closely simulated and their effects captured in evaluation of effective adsorption capacities and rates. It is likely that
these factors exert impacts on such
measurements in direct proportion to
the number of relevant components

present, and certainly the difficulty of

otherwise explicitly accounting for
such impacts so increases. It is anticipated that the use of smaller experimental designs will also permit greater flexibility in expanding ranges of system
conditions that can be examined easily
(e:g., system flow rate, bed depth, and
initial conditions) and thus effect savings of time and material in the design
process by extending model calibration
and verification capabilities at bench
scale.
In all of this must be appreciated the
possibility that no single model, regardless of how modified or manipulated,
may be universally applicable. Perhaps
a more reasonable objective for future
modeling efforts, at least in the near
term, is the structuring of a general
modeling methodology that incorporates decision-making capabilities
based upon a set of well-defined preliminary observations and feedback refinement ability.
Modeling approaches that are system-specific in nature appear to be warranted for most practical applications in
view of the variable solution and adsorption characteristics of target compounds when associated with complex
background waters. Such approaches
can accommodate both the immediate
need for establishing economical and
accurate scale-up specifications for
field-scale design as well as the longer
term scientific objective of investigating fundamental relationships between
model and system variables.

Acknowledgment
This article has been reviewed for suitability as a critical review by Francis A. DiGiano, University of N o r t h Carolina,
Chapel Hill, N . C . 27514 and by Gordon
McKay, The Queens University of Belfast,
Northern Ireland. U.K.

References
(1) Weber, W. J . , Jr.; Smith, E. H. Environ.
Sei. Technol. 1986. 20. 970-79.
(2) Weber, W. J., Jr.; Wang, C. K., Environ.
Sci. Technol., 1987,21, 1096-1102.
(3) Smith, E. H.; Weber. W. J.. Jr.. submitted
for publication in Environ. Sei. Technol.
(4) Smith, E. H.; Weber, W. J . , Jr., Evaluation of Mass Transfer Parameters for GAC
Adsorption of Complex Organic Mixtures,
Environ. Sei. Technol., in preparation.
(5) Weber, W. J . , Jr. J . Environ. Eng.
(ASCE) 1984, 110, 899-917.
(6) Mathews, A. P.; Su, C. A. Environ. Progress 1983, 2, 257-61.
(7) Ruthven, D. M. In Proceedings, First lnternational Conference on the Fundamentals
of Adsorption; Myers, A. L.; Belfort, G.,
Eds.; Engineering Foundation and AIChE:
New York, 1984; pp. 739-41.
(8) Crittenden, J. C. et al. J . Amer. Water
Works Assoc. 1987, 79 (l), 74-82.
(9) Liang, S.; Weber, W. J . , Jr. Chem. Eng.
Commun. 1985, 35, 49-61.
(10) Balzli, M. W.; Liapis, A. I.; Rippin,
D.W.T. Trans. Inst. Chem. Eng. 1978, 56,
145-56.
(11) Liang, S . Ph.D. Dissertation, University of Michigan, Ann Arbor, 1984.

(12) Weber, W. J . , Jr. Keynote Address at the

IOA-NIWR International Conference on the
Use of Ozone and Activated Carbon in Water-Wastewater Treatment, Pretoria, S.A.,
March 1984.
(13) Van Vliet, B. M.; Weber, W. J., Jr.; Hozumi, H. Water Research 1980, 14, 171928.
(14) Yen, C. Y.; Singer, P. C. J . Environ.
Eng. (ASCE) 1984, 110,976-89.
(15) Fritz, W.; Schlunder, E. V. Chem. Eng.
Sci. 1974, 29, 1279-82.
(16) Sheindorf, C . M . ; Rebhun, M . ;
Sheintch, M. J . Colloid & Interface Science
1981, 79, 136-42.
(17) Mvers. A. L.: Prausnitz. J. M. J .
AlChE1965, 11, 121-27.
(18) Radke, C. J.; Prausnitz, J . M. J. AIChE
1972, 18. 761-68.
(19) Yonge, D. R.; Keinath, T. M. J. Water
Poll. Control Fed. 1986, 58, 77-81.
(20) Tien, C. Chem. Eng. Commun. 1986,
40,265-79.
(21) Crittenden, J. C . ; Berrigan, J. K . ;
Hand, D. W. J . Water Poll. Control Fed.
1986,58,312-19.
(22) Kidnay, A. J.; Myers, A. L. J . AIChE
1966, 12, 981-86.
(23) DiGiano, E A , ; Baldaulf, G.; Frick, B.;
Sontheimer, H. Chem. Eng. Sei. 1978, 33,
1667-73.
(24) DiGiano, F. A.; Baldaulf, G.; Frick, B.;
Sontheimer, H. In Activated Carbon Adsorption of Organics from the Aqueous Phase;
Suffet, I. H.; McGuire, M. J., Eds.; Ann
Arbor Science: Ann Arbor, Mich., 1980;
Vol. I, pp. 213-28.
(25) Baldaulf, G.; Frick, B.; Sontheimer, H.
Wasser 1977, 49, 315-30.
(26) Frick, B. Extension of Simplified BiSolute Model for ADDlication in ComDlex
Mixtures; internal &port of the Institute
for Water Chemistry, Karlsruhe University,
F.R.G., 1977.
(27) Fritz. W.: Schlunder. E. V. Chem. Enp.
Sci. 1981; 36,721-30.
(28) Singer, I? C.; Yen, C. Y. In Activated
Carbon-Adsorption of Organics from the
Aqueous Phase; Suffet, 1. H . ; McGuire, M.
J., Eds. Ann Arbor Science: Ann Arbor,
Mich., 1980; Vol. I, pp. 167-89.
(29) Crittenden, J. C. et al. Environ. Set.
Technol. 1985, 19, 1037-43.
(30) Noll, K. E. et al. In Proceedings, First
International Conference on the Fundamentals ofAdsorption; Myers, A. L.; Belfort, G.
Eds.; Engineering Foundation and AIChE:
New York, 1984; pp. 411-20.
(31) Thacker, W. E . ; Crittenden, J. C . ;
Snoeyink, V. L. J . Water Poll. Control Fed.
1984,56,243-50.
(32) Crittenden, J. C. et al. Multicomponent Adsorption of Volatile Organic Chemicals onto Granular Activated Carbon, written communication, Michigan Tech.
University, Houghton, 1984.
(33) Fritz, W.; Merk, W.; Schlunder, E. U.;
Sontheimer, H. In Activated Carbon Adsorption of Organics from the Aqueous Phase;
Suffet, I. H . ; McGuire, M. J., Eds.; Ann
Arbor Science: Ann Arbor, Mich., 1980;
Vol. I, pp. 193-211.
(34) Yonge, D. R. Ph.D. Dissertation, Clemson University, Clemson, S.C., 1982.
(35) Summers, R. S. Ph.D. Dissertation,
Stanford University, Palo Alto, Calif., 1986.
(36) Weber, W. J., Jr. Physicochemical Processes for Water Quality Control;Wiley: New
York, 1972; pp. 219-29.
(37) Hsieh, J.S.C.; hrian, R. M.; Tien, C.
J . AlChE 1977, 23, 263-75.
(38) Mansour, A . ; von Rosenberg, D. U.;
Sylvester, N. D. J. AlChE 1982,28, 765-72.
(39) DiGiano, E A , ; Weber, W. J., Jr. J . San.
Eng. Div. (ASCE) 1972, 98, 1021-37.
(40) DiGiano, F. A.; Weber, W. J . , Jr. J. Water Poll. Control Fed. 1973, 45, 713-25.
(41) Masamune, S.; Smith, J. M. J . AIChE
1965, 11, 34-40.
(42) Furusawa, T.; Smith, J. M. Ind. Eng.
Chem. Fund. 1973, 12, 197-203.
Y

(43) Komiyama, H.; Smith, J. M. J. AlChE

1974, 20, 1110-17
(44) Rosen, J. B. J . Chem. Phys. 1952, 20,
387-94.
(45) Rosen, J . B. Ind. and Eng. Chem. 1954,
46, 1590-94.
(46) Mathews, A. P.; Weber, W. J . , Jr. AIChE
Symposium Series 1977, 73, 91-98.
(47) Mathews, A. P.; Weber, W. J . , Jr. Chem.
Eng. Commun. 1984,25, 157-71.
(48) Weber, W. J., Jr.; Crittenden, J . C. J .
Water Poll. Control Fed. 1975, 47, 924-40.
(49) Crittenden, J. C.; Weber, W. J., Jr. J .
Env. Eng. Div. (ASCE) 1978, 104, 185-97.
(50) Crittenden, J. C.; Weber, W. J., Jr. J .
Env. Eng. Div. (ASCE) 1978, 104,433-43.
(51) Crittenden, J. C . ; Weber, W. J., Jr. J .
Env. Eng. Div. (ASCE) 1978,104, 1175-95.
(52) Weber, W. J., Jr.; Liang, S . Environ.
Progress 1983, 2, 167-75.
(53) Crittenden, J. C. Ph.D. Dissertation,
University of Michigan, Ann Arbor, 1976.
(54) Williamson, J. E.; Bazaire, K. E.;
Geankoulis. C. J. lnd. and E. C. Fund.
1963, 2: 126-29.
(55) Hand, D. W.; Crittenden, J . C . ;
Thacker. W. E. J . Environ. Ena. ( A X E ) .
1983, 109, 82-101.
(56) Hand, D. W.; Crittenden, J . C . ;
Thacker, W. E. J . Environ. Eng. (ASCE),
1984, 110,440-56.
(57) Smith, E. H.; Tseng, S . ; Weber, W. J.,
Jr. Environ. Progress 1987, 6, 18-25.
(58) Van Vliet, B. M.; Weber, W. J., Jr. J .
Water Poll. Control Fed. 1981, 53, 1585-98.
(59) Weber, W. J., Jr.; Pirbazari, M. J. Amer.
Water Works Assoc. 1982, 74, 203-09.
(60) Weber, W. J., Jr.; Liu, K. T. Chem. Eng.
Commun. 1980, 6,49-60.
(61) Liu, K. T.; Weber, W. J., Jr. J . Water
Poll. Control Fed. 1981, 53, 1541-50.
(62) Crittenden, J. C . ; Hand, D. W. In Proceedings, First International Conference on
the Fundamentals of Adsorption; Myers, A.
L.; Belfort, G . , Eds.; Engineering Foundation and AIChE: New York, 1984, pp. 18594.
(63) Lee, M. C. et al. J . Env. Eng. Div.
(ASCE) 1983, 109, 631-45.
(64) Endicott, D. D.; Weber, W. J., Jr. Environ. Progress 1985, 4, 105-1 1.
(65) Mathews, A. F? In Proceedings, First International Conference on the Fundamentals
of Adsorption; Myers, A. L.; Belfort, G.,
Eds.; Engineering Foundation and AIChE:
New York. 1984. uu. 345-54.
(66) Crittenden, J:LC. et al. J. Water Poll.
Control Fed. 1980, 52, 2780-95.
(67) Weber, W. J.. Jr. In Proceedinas. First
International Conference on the Fuidamentals of Adsorption; Myers, A. L.; Belfort,
G., Eds.; Engineering Foundation and
AIChE: New York. 1984. DD. 679-92.
(68) Merk, W.; Fritz, W.;Schlunder, E. U.
Chem. Eng. Sei. 1980, 36, 743-57.
(69) Friedman, G. M.S. Thesis. Michiean
Technological University, Houghton, 1982.
(70) Sladek, K. J.; Gilliland, E. R.; Baddour,
R. F. Ind. Eng. Chem. Fund. 1974, 13, 10005.
(71) Kaastrup, E.; Summers, R. S . ; Roberts,
P. V. In Proceedings of the 1984 Specialty
Conference, Environmental Engineering Division; Pirbazari, M.; Devinny, J. S . , Eds.;
ASCE: New York, 1985; pp. 109-17.
(72) Summers, R. S . ; Roberts, P. V. J . Environ. Eng. (ASCE) 1984, 110, 73-92.
(73) Frick, B; Sontheimer, H. In Treatment
of Water by Granular Activated Carbon;
McGuire, M. J . ; Suffet, I. H.,Eds.; American Chemical Society: Washington, D . C . ,
1983; pp. 247-68.
(74) Fettig, J . ; Sontheimer, H. J . Environ.
Engr. (ASCE) 1987,113, 795-810.
(75) Calligaris, M. B.; Tien, C. Canadian J .
Chem. Eng. 1982, 60, 772-80.
(76) Jayaraj, K.; Tien, C. Ind. and Chem.
Proc. Des. and Dev. 1985,24, 1230-39.
(77) Tien, C. In Proceedings, First International Conference on the Fundamentals of
Adsorption; Myers, A. L.; Belfort G., Eds.;
-

Snviron. Sci. Technol.,VOI. 21, NO. 1 1 , 1987 1049

Keep current on government activities

as they aflecf the#eld ofchemistry with

WASHINGTON

Engineering Foundation and AIChE. New

York. 1984; pp. 647-56.
(78) Ramaswami. S.: Tien, C. Ind. ond
C h m . Proe. Des. and De". 1986. 25, 13339.
(79) Critlenden. J. C.; Luft. P.; Hand. D. W.
Water Research 1985. 19, 1537-48.
(80) Critrenden. 1. C.; Lufi. P.; Hand, D.
W.; Friedman, G. "Predicting the Removal
ofKnown Organic Compounds in Unknown
Mixtures and Total Organic Halogen i n
Fixed-Bed Adsorbers." writlen communication. Michigan Technological University.
Houghton. 1984.
(81) Weber, W. J.. Jr. In Proceedings. First
Imemnrional Conference on the FundomenI d s of Adsorprim; Myers. A. L.: Belfort,

This i n f o m i a t i v c Iii-wer?kl.v p ~ h l i ~ ~ a lCiI ~ W

i nI S Fedorid acIivitic!s as thev
c o n c e r n science in g e n e r a - a n d d i e m i s t i v i n p a r t i c u l a r . I t gives timely
i n f o m i a t i o n on: Programs to be Funded * New Regulations * Committee
Hearings,Activities * Proposed Legislation Reference Studies.
Publications* And much more!

Based on i n f o i m a t i n n Iimught diirctly firim Capitol Hill Iiy ACS

Washington Alert siitll r n c n i l i e i s a n d p u h l i s h e d hy the A m e r i c a n Cheniicxl
Socir:t.v's I l e p a i l n t e n t

____________________________________

of G c w e r n n l m t

Relations ant1 S c i e n c e
Policy, t h i s is p r i r n a y
journalism at its rnosf
,finctiona/-iiirect.
relevant. accurate
rO1'
infc,,ma~ion,
use the ~:ou~""l.
unite, or

G., Eds.; Engineering Foundation and

AIChE: New York. 1984; pp. 745-48.
(82) Wilke, C. R.; Chang. P. J. AIChE 1955.

.._".

I 7M-7n

(83) Wilson. E. 1.; Geankoplis, C. I . Ind.

Eng. Chem. Fund. 1966.5. 9-12.
(84) Oharhi, H. e l al. I . Chon. Eng. Japan
1981, 14.433-38.
( 8 5 ) Roberls. P. V ; Cornel. F?; Summers.R.
S. J . Environ. Eng. 1985. 111. 891-905.
(86) Kataoka. T.: Yoshida, H.; Ueyama, K.
J. Chem. Eng. Japon 1972.5, 132-36.
(87) Dwivedi. P N.: Upadhyay, S. N. Ind.
Eng. Chem. Proc. Des. ondDev. IW7.16,
157-65.
(88) Young, B. D.; van Vliel. B. M.. submitted far publication in In:. J. H m Moss

An,rrC:itn Chmlical Socic1,v

Peviotlicitls Marketing I)epatlmcnl
1155 sixl*?,?,,tl,StIPCt. W\'
Washinglon. I)(:
ZUO36

I
I

TRZ"SfW.

(89) Yonge. D. R., et al. Environ. Sci. k h nol. 1985. 19,690-94.

(90)Aguwa. A. A. Ph.D. Dissertation. lllinot5 Institute ofTechnology. Chicago. 1984.

YES! P l e a s e send nie niore i n f o m i a t i o n on the

I
I

tii-wrx:kl.v p u h l i c a t i o n , ACS Washington Alert

city. StillC..

xi,,

Dependon DRIERITE.

. ... .

the d e s i c c a n t for y o u r laboratory drying n e e d s

.-.-,

,\ "

DRlERlTE
Indicating DRlERlTE
* laboratory Gas Drying Unit
* Drying Tubes
Technical Advice
Write or call for free catalog

See our catalog in

*=

W. A. Hammond DRlERlTE Co.

P. 0. Box 460
Xenia, Ohio 45385
51 3-376-2927

1050 Environ. Sci. Technal.. Val. 21. NO. 11. 1987

Walter J . Weber, Jr., (1) is the Eornesr

Bo>?(TDisrinruished Professor of Enginerring and direrror of Environmental and
Worrr Resources Engineering ar the Universify of Michigan. He holds engineering
degrees from Brown. Rurgers. and Harvard universities. His research and teaching activiries focus on rhe characrerizarion
and modeling of physicochemical processes in narural and engineered sysrems.
He is widelv recog;rized for his work on
research and development of adsorption
processes in environmenral engineering
applicarions.

L ? d d H. S d h (r) is a research associate in the Universiry of Michigan :T Environmental and Water Resources Engineering Program. He holds bachelnr:~and
master? degrees in civil en8ineering from
the Universiry of Delaware, and a Ph.D. in
environmenral engineering from rhe Universify of Michigan. His principal areas of
research include rhe adsoplion of organic
and inorganic conraminants by activated
carbon, and rhe applicarion of adsorprion
processes ro deconraminarion of surface
and subsurface wafer supplies.