&EPA
United States
Environmental Protection
Agency
Industrial Environmental Research
Laboratory
Research Triangle Park NC 27711
EPA-600/2-79-195
October 1979
Research and Development
Hyperfiltration Processes
for Treatment
and Renovation
of Textile Wastewater
-------
RESEARCH REPORTING SERIES
Research reports of the Office of Research and Development, U.S. Environmental
Protection Agency, have been grouped into nine series. These nine broad cate-
gories were established to facilitate further development and application of en-
vironmental technology. Elimination of traditional grouping was consciously
planned to foster technology transfer and a maximum interface in related fields.
The nine series are:
1. Environmental Health Effects Research
2. Environmental Protection Technology
3. Ecological Research
4. Environmental Monitoring
5. Socioeconomic Environmental Studies
6. Scientific and Technical Assessment Reports (STAR)
7. Interagency Energy-Environment Research and Development
8. "Special" Reports
9. Miscellaneous Reports
This report has been assigned to the ENVIRONMENTAL PROTECTION TECH-
NOLOGY series. This series describes research performed to develop and dem-
onstrate instrumentation, equipment, and methodology to repair or prevent en-
vironmental degradation from point and non-point sources of pollution. This work
provides the new or improved technology required for the control and treatment
of pollution sources to meet environmental quality standards.
EPA REVIEW NOTICE
This report has been reviewed by the U.S. Environmental Protection Agency, and
approved for publication. Approval does not signify that the contents necessarily
reflect the views and policy of the Agency, nor does mention of trade names or
commercial products constitute endorsement or recommendation for use.
This document is available to the public through the National Technical Informa-
tion Service, Springfield, Virginia 22161.
-------
EPA-600/2-79-195
October 1979
Hyperfiltration Processes for Treatment
and Renovation of Textile Wastewater
by
S. M. Ko and J. A. Tevepaugh
Lockheed Missiles and Space Co.
Huntsville Research and Engineering Center
Huntsville, AL 35807
Contract No. 68-02-2614
Task No. 009
Program Element No. 1BB-610
EPA Project Officer: Max Samfield
Industrial Environmental Research Laboratory
Office of Environmental Engineering and Technology
Research Triangle Park, NC 27711
Prepared for
U.S. ENVIRONMENTAL PROTECTION AGENCY
Office of Research and Development
Washington, DC 20460
-------
ABSTRACT
A computer program for design and simulation of a multi-
stage hyperfiltration system for renovation of textile waste-
water has been developed. The program is capable of practical
design, parametric simulation and cost projection of the multi-
stage hyperfiltration system with tapered innerstages. The
mathematical model is formulated based on Sourirajan's prefer-
ential sorption and solute diffusion theory. Experimental
rejection and flux data of a test hyperfiltration module are
required as input parameters. Empirical correlations and test
results available from recent EPA-sponsored programs are uti-
lized to calculate membrane transport parameters. Computed
results for sample cases using cellulose acetate and dynamic
membranes are presented. Various design and operating para-
meters are considered in the numerical computations to show
effects of these parameters on economics of the system. This
simulation program has been developed in a general manner and
is readily adaptable for evaluation of other R0*/hyperfiltration
applications.
* Reverse Osmosis
11
-------
This report was submitted in fulfillment of Contract
68-02-2614, Task 009, by the Lockheed Missile § Space
Company Huntsville Research § Engineering Center under the
sponsorship of the U.S. Environmental Protection Agency. This
report covers the period November 19, 1978 to August 31, 1979
and work was completed as of August 31, 1979.
111
-------
TABLE OF CONTENTS
Abstract
11
Acknowledgements
1. Introduction ^
2. Conclusions 4
3. Theory of Hyperfiltration 6
4. Transport Parameters 12
5. Membrane Systems Model 18
6. Parametric Results 38
References
Appendices
A. Experimental Rejection/Flux Data ....
B. "Prediction of Osmosis Membrane Separation
Efficiencies for Solutes in Dilute Aqueous
Solutions
C. Computer Code Listing j
48
08
IV
-------
FIGURES
Number Page
1. Concentration profiles of water and
solute for steady state operation of
reverse osmosis process 7
2. A tubular hyperfiltration module ... 20
3. Concentration profile of the
differential hyperfiltration module
under steady state 26
3b. Summary of hyperfiltration model ... 27
4. Schematic diagrams of tapered multistage
hyperfiltration systems 34
5. Flow chart of the three-stage hyper-
filtration design program 36
6. Performance simulation of a single
tubular hyperfiltration module .... 39
7. Effect of design product rate factor and
rejection on unit cost 43
8. Effect of operating temperature and
pressure on unit cost 44
v
-------
TABLES
Number
Summary of Computer Design 40
Simulation Results for Two
Sample Hyperfiltration Systems
of One Million Gallons Per Day
Capacity
VI
-------
LIST OF SYMBOLS
A pure water permeability of membrane
A constant
o
B constant
C molar concentration of solution
Ci solvent concentration in membrane
C? concentration of water in solution on low
pressure side
G£ molar concentration of boundary solution
AC2 solute concentration gradient across
membrane
C, molar concentration of product solution
AC. difference in solute concentration in feed
and product solutions
CAI molar concentration of solute in bulk
molar concentration of solute in product
CAM solute concentration in membrane
D diffusivity
DI solvent diffusion coefficient
D? solute diffusion coefficient
DAB diffusivity of solute in feed solution
diffusivity of solute in membrane phase
f Fanning friction factor
F flow rate
VII
-------
g conversion factor
J-L solvent flux
^2 solute flux
K mass transfer coefficient
1 effective thickness of concentrated
boundary layer
L channel length
L^ solute permeability
Lg filtration coefficient
m^ molarity of feed solution
mj molarity of product solution
M average molecular weight of solution
MA molecular weight of component
Mg molecular weight of water
NA solute flux
Ng solvent flux
P bulk pressure
AP difference in bulk pressure across
membrane
PR product rate
PWP pure water permeability
Q permeability
r tube radius
rj rejection
R gas constant
Vlll
-------
R Reynolds number
"
S solute transport parameter
S. surface area
Sc Schmidt number
Sh Sherwood number
t channel thickness
T temperature
TCM mass transfer coefficient of solute
u average axial velocity
v permeation velocity
V1 partial molar volume of water
x axial distance
XA2 mole fraction of solute in concentrated
boundary solution on high pressure side of
membrane
X», mole fraction of solute in product solution
a constant
B constant
6 solubility parameter
-------
ACKNOWLEDGEMENTS
The authors are grateful to Dr. J.L. Gaddis of Clemson
University and Dr. C.A. Brandon of Carre, Inc., for their
contribution to this study, in particular, their comments
on the mathematical model. The EPA project officer was
Dr. Max Samfield and his support and interest in this study
is gratefully acknowledged.
-------
1. INTRODUCTION
Hyperfiltration (also termed reverse osmosis) as a textile
wastewater treatment and renovation process has been studied
under EPA/IERL/RTP* sponsorship to investigate the technical
feasibility and the economic practicability of the separation
process. A recent investigation (Ref.l) of a pilot scale hyper-
filtration facility at LaFrance Industries, a division of Riegel
Textile Corporation, successfully demonstrated the feasibility
of dynamic hyperfiltration membranes for the in-plant recycle
and reuse of composite textile dyeing and finishing wastewater.
The applicability of the concept to a variety of composite
textile wastewaters has been confirmed in a more recent study
(Ref.2) at eight different textile mills encompassing eight
different subcategories of the textile mills point source
category (Ref.3). The scope of these studies, however, was
limited to testing and evaluating a few commercial membranes
using plant composite wastewater and the mixed dyehouse effluent.
Furthermore, optimization of process parameters in regard to more
favorable economics, energy conservation, byproduct recovery and
effluent control of the process system were not investigated in
detail.
* Environmental Protection Agency. Industrial Environmental
Research Laboratory, Research Triangle Park, N.C.
-------
Textile finishing wastewater contains a variety of chemi-
cals depending upon the particular dyeing and finishing opera-
tions. The type of chemicals in the wastewater greatly affects
the performance of the membranes. The effectiveness of mem-
branes is also sensitive to the temperature of the wastewater
being treated. In the interest of energy recovery, high temp-
erature operation of the process is desirable. Thus, a para-
metric investigation of the separation process directed toward
efficient and cost effective design of the treatment system is
required to provide the optimum performance and design informa-
tion essential in the development of a full-scale system. Since
experimental investigations of these parameters are not practical,
such a parametric study necessitates a theoretical analysis of
fundamental mechanisms involved in the hyperfiltration process
and the development of a computer model of the process system.
Unfortunately, however, direct extension of general reverse
osmosis theory to mixed solute systems is complicated due to
nonavailability of pertinent fundamental physiochemical proper-
ties of both membranes and solutes, and possible complex inter-
actions of chemicals in such a system. Thus, the transport
properties of solutes in the composite wastewater must be calcu-
lated from experimental flux and rejection data. Where such
data are not available, these transport parameters may be esti-
mated using empirical correlations (Refs.4 and 5).
In the following sections a discussion of hyperfiltration
-------
theory and associated transport parameters is followed by a
description of a computer model for design and simulation of a
multi-stage hyperfiltration system. Results of a parametric
study of the multi-stage system using the computer model are
presented and significant conclusions summarized. In Appendix
A, experimental membrane rejection and flux data are presented in
tabular and graphic form. A report authored by Mr. A. Schindler
of Research Triangle Institute on osmosis membrane separation
efficiencies is included as Appendix B. A listing of the com-
puter code is contained in Appendix C.
-------
2. CONCLUSIONS
Governing equations describing the hyperfiltration of solvents
and solutes can be derived from phenomenological considerations.
These equations can be used to predict the performance of actual
membranes provided experimental data and correlations are available
for determining membrane transport parameters. These parameters
are membrane and solute dependent.
A mathematical model formulated from Sourirajan's preferential
sorption and solute diffusion theory is adequate for predicting
membrane performance. The computer code provides reasonable design
and simulation results (pressure, flow rate, rejection, recovery
factor, concentration polarization) for a single module with mul-
tiple innerstages using both cellulose acetate and dynamic mem-
branes. Given accurate cost information, the economic model real-
istically depicts the impact of various design parameters on unit
cost.
The computer model is capable of predicting system perfor-
mance as well as analyzing system economics to find an optimum set
of design parameters. Reliability of the computer results is
largely dependent on the availability of rigorous cost information
-------
for the system as well as the accuracy of the test module data and
membrane specifications.
-------
3. THEORY OF HYPERFILTRATION
Phenomenology
Hyperfiltration (reverse osmosis) is a separation process
which utilizes the selective sorbability of the solvent from a
solution by semipermeable, microporous membranes. Figure 1 shows
a schematic of steady state concentration profiles of the solvent
(water) and the solute across the membrane. Due to relatively
high affinity of the membrane to solvent molecules, a concentra-
tion gradient is established in the interface region. Adsorbed
pure water molecules are permeated through the microporous struc-
ture of the membrane at a rate which is determined by the charac-
teristics of the membrane and the pressure exerted to overcome
the osmotic pressure.
Solute, on the other hand, diffuses to both directions from
the interface. The selective permeation of water molecules will
develope high solute concentration in the vicinity of the inter-
face and thus develop a maximum concentration. This provides a
driving force for the solute diffusion in both directions from
the interface.
-------
Concentration
Interface
Water
Concentration
Profile-
Solute
Concentration
Profile
Bulk Feed Solution Under
Operating Gauge Pressure
Concentrated Boundary
Solution
Preferentially Sorbed
Interfacial Region
Dense Microporous
Membrane -Surface
. Spongy Porous Membrane
Product Solution at
Atmospheric Pressure
Figure 1. Concentration profiles of water and solute for steady state
operation of reverse osmosis process.
-------
It is necessary to define a quantity to express the overall degree
of permeation in general, regardless of the actual transport
mechanism. Permeability may be defined in terms of concentration
or pressures in the general forjn,
where
Q = permeability
S^ = area
F = flow rate
The concentration gradients of solutes and the solution pressure
are important forces primarily responsible for inducing fluxes of
solute and solvent.
Each flux can be associated with the various forces using a
finite number of the linear phenomenological coefficients. Dif-
fusion models are generally one of two types, diffusion or pore.
The diffusion model is very convenient, particularly in the case
where the molecular size of a chemical species is of the same
order as that of another species from which it is to be separated.
If the molecular sizes are quite different as in ultrafiItration,
a porous membrane with a suitable average pore size is commonly
chosen to effect the separation (pore model). The membrane is
viewed as a homogeneous medium with a finite thickness. When
the membrane is associated with a solution, the membrane phase
itself is considered as a solution in which the "component" of the
membrane segment is very sluggish and almost stationary. Thus,
8
-------
Three transport coefficients are involved in the transport
process. The relative magnitude of these coefficients will de-
termine the shape of concentration gradients and the efficiency
of the hyperfiltration process. They are:
• Water permeability constant through the membrane
• Solute transport coefficient through the membrane
• Mass transfer coefficient of the solute in the solution
These transport coefficients can be obtained from a set of ex-
perimental measurements of the product water rate, pure water
permeability of the membrane (normally given by the membrane
manufacturer) and solute separation efficiency. The mass trans-
fer coefficient of the solute can also be obtained from empirical
correlations published in the literature.
Preferential Sorption/Diffusion Theory
Permeation is a phenomena in which a species or component
is passing through another substance, usually but not necessarily
by means of diffusion. In fact, permeation is a phenomenological
definition which encompasses a variety of transport mechanisms.
Driving forces which cause permeation include:
• concentration gradient
• pressure gradient
• electric potential
• temperature gradient
-------
the well established solution theory can be applied directly to
the system.
Nonequilibrium Flow Model
Katchalsky and Curran (Ref.6) developed equations governing
membrane permeability from phenomenological considerations of
nonequilibrium flow. The solvent flux N?, was determined to be,
NB = LB (Ap-aAir)
and similarly the solute flux, N., was determined to be,
where,
L_ = filtration coefficient (represents velocity of
fluid per unit pressure difference)
L = solute permeability (measured at NR = 0)
o = Staverman reflection coefficient (measure of
membrane selectivity depending on properties
of both the membrane and solute)
AP = difference in bulk pressure across the membrane
Air = difference in osmotic pressure across the membrane
= solute concentration in membrane
AC. = diiference in solute concentration in feed and
product solutions
10
-------
Substituting for NR, the solute flux may be written,
N = CL ^PCl-o) -aAir(l-o)] + L^A
A AMB
For the special case of a nonselective membrane (^ = 0) , the solute
flux is,
NA ' CAMVP * VCA
For the special case of an ideal semipermeable membrane, o=l,
and the solute flux reduces to the expression
NA ' LAACA
The coefficients, L , Lfi, and a can be determined from ex-
perimental measurements. Thus, membrane permeability can be cal-
culated from phenomenological considerations provided the neces-
sary constants can be determined experimentally.
11
-------
4. TRANSPORT PARAMETERS
Two parameters have been suggested by Sourirajan for char-
acterizing membranes. These parameters are the pure water per-
meability constant, A, and a mass transfer coefficient, D^/Re.
The parameter, A, is a measure of the overall porosity of a film
and is independent of any solute. The parameter, DAM/K6, is a
mass transfer coefficient with respect to solute transfer.
Permeability
The parameter, A, is pressure dependent according to the
relationship,
A = Ao exp (- aP)
where A0 is A at P=0 and a is a constant. The parameter,
A (gm-moles-water/cm^-sec-atm), may be further defined,
PWP NR
A = = !
MBxSAx3600xP P- MXA )+ "(XA )
L* *J
where ,
PWP = pure water permeability (gm/hr)
M = molecular weight of water (gin/mole)
12
-------
S = surface area of membrane (cm")
r\
P = operating pressure (atm)
NL = solvent water flux (gm-mole/cm-sec)
B
TT = osmotic pressure (atmj
X = mole fraction of solute in concentrated boundary
solution on high pressure side of membrane
XA, = mole fraction of solute in product solution
f\ J
the term, NR, may be obtained from the following equation,
1
= (PR)
1-
(l-f)MA
Mg-S-3600
where ,
(PR) = product rate (gm/hr-cm2)
IR-, = molarity of feed solution (moles of solute/1000 gm
water, m]_ = ppm/1000MA where MA is the solute
molecular weight)
f = separation = 1 --- = rejection
ml
m, = molarity of product solution
Values for the terms, N_ and X. at different pressures are not
D AT
usually available for commercial membranes. In order to obtain
approximate values for A, the osmotic pressure, f (XA ) was
•D
assumed zero and the osmotic pressure, " (X ) was assumed equal
A2
to ^(X ), the i smotic pressure of the feed solution.
Al
13
-------
Solute Mass Transfer Coefficient
The parameter, D^/Kd, is pressure dependent according to
the relationship,
DAM/K6 aP
where g is a constant. For many solutes, D«,,/KS is independent
of feed concentration and feed flow rate at any given operating
pressure. The solute mass transfer coefficient is further de-
fined ,
D
AM
N
B
K6
I-XA-
where,
D
AM
= diffusivity of solute in membrane phase (ff/sec)
K = mass transfer coefficient (.cm/sec)
6 = solubility parameter
C? = molar density of boundary solution (gnTmole/cnrM
Cj = molar density of product solution (gm'mole/cm^)
Values for the term, C-,, at different pressures are not usually
available for commercial membranes. In order to obtain approxi-
mate values for DAM/K 6, it was assumed that C2 is equal to the
molar density of pure water, X. 7 is equal to X.,, and the mole
i\ £ f\ -*•
fractions, XA, are,
-6 M
X. = ppmxlO ^
( M )
MA
14
-------
Mass Transfer Coefficient
The mass transfer coefficient of the solute in solution is
given by the expression,
TCM =
ShD
2r
where,
Sh = Sherwood number (mass diffusivity/molecular
diffusivity)
D = diffusivity
r = tube radius
Sherwood number depends on membrane geometry and flow conditions,
i.e., whether the flow is turbulent or laminar. The flow is
considered turbulent for Reynolds numbers greater than 2100
(based on hydraulic diameter). For turbulent flow, the Sherwood
number is assumed independent of axial distance from the tube
entrance and for a tubular membrane is calculated with an ex-
pression developed by Gill and Sherwood,
7/81/4 ' 60 l1n'5
,,
Sh = O.lSRe Sc
where,
Re = Reynolds number
Sc = Schmidt number
0.127
.875
Re
and for a sheet membrane is calculated with an expression de-
veloped by Linton and Sherwood,
15
-------
7/12 Sct 0.333
Sh = 0.44Re ( )
2L
t = channel thickness
L = channel length
For laminar flow, the Sherwood numher is a function of axial
distance and for a tubular membrane is calculated with an ex-
pression developed by Lebeque and modified by Sourirajan,
ReScZr °'3"
Sh = 1.95 ( )
x
and for a sheet membrane,
ReScZt °'"3
Sh = 2.24 ( )
x
where x is the axial distance from the tube entrance.
Macroscopic Parameters
Several macroscopic parameters are required in order to
define the membrane transport. The diffusion model for membrane
transport assumes that Pick's law is obeyed and that uncoupled
flow occurs (the flow of one component is unaffected by the
flow of other components within the membrane I. The solvent flux,
J,, is given by the expression,
D1C1V1
J = —±~—i— (AP-A n )
1 RTX
16
-------
where, C, = solvent concentration in membrane
D = solvent diffusion coefficient
Vi = partial molar volume of water
AP = bulk pressure difference across the membrane
ATT = osmotic pressure difference across the membrane
R = gas constant
T = temperature
\ = mass transfer coefficient on high pressure side
divided by the solute mass transfer coefficient
The solute flux, J , is given by the expression,
-D2k(AC,)
where,
k = distribution coefficient tg/cnT3 of solute in
the membrane divided by the g/cnP of solute in
the surrounding solution)
D2 = solute diffusion coefficient
AC9 = solute concentration gradient across the membrane
Performance of the membrane is also characterised by the rejec-
tion, r j ,
D2k
1
+
D,C1V1 f A P- ATT )
where,
C* = concentration of water in solution on the low
1 pressure side of the membrane
The utilization of the preceding parameters in the membrane
systems model is described in the following chapter.
17
-------
5. MEMBRANE SYSTEMS MODEL
In the following paragraphs a mathematical model of hyperfiltra-
tion processes is developed. The model includes a mathematical
description of membrane behavior and the economics of a hyper-
filtration system. A description is presented of the hyperfil-
tration system to be simulated and the computer code developed
from the mathematical model.
Mathematical Model
The model described herein is based on Sourirajan's prefer-
ential sorption and solute diffusion theory (Ref.7) which applies
simple boundary film theory (Refs. 8 and 9) to obtain the con-
centration polarization of the solute (Ref. 10) . The effect of
pressure drop due to friction losses and momentum changes on the
performance of the system (Refs. 11 and 12) is included in the
model.
Membrane model: The model described herein is based on Sourirajan's
preferential sorption and solute diffusion theory (Ref. 1 ) which
applies simple boundary film theory (Refs. 2 and 3) to obtain
18
-------
the concentration polarization of the solute (Ref.4). The effect
of pressure drop due to friction losses and momentum changes on
the performance of the system (Refs.5 and 7) is included in the
model.
Consider a tubular hyperfiltration module shown in Fig. 2.
From the overall material balance for the differential control
volume, the change in average axial velocity (u) can be written
as
.du Z_ (5.la)
dx ~ r w
where v is the permeation velocity of product water through the
w
membrane in ft/sec. Similarly, the solute material balance can
be written as , ._CA,» ->
d (u Al) Z_ v r
to " r w °A3 (5.2a)
where C and C are the molar densities of solute in the hulk
Al A3
and the product, respectively.
Now, an expression for the pressure drop in a tube in terms
of friction losses and momentum changes may be obtained from an
energy balance (Ref. 5).
u2f u_ du. (5.3a)
7 + gc
.UJL , 2 Vw 5 (5.3b)
gcr gcr
or substituting (5.]a)yields
19
-------
where C is the molar density of solution (Ib mole/ft3), M is the
average molecular weight of solution, f is the Fanning friction
factor, and gc is the conversion factor.
Membrane
\
Product
Water
Feed
P'CA1.U
\ \
w
dx
t t
Reject
'A 2
\ I
Product
Water
Fig. 2 - A Tubular Hyperfiltration Module
20
-------
By solving Eqs. ' 5J.a) and ( 5. 3h)simul taneous ly, one can com-
pute output variables, u, P and CA] for the differential volume
for a given set of dependent variables, prod .> f rate i v i and
solute transport (C.-). These dependent v;in thles car. he related
A J
to the feed rate (u), the system operating pressure iP and the
concentration of feed streani 'f^i' by transper4 mechanisms of
solute in the high pressure side of the membraru .
The transport of product water through The membrane (Refs.
1 and 4) is proportional to the effective opei.-iting pressure of
the system (see Fig. 2a),
N
B
[" - «lx - 'Ix |]
A 2 A 3 -J
4ai
where A is the pure water permeability of t;ie membrane, n is the
osmotic pressure at given solute concentrat; n .-inJ r i- the sv^-
tem operating pressure, NR is the product K.itei flu.x through the
membrane. XA2 and XA3 are the mole fraction- nf the solute at
the high pressure and product water sides of the membrane. l-.qua-
tion ( i.4a)can be written in terms of mole fraction by assuming
a linear dependency of osmotic pressure on mole fraction. This
is true for most dilute solutions.
ir = BX i 5. 4b)
where B is a constant with units of pressure. Substituting, we
have T B 1
NB = AP I1 - & (CA2 - CA3>J (3.40
21
-------
or dividing both sides of Eq. (5.4c)by C,
v - !!§. _ AP r B ,r ., n
w- C - C V ~ PC (CAZ ~ CA3}\ (5-4d)
The transport of the solute through the membrane is given
by (Ref. 1)
NA ' ft?) (5.5a)
where CA2* a"d CA3* are the molar densities of the solute in the
membrane phase in equilibrium with the respective solution phases,
"\
DAM is the diffusivity of solute in the membrane phase in ft /sec,
and &tis the effective thickness of membrane. The equilibrium
concentration can be written as
CA = KCA* (5.5b)
where K is a characteristic constant determined by the properties
of membrane and solute. Substituting (5.5b)into (5. 5a)yields
(5.5C)
. v
I K^ ) is the solute transport parameter (ft/sec) uniquely
determined by the characteristics of specific membrane/solute
combination.
Dividing ( 5. 5c)by ( 5. 4c)yields
22
-------
or
CA2 " CA3
D
Now, for a simultaneous solution of equations (5-4d)and fS.SD
one must have an expression for the solute concentration in the
boundary layer (CA2) . The net rate of solute transport to the
membrane within the thin boundary layer is given by (Refs. 1 and
4),
C
N -- A_3_ /N + N ) = Bulk flux — solute diffusion flux back
A C A B „ to the bulk
CA d A
—
*-'•*• UL» 1 A** .. I A 1 -r-\ I I - ' A ' ~ ' U 1 / — ,
;A3 (5'6al
with boundary conditions
CA = CA1 at y=0
CA = CAZ at y = ^
where DATD is the diffusivity of solute in the feed solution, and
AD
/ is the effective thickness of the concentrated boundary layer
Solving the differential equation ( 5. 6a )and defining the
mass transfer coefficient of solute in the feed solution,
(5.6bj
23
-------
we have
log.
'CAZ-CA3\ NA+NB
kC
(5.6c)
or solving for N.
NA= k
From ( 5. 5c)and( 5. 6d)we write
- C
CA2 - CA3
CA1 * CA3
CA , - C
A3
A3
Substituting( 5. 5d)into( 5- 6e)and solving for C., yields
where
CA = C q ,
Al A3
Al
(5.6d
(5.be
q = 1 +
y cA31 0
exP
CA3+ 9)
( 5.6f
y
9
B
CP
w
(5.6gJ
K6
v *= AP.
w C
The set of equations summarized in Fig. 3b can be solved
simultaneously to obtain output conditions of the differential
section for a given input (or initial) conditions at the inlet
of the differential section. By repeating the procedure one can
24
-------
compute the overall performance of the module.
To predict the overall performance of the hyperfiltration
system, the following input data are required:
1. Design specifications for the system, i.e., geometry of the
module,
2. Initial process conditions at the inlet of the system, i.e.,
flow rate, feed concentration, pressure, temperature and pH,
3. Physical properties of feed stream, i.e., viscosity and
osmotic pressure (IT) as functions of solute concentration
and temperature,
4. An empirical expression for the mass transfer coefficient
(k) as a function of Reynolds number, arid Schmidt number
(Section 4),
5. Membrane parameters:
a. Permeability constant (A) as a function of pressure and
temperature, (Section 4), and
b. Solute transport parameter (DAM/K6) as a function of
pressure, temperature and pH (Section 4).
Items 1 and 2 are defined by the system design specifica-
tions. Item 3 may be obtained from published data in the litera-
ture. In the event when these data are not available from the
literature, experimental measurements are necessary. Item 4 can
be obtained from the literature. Item 5, membrane parameters,
must be obtained through a series of experimental measurements.
Variables to be monitored in an experimental system for this
purpose are solute rejection, product rate and system pressure.
The experimental data then can be used to compute membrane para-
meters using the system of equations presented in Fig. 3b.
25
-------
Preferentially Sorbed
Interfacial Region
Concentrated Boundary
Solution
Bulk Feed Solution
Under Operating
Pressure
c
o
in
O
cu
fl)
«
7^=*
x *->
'A3 /
A 2
/
3A3
Concentration
Product Solution at
Atmospheric Pressure
Spongy Porous Membrane
or Membrane Support
Dense Microporous
Membrane Surface
Fig. 3- Concentration Profile of the Differential Hyperfiltration
Module Under Steady State (c^2 and CA \ denote solute
concentrations in the membrane phase in equilibrium
with respective concentrations in the liquid phases.)
26
-------
List of Equations
.
dx r
dx
• CM
e r
ec
2 Vw u
re
v = v
A3
CA! = CA3 '+^
J exp (-
X =
* B
Fig.Sb- Summary of the Hyperfiltration Model
27
-------
Consider a tubular hyperf iltration module shown in Fig. 1.
From the overall material balance for the differential control
volume, the change in average axial velocity (uj can be written a;
du" 2
dx- ~- 7 vw
where v^ is the permeation velocity of product water (permeate
through the membrane (ft/sec). Similarly, the solute material
balance can be written as
_ Al) 2
cfx ~ r vv A3 (S.2a)
where C^, and C^ are the molar concentrations of solute in the
bulk and the product, respectively.
Now, an expression for the pressure drop in a tube in terms
of friction losses and momentum changes may be obtained from an
energy balance (Ref. 11).
dp u2f u du
- j — = CM - + - -= —
dx g(,r gc dx
or substituting (1) yields
£ -2V™U (5.3a,
where C is the molar density of solution (Ib mole/ft ), M is the
average molecular weight of solution, f is the Fanning friction
factor, and g is the conversion factor.
By solving Eqs. (5. la), ( 5. 2a) and (5.3a) simultaneously, one
can compute output variables, u, P and CAI for the differential
28
-------
volume for a given set of dependent variables, product rate (v )
W
and solute transport (CA3) • These dependent variables can be
related to the feed rate (u) , the system operating pressure (P)
and the concentration of feed stream (CA1) by transport mechan-
isms of solute and permeate (product water) through the membrane
and associated concentration polarization of solute in the high
pressure side of the membrane. To obtain the expressions for vv
and CA7, let us consider a differential section of the tubular
f\ O
hyperfiltration module shown in Fig.3a. The figure shows a
schematic of steady state concentration profile of the solute
across the membrane. Due to relatively high affinity of the mem-
brane to solvent (water) molecules, a concentration gradient is
established in the interface region. Adsorbed water molecules
are permeated through the microporous structure of the membrane
at a rate which is determined by the characteristics of the mem-
brane and the pressure exerted to overcome the osmotic pressure.
Solute, on the other hand, diffuses to both directions from the
interface. The preferential sorption of water molecule will
develop high solute concentration right in the vicinity of the
interface and thus develop a maximum concentration. This pro-
vides a driving force for the solute diffusion to both directions
from the interface.
The transport of product water through the membrane (Refs.
7 and 10) is proportional to the effective operating pressure of
the system. The effective pressure is defined to he the differ-
ence between the operating pressure and the osmotic pressure
29
-------
exerted by the concentrated boundary layer. Assuming a linear
dependency of osmotic pressure on mole fraction, we obtain the
following expression. Detailed derivation of the expression is
available in Ref. 13.
N
B AP f B
Vw = — = ~C~ L1 " PC (CA2 ' CA
NB is the permeate water flux through the membrane (lh mole/ft 2
-sec), C is the molar concentration of the bulk solution (lb-mole/
ft ), A is the pure water permeability of the membrane (Ib mole/
ft -sec-psi), P is the operating pressure (psi), and B is a pro-
portionality constant for osmotic pressure (psi)*.
The expression for CA3 can be obtained by integrating the
differential equation obtained from a differential solute material
balance within the concentrated boundary layer. A detailed- deriv-
ation and solution of the boundary equation are presented in Ref.
13. The resulting expression is:
C
B/CP + CS/AP |" k/S (C B/CP + SC/AP)
S is the solute transport parameter* (ft/sec) and k is the mass
transfer coefficient of the solute in the solution phase (ft/sec)
Note that the constant B is a very small number for organic
solutes found in textile wastewaters. For such cases, the
osmotic pressure term can be neglected and the flux can be
written simply as: v = AP/C.
*S is used to denote the solute transport parameter which is rep-
resented by (DAM/K6) by Sourirajan (Ref. 7). DAM is the diffu-
sivity of solute in the membrane phase (ft2/sec). K is a charac-
teristic constant which represents preferential sorbability of
the solute on a given membrane surface. S is uniquely determined
by the characteristics of specific membrane/ solute combination.
30
-------
The set of equations, (5.la) - (5.3a), (5.4d), and (5.7),
can be numerically solved to obtain output conditions from the
differential section of the module for a given input (or initial)
conditions at the inlet of the differential section. By re-
peating the procedure one can compute the overall performance of
the module.
The relative magnitude of the three transport coefficients
v,ill determine the shape of concentration profile and the
efficiency of the hyperfiltration process. The transport coef-
ficient can be obtained from a set of experimental measurements
of the product water rate, pure water permeability of the mem-
brane (normally given by the membrane manufacturer) and solute
rejection efficiency. These parameters are temperature and
pressure dependent. The dependencies can be expressed by the
following equations (Refs. 14 and 15.):
A oc exp(-aP)
S « P~P exp(-y/T)
a , 3 and T are constants, which can be calculated from experi-
mental results. Empirical correlations published in the litera-
ture can be used to calculate the solute mass transfer coeffi-
cient (Ref. 16).
31
-------
Economic model: The incremental cost for producing unit quan-
tity of permeate water is considered to analyze the system eco-
nomics. The cost elements contributing to the incremental product
cost are amortized capital cost, UCC ($/Kgal), and 0§M costs
($/Kgal). The amortized capital cost is calculated based on the
installed system capital cost per unit membrane area ($/ft2 of
membrane surface). The 0§M costs include membrane replacement
cost. UMRC ($/Kgal), pumping power cost, UPP ($/Kgal), and other
0§M costs, UMOMC ($/Kgal). The credits to the product cost are
credit from recovered water, CRW ($/Kgal), and credit from recov-
ered energy, CRE ($/Kgal). The credit from recovered chemicals
from the concentrate usually requires additional process modi-
f i cat ion<^.
The unit incremental cost for product uater, UCPW ($/Kgall,
is written as:
UCPW = UCC + UMRC + UPP + UMOMC - CRW - CRE (6)
The unit costs can be obtained from vendors or actual estimations.
Description of System and Method
In practical applications of the reverse osmosis process, a
certain multistage design concept is desirable to achieve desired
levels of product recovery and solute rejection. Three such con-
cepts with tapered inner stages are presented in Fig. 4.
32
-------
Note that the innerstages are tapered with respect to the number
of modules rather than the physical shape of the membrane support
structure. The tapered innerstage design increases efficiency
of the system since it reduces the concentration polarization
in the boundary layer. The feed rate at the inlet of each
module is maintained at the design value by reducing the number
of modules within an innerstage. The number of modules for an
innerstage is determined by the flow rate fed to the particular
innerstage. As the bulk flow rate decreases due to permeation
of water through the membrane, the level of polarization in a
module increases. Minimizing the concentration polarization
increases the efficiency of solute rejection as well
as the product rate. The rejection increases due to reduced
solute concentration in the boundary layer while higher effective
operating pressure is responsible for the improvement in the
product rate.
The single stage concept employs direct recycle of a portion
of the reject stream to concentrate the reject to the specified
design value. In this case, the system is operated at a some-
what higher concentration than the concentration of the feed. In
the two-stage concept shown in Fig. 4b, the first stage is used
as a purification stage, and the second as a concentration stage.
Since the permeate from the second stage has a higher concentra-
tion than the design product concentration, it is recycled to
the first stage. In the two-stage case, the concentration of
the combined feed to the first stage is lower than the incoming
33
-------
a. Single Stage
Recycle
Feed
1
r
n
y
r=i j=i
£r L^T
— i
j
^Reject
(Concentrate)
Permeate
b. Two Stage
Feed.
Recycle
Reject
(Com i • n •
trate)
1 'Permeate
c. Three Stage
Feed
Reject
(Concen-
t ratri
Permeate
Fig. 4- Schematic Diagrams of Tapered Multistage Hyperfiltration Syst
34
ems
-------
feed concentration to the hyperfiItration system. An additional
stage is employed in the three-stage concept. The purpose of
this stage is to concentrate the permeate from the second stage
hack to the feed concentration. This scheme provides higher
Astern efficiency because of the reduced numher .it' modiues
required for the system.
For a given design rejection and product recovery factor,
the best system efficiency can be obtained when the recycle flow
rate is the minimum and the concentration of the recycle stream
is the same as the feed stream. Numerical results indicated that
the three-stage concept reduces the total number of modules by
30 and 10 percent over the single and two-stage system, respec-
tively. For this reason, the three-stage concept is chosen as
the system model.
For the reasons discussed above the tapered innerstage
design is chosen for the system design purposes. Note that the
innerstages are tapered with respect to the number of modules
rather than the physical shape of the membrane support structure.
A schematic flow chart for the design simulation of the
three-stage tapered hyperfiltration system is shown in Fig.5.
A computer program developed to solve the system of equations
presented in the previous section was utilized to obtain design
and economic results for the staged system. The numerical
procedure for simulation of a single module employs an iterative
35
-------
( Start J
Input Module Geometry,
Operating Conditions and
Test Module Data.
Calculate Membrane
Parameters
Assume a Value for
Stage 1 Permeate
Water Concentration,
Calculate Performance
of Stage 1 Innerstages.
Calculate Performance
of Stage 2 Innerstages.
Calculate Performance
of Stage 3 Innerstages.
Calculate Total
Number of Modules
and Costs.
C End J
Fig. 5 - Flow Chart of the Three-Stage Hyperfiltration Design Program
(Subscripts P, R and F denote permeate, rejected concentrate,
and feed, respectively.)
36
-------
scheme based on the Newton-Raphson method over a number of
finite subsections of the module. A similar iterative procedure
is utilized to match the concentrations and flow rates to and
from each stage.
37
-------
6. PARAMETRIC RESULTS
The numerical results of a single module performance simula-
tion are shown in Fig. 6. The case shown in the figure simulates
a tubular Westinghouse module with a cellulose acetate membrane on
the inside tube wall. The operating conditions and the geometry
are indicated in the figure. The pressure drop for the 100 ft.
module is approximately 1 percent. The initial rejection is 96
percent and slightly decreases as the concentration polarization
increases by 8 percent. Approximately 10 percent of the feed i«
recovered as permeate water and the resulting decline of the tl
-------
£ 4.0
£ *""
Rejection
1.0
0,95
-
-
0.1
S °
> 4->
O U
U ft
4) [^
c
.2 O 74
4_J ._. &. t
rt ti
SM 2.3
2.2
_L
JL
20 40 60
Tube Length (ft)
80
100
Fig. 6 - Performance Simulation of a Single Tubular Hyperfiltration Module
(Feed Rate = 10 gpm, Temperature = 70 F, Tube Diameter = 1 in.,
Cellulose Acetate Membrane (by Westinghousc) with Flux r 4.75 x
10"5 ft/sec, Feed Concentration - 2.0 x 10'5 lb-molc/ft3)
39
-------
Table 1
SUMMARY OF COMPUTER DESIGN SIMULATION RESULTS FOR TWO
SAMPLE HYPERFILTRATION SYSTEMS OF ONE MILLION
GALLONS PER DAY CAPACITY
Item
Design Parameters
Test Tube Rejection
Test Tube Flux (ft/sec) ^
Permeability (Ib-mole/ft -sec-psi)
Solute Transport Parameter (ft/sec)
Design Tube Diameter (in.)
Design Tube Length (ft)
Design Product Recovery Factor
Design Rejection
Design Feed Concentration (Ib-mole/ft )
Design Temperature (F)
Design Pressure (psi)
Design Results
Number of Innerstages, Stage 1
Stage 2
Stage 3
Number of Modules, Stage 1
Stage 2
Stage 3
Total Number of Modules
Total Membrane Area (ft^)
Economic Results'""
Total Installed Capital Cost ($)
Capital Amortization Cost (cents/Kgal
Membrane Replacement Cost (cents/Kgal)
Pumping Power Cost (cents/Kgal)
Other O&M Costs (cents/Kgal)
Credit for Recovered Water (cents/Kgal)
Credit for Recovered Energy (cents/Kgal)
Total Unit Cost (cents/Kgal)
Cellulose
Acetate
Membrane
0.96
4.75 x 10 Q
2.07 x 10"^
1.23 x 10
1
100
0.8
0.95
2.0 x 10
70
600
19
1 1
42
846
213
204
1,263
31,727
539,350
24
79
15
10
40
0
88
ZR(IV)-PAA
Dynamic
Membrane
0.96 ,
11.3 x 10 "J
2.88 x 10"7
2.53 x 10"
1
100
0.8
0.95 ,
2.0 x 10
150
1,000
7
5
15
339
101
99
539
13,540
2,301,700
102
2
25
20
40
67
42
"The unit cost basis are obtained from a recent study estimation (Ref. 2). More
detailed cost breakdown is described in the reference
40
-------
for two sample one million gallons per day hyperfiltration systems.
One is a Cellulose Acetate (CA) membrane system (Westinghouse)
operating at 600 psi and 70 F. The other is a Zr(IV)-PAA (poly-
acrylic acid) dynamic membrane system (Selas) operating at 1000
psi and 150 F. These two membranes are selected because of readily
available design and cost information (Refs. 1 and 2) . The trans-
port parameters are calculated from experimentally measured rejec-
tion and flux data from the references. The two systems are de-
signed for a product recovery factor and rejection of 0.8 and 0.95,
respectively. The unit cost information was obtained from vendors
and recent study estimations (Refs. 1 and 2) .
The computed total numbers of modules of 1 in. in diameter
and 100 ft long for the CA and dynamic membranes are 1263 and 539,
respectively. The corresponding membrane areas required are 31,727
and 13,540 sq. ft. The higher surface area requirement for the CA
membrane system is due to the smaller permeability of water through
the membrane. The unit cost for producing a thousand gallons of
permeate water is 88 cents for the CA system and 42 cents for the
dynamic membrane system. It is noted that the major cost element
for the CA system is the membrane replacement cost, while the capi-
tal amortization cost makes the largest cost contribution. The
credit from recovered energy is significant enough to cause the
dynamic membrane to be attractive. It should be noted that the
design and operation conditions used are for illustration purposes
and do not represent a typical case.
41
-------
To demonstrate the range of applicability of the developed
model and to investigate the effects of various design parameters
on the economics of the system, a series of parametric studies wa:
performed. The results are shown in Figs. 7 and 8.
Figure 7 shows the effects of design product rate factor and
solute rejection on the unit cost. The effect of design product
rate factor on the unit cost is relatively moderate. The design
rejection factor has a more pronounced effect on the cost of the
dynamic membrane system. Significantly lower unit cost can be
obtained at slightly lower design rejection for the dynamic mem-
brane. On the other hand, the effect of design rejection is less
significant in the CA membrane. Thus, a proper combination of
these two membrane systems may be desirable when a higher design
rejection is necessary due to more stringent quality requirements
for reuse of the permeate water.
Figure 8 shows the effects of operating temperature on the
unit cost with operating pressure as a parameter. Due to charac-
teristics of the membranes, the effects are shown only for the
respective applicable operating temperature and pressure ranges.
It is shown that the unit cost is generally lower at higher temp-
erature and pressure. Unit cost is a stronger function of- temp-
erature and pressure than product recovery factor and rejection
(Fig.7).
The unit cost does not include credit from possible reuse of
the chemicals contained in the concentrated reject stream. A net
42
-------
1.4
l.Z
1.0
•o
o
w>
0.6
tn
O
'J
2 0.4
c
D
0.2
Cellulose Acetate Membrane (Westinghouse)
Zr(IV)-PAA Dynamic Membrane (Selas)
DREJ = Design Rejection DREJ
-DREJ
0.97^
°-95^^— -*
0.95
0.93
0.6 0.75 0.8 0.85
Design Product Rate Factor
0.9
Fig. 7 - Effect of Design Product Rate Factor and Rejection on Unit Cost
(Feed Rate = 10 gpm, Tube Diameter = 1 in., Temperature - 70 F
for CA Membrane and 150 F for Dynamic Membrane, Pressure =
600 psi for CA Membrane and 1000 psi for Dynamic Membrane)
43
-------
3.5
3.0
2.5
3
T3
O
nl
cm
1.5
CO
O
O
X 1.0
c
P
0.5
Cellulose Acetate Membrane (Westinghouse)
Zr(IV)-PAA Dynamic Membrane (Solas)
P = Operating Pressure (psi)
1000
60 80 100 120 140 160
Operating Temperature (F)
Fig. 8- Effect of Operating Temperature and Pressure on Unit Cost
(Feed Rate = 10 gpm. Tube Diameter = 1 in., Design Rejection
= 0.95, Design Product Rate Factor = 0.8)
44
-------
saving can be realized if the chemical credit is included in the
computation of the unit cost. Although high temperature and
pressure operation is desirable in the interest of achieving lower
unit cost or even net savings, it is limited by excessive main-
tenance costs at high operating temperature and pressure.
Membrane lifetime is diminished by operation at high temperatures
and very high pressures (several hundred atmospheres) thus in-
creasing maintenance costs. For the range of temperatures and
pressures investigated, a flat estimated average maintenance cost
was assumed. At higher temperatures and pressures, a functional
relationship between these operating variables and maintenance
cost should be developed and included in the computations.
45
-------
REFERENCES
1. Brandon, C.A., and J.J. Porter, "Hyperfiltration for Renovation
of Textile Finishing Plant Wastewater," EPA-600/2-76-060, March
1976.
2. Brandon, C.A., J.J. Porter and D.K. Todd, "Hyperfiltration for
Renovation of Composite Wastewater at Eight Textile Finishing
Plants, " EPA-600/2-78-047, March 1978.
3. Development Document for Effluent Limitations Guidelines and
New Source Performance Standards for the Textile Mills Point
Source Category," EPA-440/l-74-022-a, June 1974.
4. Shindler, A., Unpublished results developed under EPA/RTP spon-
sorship.
5. Spencer, H.G., and J.L. Gaddis, "Hyperfiltration of Nonelectro-
lytes: Dependence of Rejection on Solubility Parameters," Paper
presented at EPA Symposium on Textilw Industry Technology, 5-8
December 1978.
6. Katchalsky, A. and P.P. Curran, Nonequilibrium Thermodynamics
in Biophysics, Harvard University Press, Cambridge Massachu-
setts, 1965.
7. Sourirajan, S., Reverse Osmosis, Academic Press, New York,
1970, pp. 176-184.
8. Sourirajan, S., Reverse Osmosis, Academic Press, New York,
1970, pp. 185-188T
9. Brian, P.L.T., "Mass Transport in Reverse Osmosis," in Desalin-
ation by Reverse Osmosis, edited by Ulrich Merten, The MIT
Press, Cambridge,1971.
10. Pusch, W., "Concentration Polarization in Hyperfiltration
Systems," in Reverse Osmosis Membrane Research, edited by
H.K. Lonsdale and H.E. Podall, Plenum Press, New York, 1972,
pp. 43-57.
\
11. Griffith, W.L., R.M. Keller, and K.A. Kraus, "Parametric Study
of Hyperfiltration in Tubular Systems with High Permeability
Membranes," Desalination. Vol. 4, 1968, pp. 203-308.
46
-------
12. Johnson, J.S., Jr., L. Dresner and K.A. Kraus, "Hyperfiltra-
tion (Reverse Osmosis)," in Principles of Desalination, edited
by K.S. Speigler, Academic Press, New York, 1966, pp. 345-439.
13. Ko, S.M., and P.G. Grodzka, "Study of Hyperfiltration Proces-
ses for Treatment and Renovation of Textile Wastewater," to
be published.
14. Sourirajan, S., Reverse Osmosis, Academic Press, New York,
1970, pp. 191-ZOT;
15. Gaddis, J.L., Private communication dated 24 May 1978.
16. Hwang, S.T., and K. Kammermeyer, Membranes in Separations,
Wiley-Interscience, New York, 1975, pp. 351-359.
47
-------
Appendix A
EXPERIMENTAL REJECTION/FLUX DATA
The experimental rejection data presented in Tables 1-8 and
Figs. A-l through A-18 were obtained from: CARRE, Inc.,
"Compilation of Toxic Rejection Data for Membranes". Prepared
for EPA, December, 1977.
48
-------
Table 4-1
REJECTION OF SPECIFIC COMPOUNDS BY THE SELAS DYNAMIC Zr(IV)-PAA MEMBRANE
OPERATING ON TEXTILE WASTEWATER
Solution (underlined
solutes in consent
decree)
COD
BOD
TOC
Dissolved Solids
Total Solids
Volitile Solids
Color
Phenol
Iron
Nickel
Chromium
Zinc
Copper
Manganese
Rejection
(%)
71 - 99
74 - 99
82 - 98
62 - 99
63 - 96
70 - 99
89 - 100
86 - 100
97 - 99
80 - 98
89 - 99
94 - 99
92 - 99
90 - 98
pH
5.2 - 12
5.2 - 12
5.2 - 12
5.2 - 12
5.2 - 12
5.2 - 12
6.6 - 9.6
5.2 - 9.6
5.2 - 9.6
5.9 - 12
5.2 - 7.5
5.2 - 9.3
6.1 - 7.4
Concentration
(mg/1)
1600 - 7100
25 - 2300
175 - 2000
670 - 128000
2500 - 215000
370 - 2700
120 - 3400*
0.66 - 315
8.25 - 20
0.7 - 3.87
0.7 - 23
2.1 - 18
1.2 - 5.5
0.5 - 1.02
70-80%
1
3
0
2
2
1
1
0
0
0
0
0
0
Number
80-90%
7
6
6
8
6
1
2
0
3
1
0
0
0
of Data
90-100%
27
29
26
21
21
19
4
6
7
7
13
14
4
Total
35
38
32
32
40
29
21
7
6
10
8
13
14
4
*Concentration in Pt-Co Units
Data compiled from: Brandon, C. A., Porter, J. J., and Todd, D. K., "Hyperfiltration for
Renovation of Composite Wastewater at Eight Textile Finishing Plants."
Final Report, EPA Grant No. S802973, Clerason University report in
preparation.
-------
Table A-2
REJECTION OF SPECIFIC COMPOUNDS BY THE WESTINGHOUSE
TUBULAR CELLULOSE ACETATE MEMBRANE
OPERATING ON TEXTILE WASTE WATER
Solute (underlined
. solutes in consent
decree)
COD
BOD
TOG
Dissolved Solids
Volatile Solids
Color
Phenol
Iron
Nickel
Chromium
Zinc
Copper
Manganese
Rejection
<%)
89-99
87-99
82-96
82-97
90-98
91-98
9-99
96.5
_
-
_
-
67-99
PH
5.8-7.1
5.8-7.1
5.8-6.2
5.8-7.1
5.8-7.1
5.8-7.1
6.8-7.1
5.8
-
-
5.8-7.1
Concentration
(mg/1)
1765-8664
128-1800
345-1800
2804-6303
630-3504
220-2000*
0.54-0.65
8.48
—
-
0.12-1.34
70-80%
0
0
0
0
0
0
0
0
_
_
1
Number
80-90%
2
2
1
0
0
0
0
0
_
_
2
of Data
90-100%
6
6
4
3
5
5
1
1
_
—
1
TOTAL
8
8
5
3
5
5
2
1
_
_
5
Concentration in Pt-Co Units
Data compiled from:
Brandon, C. A., Porter, J. J., and Todd, D. K., "Hyperfiltration for
Renovation of Composite Wastewater at Eight Textile Finishing Plants."
Final Report, EPA Grant No. S802973, Clemson University report in
preparation.
-------
Table A-3
REJECTION OF SPECIFIC COMPOUNDS BY THE UOP
SPIRAL-WOUND CELLULOSE ACETATE MEMBRANE
OPERATING ON TEXTILE WASTE WATER
Solute (underlined
solutes in consent
decree)
COD
BOD
TOC
Dissolved Solids
Volatile Solids
Color
Phenol
Iron
Nickel
Chromium
Zinc
Copper
Marganese
*Concentration in Pt
Rejection
(%)
89-99
89-100
83-99
90-99
89-99
98-100
20-98
94-99
80-83
97
89-99
92-99
98-99
-Co Units.
PH
5.2-9.4
4.9-9.4
4.9-9.4
4.9-8.0
4.9-7.2
3.6-8.0
6.8-7.1
5.2-6.8
5.2-6.8
6.8
3.6-7.5
3.6-7.5
4.9-6.2
Concentration
(mg/1)
27-2303
500-8664
232-2000
670-12120
630-3504
65-2000*
0.54-35.0
8.25-12.0
1.02-1.20
3.6
3.02-18.3
1.78-5.51
0.47-1.32
70-80%
0
0
0
0
0
0
0
0
0
0
0
0
0
Number o
80-90%
2
1
3
0
1
0
1
0
2
0
1
0
0
f Data
90-100%
18
21
11
22
13
17
1
5
0
1
9
6
5
TOTAL
20
22
14
22
14
17
3
5
2
1
10
6
5
Data compiled from:
Brandon, C. A., Porter, J. J., and Todd, D. K., "Hyperfiltration for
Renovation of Composite Wastewater at Eight Textile Finishing Plants.
Final Report, EPA Grant No. S802973, Clemson University report in
preparation.
-------
Table A-4
REJECTION OF SPECIFIC COMPOUNDS BY THE DUPONT
MEMBRANE OPERATING ON TEXTILE WASTE WATER
Solute (underlined
solutes In consent
decree)
COD
BOD
TOC
Dissolved Solids
Volatile Solids
Color
Phenol
Iron
Nickel
Chromium
Zinc
Copper
Manganese
Rejection
(%)
98.3
98.4
97.2
99.2
96.7
99.1
85
-
95.7
_
*Concentration in Pt-Co Units.
PH
6.4
6.4
6.4
6.4
6.4
6.4
6.4
_
6.4
—
Concentration
(mg/1)
2776
1138
1058
8906
1950
1571*
1.0
_
0.92
Numbe
70-80%
0
0
0
0
0
0
0
^
0
r of Data
80-90% 90-100%
0
0
0
0
0
0
1
_
0
1
1
1
1
1
1
0
_
1
TOTAL
1
1
1
1
1
1
1
1
Data compiled from:
Brandon, C. A., Porter, J. J., and Todd, D. K., "Hyperfiltration for
Renovation of Composite Wastewater at Eight Textile Finishing Plants."
Final Report, EPA Grant No. S802973, Cletnson University report in
preparation.
-------
Table A-5
REJECTION OF SPECIFIC COMPOUNDS BY THE
UNION CARBIDE 3NJR MEMBRANE
OPERATING ON TEXTILE WASTE WATER
Solute (underlined
solute In consent
decree)
COD
BOD
TOC
Dissolved Solids
Volatile Solids
Color
Phenol
Iron
Nickel
Chromium
Zinc
Copper
Manganese
Rejection
(%)
mm
mm
15.2
7.3
55.7
/
61.0
5A.1
55.9
PH
_
-
8.1
8.1
8.1
8.1
8.1
8.1
Concentration
(mg/D
-
330
23853
804
0.172
1.6A
0.397
Number
70-80%
—
-
0
of Data
80-90%
-
-
0
90-100%
-
-
0
TOTAL
-
-
1
1
1
1
1
1
O«
*Concentration in Pt-Co Units.
Data compiled from:
Brandon, C. A., Porter, J. J., and Todd, D. K., "Hyperfiltration for
Renovation of Composite Wastewater at Eight Textile Finishing Plants."
Final Report, EPA Grant No. S802973, Clemson University report in
preparation.
-------
Table A-6
REJECTION OF SPECIFIC COMPOUNDS BY THE
SELAS Zr(IV)-Na2S103 MEMBRANE
OPERATING ON TEXTILE WASTE WATER
Solute (underlined
solute in consent
decree)
COD
BOD
TOG
Dissolved Solids
Volatile Solids
Color
Phenol
Iron
Nickel
Chromium
Zinc
Copper
Manganese
"Concentration in Pt-C
Rejection
(%)
71-98
88-98
85-92
41-97
81-98
99-100
16-97
67-95
90.6
93-100
89-99
94-100
99.7
o Units
PH
7.1-10.7
7.1-10.7
8.0-10.7
7.1-10.7
8.0-10.7
7.4-10.7
7.1-8.1
7.1-10.7
7.1
7.1-10.7
7.1-10.7
7.1-10.7
7.1
Concentration
(n.R/1)
933-17,800
295-6200
330-440
17810-23853
213-10870
4000-30,000*
1.4-241
2.9-9.02
2.12
0.129-1.38
1.64-7.76
0.243-1.59
3.97
Nu
70-80%
1
0
0
0
0
0
0
1
0
0
0
0
0
mber of
80-90%
1
1
2
0
1
0
0
1
0
0
2
0
0
Data
90-100%
2
4
2
1
2
3
1
1
1
5
3
5
1
TOTAL
4
5
4
5
3
3
3
4
1
5
5
5
1
Data compiled from:
Brandon, C. A., Porter, J. J., and Todd, D. K., "Hyperfiltration for
Renovation of Composite Wastewater at Eight Textile Finishing Plants."
Final Report, EPA Grant No. S802973, Clemson University report in
preparation.
-------
Table A-7
REJECTION OF SPECIFIC COMPOUNDS BY THE
WESTINGHOUSE TUBULAR POLYSULFONE MEMBRANE
OPERATING ON TEXTILE WASTE WATER
Solute (underlined
solute in consent
decree)
COD
BOD
TOC
Dissolved Solids
Volatile Solid
Color
Phenol
Iron
Nickel
Chromium
Zinc
Copper
Manganese
*Concentration in Pt-
Rejection
(%)
17-75
65-71
A9-82
23-69
30-72
25-96
11-24
48-97
15.8
82-93
33-99
76-97
50-75
Co Units
PH
3.6-12
6.4-12
5.9-12
3.6-12
3.6-11.6
5.9-12
6.4-7.0
3.6-9.2
3.6
5.9-12.0
3.6-7.0
3.6-12.0
5.9-7.0
Concentration
(mg/1)
468-5020
1180-1800
762-1338
3705-12840
370-2698
200-3409*
0.140-0.210
3.80-7.75
0.76
1.4-23.0
3.60-9.91
0.92-5.51
0.08-0.40
Number of Data
70-80% 80-90% 90-100%
2
1
2
0
1
1
0
1
0
0
0
2
1
0
0
1
0
0
2
0
0
0
2
0
0
0
0
0
0
0
0
2
0
1
0
3
4
6
0
TOTAL
8
2
8
9
8
8
3
3
1
5
6
8
5
Data compiled from:
Brandon, C. A., Porter, J. J., and Todd, D. K., "Hyperfiltration for
Renovation of Composite Wastewater at Eight Textile Finishing Plants."
Final Report, EPA Grant No. S802973, Clerason University report in
preparation.
-------
Table A-8
REJECTION OF SPECIFIC COMPOUNDS BY THE
Zr(IV)-PAA MEMBRANE ON STAINLESS STEEL
OPERATING ON TEXTILE WASTE WATER
Solute (underlined
solute in consent
decree)
COD
BOD
TQC
Dissolved Solids
Total Solids
Volatile Solids
Color
Phenol
Iron
Nickel
Chromium
Zinc
Copper
Manganese
Rejection
(%)
38-88
0-97
11-100
48-99
PH
Concentration
(mg/1)
3000-9200
0.65-7.10
12-281
0.07-0.96
Numbe
70-807.
6
0
3
0
r of Dat
80-90%
2
3
0
2
a
90-100%
0
5
4
5
TOTAL
10
10
9
9
Cn
*Concentration in Pt-Co Units
Data compiled from:
Brandon, C. A., Porter, J. J., and Todd, D. K., "Hyperfiltration for
Renovation of Composite Wastewater at Eight Textile Finishing Plants."
Final Report, EPA Grant No. S802973, Clemson University report in
preparation.
-------
100
90
80
70
c
o
73 cr»
.£, **0
ec
S« 50
40
30
20
10
O SELAS Zr (IV)-PAA
• WESTINGHOUSE TUBULAR CELLULOSE ACETATE
AUOP
D WESTINGHOUSE TUBULAR POLVSULFONE.
X DUPONT
O SELAS Zr (IV)-No2SI03
500
1000
1500 2000 2500
Concentration (mg/l)
3000
6200
Figure A-l Percent Rejection versus Concentration for Biochemical Oxygen Demand (BOD ).
-------
00
100
90
80
70
c
o
'•C 60
O)
-—^
0)
cr
jo 50
40
30
20
10
O SELAS Zr (IV)-PAA
• WESTINGHOUSE TUBULAR CELLULOSE ACETATE
A UOP
D WESTINGHOUSE TUBULAR POLYSULFONE ULTRAFILTRATION
X OUPONT
0 SELAS Zr (IV)-No2SI03
A o A CD
00
o
O
00 $ AO
00
I // I //I
1000
2000
3000
4000 5000
Concentration (mg/l)
6000
7000
8000
Figure A-2 Percent Rejection versus Concentration for Chemical Oxygen Demand (COD).
-------
100
90
60
70
c
o
I 60
«T
cc
5? 50
40
30
20
10
0
O SELAS Zr (IV)-PAA
• WESTIN6HOUSE TUBULAR CELLULOSE ACETATE
A UOP
O WEST1NGHOUSE TUBULAR POLYSULFONE ULTRAFILTRATION
O A
• 3 NJR ON UCC
X OUPONT
e SELAS Zr (IV)-Na2Si03
o _ O A
250
500
750
1000 1250
Concentration (mg/l)
1500
1750
2000
Figure A-3 Percent Rejection versus Concentration for Total Organic Carbon (TOC).
-------
100
90
80
c 70
g
.£. 60
o>
tr.
& 50
40
30
20
10
O SELAS Zr (IV)-PAA
• WESTINGHOUSE TUBULAR CELLULOSE ACETATE
A UOP
0 WESTINGHOUSE TUBULAR POLYSULFONE ULTRAFILTRATION
B 3 NJR ON UCC
X DUPONT
« SELAS Zr (IV)-No2SI03
a
J 1 ' '
O
O
O
0
O
Q
J_
A A
0O O
J L
_L
J L
i //t
0
5000
10000
Concentrotion (mg/l)
15000
Figure A-4 Percent Rejection versus Concentration for Dissolved Solids.
-------
100
90
60
70
ON OC
u
•>
•5-60
50
40
30
20
10
O SELAS Zr (IV)-PAA
• WESTINGHOUSE TUBULAR CELLULOSE ACETATE
A UOP
O WESTINGHOUSE TUBULAR POLYSULFONE ULTRAFILTRATION
00
o /^o
• 3 NJ R ON UCC
X DUPONT
a SELAS Zr (IV)-No2Si03
it. I
500
1000
1500 2000 2500
Concentration (mg/l)
3000
3500 " 10870
Figure A-5 Percent Rejection versus Concentration for Volatile Solids.
-------
O SELAS Zr (IV)-PAA
• WESTINGHOUSE TUBULAR CELLULOSE ACETATE
A UOP
0 WESTINGHOUSE TUBULAR POLYSULFONE ULTRAFILTRATION
3NJR ON UCC
SELAS Zr (IV)-Na2Si03
100
90
80
• • D
o
a
O 9
a
o
O* ^* en
NJ 'ST 60
a:
^o
^ 50
40
30
20
10
0
0
JL
_L
_L
_L
JL
J_
1000
2000
3000 4000 5000
Concentration (mg/l)
6000
7000
eood^soooo
Figure A-6 Percent Rejection versus Concentration for Color.
-------
a
100
90
80
70
o
.£. 60
«>
50
40
30
20
10
0
O SELAb Zr(IV)-PAA D WESTINGHOUSE TUBULAR POLYSULFONE ULTRAFILTRATION
• WESTINGHOUSE TUBULAR CELLULOSE ACETATE «SELAS Zr (IV)-No2Si03
A UOP
A
0
A O
0
0.25
0.50
0.75
J_
JL
1.00 1.25
Concentration (mg/l)
\
1.50'
-J 1
50 100 150 200 250"3I5
Figure A-7 Percent Rejection Versus Concentration for Phenol.
-------
100
90
80
c 70
o
40
30
20
10
•
0
O SELAS Zr (IV)-PAA
• WESTINGHOUSE TUBULAR CELLULOSE ACETATE
A UOP
1 1 1 1 ( i
0 0.001 0.002 0.003
Concentration (mg/l)
Figure A-8 Percent Rejection versus Concentration for Mercury.
-------
tn
100
90
80
70
c
o
.I60
cc
& 50
40
30
20
10
0
O SELAS Zr (IV)-PAA
• WESTINGHOUSE TUBULAR CELLULOSE ACETATE
0 SELAS Zr (IV}-No2Si03
O WESTINGHOUSE TUBULAR POLYSULFONE ULTRAFILTRATION
O AOA A A
O O
A O .
OQ A
D •
Q
0.25
0.50 0.75 1.00 1.25
Concentration (mg/l)
3.97
Figure A-9 Percent Rejection versus Concentration for Manganese.
-------
OSELAS Zr (IV)-PAA
• WESTIN6HOUSE TUBULAR CELLULOSE ACETATE
AUOP
OWESTINGHOUSE TUBULAR POLYSULFONE ULTRAFILTRATION
c
o
100*
90
60
70
S! 60
V
DC
5« 50
40
30
20
10
A D
O
O
h
X DUPONT
OSELAS Zr (IV)-No2Si03
n o
0
10
15
20
25
Concentration (mg/l)
30
35
40
Figure A-10 Percent Rejection versus Concentration for Iron.
-------
OSELAS Zr (IV)-PAA
• WESTINGHOUSE TUBULAR CELLULOSE ACETATE
AUOP
OWESTINGHOUSE TUBULAR POLYSULFONE ULTRAFILTRATION
100
90
80
. 70
o
"o
o
A
o
A
o
o
eSELASZr (IV)-PAA
^
50
40
30
20
10
0
0.25
0.50
0.75 1.00 1.25
Concentration (mg/l)
1.50
1.75
2.00
Figure A-ll Percent Rejection versus Concentration for Nickel.
-------
OO
100
90
80
; 70
"o
5-60
cc
50
40
30
20
10
• WESTINGHOUSE TUBULAR CELLULOSE ACETATE
AUOP
DWESTINGHOUSE TUBULAR POLYSULFONE ULTRAFILTRATION
e e
• 3NJR ON UCC
e SELAS Zr (IV)-Na2SI03
J- °A
O
o
o
0
Oo
o
_L
-L
o
D
_L
_L
0
0.25
050
0.75 1.00 1.25
Concentration lmg/1)
I.507
10 15 20 25
Figure A-12 Percent Rejection versus Concentration for Chromium.
-------
100
90
80
70
'=-60
-------
"100
90
80
o 70
-0 o
o «>
60
50^
40
30
20
10
0
0 SELAS Zr(IV)-PAA
• WESTINGHOUSE TUBULAR CELLULOSE ACETATE
A UOP
• WESTINGHOUSE TUBULAR POLYSULFONE ULTRAFILTRATION
0
0
A
M
A*
o
h
• 3NJRONUCC
X OUPONT
0 SELAS Zr(IV)-Na2SI03
0
o a
i
i
-//-
O50
1.00
1.50 2.00 2.50
Concentration (mg/l)
3.00
3.50
5.51
Figure A-1A Percent Rejection versus Concentration for Cooper.
-------
100
90
80
I 70
u
o
0>
60
50
40
30
20
10
O 2000 SERIES SELAS Zr(IV)-PAA
A 4000 SERIES SELAS Zr (IV)-PAA
A 4000 SERIES STAINLESS STEEL-PAA
_L
JL
J_
J_
0°
JL
0
2500
5000
7500 10000 12500
Concentration (mg/l )
15000
17500
20000
Figure A-15 Percent Rejection versus Concentration for Total' Solids.
-------
100
90 °*
80
70
o
55
A
O 2000 SERIES SELAS Zr(IV)-PAA
A 4000 SERIES SELAS Zr(IV)-PAA
A 4000 SERIES STAINLESS STEEL Zr (IV)-PAA
AO
A ° ° ° °
8
60
50
40
30
A
20U>
10
0 A
0 0.25 0.50 0.75 1.002 345 10 15 20
Concentration (mg/l)
Figure A-16 Percent Rejection versus Concentration for Copper.
-------
100
90
80
O 2000 SERIES—SELAS 2r(IV)-PAA
A 4000 SERIES —SELAS Zr(IV)-PAA
A 4000 SERIES-STAINLESS STEEL Zr (IV)-PAA
AO
O O
A
70
60
50
40
30
20
o
o
10
0.50
234
Concentration (mg/e)
Figure A-17 Percent Rejection versus Concentration for Iron.
-------
90
80
| 70
O
£ 60
5*
50
40
30
20
10
O 2000 SERIES "-SELAS Zr(IV)-PAA
A 4000 SERIES—SELAS Zr(IV)-PAA
A 4000 SERIES —STAINLESS STEEL Zr(IV)-PAA
A O
A
A A
O A
O
O O
50
100
150 200 250
Concentration (mg/l)
300
350
400
Figure A-18 Percent Rejection versus Concentration for Chromium.
-------
Appendix B
PREDICTION OF OSMOSIS MEMBRANE SEPARATION
EFFICIENCIES FOR SOLUTES IN DILUTE AQUEOUS SOLUTIONS
75
-------
V
1 • 3-
- < j -. 1
I
RTI/1430/29-01F
. .. A...I :,•. .jj e. L ;r r» ;. - - > jj. y
.>.-..;:iK.jJ ^ij^ii^.y^r»^S2a^^it<,^>^%>i^^t^^
PREDICTION OF OSMOSIS MEMBRANE SEPARATION
EFFICIENCIES FOR SOLUTES IN DILUTE AQUEOUS SOLUTIONS
by
Anton Schindler
Research Triangle Institute
P. 0. Box 12194
Research Triangle Park, North Carolina 27709
Contract No. 68-02-2612, Task 29
EPA Task Officer: Dr. Max Samfield
Industrial Environmental Research Laboratory
Office of Energy, Minerals, and Industry
Research Triangle Park, NC 27711
Prepared for
.U. S. ENVIRONMENTAL PROTECTION AGENCY
Office of Research and Development
Washington, DC 20460
76
RESEARCH TRIANGLE PARK, NORTH CAROLINA 277
09
-------
CONTENTS
Page
List of Figures 78
List of Tables 79
List of Symbols 80
Sections
I Abstract 81
II Conclusions and Recommendations 82
III Introduction ^4
IV Solute Parameters 89
V Prediction of Solute Separation 9G
References
77
-------
LIST OF SYMBOLS
F . cohesive energy
con
e . cohesive energy density
con
K dissociation constant of an acid
K . dissociation constant of a base
b
M molecular weight of solute
p pressure
R solute rejection in reverse osmosis
RO reverse osmosis (hyperfiltration)
S Small number
V molar volume
a degree of dissociation
6 solubility parameter (general)
6 solubility parameter of solvent
6 solubility parameter of polymer
5. . . solubility parameter of solute
5 solubility parameter of membrane
A difference between solubility parameters
iin ' , . , . i \
of solute and membrane (absolute value;
o Hammet constant
o* Taft constant
Av band shift for hydroxyl groups in infrared
spectra
80
-------
SECTION I
ABSTRACT
The EPA has set maximum concentration levels for many environmental
pollutants in water. Most of these pollutants are nonelectrolytes being
present in low concentrations albeit still exceeding the maximum permissible
level. Hyperfiltration possesses a high potential for large volume separation
of pollutants in industrial unit operation effluents. The objective of this
report was to present a critical evaluation of methods permitting the predic-
tions of separation efficiencies in hyperfiltration for different ncn-elec-
trolytes from their chemical structure. The evaluation was based on two
criteria important for practical applications: universality and facile
accessibility of the required correlation parameters.
81
-------
SECTION II
CONCLUSIONS AND RECOMMENDATIONS
The most extensive investigations of establishing correlations between
solute parameters and hyperfiltration efficiency (solute rejection) were
carried out by Sourirajan and Matsuura on cellulose acetate membranes. Al-
though this work represents extremely valuable contributions to the problem
by initiating and stimulating the search for predictibi1ity parameters in
hyperfiltration, the choice of solute parameters was very unfortunate for
practical applications. Generally, the parameters lack universality, i.e.,
solutes possessing the same parameter value but being of different chemical
structure exhibit different behavior in hyperfiltration. As a consequence,
as many individual correlations have to be established as there are structural-
ly distinguishable groups of solutes. In addition most of the solute para-
meters proposed are not accessible for compounds of which only the chemical
structure is known.
The above objections were circumvented in the correlation studies of
Spencer and Gaddis by selecting the solubility parameter of the solute as the
correlation parameter for predicting hyperfiltration efficiencies. These
correlations represent the best and simplest approach toward predicting
membrane performance so far obtained. For practical applications the method
only requires knowledge of the solubility parameters of solute and membrane.
The latter value, which would be nearly impossible to calculate, can be
obtained from a small number of hyperfiltration tests performed with solutes
covering a fair range of solubility parameter values. Solubility parameters
of solutes are either accessible from extensive compilations in the literature
or can be calculated from the chemical structure and the density of the
solute.
In all fairness toward Sourirajan and Matsuura it has to be pointed out
that the correlation by Spencer and Gaddis practically fails in the case of
-------
cellulose acetate membranes. One may conclude that the hyperfiltration
properties of cellulose acetate membranes cannot be described by one-para-
metric correlations comprising solutes of widely differing chemical structures.
Since cellulose acetate membranes represent the most commonly used membrane
type, further investigations, especially of commercial membranes, are required.
Presently, predictions of membrane performances according to Spencer and
Gaddis are solely based on estimates of solute distributions between membrane
and feed solution. Transport properties of the hyperfiltration system are
not considered in this treatment. This deficiency permits the use of a
single parameter in the correlation but necessarily restricts the applicabil-
ity of the correlation to a limited pressure range.
Despite this minor deficiency the correlation proposed by Spencer and
Gaddis presently represents the simplest and the most generally applicable
method for predicting hyperfiltration performance of membranes with the
exception of cellulose acetate membranes. With the latter type of membranes
the correlation is not satisfactory although it is still applicable with some
reservations. Especially the facile accessibility of the required solubility
parameters of the solutes should make the correlation method of Spencer and
Gaddis the preferred one for practical applications.
83
-------
SECTION III
INTRODUCTION
Reverse osmosis (RO) is an efficient water purification process of low
energy consumption. In the application of RO for water desalination membrane
characterization with respect to solute separation does not present too
severe a problem due to the predictable composition of the feed solution.
Complications arise in the application of RO for purification of industrial
wastewaters containing organic impurities. The great number of possible
organic compounds which can be present in industrial wastewater precludes, at
least at the present time, the application of experimental data to describe
membrane performance. It is therefore pertinent to investigate the possibil-
ity of predicting membrane performance in RO from easily accessible solute
parameters which can be related to the chemical structure of the solute.
The most extensive investigations in this direction were carried out by
Sourirajan and Matsuura. These authors based the choice of characteristic
solute parameters on their sorption-capillary flow model describing membrane
performance (Figure 1). According to this model, RO is governed by inter-
facial phenomena on the feed side of the RO membrane, Ke., either the water
or the solute is preferentially sorbed on the feed-membrane interface. As a
consequence, the membrane surface will be in contact with a solution differing
in concentration from the bulk of the feed. If water is preferentially
sorbed then an interfacial layer results on the feed side of the membrane
which is depleted in solute with respect to the bulk of the feed (Figure 1A).
This interfacial layer of low solute concentration is then transported through
the capillaries of the membrane by the applied pressure (Figure IB). Contin-
uous removal of the interfacial layer by flow under pressure (hydraulic flow)
and reformation of the layer by preferential sorption of water results in a
permeate of lower solute concentration than the feed, _Le. , the membrane
rejects the solute.
84
-------
THE SOLUTION \
nuM "
PRESS
4
H,O No'Cr H,0
HjO No*Cr H,0
HjO No*Cr HjO
HjO No*Cr HjO
'HjO H,0 HjO
k HjO H,0 HZO
''»olou» »br»;»«ct M'
- S*!1^0"^., H,
JURE
No4CI" HtO
No^cr H,O
No*CI~ H,0
No*CI~ HZ0
HtO H,0
H^O H?0
0 ^,0,, ,,IV13'«:
o ^^"^vj-t
PORE v
\ CRITICAL
\ SIZE
(A)
DEMINERAUZEO WATER AT THE
j[ INTERFACE _J[
POROUS POROUS \T
FILM FILM
(B)
CRITICAL PORE DIAMETER
ON THE AREA OF THE FILM
AT THE INTERFACE
Figure 1. Schematic representation of the sorption-capillary flow mechanism
in hyperfiltration. (A) Sorption mechanism on feed-membrane
interface, (B) Transport mechanism.
85
-------
Solutes which exhibit sorption behavior comparable to that of water or
solutes which are even more strongly sorbed than water will not be separated
under RO conditions.
According to the model of Sourirajan and Matsuura the membrane perform-
ance in RO is a function of solute-solvent-membrane interaction at the inter-
face. These interactions are assumed to arise from the polar-, steric, non-
polar- and/or ionic character of each of the three components in the RO
system. Proper selection of interaction parameters should provide means for
predicting RO separations for a wide range of different solutes. Indeed,
Surirajan and Matsuura succeeded in establishing correlations between RO
performance and solute parameters by restricting themselves to water as the
solvent and cellulose acetate as the membrane material. With these constraints
RO performance should be a sole function of solute properties characterized
by some physicochemical parameters.
Solute parameters for predicting solute separations in RO should meet
two conditions. First, in agreement with the proposed model of membrane
transport, the parameters should be relevant in describing sorption phenomena.
Secondly, it should be possible to calculate the parameters from the known
chemical structure of the solute, .i.e., the parameters should be accessible
from group contributions.
Five solute parameters were investigated by Sourirajan and Matsuura in
their correlation studies.
1. Band shifts in IR spectra of OH groups (Av)
2. Dissociation constants of carboxylic acids (K )
3. Hammett constants of aromatic compounds (a)
4. Taft constants of aliphatic compounds (a*)
5. Small's number (S))
All of the parameters are not applicable to all solutes but the latter
have to be classified into chemically related groups. Inside each group of
solutes good correlation between the applicable solute parameter and solute
separation in RO could be experimentally established.
It must be pointed out that the good correlations between solute para-
meters and solute separations in RO observed for many solutes does not repre-
sent a proof for the validity of the sorption-capillary flow mechanism proposed
86
-------
by Sourirajan and Matsuura. Indeed, this mechanism has met strong criticisms
by several authors. Nevertheless, the established correlations are experi-
mental facts and consequently demonstrate the existence of relationships
between specific solute parameters and solute properties under RO conditions.
From this point of view the investigations of Sourirajan and Matsuura repre-
sent extremely valuable contributions by initiating and stimulating the
search for predictbility parameters in RO although the underlying theory can
be considered only as a working hypothesis.
According to the sorption-capillary flow model of the authors it is only
the sorbed layer on the membrane interface at the feed side which participates
in the hydraulic flow through the membrane, ^.e., the composition of the
sorbed layer determines the composition of the effluent.
According to hydrodynamic theories the first molecular layers on a
liquid-solid interface do not particpate in the flow of the liquid but remain
at rest, _i.e., the flow of the liquid encompasses only subsequent layers.
One has now to consider that forces which govern preferential sorption of
either water or solute on the membrane interface are of short-range, i.e.,
their acting distance will not extend beyond one or two molecular layers. At
larger distances the solution will be of uniform concentration since at this
distance a solute molecule will be unaware of the presence of the membrane
surface. Since the layer being affected in its composition by sorption will
not participate in the hydraulic flow, it is difficult to understand how the
composition of this layer should determine the composition of the solution
flowing through the membrane capillaries (pores) under hydraulic pressure.
The solute parameters on which the correlation studies of Sourirajan and
Matsuura are based are not specific for sorption processes but are of general
validity in describing interactions between solutes and solvents. Consequently,
the same correlations could be expected to exist if the RO process were based
on a solution-diffusion model in which the partition of the solute between
aqueous solution and membrane plays the decisive role. A solute which is
strongly sorbed on the membrane surface will also migrate into the membrane
due to the favorable interaction with the membrane material, i.e., the parti-
tion coefficient of such a solute will be relatively high. As a consequence
of the relatively high solute concentration in the membrane the solute rejec-
tion will be low.
87
-------
Solutes which are less compatible with the membrane material will possess
a low partition coefficient, j..e., a low equilibrium concentration will
establish inside the membrane. Such solutes will be highly rejected by the
membrane.
Compatibility (miscibility) of solvents and polymers is described best
by their respective solubility parameters. Although these parameters are not
providing quantitative relationships with respect to the amounts of solute
imbibed by the membrane, they permit a relative classification of different
solutes with respect to their compatibilities with a given membrane material.
It is somewhat surprising that no direct use of solubility parameters was
made by Sourirajan and Matsuura in their correlation studies inasmuch as
2
Spencer and Gaddis could demonstrate good correlations to exist for a number
of different membranes. Indeed, the correlations obtained by Spencer and
Gaddis for different membranes could be unified thus permitting a comparison
of rejection properties of different membrane materials with respect to a
given solute.
88
-------
SECTION IV
SOLUTE PARAMETERS
A considerable number of different solute parameters were investigated
by Sourirajan and Matsuura1'3'6 with respect to their correlation with solute
rejection under RO conditions. Unfortunately, nearly all of these parameters
do not meet the most important conditions required for practical applica-
tions: universality and facile accessibility.
The condition of universality of a parameter would mean that solutes
differing in their chemical structure but possessing the same parameter value
should exhibit very similar rejection properties with respect to the same
membrane. If this condition is not met but correlations exist only for
solutes belonging to the same structurally related group (e.g., alcohols,
aldehydes, ketones, esters, ethers, etc.) prediction of membrane performance
requires as many individual correlations as there are structurally distin-
guishable groups. This unfortunate situation exists for nearly all of the
solute parameters investigated by Sourirajan and Matsuura. For example a
Taft number of zero corresponds to solute separations of 60% with aldehydes,
50% with ethers, 30% with esters, and 20% with ketones. The same pronounced
differences are observed with Small numbers in the case of hydrocarbons.
The second condition of facile accessibility of the solute parameter is
also not always met by the parameters selected by Sourirajan and Matsuura.
Very often practical applications of RO will be concerned with compounds of
which only the chemical structure is known, and this sparse information has
to suffice for the estimation of the correlation parameters. Under these
conditions the choice of solubility parameters by Spencer and Gaddis is of
advantage. Solubility parameters for an extremely large number of solvents
are listed in the literature, e.o.. , Hoy7 lists close to 700 compounds.
Values for unlisted compounds can be estimated from the chemical structure by
means of the additivity rule for group " or atomic contributions to the
molar attraction constant.
89
-------
The different solute parameters investigated by Sourirajan and Matsuura
and by Spencer and Gaddis are discussed in subsequent sections.
A. Polar Parameters
The polar parameters describe the proton-donating or the proton-accepting
properties of the solute. Numerical values related to these properties can
be derived from (i) infrared spectra of alcohols and phenols, (ii) dissocia-
tion constants of carboxylic acids, and (iii) reaction rates under standard
conditions.
(1) Infrared Spectra
Infrared spectra supply numerical values for the strength of hydrogen
bond formation under standard conditions. If the solute is a proton-donor
(alcohols or phenols) spectroscopic measurements are performed with diethyl
ether as the proton-acceptor. The relative shift, 6v (acidity), in the
absorption maximum of the hydroxyl group represents a quantitative measure for
the hydrogen bonding ability of the solute, j^.e., the larger Av, the stronger
the acidity of the solute.
If the solute is a proton-acceptor (aldehydes, ketones, ethers, or
esters) deuterated methanol (ChUOD) is used as the standard proton-donor. In
this case the relative shift in the absorbance maximum of the OD group, Av
(basicity), expresses the hydrogen bonding ability of the solute, j..e. , the
larger Av, the stronger the basicity of the solute.
(2) Dissociation Constants
In the case of carboxylic acids, amines, and aminoacids, the acidity or
the basicity of the solute is quantiatively described by its dissociation
constants, Kg or Kb, respectively.
Acid: HA + H20 ^ H30+ + A (1)
Ka = LH30+][A"]/[HA] (2)
Base: B + H20 " HB+ + OH~ (3)
Kb = [HB+][OH"]/[B] (4)
The degree of dissociation, a, expressed as the fraction of total acid
«r base being present in ionized form is then given by
K = a2c/(1"a) (5)
-------
The degree of dissociation of weak acids and weak bases is pH dependent
which has to be considered in the prediction of solute separation by RO.
(3) Hammet or Taft Numbers
Hammett numbers describe the polar effect of substitutents in m- or p-
position of aromatic compounds. The values are obtained either from equili-
brium determinations or from rate measurements under standard conditions. In
the first case dissociation constants of differently m- or p-substituted
benzoic acids are compared. If K and K are the dissociation constants of
unsubstituted and substituted benzoic acid, respectively, the Hammett constant
o, is given by:
a = (l/p)lg(K/K0) (6)
where p is a constant, in this case referring to measurements of dissociation
constants. The value of p changes with the kind of physico-chemical measure-
ment, e.cj., if one compares rate constants of ester hydrolysis, but taking
this change into account the same o value is obtained for each substituent.
Taft numbers are defined analogously for aliphatic and o-substituted
aromatic compounds. Also in this case the Taft number o* is a measure of the
polar effect of the substituent.
Both solute parameters, Hammett and Taft numbers represent group contri-
butions, _[•£., for solutes possessing more than one substituent, Hammett or
Taft number of the solute are equal to the sum of the Hammett or Taft numbers
of each of the substituent groups. Consequently, it is possible to calculate
Hammett or Taft numbers if the chemical structure of the solute is known.
Since the dissociation constants, K,, represents the total polar effect
a
of the acid molecule and a or o* represent the contribution of the substitu-
tent groups to this total effect, a relationship exists between K and o or
d
a* for aromatic and aliphatic monocarboxylic acids, respectively. From this
relationship dissociation constants of monocarboxylic acids can be estimated
from the chemical structure of the acid.
There exists also a correlation between o or o* numbers and spectro-
scopically determined Av values for reasons analogous to those given for the
correlation between K and o or o*.
d
Because of the existing relationships between different polar parameters
correlation of solute separation in RO with either the Hammett or the Taft
number of the solute will suffice.
91
-------
B. Small Numbers
Hydrocarbon solutes are hydrophobia and essentially non-polar. The
cellulose acetate membrane has both polar and non-polar character. The polar
character of the membrane arises from hydroxyl and ester groups and is respon
sible for the ability of hydrogen bond formation. The non-polar character
arises from the hydrocarbon backbone and the hydrocarbon groups in the ester
part. Consequently, the membrane may be expected to attract both the polar
solvent (water) and the non-polar solute (hydrocarbon). According to the
sorption-capillary flow mechanism of Sourirajan and Matsuura the relative
extents of these attractions determine the solute separation in RO.
For the purpose of correlation of RO data both the molar solubility of
the solute in water or the Small number of the solute were used to express
the degree of hydrophobicity or of non-polar character of the solute. From a
practical point of view, it was desirable to use a solute parameter which can
be calculated from the chemical structure of the solute. This is the case
with Small numbers but not with solubilities which require experimental
determinations.
The Small numbers are related to the cohesive energies of liquids, E
which are defined as the energies necessary to break all the intermolecular
contacts per mole of the liquid. The cohesive energy of a liquid is closely
related to its molar heat of evaporation, AH
Ecoh = AHvap ' PAV V3P (7)
Quantities derived from the cohesive energy of a liquid are, vis.:
the cohesive energy density: e . = (E ./V) , (8)
the solubility parameter: 6 = (E/V)1/2, and
con
(9)
the molar attraction constant (Small Number):
S = (VEcoh)1/2= V6
In the above expressions V is the molar volume at standard temperature
Of 298 K (25°C) and is given by molecular weight divided by density. The
dimension of the solubility parameters is [J1/2 m3/2] with 1.0 (J/m3)1/2 =
4.88 x 102 (cal/cm3)172.
In the final analysis the calculation of solubility parameters or Small
numbers revolves around obtaining the value of the heat of evaporation from
vapor pressure data. The procedure used in the calculation of solubility
parameters for 680 compounds from vapor pressure data is outlined by Hoy.7
92
-------
The additive property of the cohesive energy, E ., was demonstrated as
11 8
early as 1928 by Dunkel. Small demonstrated the usefulness of the molar
attraction constant as an additive quantity. His set of values is still
7-9
frequently used with some refinements by other authors. A comparison of
values derived by different authors is given by Van Krevelen. The latter
author also devised a list of atomic attraction constants which considerably
simplify calculations.
Generally, solubility parameters derived from vapor pressure data should
be preferred in correlation studies over those calculated by the additivity
method from group contributions. The total cohesive energy, E, which holds a
liquid together and which is derived from vapor pressure data, can be divided
into contributions from dispersive forces (London forces), E., dipole-dipole
forces, E , and hydrogen bonding forces, E..
E - Ed+ EP * Eh <">
Multiplying this equation by the molar volume of the compound one obtains
VE = VEd + VEp + VEh (12)
or with VE = S from equ. 10
s2 = sd + Sp + Sy (13)
The value of S calculated from group contributions strongly depends on
the way the individual group values were derived with respect to equ.(13).
For example, Small obtained a value of 170 for the OH group which was derived
from the contribution of an ether oxygen (70) and a hydrocarbon hydrogen
(100). This value represents only the contribution to the cohesive energy
derived from dispersion forces but neglects the considerably high contribution
resulting from hydrogen bonding which does not exist in hydrocarbons from
which the value of 100 for hydrogen was derived. According to Konstam and
Q
Feairheller the value for a single OH group is 399 which yields good agree-
ment between calculated and experimental solubility parameters of alcohols
(see Table I).
Problems arise with compounds containing more than one functional group.
In some cases, e.g., OH in diols, Cl in CC12 or CC13, modified group contibu-
tions are reported, but generally the additive method yields ambigous results
if applied to polyfunctional compounds.
93
-------
Table 1. COMPARISON OF SOLUBILITY PARAMETERS OF SELECTED
COMPOUNDS OBTAINED BY DIFFERENT METHODS
Compound
Alcohols
Methanol
Ethanol
n-Propanol
i-Propanol
n-Butanol
n-Pentanol
3-Pentanol
Ethyl eneglycol
Propyleneglycol
Glycerol
Benzylalcohol
Esters
~~ Methyl acetate
Ethylacetate
Ethylbenzoate
Ketones
Acetone
Butanone
Aldehydes
Acetaldehyde
Propionaldehyde
Acids
Acetic acid
Propionic acid
Butyric acid
Acrylic acid
Ethers
Diethylether
Isopropylether
Amines
DTethylamine
Tri ethyl ami ne
Tri ethyl enetetramine
Ethylenediamine
fjydrocarbons
"ii-Hexane
Cyclohexane
Benzene
Toluene
Molac Volume
(mVkmol)
40.66
58.60
75.11
78.8?
91.53
99.41
104.27
55.92
73.70
73.19
103.82
79.88
98.54
144.20
74.01
90.19
57.13
73.44
57.26
74.99
92.45
69.30
104.78
142.33
104.30
139.95
149.83
66.57
131.61
108.79
89.42
106.88
Solubility Parameter in (J/m3)1/2 x 103
Hoy
(Ref. 7)
29.7
26.2
25.0
23.4
23.8
22.8
20.8
34.9
30.7
36.3
24.7
19.4
18.3
20.0
19.7
19.4
20.2
19.3
26.7
25.6
24.5
26.4
15.4
14.5
16.5
15.2
22.8
25.3
14.9
16.8
18.8
18.3
VanKrevelen
(Ref. 10)
31.9
27.0
24.8
22.0
23.4
24.4
21.0
34.5
30.0
34.9
25.8
17.0
16.6
19.0
20.7
20.1
21.8
20.8
-
-
-
•
15.8
15.6
17.6
15.0
25.1
23.0
14.9
16.6
18.6
18.2
Konstam et. al .
(Ref. 9)
30.9
26.1
24.1
22.2
22.7
23.6
21.0
32.5
27.7
34.3
25.0
18.9
18.1
19.8
21.1
20.3
22.1
20.9
26.4
23.8
22.3
24.4
19.1
17.2
-
-
-
29.6
14.9
16.9
17.0
18.2
94
-------
The usefulness of solubility parameters for predicting solute rejections
p
in RO was pointed out by Spencer and Gaddis and is based on the fact that
these parameters describe the solubility of a solute in a polymer. According
to theoretical concepts developed by Hildebrand the solubility parameters
of solvent, 6 , and polymer, 6 , determine the compatibility of both. The
larger the difference 6-6 the less compatible the components whereas
close similarity of 6 and 6g assures complete miscibility.
95
-------
SECTION V
PREDICTION OF SOLUTE SEPARATION
A. Polar Parameters of Sourirajan and Matsuura
Although several of the polar parameters selected by Sourirajan and
Matsuura show good correlations with solute rejections in RO, these results
are more of academic interest and are not useful for practical applications.
Band shifts in infrared spectra cannot be predicted from the chemical
structure of the solute but would require individual determinations. Further-
more, the range from zero rejection to over 80% rejection is compressed
inside a range of 20 units in Av, thus small changes in the parameter cause
extremely large changes in the rejection value.
A similar criticism applies to the use of Taft numbers. Also in this
case the Taft number would have to be known to three significant digits to be
useful for predicting solute rejections. In addition, as pointed out previous-
ly, structurally different compounds possessing the same Taft number exhibit
widely differing rejection properties. Consequently, a great number of
correlation curves would be required for practical applications.
B. Small Numbers
Small numbers of different aliphatic and aromatic hydrocarbons were
correlated with their rejection properties by Sourirajan and Matsuura. No
correlation existed between Small numbers and rejections if all of the com-
pounds were considered. Some improvement could be obtained by grouping the
solutes according to chemically related groups, e.g., paraffins, olefins,
aromatics, etc.
The usefulness of Small numbers as correlative parameters is doubtful.
With structurally related compounds for which correlations are observed, the
Small numbers generally increase linearly with molecular weight as a conse-
quence of their additive nature. Such a dependence is shown in Figure 2 for
hydrocarbons, but also exists for alcohols, esters, ketones, etc. Because of
96
-------
130
120
110
100
90
BO
70 -
(SMALL NUMBER; x io"2
10 11 12
13
Figure 2. Molecular weight dependence of Small numbers for structurally
related compounds, Aliphatic (+), cycloaliphatic (A), and .
aromatic (°) hydrocarbons.
97
-------
this linear relationship molecular weights of solutes can be substituted for
Small numbers without change in the shape of the correlation curve, i.e.,
correlation of solute rejection with molecular weight of the solute will be
as good as correlation with Small numbers.
C. Solubility Parameters
Correlation of solute rejections, R., in RO with solubility parameters
of the solute, 6^, was proposed by Spencer and Gaddis (2). This correlation
is based on the assumption that the concentration of solute available for
transport across the membrane depends on the difference between the solubility
parameters of solute, 6^, and membrane material, 6 .
*1. • ai - 6M " 04)
Consequently, R. should be a function of A. and a membrane should exhibit
i im
maximum rejection properties for solutes which differ strongly in the 6^
value from that of the membrane material. Contrarily, minimum rejection or
no rejection at all will be observed with solutes possessing 6. values equal
to that of the membrane.
Correlation plots of R. vs 6. for three different membranes were found
in good agreement with the predicted dependence of R. on 6-. The plots
exhibited minima which permitted the estimation of the 6 values of the three
membranes. It was further observed .that close to 100% rejection is obtained
01/0
if the 6^ value of a solute differed ±0.01 (J/m ) from the 6m value of the
membrane. Inside the intermediate range of 6. values the R. values showed a
linear correlation with respect to 6..
Since the A^ values observed for onset of 100% rejection were found to
be the same for the three membranes, their correlation plots could be super-
imposed by changing the independent variable 6. to A. , I.e., by using the 6
values as the common origin. The resulting correlation plot in its gener-
alized form is shown in Fig. 3.
The correlation by Spencer and Gaddis between R. and 6- represents the
best and simplest correlation so far obtained for predicting membrane per-
formance in RO. For practical applications the method only requires know-
ledge of 6.j of the solute (See Section IV-B) and of 6m of the membrane. The
latter value, which would be nearly impossible to calculate, can be obtained
from few hyperfiltration tests with solutes of differing 6. values.
-------
0.4 -
0.2 '
16
Figure 3. Dependence of solute rejection on the difference between the
solubility parameters of solute and membrane. Data from
Ref. 2.
99
-------
The correlation shown in Fig. 3 does not include f^ values obtained with
cellulose acetate membranes. Such values are shown in Fig. 4. Also in this
case 100% rejection approximately corresponds to a A- value of 0.01 (J/m3)1''2
but the scatter of data is considerably stronger than for the three combined
membranes of Fig. 3. Solute rejection by cellulose acetate membranes can be
predicted only with some reservations.
As pointed out by Spencer and Gaddis (2) their approach toward predicting
membrane performance in RO is solely based on estimating the distribution of
the solute between the separating membranes and the bulk of the feed. This
treatment does not include a parameter describing the transport properties of
the hyperfiltration system.
A possible effect of a changing transport parameter is indicated if one
compares rejection data obtained with the same membrane at widely differing
pressures. In Table II, data by Chian and Fang (13) are presented which were
obtained with a polyamide membrane at three different operating pressures.
In Fig. 5 the rejection values are plotted versus 6. values of the solutes.
Although data are insufficient to permit the estimation of the 6 value of
r m
the membrane, they clearly demonstrate that at low pressure the rejection is
much less sensitive to changes in 6. than at high pressure. Probably for this
reason the correlation between Ri and 6, is somewhat lost at higher pres-
sures.
0. Molecular Weight
The value of solute molecular weights, Mi, as a parameter for predicting
the rejection of non-electrolytes in RO was pointed out by Spencer et aj
(14). Most highly rejecting hyperfiltration membranes effectively reject
non-electrolytes in the molecular weight range exceeding a limiting value of
about 70. This is demonstrated for data by Cadotte et al_ (15) shown in Fig.
5. Solutes with molecular weights exceeding a value of 60 exhibit an average
rejection of 93.2 ± 4.1%.
Hyperfiltration results presented in Fig. 5 are replotted in Fig. 7 in
dependence of solute molecular weights. Data obtained at lowest operating
pressure show very weak correlation but the correlation improves by going to
higher pressures. If one neglects two solutes, methanol and methyl acetate,
which show decreasing rejection with increasing pressure, the average re-
jection for solutes with M^ >60 is 91%.
100
-------
Table 2. EFFECT OF OPERATING PRESSURE ON SOLUTE REJECTION
BY A POLY(AMIDE)MEMBRANE.
Solute
Methanol
Ethanol
i-Propanol
Acetone
Diethylether
Glycerol
Aniline
Me thy! acetate
Acetic acid
Phenol
Formaldehyde
Molecular
Weight
(kg/kmol)
32
46
60
58
74
92
93
74
60
94
30
Solubility
Parameter9
(J/m3/1/2 x 103
29.7
26.2
25.0
19.7
15.4
36.3
(21.3)
19.4
26.6
(23.5)
(28.5)
Percent Solute Rejection at
Operating Pressure in MPa
2.75
28
36
90
53
58
88
47
57
31
45
21
4.13
19
28
89
72
90
88
78
54
70
80
52
10.34
5
57
95
72
92
88
82
45
82
89
69
a Solubility parameters from Ref. 7, values in parenthesis calculated
according to Ref. 10.
Rejection values from Ref. 13.
101
-------
0.4
0.2 •
12 14 16 18 20 22 24 26 28 30
(SOLUBILITY PARAMETER) X 103
32 34
Figure 4. Solute rejection by cellulose acetate membranes in dependence
of the solubility parameters of the solute. Data from Ref. 2.
102
-------
.6
.4
.8
.6
.2
O
5
01
LU
ec
PRESSURE
2.U MPa
PRESSURE
' 1
O
3
c
PRESSURE
(SOLUBILITY PARAMETER) X 103
14
16 18 20 22 24 26
28
30 32 34 36
Fiaure 5 Solute rejections for three operating pressures in dependence
' on the solubility parameters of the solutes. Data from
Table II.
103
-------
1.0
MOLECULAR WEIGHT
30
40
SO
60
70
80
90
100
110
120
Figure 6. Solute rejection in dependence on the molecular weight of solutes
for NS-1 membrane at 5.52 MPa operating pressure. Data from
Ref. 15. Alcohols (<>), aldehydes and ketones (•), amines (A),
and esters (A).
104
-------
PRESSURE: 2-76 MP»
PRESSURE: 4.13 MPi
PRESSURE: 10.34 MPi
. O
i
U 3
c
MOLECULAR WEIGHT
20
40
60
80
100
Figure 7. Solute rejection for three operating pressures in dependence
nn thp mnlprular weiaht of the solute. Data from Table II.
105
-------
As demonstrated by Spencer and Gaddis, (2) no correlation between R. and
M. exists for cellulose acetate membranes. With other membrane systems the
molecular weight of the solute can be considered to represent a useful para-
meter for estimating solute rejection in RO.
106
-------
References
1. S. Sourirajan and T. Matsuura, in "Reverse Osmosis and Synthetic Membranes",
S. Sourirajan, Ed., National Research Council of Canada Publication,
Ottaw, Canda, 1977, Chapter 2.
2. H. G. Spencer and J. L. Gaddis, EPA Grant Number R805777-1.
3. T. Matsuura and S. Sourirajan, J. Appl. Polymer Sc. , 1_5, 2905 (1971).
4. T. Matsuura and S. Sourirajan, J. Appl. Polymer Sci., V7, 1043 (1973).
5. T. Matsuura and S. Sourirajan, J. Appl. Polymer Sci., ±6, 1663 (1972).
6. T. Matsuura and S. Sourirajan, J. Appl. Polymer Sci., ]7, 3683 (1973).
7. K. L. Hoy, J. Paint Techn., 42, 76 (1970).
8. P. A. Small, J. Appl. Chem., 3, 71 (1953).
9. H. H. Konstam and W. R. Feairheller, AIChE J. , J_6, 837 (1970).
10. D. W. Van Krevelen and P. J. Hoftyzer, "Properties of Polymers", Elsevier
Publishing Co., Amsterdam, Netherlands, 1972, Chapter 8.
11. M. Dunkel, Z. physik. Chem., A138. 42 (1928).
12. J. H. Hildebrand and R. L. Soctt, "The Solubility of Non-Electrolytes",
Reinhold, New York, 1959.
13. E. S. K. Chian and H. H. P. Fang, Annual Report 1973, U. S. Army Medical
Res. and Development Command, Contract No. DADA 17-73-C-3025.
14. H. G. Spencer, J. L. Gaddis, and C. A. Brandon, Membrane Separation
Technology Seminar, Clemson University, Clemson, S.C., 1977.
15. J. E. Cadotte, C. V. Kopp, K. E. Cobian, and L. T. Rozelle,
Progress Report June 19 to Office of Water Research and Technology, U.
S. Dept. of the Interior.
107
-------
Appendix C
COMPUTER CODE LISTING
108
-------
REAL LTU»E.LCHNL.LCHNT.LTU1T
REAL f1AREA.nRC.HOnC
DlfltNSION PIN(9«.3).CIN(99.3>.POUT(99.3>,COUT(99.3>.CPRI1<99,3>.
* RF<99.3>.FTRJC99.3>.FTPm99.3>.CPHOA(99.3>.ISTG<3>.FINC99.3).
X NTB(99,3>,NTBS<99.3>
C
COflnON/PRAH'IftDUl,DIF.I
COimON/TGEOn/ LTUIT.RTUIT.UCHNT.TCHNT.LCHNT
COfHION/TCOND/CFDT.OFDT.TFDT.PFDT
COWtONsCOND'OFEED, TFEED
COnnON/GEOfl/RTUIE. LTUBE, RFCTH, UCHNL. TCHNL. LCHNL, ATUBE. ACHNL. RMVDR
COn«ONxHSPEC/PERH,STP
C
C INPUT SECTION
C READ TVPE OF HODUU(TUWU« HEfWRANE-l. SHEET
!••• CONTINUE
RCAD(S,3> INDUL
IF (IHDUL.EO.*) CO TO 8M«
READ(5.4) TREJ.TFLUX
CO TO , inOUL
Z» READ(5,a)LTUlT.RTUIT
CO TO ZZ
81 READCt.2) UCHNT,TCHNT,LCHNT
22 COMTINUE
RCAD(f.E) DIF.CFDT,QFDT,TFDT,PFOT,I
C READ INITIAL CONDITIONS
RIAD<».1 )PTEED,OfEED.CFEE0.TFEED
C READ *V«TCH fPfCIFICATION
00 TO (3»,31). IRDUL
M RIADd.t) RTUM.LTUtf.tFCTR
-------
QO TO 38
31 READ(S,2) UCHNL.TCHNL.LCHhL.RFCTR
38 CONTINUE
READtS,2> DPRF.DFLOU.DREJ
C READ COHSTANTS FOR TRANSPORT COEFFICIENTS
READ(5,3> UCC.UnRC.UflOHC.EPUflP.ERATE.UUC.UEC
C
1 FORRAT<2F1«.4,E1«.4,F1».4)
a FORrtAT<8Fl«.4>
3 FORRATU6IS)
4 FORIV>T(F1».4.E1«.4>
5 FORflAT<3Eie.4.4FI».4>
TFLUX-TFLUX«',F1».4)
00 TO f0
97 IMITC(C,M> UCHNT.TCHNT.LCHNT
-------
6* FORHATC WIDTH- '.Fl». 4. 3X. 'THICKNESS-'. Fl«. 4. ax," LENGTH- -,F1». 4)
b9 CONTINUE
URITE(6,fil>
61 FORHATI/.' TEST CONDITIONS')
URITE<6,62> CFDT.OFDT.TFDT.PFDT
62 FWMATC CFDT-'.E1«.4.'OFDT-'.F1».4. 'TFDT- • ,F1».4. 'PFDT- ' .F19.4 )
URITE(6,5»)
URITE(6.51> PERH.STP
59 FORHAK/. ' TRANSPORT PARAHETER')
51 FOUfWTC PERn€AiILITV'.E15.g,3X. 'SOLUTE TRNS. PARA. •' .ElS.B.ax,
> 'FT/SEC')
C CALCULATE TU»E CROSS SECTION AREA AMD HYDROLIC RADIUS
GO TO (33.34), IHDUL
33 ATUlE-3.141S93*RTU»E**a.
RHVM-RTUBEX3.
60 TO 35
34 ACHNL'UCHNLSTCHNL
RHVDR-TCHHL/B.
35 CONTINUE
C
C
CR-CFEED«( I .-DPRF+DPRP*DREJ )/< 1. -DPRF )
CP-CFEEDt(l.-DREJ)
C
1*1
CIN(I.J)-CFIED
C
1*4 COMTZNUI
-------
c
c
CALL nODULCPlN.
« RFCI.J))
ircl.CT.l ) CO TO 151
FTRJ(1,J>-1.-RFC1.J>
FTPHCl,J)-l.-FTRJ«(1.-RF(I,J»
FTPI1CI,J>-1.-FTRJ(I.J>
CPPKV»(I,J)-FTPH(I-l,J)tCPnOA(I-l.J)/FTPM(I.J)
s +j))«cpRHti,j)-'FTpn(r. j)
c
153 CONTINUE
IFCJ.E0.3) CO TO 15*
IF(CPHO«(I.J).LT.CP) IP-I
IF(COUT(I,J).GT.CR) GO TO 153
CO TO 149
150 IF(COUT(I,J).6T.CFEED> CO TO 158
140 CONTINUE
PIN(I*1.J)-POUT(I.J)
CIN
I-I+l
CO TO 154
C
1B3 CONTINUC
III-I
-------
155 CONTINUE
ISTG(8)-III-IX
ISTG(1)-IX
C
c
ISTCl-ISTC(l)
ISTOa-ISTC(S)
c
00 157 I-1.ISTG2
PIN
COUT(I.2)-COUT(IX+I.l >
IF(I.CT.l) GO TO 158
FTRJ(1,2»-1.-RF(1,2)
FTPHC1.8)-1.-FTRJ(1.8)
60 TO 157
158 FTRJ(I.8)«FTRJ(I-1.8)«C1.-RF(I.8))
FTPB(I.8)-1.-FTRJ(I,8)
CPflOA(I.3)-FTPPI(I-l,2)«CPnOA(I-l,a)/F7Pn(I,2)
* *1
JO
PJH
-------
CIH< I. J >-CPHOA< ISTOa, 8 )
GO TO 154
1C9 CONTINUE
MSTC-J
ItTQU)-!
IST63-ISTQO)
C
C
***-FTPmiSTGl,l>/U.-FTRJCISTCl.l>*FTPf1
Mi-FTRJ( ISTOi, 1 >*FTPfU ISTOZ,8)*FT»»m ISTC3. 3 )/
S (l.-FT»J(I«T01.1)*FTPH(I6TCa.a)«FTRJ(IST03.3))
CX'AAA/CCOCPnOA ( IST01 , 1 )+MtxCCCICPfKW( ISTC3, 3 >
E-CX-CF
TOL*«.*1«*CFEED
IF(MS(E).LT.TOL) 00 TO 1M
IF(E.LT.«.) IX-IX+1
IFCE.OT.t.) IX-IX-1
00 TO 156
C
C
16* CONTINUE
FINU.l).DFLOU'tFTPmiSTG2,2>SFTRJ{IST03,3>)
FIM(1.8).FTdJ(I«T01.1>«FIN(l,l)
MTMT-*
BO 197 J-1,3
M IM i-i,irroj
tra.io.t) oo TO
-------
188 CONTINUE
NTBCI.J)*FIN(I.J)/OFEED>1
IF(I.GT.l) 00 TO 17*
NTMU.JI-NTBd.J)
00 TO 1M
17* NTBS
1S7 CONTINUE
C
C
gRITtca.M)
94 FORNAT<'ISUHNARV OF DESIGN AND PERFORMANCE OF THE SYSTEM',///.
X ' MODULE DESIQN SPECIFICATIONS AND OPERATING CONDITIONS'.
S IX.'SPECIFIED'>
00 TO <9*.9*>. INDUL
96 URITECS.W) RTUBE,LTUBE.RFCTR.OFEED,TFEED.PFEED
9S FORHATt/.' TVPE OF MEMBRANE-',11X.'TUBULAR',/,
X ' RADIUS OF TUB£-',1*X.E18.K.' FT'./.
X ' LENGTH OF TUBE-'.1»X.E18.S.' FT'./,
X ' TUBE UALL ROUGHNESS FACTOR-'.FS.I,/.
X ' TUBE FEED RATE-',1«X,E1S.S.' QPM',/.
X ' FEED TEMPERATURE-', W,FB.I.' DEO F',/.
X ' PUMP OUTLET PRESSURE-'.RX.FB.I,' PSIO')
00 TO 97
C
9« URITE(S,M> MCHNL.TCHNL.LCMNL.RFCTR.OFEED.TFEED.PFEED
M FORfMTC/,' TVPE OF NEHBRANE-SMEET OR SPIRAL MOUND'./,
X ' WIDTH OP CHANNEL-'.Ell.f,' FT'/,
-------
* ' LENGTH Or CHANNEL-'.*«.«.' FT','.
* ' TU»E WALL WUOHICM FACTOR-',Ff.a./,
* ' row rccB MTE-'.CIB.*. - OPK'.X.
* ' FEED TEMPERATURE-',Fi.«,- DEO f,/,
» ' PU«P OUTLET PRESttttE-'.FS.t,' P»Ifl'»
•7 CONTINUE
C
IMITCCC,!?)
•7 rowwTty.' svrren DCSjOH SPECIFICATIONS «NO opgRATiwe CONDITIONS'.
t IX,'SPECIFIC*'>
URlTE(«,ft> DrLOW.DKEJ.
M FOI»MT(^,' K»I8N FtOU
S ' DESIGN REJECTION-',J4X,Ft.2,/.
* ' DESIGN PRODUCT RECOVERY FACTOR-'.FH.3>
C
URlTEIf.H)
>9 FOtlfMTt UWWMttV OF NUMERICAL RESULTS*,/,
* 3X.'J',4X,'I'.BX.'NTt',4X,'NT»S',5X,'FIN',1»X,'PI«',»X,
* 'POUT', iX,'CIN'.9X.'COUT',*X,'CP«OA'.ex,'FTPK'./,eBX.
* '(CIOU/CUFT)',/)
DO 9* J-1,HSTO
90 91 I«t.I»T8J
WRITt(l,Mi /,Z.
* CIH(I,J>, COUTt t,4»,CP«0*
»S CONTtNUC
*• CONTINUt
9
-------
WRITE (6, 93) CX.AAA.Mft
93 FORMATC//. ' CX-'.Eia.C./,
X 'PERMEATE FRACTION FROM STAGE 1 -'.FS.a./.
* 'PERMEATE FRACTION FROM STAGE 3 -'.FS.8)
nAREA-RTU»C*«.attLTU*E«FLOAT
TCC-UCCSfMKA
C «* CAPITAL AMORTIZATION FACTO*. CAF-«.»1334S FOR IS PERCENT INTEREST
C *»* RATE AND TEN VEAR LIFE.
CACOST-B.31S*CAF*TCC/DFLOU
C *M CALCULATE NEM1RAME REPLACEMENT COST, CENTS/KCAL.
nRC-UnRC*I1AREAx(DFLOU>144*. )
C «» CALCULATE OTHER 0*11 COSTS. CEKTS'KGAL.
C «« CALCULATE PUHPINC POWER REOUIREHENT, KU.
PWJ-(PFEED^3. li7*CDFLOM*FIN(1.3) )tDFLOU»OFEED*«2.*l .
* RTU§EM4.*
7* FOMMTClMMfMNV Of SVSTEH ECONOMICS')
URITI (§,?! > TCC,KT»rr,nAREA,«RC,«OHC,PCOrr,CACOST.UCRDT,
-------
00
» ECRDT.TCMT
71 FORnAT',Fi*.t,
« x.' punpiNQ POUEW corr.cENT«xiceAL-',Fi«.a.
t x.' CAPITAL AMORTIZATION C08T,CEHTSxKCAL-',Fl».8.
* x.' CREDIT FOR RECOVERED UATER,CENTS'K6AL-',F1«.3,
S /.' CREDIT FOR RECOVERED ENERGY.CENTSxKGAL-'.Fit.a,
S ',' TOTAL UNIT COST,C£MTS/1C«AL-'.Fl».3>
00 TO
CONTINUE
STOP
END
SUBROUTINE FIBSPEC (TREJ,TFLUX,PERn,STP>
COWION/TOEOn/ LTUi«.RTU»E,UCHriL,TCHNL.LCHNL
COWWNXTCOND/ CFEED.OFEED.TFEED.PFEED
COf*WN/PRAf|xlHDUL.DIF,l
REAL LCHhL,LTU»E
C
C CALCULATE TUBE CROSS SECTION AREA AND HVDROLIC RADIUS
00 TO (33,34), IRDUl
33 ATU*t-3.i41§93*RTU»E»«t.
RHVDR.RTUttxt.
00 TO 3C
34 ACMHL-WCHNL«TCHNl
RHVDR-TCHNLXI.
-------
36 CONTINUE
C
00 TO (36.37>. II1DUL
3S UBAR*OFECD'ATUBE'44I.«:C
00 TO tt
37 UBAR-OFEED/'ACHNL/44».i«
3t CONTINUE
C CALL PROPU AND CALCULATE DENSITY AND VISCOSITVOOF SOLUTION. IT IS
C ASSUNflED THAT THESE PROPERTIES ARE THE SAME AS THOSE FOR PURE WATER.
CALL PROPU(TFEED,DENS.UIS>
C
C CALCULATE REYNOLDS NUMBER
RENR-4.»RHYDR»UBAR«DENS A> IS
M C CALCULATE SCHMIDT NUMBER
i-1
10 SCMi-UIS'C DENSSDIF)
CO TOC39.4*), IHDUL
39 X-LTUBE
00 TO 41
4« X'LCHNL
41 CONTINUE
C
C CALCULATE SHERWOOD NUMBER
CALL SHUDNRUL,X,RENR,SCN*,SHNR>
C
C CALCULATE MASS TRANSFER COEFFICIENT
TCM-SMNUDIF/ (4. SRHYDft)
P-PFEEDXiM.
C-DOffvlf.
e
rrp-rrtuxsti.-TRCJIXTREJCEXPI-TTLUX/TCM>
-------
PEim-STPtTFLUXtCtta./>
RETURN
END
SUBROUTINE nO»Ul(PINTL.ClNTL.P.CAl,CA3AU.RF>
COnHOrfCOND'OFEED. TFCED
, LTUWE.RFCTR. UCHNL. TCHNL, LCHML. ATU»E. ACHML.RHVDH
1.STP
REAL LTU»E,LCHNL,LAnDA
DIMENSION TCA1(3),TCA3(3),E<3).GTHET(3)
C
C INITIALIZE VARIABLES
X-».
XPRNT-9.
CA3UUS-*.*
P-PINTL
CO TO C3fl,4«). IHDUL
39 UINTl-QFEED/ATUIE/448.1*
00 TO 41
4« UIim.*OFEED'ACHNL/44l.t6
41 CONTINUE
UIMt-UINTL
CA1-CINTL
UC-UMN*CA1
DEFINE INTEOMTIOM fTEP SIZE
DXX-I.1H
PMINT INTMVAL
DXP-t.ff
-------
7 FORNATC1 MODULE PERFORMANCE SIMULATION RESULTS'./.ax, 'X'. 1«X.
It 'P'.18X.'U1AR',1BX.'CA1'.18X, 'CAHMax, 'C*3'.11X,'CA3AV'.BX.
S 'REJ'.4X.'REC',/.' '.4X.'(LI'SOFT)',iX. '(FT/SEC)'. SX.
S 'CHOLE/CUFT)',4X.'(HOLE/CUFT)',4X.'(HOLE/CUFT)' ,4X.
S MBOLEXCUTT)'./)
C
C INITIALIZE RKQ
L-l
CALL RKO(X.DXX,L)
C
C CALL PROPU AND CALCULATE DENSITV AMD WISCOSITVOOF SOLUTION. IT IS
C ASSUMED THAT THESE PROPERTIES ARE THE SAKE AS THOSE FOR PURE WATER.
CALL PROPU(TFEED.DENS,UIS>
C
C CALCULATE REYNOLDS HUNKER
1*3 REN*-4.»RHYDR«UIAR»D£NS/UIS
C CALCULATE FANNING FRICTION FACTOR
FFF-t.M«RFCTR/RCNRM« .8S
C CALCULATE SCHMIDT NUMBER
SCNR'UIS/( DENSkDIF)
C
C CALCULATE SHCRUOOD NUflBER
CALL SHUDNR(RTUBE,TCHm.>LCHNL.INDUL,X,RENR.SCNR,SHNR)
C
C CALCULATE IMSfl TRANSFER COEFFICIENT
TCN-SHNRSDIF/(4.«RHVDR)
e
C CALCULATE HOLAR DENflTY OF fOLUTION
C* DCNfxll.
-------
CALCULATE DIHENSIONLESS UAftIA*LC0
UAHDA-TCfl'STP
WUST«-P€R««P/C
THETA-STP/VUSTR
C
C
C CALCULATE CA3 »V NEUTON RAPHSON METHOD
C INITIALIZE TRIAL FOR CA3
TCA3U)- ».5*CA1
TCA3(g>- ».4*CA1
C TOLERENCE FOR N-R CONVERGENCE
TOL-t.MStCAl
!!••
DO S« I • 1,8
TCAKI )-TCA3(I )•(!.*!. /GTHETC I >SEXP(-l.'LAflDA/GTHETCI )))
5* Ed)-CAl-TCAl(I)
sa n-im
C URITE(S.B) 00 TO (1
IF(II.ta.l*«> 00 TO (••
TCA3(3)-TCA3
-------
ECB»-CA1-TCA1(8>
00 TO 58
51 CA3-TCA31B>
UW-WOSTRt(1.-OAfinASCA3/
R£J-(CA1-CA3)/CA1
RF< (UINTL-UMM >XUINTL
C LOGIC TO STOP PHOQRAfl
00 TO (36,37). IHOUL
36 IF(X.Ce.LTUlE) 60 TO 1M
M 00 TO 3S
W 37 IF(X.BE.LCHNL) 60 TO 1M
38 CONTINUE
C
C OUTPUT SECTION
IF 60 TO 101
C URITE(6,4>X.P.UBAR,CA1.CAS,CA3,CA3AU.REJ,RF-
4 FOKHAT( F«.1.6EJS.i,EF7.3 )
XP«NT-XPR«T»OXP
1*1 CONTINUE
C EXECUTE MCO 4 TIRES
00 1M 1-1,4
L-8
C
C
C DEFINE DIFFERENTIAL EQUATIONS
-------
cc-aa.a
DX-l.
DUBAft— 1 .
DUG — 1 . /RHVWWC A3*UU
DP--DEHS*(U»ARM2.«FFFx
CALL mCO(P,DP,L>
CALL RKGtUBAR.DUBAR.L)
1*3 CALL RKG(UC,DUC,L>
C URITE(C,Z9> X,DX.P,DP.U»AR.DUBAR,UC.DUC
CAl-UCxUIAR
C GO TO NEXT STEP
GO TO 1*3
1M CONTINUE
C UMITE(8,4) X.P.UBAR.CA1.CA2,CA3,CA3AU,REJ,RF
4^ RCTURN
SM IMITE(6,fi)
6 FORTMTC/,' N-R NOT CONVERGING')
STOP
END
SUBROUTINE RKGCZ.DZ.U
DIMENSION Q(M)
00 TO <1.8.3).L
1 N-M
A3-I.-AI
M-l.x*.
00 4 I-1,N
-------
4
H-D2
J-l
RCTUftN
e oo TO.J
i« *••.$
I-B.
GO TO 11
8* A-A8
00 TO 12
3* A*A3
ia 1-1.
11 C*A
00 TO 13
4« A-M
••a.
13 J-J+1
L-3
!••
3 1-1*1
DX-HXDZ
UC-M(DX-ltQ(I»
Q(t)-«(I)*3.«US-CXDM
Z-Z*M«
RETURN
•UMOUUNi PROPU(T,MD«,UI»)
Cttt
-------
C«»THIS SUBROUTINE CALCULATES UISCOSITV AND DENSITV OF UATER AS A FUNCT
ION
CSSXOF TEMPERATURE.
CIM
REAL LOQT.LOOT8
01-M. 717753
08—•.U1S89MC-I
03—•.4M38777E-4
DENS*D1+D8XT+D3STX*B.
C
C
V1-82.81M3C
ve—It. 46959*
V3O.M4958S
IOCT-ALOC1»(T)
_ lOOTa-LOOTXU.
05 VIS-(Ul«ttSLOCT+U3*LOGT8>/36M.
RETURN
END
SUWtOUTINE SHUDNR
-------
c
C TURBULENT FLOU CASE
C CHECK TYPE OF NODULE
00 TO <1«.8*>,I
C
C FOK TUBULAR HENBRANC
SHNR*«.ll*AlMENRtS(7./B. >*SCNH«*.8S
RETURN
C
C FOR SHEET NEWKANE
2« SHNR-».44*RENR«<7./1Z. >«(SCNRITCHNl/3./LCHW.>«(l.'3.
RETURN
C
C LAMINAR FLOU CASE
C CHECK TYPE OF HODULE
!•• 00 TO (3»,4«>,I
C
C FOR TUBULAR HEHBRANE
3« A3-1.9S
SHNR-A3SCRENR*SCNR*2.SRTUBE/X)S*(1.'3. )
RETURN
C
C FOR SHEET FWttRAME
4» M>l.>4
8MNR-A4«(tEHR*»CNR«t.«TCHNl/X )«•(! ./3. )
RETURN
END
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing]
1 REPORT NO.
EPA- 600/2-79-195
2.
4 TITLE AND SUBTITLE
Hyperfiltration Processes for Treatment and Renovation
of Textile Wastewater
5. REPORT DATE
October 1979
6. PERFORMING ORGANIZATION CODE
7 AUTHOR(S)
S.M. Ko and J.A. Tevepaugh
8. PERFORMING Ol
9 PERFORMING ORGANIZATION NAME AND ADDRESS
Lockheed Missiles and Space Co.
Huntsville Research and Engineering Center
Huntsville, AL 35807
10. PROGRAM ELE!
1BB-610
11. CONTRACT/GRANT NO.
68-02-2614, Task 009
12 SPONSORING AGENCY NAME AND ADDRESS
EPA, Office of Research and Development
Industrial Environmental Research Laboratory
Research Triangle Park, NC 27711
TMnal; 10/78 - 8/7Q
•^ -i_im.-i~ | ^ ._. f j .-. 1 i f , ,j
14. SPONSORING AGENCY CODE
EPA/600/13
15 SUPPLEMENTARY NOTES IERL_RTP project officer is Max Samfield, Mail Drop 62, 919/
541-2547.
16. ABSTRACT
--— - The report describes a computer program developed for the design and
simulation of a multistage hyperfiltration system for renovation of textile waste-
water The program is capable of practical design, parametric simulation, and
cost proiection of the multistage hyperfiltration system with tapered innerstages.
The mathematical model is based on Sourirajan's preferential sorption and solute
Diffusion theory. Experimental rejection and flux data of a test hyperfiltration
£dSe are required as inputs. Empirical correlations and test results available
from recent EPA-sponsored programs are used to calculate membrane transport
parameters Computed results for sample cases using cellulose acetate and
5£SI membranes are presented. Various designs and operations are considered
in the computations to show their effects on system economics The Program
is readily adaptable for evaluation of other reverse osmosis/hyperfiltration
applications.
KEY WORDS AND DOCUMENT ANALYSIS
b.lDENTIFIERS/OPEN ENDED TERMS
Pollution
Textiles
Waste Water
Filtration
Water Treatment
Renovating
Computer Programs
^DISTRIBUTION STATEMENT
Release to Public
"
EpA Form 2220-1 (9-73)
Design
Cost Analysis
Osmosis
Pollution Control
Stationary Sources
Hyperfiltration
Reverse Osmosis
i?. SECURITY CLASS (This Report I
20. SECURITY CLASS (This page)
14A
07D
128
------- |